Extrahepatic Deposition and Cytotoxicity of Lithocholic Acid:Studies in Two Hamster Models of Hepatic Failure

and in Cultured Human Fibroblasts

SUSAN CERYAK,1 BERNARD BOUSCAREL,1,2 MAURO MALAVOLTI,3 AND HANS FROMM1

Effects of bile acids on tissues outside of the enterohe-patic circulation may be of major pathophysiological signifi-cance under conditions of elevated serum bile acid concen-trations, such as in hepatobiliary disease. Two hamstermodels of hepatic failure, namely functional hepatectomy(HepX), and 2-day bile duct ligation (BDL), as well ascultured human fibroblasts, were used to study the compara-tive tissue uptake, distribution, and cytotoxicity of lithocho-lic acid (LCA) in relation to various experimental condi-tions, such as binding of LCA to low-density lipoprotein(LDL) or albumin as protein carriers. Fifteen minutes afteriv infusion of [24-14C]LCA, the majority of LCA in sham-operated control animals was recovered in liver, bile, andsmall intestine. After hepatectomy, a significant increase inLCA was found in blood, muscle, heart, brain, adrenals, andthymus. In bile duct–ligated animals, significantly moreLCA was associated with blood and skin, and a greater thantwofold increase in LCA was observed in the colon. In thehepatectomized model, the administration of LCA bound toLDL resulted in a significantly higher uptake in the kidneysand skin. The comparative time- and concentration-dependent uptake of [14C]LCA, [14C]chenodeoxycholic acid(CDCA), and [14C]cholic acid (CA) in cultured humanfibroblasts was nonsaturable and remained a function ofconcentration. Initial rates of uptake were significantlyincreased by approximately tenfold, with decreasing hydrox-ylation of the respective bile acid. After 1 hour of exposureof fibroblasts to LCA, there was a significant, dose-dependent decrease in mitochondrial dehydrogenase activ-

ity from 18% to 34% of the control, at LCA concentrationsranging from 1 to 20 mmol/L. At a respective concentrationof 100 and 700 mmol/L, CDCA caused a 35% and 99%inhibition of mitochondrial dehydrogenase activity. None ofthe bile acids tested, with the exception of 700 mmol/LCDCA, caused a significant release of cytosolic lactatedehydrogenase into the medium. In conclusion, we showthat bile acids selectively accumulate in nonhepatic tissuesunder two conditions of impaired liver function. Further-more, the extrahepatic tissue distribution of bile acidsduring cholestasis may be affected by serum lipoproteincomposition. At a respective concentration of 1 and 100mmol/L, LCA and CDCA induced mitochondrial damage inhuman fibroblasts, after just 1 hour of exposure. Therefore,enhanced extrahepatic uptake of hydrophobic bile acidsduring liver dysfunction, or disorders of lipoprotein metabo-lism, may have important implications for bile-acid inducedcytotoxic effects in tissues of the systemic circulation.(HEPATOLOGY 1998;27:546-556.)

Under physiological conditions, bile acids are chieflyconfined to the enterohepatic circulation. Their concentra-tion in the systemic circulation is unique to each bile acid andcan increase from two- to threefold postprandially from afasting level of about 1 to 3 µmol/L.1-2 Serum bile acids aremainly present in the amidated form; however, in normalman, the level of unconjugated bile acids exhibits a diurnalvariation, attaining a rapid peak concentration of around 30%of the total serum bile acids after breakfast.2 In cholestatichepatobiliary disorders, bile acids accumulate in the systemiccirculation, resulting in a 20- to 100-fold increase in theirconcentration.3 Under these conditions, serum levels ofunconjugated bile acids, in particular that of chenodeoxycho-lic acid (CDCA), can increase dramatically if portal cirrhosisis present.3 In patients with the stagnant loop syndrome,serum unconjugated bile acid levels increase because ofbacterial overgrowth in the small intestine.4 Thus, a numberof physiological and pathological conditions exist in whichtissues outside of the enterohepatic circulation are in contactwith high concentrations of both unconjugated and conju-gated bile acids.

Of particular interest and possible pathognomonic signifi-cance is lithocholic acid (LCA), a monohydroxylated second-ary bile acid, which is formed in man mainly from theintestinal bacterial 7-dehydroxylation of the primary bileacid, CDCA. This bile acid comprises less than 5% of the totalbile acid pool in man and is one of the most hydrophobicnaturally occurring bile acids. Systemic LCA concentrations

Abbreviations: LCA, lithocholic acid (3a-hydroxy-5b-cholan-24-oic acid); LDL,low-density lipoproteins; CDCA, chenodeoxycholic acid (3a,7a-dihydroxy-5b-cholan-24-oic acid); CA, cholic acid (3a,7a,12a-trihydroxy-5b-cholan-24-oic acid);BDL, bile duct–ligated; HepX, functionally hepatectomized; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyldehydrogenase; HDL, high-density lipoproteins; VLDL, very-low-density lipoproteins.

From the 1The Division of Gastroenterology and Nutrition, Department of Medicineand 2The Department of Biochemistry and Molecular Biology, The George WashingtonUniversity Medical Center, Washington, DC; and 3Istituto di Clinica Medica eGastroenterologica, University of Bologna, Bologna, Italy.

Received March 7, 1997; accepted November 3, 1997.S.C. was supported in part by a postdoctoral fellowship from the American Institute

for Cancer Research. B.B. is supported in part by grants DK46954 and CA53793 fromthe National Institutes of Health.

Presented, in part, at the Annual Meeting of the American GastroenterologicalAssociation in San Francisco, CA, in May 1992.

Address reprint requests to: Susan Ceryak, Ph.D., Division of Gastroenterology andNutrition, Department of Medicine, The George Washington University Medical Center,2300 I Street, N.W., 523 Ross Hall, Washington, DC 20037. Fax: (202) 994-3435.

Copyright r 1998 by the American Association for the Study of Liver Diseases.1270-9139/98/2702-0032$3.00/0

546

are normally less than 1 µmol/L; however, during certaincholestatic hepatobiliary disorders in which obstruction ofbile flow is incomplete, serum LCA concentrations up to 4µmol/L have been measured.5-6

Sodium-dependent bile acid transporters are reported asfunctionally present only in the liver, ileum, and kidney.7-8

Further evidence points to carrier-mediated uptake of conju-gated bile acids in the jejunum, which exists in addition topassive diffusion.9 Unconjugated bile acids can enter cells bya sodium-independent transporter as well as by passivediffusion, which is a function of their respective hydrophobic-ity.10-13 Through this latter mechanism, certain bile acids cantraverse the cellular plasma membrane of tissues both insideand outside of the enterohepatic circulation. We previouslyreported that, in contrast to its taurine conjugate, therelatively hydrophilic ursodeoxycholic acid was taken up bycultured human fibroblasts in a dose- and time-dependentfashion and with a rate constant of approximately 4.6 3 1024

nmol/mg protein/sec/µmol/L.14

Bile acids are transported in serum bound to carrierproteins, predominantly albumin.15 Previously, we demon-strated that serum lipoproteins exhibit a similar, or slightlygreater, affinity for bile salts compared with the low affinitysite of albumin.16 Furthermore, lipoproteins present 5 to 9times more bile salt binding sites than does albumin, whereaslipoprotein affinity increases with decreasing hydroxylation(polarity) of the steroid nucleus of the bile acid.16 Conse-quently, under conditions in which the serum bile acidconcentration is higher than the capacity of the high affinityclass of albumin binding sites for bile acids, bile acids wouldbe bound to lipoprotein carriers. Because low-density lipopro-tein (LDL) is internalized by most cells, it may play a majorrole in the transport of bile acids into the cell.

The physicochemical properties of bile acids are correlatedwith their physiological functions17 and may be implicated intheir pathophysiological effects. Bile acid-induced membranedamage has been reported not only to positively correlatewith the degree of hydrophobicity of the bile acid18 but also tobe influenced by plasma membrane attributes, such ascholesterol and phospholipid composition and concentra-tion.19 In addition to detergent effects on the cell membrane,bile-acid induced cytotoxicity has been reported to be causedby the impairment of mitochondrial function.20-21 Further-more, we have shown that at submicellar concentrations bileacids modulate second messenger pathways involved in theregulation of cellular metabolism not only in the liver22-23 butalso in extrahepatic tissues.14 In the latter case, the regulatoryrole of bile acids is dependent upon their ability to cross theplasma membrane of the cell.14 Both the physiologic and thepathophysiologic effects of bile acids have been well docu-mented for the tissues involved in their enterohepatic circula-tion. However, little information exists regarding the uptakeand toxicity of bile acids, under conditions of cholestasis intissues outside of the enterohepatic circulation, despite theirexposure to bile acids.

Therefore, the goal of the present study is to compare theacute tissue distribution of LCA, following iv administration,in a hamster model with an intact enterohepatic circulationand in two pathological models of impaired hepatobiliaryfunction, namely, the 2-day bile duct–ligated (BDL) hamsterand the functionally hepatectomized (HepX) hamster. Liga-tion of the common bile duct represents a well-definedexperimental model of extrahepatic cholestasis in which the

systemic concentration of bile acids is significantly elevated.In the HepX animal model, the hepatic bile-acid transport iscompletely abolished. Studies to determine the comparativerole of LDL and albumin as bile-acid carriers in the distribu-tion of LCA among extrahepatic tissues during impaired liverfunction were also performed. Finally, the comparative initialrate of uptake, as well as the short-term relative cytotoxicityof bile acids, was determined as a function of their hydropho-bicity using LCA, CDCA, and cholic acid (CA), respectively.The latter studies were conducted with cultured humanfibroblasts, which, similar to other cells outside of the entero-hepatic circulation, do not possess a bile acid transporter.

MATERIALS AND METHODS

MaterialsReagents. LCA, as well as CDCA and CA, were obtained from

Steraloids (Wilton, NH) as free acids and were 98% to 99% pure, asdetermined by gas-liquid chromatography. [24-14C]LCA (sp act, 55mCi/mmol) was purchased from Amersham Corporation (ArlingtonHeights, IL). Both [24-14C]CDCA (sp act, 50 mCi/mmol) and[24-14C]CA (sp act, 52mCi/mmol) were obtained from DuPont NENResearch Products (Boston, MA). Radiolabeled bile acids were 98%to 99.9% pure, as judged by thin-layer chromatography. [3H]inulin(sp act, 5 Ci/mmol) and [125I]sodium iodide (sp act, 16-20 mCi/µg)were purchased from Amersham. Hamster albumin was fromResearch Plus, Inc. (Bayonne, NJ). All other reagents used were ofanalytical grade available from commercial sources.

Experimental Models. Age- and weight-matched adult male GoldenSyrian hamsters (range, 100-120 g body weight; Harlan SpragueDawley, Indianapolis, IN) were used to study the comparative tissuedistribution of LCA. Animals underwent one of three surgicalprocedures, as follows: 1) BDL; 2) HepX; or 3) sham operation.During all surgical and experimental procedures, animals wereanesthetized with diethyl ether. All animals received humane care incompliance with The George Washington University guidelines.

Methods

BDL. BDL was performed 2 days before the experiment, asdescribed by Matsuzaki.24 The surgery was performed under asepticconditions. The abdomen was opened with a midline incision, andthe common bile duct was ligated in two positions, proximal to thepancreatic duct, and cut between the sutures. The abdomen wasclosed with sutures and the animals were allowed to recover. Duringrecovery, the animals were permitted free access to food and water.

Functional Hepatectomy. For the HepX model, a functional hepatec-tomy was performed using a modification of the method describedby Redgrave.25 The abdomen was opened with a midline incision,and the common bile duct (proximal to the pancreatic duct) wasligated. The inferior vena cava, proximal and distal to the junction ofthe hepatic veins, as well as the portal vein and the hepatic artery,were ligated. The animals were maintained under anesthesia for theensuing experiment.

Sham Operation. In the sham-operated control animals, the abdo-men was opened with a midline incision immediately before theexperiment.

Preparation of [125I] Hamster Albumin. Hamster albumin was dilutedin normal saline to a concentration of 1.5 mg/mL and labeled with[125I]sodium iodide to a specific radioactivity of 0.4 µCi/µg, aspreviously described.26

Determination of Tissue Distribution of LCA. Stock solutions of thesodium salt of LCA were prepared freshly in the morning of eachexperiment and were diluted accordingly in 150 mmol/L NaCl. Onemilliliter of a solution of 5 µmol/L unlabeled LCA containing 1.5 µCi[24-14C]LCA and 1.5 µCi [3H]inulin, as an extracellular marker, wasadministered in a bolus via a jugular vein catheter and followed by anormal saline infusion in 1 animal of each respective experimental

HEPATOLOGY Vol. 27, No. 2, 1998 CERYAK ET AL. 547

group (n 5 5). Fifteen minutes after the injection, blood and urinewere taken and the animals were perfused with 60 mL of phosphate-buffered saline. Gallbladder bile was aspirated; all organs wereremoved, rinsed in normal saline, blotted on filter paper, andweighed. The organs of the gastrointestinal tract were gently flushedwith normal saline to remove their contents. All organs, except forthe muscle, skin and fat, were combusted in toto in a PackardOxidizer (Packard Instrument Co., Meriden, CT) for the determina-tion of 14C and 3H radioactivity.27 Aliquots of plasma, urine, and bilewere also combusted. The radioactivity in the combusted organs, aswell as in the plasma, urine, and gallbladder bile, were determinedby scintillation counting.

The 14C recovered in the tissues was corrected by the ratio of 14Cto 3H in the blood to obtain the true tissue-associated 14C, which wasexpressed as the percent dose administered per gram of tissue. Thepercent dose per organ was calculated by multiplying the percentdose per gram of tissue by the respective organ weight, normalizedfor a 100-g animal. Hamster organ weights of muscle, skin, and fat,respectively, were determined to be 45%, 11%, and 10% of the totalbody weight, whereas blood volume was calculated to be 7.5% oftotal body weight.27

Because of the cellular permeability to inulin in the kidneys, skin,and the lungs, the ratio of 14C to 3H was not used to correct forvascular space in these tissues. However, in a parallel series ofexperiments, the tissue distribution of 1 µCi of [125I]hamsteralbumin was determined in each experimental model (n 5 3 permodel) in essentially the same manner, as described for [3H]inulin,except that the 125I radioactivity was directly determined in a gammacounter (Beckman Instrument Corp., Palo Alto, CA). Tissue distribu-tion of albumin was similar to that of inulin with the exception ofthe kidneys, skin, and the lungs in which the 14C to 125I ratio wasused to correct for entrapped blood in these tissues.

Identification of the Radioactive Metabolites of [14C]LCA. The chemicalform of [14C]LCA in serum, 15 minutes after iv administration, wasanalyzed by thin-layer chromatography in 2 to 3 animals of eachexperimental group. Bile acids were extracted from serum aspreviously described.28 The extracted bile acids were applied toSilica Gel H thin layer plates (Analtech, Newark, DE), and weredeveloped in chloroform:methanol:acetic acid:water, 65:24:15:9(vol/vol/vol/vol), as described by Cass et al.29 The area on the platecorresponding to the different bile acid standards was scraped in0.5-cm sections, and the radioactivity was determined in a scintilla-tion counter (Beckman).

Separation of Plasma Lipoprotein and Albumin Fractions. Hamsterplasma lipoprotein and albumin fractions were separated by density-gradient ultracentrifugation, as previously described.26-27 The differ-ent plasma protein fractions were removed by needle aspiration oflayers corresponding to the following densities: very-low-densitylipoproteins (VLDL), d , 1.006 g/mL; LDL, d 5 1.019 to 1.063g/mL; high-density lipoproteins (HDL), d 5 1.063 to 1.21 g/mL; andthe lipoprotein-free, albumin-rich fraction, d . 1.210 g/mL.

Relative distribution of [14C]LCA among the different plasmaprotein fractions was determined by scintillation counting. The totalprotein content of the different lipoprotein- and albumin-containingfractions was determined with the Biorad protein assay (BioradLaboratories, Richmond, CA). Total plasma cholesterol was mea-sured by a standard enzymatic technique (Sigma Chemical Com-pany, St. Louis, MO).

Preparation of LDL- and Albumin-Bound LCA. Isolated hamster LDLand hamster albumin were incubated with 10 µCi [24-14C]LCA, for3 hours at 37oC to reach binding equilibrium. [14C]LCA bound tothe respective protein fractions was separated from that unbound byultracentrifugation and dialysis, as previously described.16,30

Determination of Tissue Distribution of Protein-Bound LCA. One µCi14C-LCA, bound to either 100 µg of LDL or 600 µg of albumin, wasadministered via jugular vein infusion to either sham-operated(control) or HepX hamsters in a bolus of 5 µmol/L LCA containing 1µCi [3H]inulin, in 1 mL of 150 mmol/L NaCl (pH 7.4), followed by a

15-minute normal saline infusion (n5 3-4 animals per respectiveexperimental model and infusate). Animals were exsanguinated,perfused with phosphate-buffered saline, and the distribution ofradioactivity was determined as described for the administration ofunbound LCA.

Culture of Human Skin Fibroblasts. Human skin fibroblast cells(GMO3377B) obtained from forearm skin biopsy were purchasedfrom the Coriell Institute for Medical Research (Camden, NJ). Cellswere cultured in Dulbecco’s modified Eagle’s medium containing10% fetal calf serum.14 The cells were plated at a density of 3 to 5 x104 cells/well in six-well plates. The medium was changed every 2 to3 days for 6 days or until the cells reached confluence. Two to 4hours before the experiment, the cells were cultured in Dulbecco’smodified Eagle medium containing 0.1% fetal calf serum.

Uptake of [14C]LCA, [14C]CDCA, and [14C]CA by Human Skin Fibro-blasts. The uptake of bile acids by cultured human fibroblasts wasmeasured as previously described.14 Briefly, cells were incubated at37oC with increasing concentrations (1-100 µmol/L) of either[14C]LCA, [14C]CDCA, or [14C]CA. After 30, 60, or 120 seconds ofincubation, cells were washed three times with phosphate-bufferedsaline and removed from the wells; radioactivity was then determined.

Initial rates of bile acid uptake (nmol/mg protein/sec) werederived from the equation of the respective regression line obtainedfrom the plot of bile acid uptake as a function of time.16 Initial rate ofbile acid uptake was plotted as a function of the correspondingrespective bile acid concentration, and data were analyzed by linearcurve fitting.31

Effect of LCA on Lactate Dehydrogenase Release and Mitochondrial3-(4,5-dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide Hydrolysis inHuman Skin Fibroblasts. The ability of cells to hydrolyze the yellowtetrazolium salt 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetra-zolium bromide (MTT), to a dark blue formazan compound is anindex of mitochondrial dehydrogenase activity and is used as ameasure of cell growth and viability.32 Following incubation withdifferent bile acids, the medium was removed and the cells wereincubated for 4 hours with 0.5 mg/mL of MTT in RPMI medium.After this period of time, the blue formazan crystals were solubilizedby incubation with 10% sodium dodecyl sulfate in 0.01 N HCl. Theformation of the blue formazan compound was determined spectro-photometrically at 560 nm with 670 nm as a reference.

In addition, the release into the medium of the cytosolic enzyme,lactacte dehydrogenase, was assessed by measuring the lactatedehydrogenase activity as previously described,22,26 using a standardtechnique (Boehringer Mannheim, Indianapolis, IN).

Statistical Analysis

All data are represented as the mean 1 SE. ANOVA was used toassess a significant difference among the experimental models, andsignificant differences between the models were determined by theStudent Neuman-Keuls test.

RESULTS

Tissue Distribution of [14C]LCA. BDL for 2 days resulted ineight- to tenfold increases in serum alkaline phosphatase andalanine aminotransferase levels, from , 20 U/L to 166 U/Land from 49 U/L to 539 U/L, respectively (data not shown).Serum total bilirubin concentration increased 26-fold, from0.4 to 10.8 mg/dL.24

The total calculated recovery of 14C in the combustedorgans and blood was 78% in the HepX model, 70% in theBDL model, and 32% in the sham-operated control model. Inthe respective BDL and control animals, an additional 1.4%and 2.7% of the administered 14C was recovered in thegallbladder bile. The percentage of [14C]LCA recovered in theurine collected during the period of infusion was variable,averaging around 0.1% (range, 0.004%-0.30%) and did not

548 CERYAK ET AL. HEPATOLOGY February 1998

differ among the different experimental models. The recoveryof radioactivity in the flushed-out gastrointestinal contentsand in the carcass was not determined.

The tissue distribution of LCA in the control animals andin the two models of hepatic failure, presented as percentdose recovered per organ, is shown in Fig. 1A-1C. With littleLCA in the blood, the majority of this bile acid in the controlanimals was found in the liver, bile, and small intestinal tissue(Fig. 1A). In contrast, significantly higher blood levels ofLCA were observed in both HepX and BDL hamsters (Fig.1A). Extrahepatic LCA uptake was increased in many organsin both models of hepatic failure and appeared even higher inthe HepX animals, which would be expected because of therelative absence of LCA uptake by the liver in that model(Fig. 1A). Significantly higher amounts of LCA were associ-ated with the muscle (Fig. 1A), heart, brain (Fig. 1B),adrenals, and thymus (Fig. 1C) of the HepX animals. In theBDL animals, significantly more LCA was associated with theskin (Fig. 1B) than in the control animals. In addition, BDLresulted in a greater than twofold increase in the colonicdeposition of LCA when compared to both the control and

HepX animals (Fig. 1C). The organs displaying significantlydifferent LCA distribution among the different experimentalmodels are summarized in Table 1.

The comparative distribution of LCA, when calculated aspercent of dose recovered per gram of tissue, presents apattern similar to that calculated per organ, and is shown inFig. 2A-2C. Significant differences among the different studygroups are summarized in Table 2. As expected, the LCArecovery per gram of tissue in the liver was higher in thecontrols than in the HepX animals. The respective recovery inthe small intestinal tissue was significantly higher in thecontrols than in both the HepX and BDL animals. Whencalculated per gram of tissue, the lungs, adrenal glands, andheart contained the highest extrahepatic concentration ofLCA in the two models of hepatic failure. The LCA recoveryper gram of tissue for these organs was as follows: 9.8% and5.6% for the lungs; 2.3% and 0.9% for the adrenal glands; and1.9% and 0.5% for the heart, in the HepX and in the BDLmodels, respectively (Fig. 2A-C; Table 2).

The relative amidation of the [14C]LCA recovered in serumwas assessed by thin-layer chromatography. In the HepX and

FIG. 1. Tissue distribution of iv administered [24-14C]LCA after 15minutes in sham-operated control [SHAM], hepatectomized [HEP], and2-day BDL [LIG] hamsters. A solution of 5 µmol/L unlabeled LCA,containing 1.5 µCi [24-14C]LCA and 1.5 µCi [3H] inulin as an extracellularmarker, was administered in a bolus via a jugular vein catheter, followed bynormal saline infusion, in 1 animal of each respective experimental group(n55). Fifteen minutes after the injection, the animals were exsanguinatedand tissues were combusted in a Packard Oxidizer (Packard Instrument Co.,Meriden, CT) to determine 14C and 3H radioactivity. The 14C recovered wascorrected by the ratio of 14C to 3H in the blood to obtain the truetissue-associated 14C, expressed as percent dose administered per gram oftissue. Percent dose per organ was calculated by multiplying the percent doseper gram of tissue by the respective organ weight, normalized for a 100-gramanimal. In a parallel series of experiments, the tissue distribution of 1 µCi of[125I]hamster albumin was determined in each experimental model (n53 permodel), and the ratio of 14C to 125I was used to correct for entrapped blood inthe kidneys, skin, and the lungs. Significantly different from control(sham-operated) hamsters: *P , .05; **P , .02; and ***P , .01.Significantly different between BDL and hepatectomized hamsters: 1P , .05;11P , .02; and 111P , .01.

HEPATOLOGY Vol. 27, No. 2, 1998 CERYAK ET AL. 549

sham-operated animals, respectively, 99% and 76% of theradioactivity was in the form of unconjugated LCA. In thesham-operated group, around 9% and 10% of the radioactiv-ity was associated with the respective glycine and taurineconjugates. The remaining 5% of the serum radioactivity ofthe sham animals did not co-migrate with any of the commonbile acid standards; however, 2 radioactive peaks, comprising2% to 3% and 1% to 2%, had a respective Rf value of 0.43 and0.24. The most significant degree of amidation of serum LCAwas observed in BDL animals, as follows: 22% and 21%,respectively, was associated with the glycine and taurineconjugates; 49% of the LCA remained unconjugated and 8%of the radioactivity was not identified; however, around 5% ofthis remnant migrated with an Rf value of 0.34 (data notshown).

Distribution of [14C] LCA Among Plasma Protein Fractions. Arespective total recovery of 3.2%, 23.6%, and 32.3% of theadministered LCA was found in the blood of the control,HepX, and BDL animals (Fig. 1A-1C). Fifteen minutesfollowing iv administration, [14C]LCA displayed a similarrelative distribution among the plasma protein fractions ofthe HepX and BDL animals, respectively, as follows: 1.0% and1.2% in the VLDL fractions; 2.6% and 4.0% in the LDLfractions; 18.1% and 25.2% in the HDL fractions; and 78.3%and 69.5% in the albumin-containing fractions (Fig. 3). Theamount of LCA associated with the VLDL fraction of theHepX animals was small but significantly lower than that ofthe control animals (2.2%). The remainder of the LCA in the

plasma of the control animals was associated with the LDL,HDL, and albumin fractions, with values of 3.5%, 20.0%, and74.3%, respectively (Fig. 3).

Furthermore, the relative protein concentration of thedifferent plasma protein fractions, as expressed as a percent oftotal plasma protein, was not significantly different amongthe three experimental models, and ranged from 0.3% to0.8%, 0.3% to 0.8%, 2.4% to 3.5%, and 96% to 97% for theVLDL, LDL, HDL, and the albumin-containing fraction,respectively (data not shown).

The total cholesterol concentration was determined inplasma 15 minutes after LCA infusion. The respective concen-trations were 60.8 1 9.1, 55.9 1 7.6, and 95.2 1 6.3 mg/dL inthe sham-operated, HepX, and BDL hamsters. BDL for 2 dayswas associated with a significant increase in the total serumcholesterol.

Respective Effect of LDL and Albumin on the Tissue Distribution of[14C]LCA. In another set of experiments, the tissue distribu-tion of either albumin- or LDL-bound [24-14C]LCA wasdetermined exactly as described earlier, using the HepXmodel and, as control, sham-operated animals.

As observed with the infusion of 14C-LCA alone (Figs. 1and 2A-2C), hepatectomy again resulted in an increasedextrahepatic tissue uptake of radioactivity (data not shown).The tissues displaying a marked or significant difference inthe respective percent of LCA dose recovered per organ andper gram of tissue, which is attributable to the protein carrier,are indicated in Tables 3 and 4.

In the sham-operated control animals, there were noobserved differences in LCA uptake for that administeredbound to albumin when compared to that administered LCAbound to LDL. The organ uptake of LCA when administeredbound to albumin was virtually identical to that measuredwithout a protein carrier, with the exception of the lungs. Inthe lungs, there was an approximate 30-fold increase in LCAuptake, when LCA was administered without a proteincarrier, compared with LCA administered bound to eitheralbumin or LDL.

In the HepX model, the absence of a protein carrier againresulted in a dramatic, significantly increased LCA uptake inthe lungs when compared with either the albumin- orLDL-bound LCA. However, in the kidneys, the binding ofLCA to LDL resulted in a significantly higher uptake.Furthermore, when administered bound to LDL, a signifi-cantly higher LCA uptake was exhibited in the skin whencompared with the administration of LCA bound to albumin.Finally, after 15 minutes of infusion, there was a trend in thecolon and thyroid gland for LCA binding to LDL, to cause amarked increase in its uptake in that tissue compared withLCA bound to albumin; however, this was not statisticallysignificant.

Distribution of LDL-and Albumin-Bound [14C]LCA Among PlasmaProtein Fractions. Table 5 shows the relative distribution of[14C]LCA among the different plasma protein fractions 15minutes after iv administration of LCA bound to eitheralbumin or to LDL, as well as in the absence of a proteincarrier (Fig. 3). In the plasma of the sham-operated hamsters,a significantly increased proportion of LCA was redistributedto HDL 15 minutes after infusion of [14C]LCA bound toeither albumin or to LDL when compared with the distribu-tion of LCA infused in the absence of a protein carrier. After

TABLE 1. Organs Displaying Significant Differences in % LCA DoseRecovered, As a Function of the Experimental Model

Sham-Operated Hepatectomized

Bile-DuctLigated

Blood 3.17 6 1.05 23.57 6 6.36* 32.29 6 9.47‡Liver 13.74 6 0.99 0.86 6 0.31‡,# 18.55 6 4.73Bile 2.69 6 0.62 0.01 6 0.01‡ 0.30 6 0.28‡Small intestine 6.68 6 2.22 0.50 6 0.35† 0.62 6 0.16‡Muscle 4.34 6 1.38 42.09 6 8.96‡,# 9.98 6 2.04Heart 0.02 6 0.01 0.51 6 0.17‡,§ 0.16 6 0.09Brain 0.05 6 0.02 0.48 6 0.08‡,# 0.15 6 0.05Skin 1.04 6 0.15 2.06 6 0.58 2.77 6 0.42‡Colon 0.05 6 0.01 0.06 6 0.04 0.21 6 0.06†,§Adrenals 0.01 6 0.002 0.05 6 0.02† 0.02 6 0.01Thymus 0.008 6 0.002 0.038 6 0.009‡,§ 0.015 6 0.004

NOTE. A solution of 5 µmol/L unlabeled LCA, containing 1.5 µCi[24-14C]LCA and 1.5 µCi [3H] inulin as an extracellular marker, wasadministered in a bolus via a jugular vein catheter followed by normal salineinfusion, in 1 animal of each respective experimental group (n 5 5). Fifteenminutes after the injection, the animals were exsanguinated, and tissues werecombusted in a Packard Oxidizer (Packard Instrument Co., Meriden, CT) todetermine 14C and 3H radioactivity. The 14C recovered was corrected by theratio of 14C to 3H in the blood to obtain the true tissue-associated 14C,expressed as percent dose administered per gram of tissue. Percent dose perorgan was calculated by multiplying the percent dose per gram of tissue bythe respective organ weight, normalized for a 100-gram animal. In a parallelseries of experiments, the tissue distribution of 1 µCi of [125I]hamsteralbumin was determined in each experimental model (n 5 3 per model), andthe ratio of 14C to 125I was used to correct for entrapped blood in the kidneys,skin, and the lungs. All % of dose recovered per organ. Mean 6 SEM.

Significantly different from control (sham-operated) hamsters: *P , .05;†P , .02; ‡P , .01.

Significantly different between BDL and hepatectomized hamsters. §P ,

.05; #P , .01; ##P , .02.

550 CERYAK ET AL. HEPATOLOGY February 1998

administration of LCA bound to LDL, a tenfold significantlyhigher proportion of LCA associated to this lipoproteinfraction remained, with a significantly decreased proportionof LCA associated to the albumin fraction. Likewise, whenLCA was infused bound to albumin there was a significantlyhigher proportion of LCA in plasma associated with thealbumin fraction compared with that after the administrationof LCA bound to LDL. However, the percentage of LCAassociated to the albumin fraction was significantly less whenLCA was administered bound to albumin than when LCA wasadministered without a protein carrier. In contrast, in theplasma of the HepX animals, with the exception of asignificantly lower association of LCA with the LDL fractionin those hamsters in which LCA was administered bound toalbumin, there were no differences in LCA distributionamong the different plasma protein fractions, as a function ofthe protein carrier. Finally, the administration of the respec-tive [14C]LCA protein carrier did not significantly affect therelative protein concentration of the different plasma proteinfractions (data not shown).

FIG. 2. Concentration of iv administered LDL-bound [24-14C]LCA,expressed per gram of tissue after 15 minutes in sham-operated control[SHAM], HepX [HEP], and 2-day BDL [LIG] hamsters. For experimentaldetails, and explanation of symbols, see legend of Fig. 1 and Materials andMethods.

TABLE 2. Organs Displaying Significant Differences in % LCA Dose(recovered per gram of tissue) As a Function of the Experimental Model

Sham-Operated Hepatectomized BDL

Blood 0.42 6 0.14 3.14 6 0.84* 4.30 6 1.26‡Liver 3.78 6 0.37 0.22 6 0.08‡,# 5.02 6 1.37Small intestine 3.34 6 1.16 0.19 6 0.11† 0.30 6 0.07†Muscle 0.10 6 0.03 0.94 6 0.20‡,# 0.22 6 0.04Heart 0.06 6 0.02 1.86 6 0.64‡,§ 0.50 6 0.29Brain 0.06 6 0.02 0.58 6 0.02‡,## 0.24 6 0.13Skin 0.09 6 0.01 0.19 6 0.05 0.25 6 0.04‡Colon 0.07 6 0.02 0.08 6 0.03 0.22 6 0.06*,§Adrenals 0.22 6 0.08 2.27 6 0.78* 0.91 6 0.56Thymus 0.20 6 0.04 0.64 6 0.14* 0.42 6 0.18

NOTE. For experimental details and explanation of symbols, see legend ofTable 1 and Materials and Methods. See symbol definitions from Fig. 1. All %of dose recovered per gram of tissue. Mean 6 SEM.

HEPATOLOGY Vol. 27, No. 2, 1998 CERYAK ET AL. 551

Uptake of LCA, CDCA, and CA by Fibroblasts. We havepreviously shown that unconjugated bile acids, such asursodeoxycholic acid, are able to cross the plasma membraneof cultured human fibroblasts, by a mechanism that isconsistent with passive diffusion.14 To assess the relativeuptake of different bile acids by cells of non-hepatic originand that lack a known bile acid transporter, the comparativetime- and concentration-dependent uptake of [14C]LCA,

[14C]CDCA, and [14C]CA was determined in cultured humanfibroblasts. In the case of all the bile acids studied, therespective cellular uptake was nonsaturable at the concentra-tions studied and remained a function of their concentration.Furthermore, the initial rates of uptake (Table 6) weresignificantly increased by approximately tenfold, as a func-tion of decreasing polarity of the steroid nucleus of therespective bile acid (i.e., decreasing hydroxylation).

Effect of Acute Exposure to Bile Acids on Cell Viability. It is wellappreciated that long-term exposure of cells to non-polar bileacids is associated with cytotoxicity. In the present study, wefound that bile acids exhibit a rapid, probably diffusible,uptake in cells that lack a bile acid transporter. To determineif acute exposure of the same cell type to these bile acids wasassociated with any toxic effects, fibroblasts were incubatedwith increasing concentrations of LCA, CDCA, and CA for 1hour and toxicity was assessed by measuring the mitochon-drial dehydrogenase activity, as well as the release of thecytosolic enzyme, LDH. The results are shown in Table 7.

After 1 hour of exposure to LCA at 37oC, there was asignificant and dose-dependent decrease in mitochondrial

FIG. 4. The rate of uptake of LCA, CDCA, and CA in cultured humanskin fibroblasts. Fibroblasts were incubated with increasing concentrationsof either [14C]LCA (1-20 µmol/L), [14C]CDCA (1-100 µmol/L), or [14C]CA(10-100 µmol/L). After 30, 60, or 120 seconds, cells were washed, removedfrom the wells, and radioactivity was determined. Initial rates of bile aciduptake (nmol/mg protein/sec) were derived from the equation of therespective regression line obtained from the plot of bile acid uptake as afunction of time. The initial rate of bile acid uptake was plotted as a functionof the corresponding respective bile acid concentration, and data wereanalyzed by linear curve fitting.

TABLE 3. Organs Displaying Significant Differences in % LCA DoseRecovered, As a Function of the Protein Carrier

Model

Protein Carrier(% of Dose Recovered Per Organ)

(Mean 6 SEM)

Albumin LDL None

Lungs Sham 0.07 6 0.02‡ 0.10 6 0.02‡ 2.18 6 0.42Lungs Hep 1.00 6 0.25* 1.01 6 0.30* 4.62 6 1.46Kidney Hep 1.06 6 0.19 1.56 6 0.26* 0.76 6 0.05Skin Hep 1.31 6 0.44 3.50 6 0.74§ 2.06 6 0.58Colon Hep 0.05 6 0.01 0.19 6 0.08 0.06 6 0.04Thyroid Hep 0.18 6 0.10 0.51 6 0.11 0.24 6 0.11

NOTE. One µCi [14C]LCA (bound to either 100 µg of LDL or 600 µg ofalbumin) was administered via jugular vein infusion to either sham-operated(SHAM, control) or HepX (HEP) hamsters in a bolus of 5 µmol/L LCAcontaining 1 µCi [3H]inulin, in 1 mL of 150 mmol/L NaCl (pH 7.4), followedby a 15-minute normal saline infusion; the distribution of radioactivity wasdetermined as described in the legend of Table 1 and in Materials andMethods. Values determined in the absence of a protein carrier (None) arederived from Fig. 1.

Significantly different between LCA administered bound to either albuminor LDL, and LCA administered without a protein carrier: *P , .05; †P , .02;‡P , .01.

Significantly different between LCA administered bound to albumin andLCA administered bound to LDL: §P , .05; #P , .02; ##P , .01.

TABLE 4. Organs Displaying Significant Differences in % LCA DoseRecovered (Per Gram of Tissue), As a Function of the Protein Carrier

Model

Protein Carrier(% of Dose Recovered Per g of Tissue)

(Mean 6 SEM)

Albumin LDL None

Lungs Sham 0.12 6 0.04‡ 0.15 6 0.02‡ 4.51 6 0.92Lungs Hep 1.83 6 0.22‡ 1.85 6 0.49† 10.56 6 2.90Kidney Hep 1.28 6 0.13 1.75 6 0.14‡,§ 0.85 6 0.05Skin Hep 0.12 6 0.04 0.32 6 0.07§ 0.19 6 0.05Colon Hep 0.07 6 0.01 0.21 6 0.07 0.08 6 0.03Thyroid Hep 0.26 6 0.13 0.77 6 0.10 0.47 6 0.20

NOTE. For experimental details, and explanation of symbols, see legendof Table 3 and Materials and Methods.

FIG. 3. Distribution of iv administered [24-14C]LCA among plasmaprotein fractions 15 minutes after infusion in sham-operated control[SHAM], HepX, and 2-day BDL hamsters. One microcurie of [14C]LCA(bound to either 600 µg albumin or 100 µg LDL or in the absence of a proteincarrier) was administered via jugular vein infusion to either sham-operatedcontrol [SHAM] or functionally HepX [HEP] hamsters, as described in thelegend of Fig. 1 and Materials and Methods. Hamster plasma lipoprotein andalbumin fractions were separated by density gradient ultracentrifugation,and the relative distribution of [14C]LCA was determined by scintillationcounting. Significantly different from control (sham-operated) hamsters:*P , .05. ALB, lipoprotein free-, albumin-rich fraction.

552 CERYAK ET AL. HEPATOLOGY February 1998

dehydrogenase activity from 18% to 34% of the control at allLCA concentrations tested, ranging from 1 to 20 µmol/L. At aconcentration of 100 µmol/L, CDCA caused a 35% inhibitionof mitochondrial dehydrogenase activity, while 700 µmol/LCDCA completely abolished all mitochondrial activity. Incontrast, CA, even at a concentration of 700 µmol/L, causedno significant reduction in mitochondrial dehydrogenaseactivity.

None of the bile acids tested, with the exception of 700µmol/L CDCA, caused a significant release of cytosolic lactatedehydrogenase into the medium. After a 1-hour incubationwith 700 µmol/L CDCA, cell viability was significantlydecreased to 79% of the control, as measured by a significantincrease of 21% of the total cellular lactate dehydrogenaseinto the medium (Table 7).

DISCUSSION

The present study is the first to compare the acute tissuedistribution of a cytotoxic, hydrophobic bile acid in a hamstermodel with an intact enterohepatic circulation, and in twomodels of impaired hepatobiliary function. In the sham-operated hamsters, with an intact enterohepatic circulation,

around 75% of [14C]LCA recovered 15 minutes after ivinfusion was confined to the liver, biliary tract, and smallintestine, in agreement with its reported rapid plasma clear-ance and efficient hepatic uptake.34

As shown in man and in a rodent BDL model, cholestasis isassociated with a marked increase in the concentration of bileacids in both the portal and systemic circulation,3,24,33 thereby,providing significant exposure of extrahepatic tissues to bileacids. In the present study, 15 minutes after infusion, LCAexhibited a more pronounced extrahepatic uptake than didglycocholic acid, as used by Ng and Hofmann in BDL rats,33

which probably reflects the relative permeability of the cellmembrane to these bile acids.11-12 This finding is alsosupported by the study of Oh and Dupont,35 who haveidentified a measurable quantity of the more hydrophobicdihydroxy bile acids as compared to CA, in muscle andadipose tissue extracts of healthy rats. This is probably causedby differences in the transbilayer mobility (‘‘flip-flop’’) of bileacids, which is inversely related to the number of theirhydroxyl groups.10-13 Indeed, this has also been shown in thepresent study, in which the rate constant for the cellularuptake of LCA in fibroblasts was 100-fold higher than thatof CA.

There is considerable evidence in man and in animalmodels that cholestasis is associated with increased deposi-tion of bile acids in the skin.33,36-37 The results of these studiesunderline the significance of the present findings, as in bothmodels of cholestasis there was an increased recovery of LCAin the skin, which was significant only for the BDL hamster.We also observed a striking increase in LCA uptake in thecolon of BDL hamsters, in contrast to both the control andHepX animals. The enhanced uptake of LCA in the colon maybe related to selective changes in the colonic cell membranepermeability, which are associated with the absence of bile inthe intestine, as would be anticipated 2 days after BDL.38

TABLE 5. Distribution of [14C]LCA Among Plasma Protein FractionsFollowing IV Administration

Model

Protein Carrier(% of Total [14C]LCA in Plasma)

(Mean 6 SEM)

Albumin LDL None

VLDL Sham 4.23 6 0.27 4.80 6 2.59 2.19 6 0.34LDL Sham 2.37 6 1.92 21.90 6 8.49*,§ 3.51 6 1.05HDL Sham 31.90 6 2.20‡ 30.37 6 1.46‡ 19.97 6 1.30Albumin Sham 61.40 6 2.60* 42.70 6 4.85‡,## 74.34 6 2.54VLDL Hep 1.37 6 0.12 1.37 6 0.32 0.99 6 0.23LDL Hep 0.97 6 0.03* 1.83 6 0.27 2.62 6 0.54HDL Hep 17.30 6 1.90 18.73 6 1.87 18.07 6 1.66Albumin Hep 80.40 6 1.93 78.07 6 1.85 78.31 6 2.25

NOTE. For experimental details, see legends of Tables 1 and 3 andMaterials and Methods. Hamster plasma lipoprotein and albumin fractionswere separated by density gradient ultracentrifugation, and the relativedistribution of [14C]LCA was determined by scintillation counting. Forexplanation of symbols, see legend of Table 3.

TABLE 6. Dose-Dependent Uptake of [14C]LCA, CDCA, and CA,Respectively by Cultured Human Fibroblasts

Initial Rate of Bile Acid Uptake(nmol/mg protein/sec/mmol/L)

Mean SEM

Lithocholic acid 1.19 3 1023 6 0.29 3 1023

Chenodeoxycholic acid 7.39 3 1024 6 1.25 3 1024

Ursodeoxycholic acid13* 4.63 3 1024 6 0.21 3 1024

Cholic acid 4.23 3 1025 6 1.66 3 1025

NOTE. Human fibroblasts were incubated with increasing concentrationsof either [14C]LCA (1-20 µmol/L), [14C]CDCA (1-100 µmol/L), or [14C]CA(10-100 µmol/L). After 30, 60, or 120 seconds of incubation, cells werewashed and removed from the wells, after which radioactivity was deter-mined. Initial rates of bile acid uptake (nmol/mg protein/sec) were derivedfrom the equation of the respective regression line obtained from the plot ofbile acid uptake as a function of time. Initial rate of bile acid uptake wasplotted as a function of the corresponding respective bile acid concentrationand data were analyzed by linear curve fitting.

*Value was obtained from a previous study.13

TABLE 7. Effect of Acute Exposure to Bile Acids on Fibroblast Viability

Bile AcidConcentration

(mmol/L)

Relative Viability(% of control)

(Mean 6 SEM)

MTT LDH

None (control) — 100.0 6 4.8 99.6 6 0.4LCA 1 81.6 6 0.7* 99.2 6 0.8LCA 2 76.7 6 3.1* 99.2 6 0.8LCA 5 70.7 6 0.5* 100.0LCA 10 70.5 6 0.6* 100.0LCA 20 66.4 6 2.2* 100.0CDCA 50 86.1 6 5.0 100.0CDCA 100 64.8 6 1.7* 100.0CDCA 700 1.4 6 0.1* 78.9 6 12.2CA 50 104.0 6 0.6 ndCA 100 95.3 6 1.3 100.0CA 700 94.9 6 2.8 97.2 6 2.8

NOTE. Cells were incubated at 37°C for 1 hour in the absence andpresence of different bile acids. Then, the hydrolysis of MTT was measuredcolorimetrically, and cell viability, as assessed from mitochondrial dehydroge-nase activity, expressed as % formazan formation in the absence of bile acids.LDH activity in medium was measured and viability, as assessed by thepermeability of the cell membrane to LDH, expressed as 100 2 % totalactivity after incubation with 1,500 µmol/L CDCA. Results are the mean 6

SEM of 3 to 6 replicates.Abbreviations: nd, not determined; LDH, lactate dehydrogenase.Significantly different in the presence of bile acids: *P , .01.

HEPATOLOGY Vol. 27, No. 2, 1998 CERYAK ET AL. 553

Indeed, Cucchiaro, et al.39 have shown that bile acids canenter the intestine from plasma by reverse transport underconditions in which there is no canalicular bile flow.

The absence of hepatic uptake of bile acids in the HepXmodel did not result in higher serum levels of LCA ascompared with the BDL hamster, but rather, increased tissuedistribution, most notably in the muscle, heart, and brain. Asimilar increase of tissue recovery of 14C-taurocholate in theskeletal and cardiac muscle was reported in anhepatic rats.39

A surprising discovery was the dramatic tenfold increase of[14C]LCA in the brain of the HepX hamster. Spiegelman etal.40 showed that dehydrocholate (3,7,12-trioxo-5b-cholan-24-oic acid) infusion increased the permeability of the bloodbrain barrier at both submicellar and micellar concentrations.In the present study, the concentration of LCA infused issubmicellar, however, the mechanism of blood brain barrierdisruption by dehydrocholate is unknown; it is possible thatLCA acts by a similar mechanism. In contrast, in the BDLhamster, there was significantly less cerebral LCA uptake.Wahler et al.41 reported that during cholestasis, accompaniedby hypercholesterolemia, blood brain barrier permeability isdecreased, probably as a result of enhanced membranestabilization by the increased cholesterol content of themembrane. Because, in the present study, a two-day BDL wasaccompanied by a significantly increased serum cholesterollevel, this could account for the difference in cerebral LCAuptake between these two cholestatic models.

The adrenal glands of the HepX hamster displayed thesecond highest concentration of LCA when expressed pergram of tissue. Whereas there is no apparent precedent forbile acid transport to this organ, perhaps it is merely the resultof a particular affinity of the adrenal cortical membrane forthe unconjugated bile acid, which was the most abundantform of the administered bile acid in this model. Alternatively,the presence of cytosolic sterol binding proteins, which mayshow binding affinity for LCA, could sequester this bile acidand increase the concentration gradient from the extracellu-lar to the intracellular space, thus promoting further LCAuptake.42 Finally, there was a two- to fourfold increase of LCAin the thymus of the HepX hamster when compared with theother experimental groups. The potential accumulation ofbile acids in tissues of the immune system may play animportant role in the observed respective effect of cholestasisin vivo, and of hydrophobic bile acids in vitro on the biologicalactivity of interferon, as well as on the activity of natural killercells.43

In the present study, the majority of [14C]LCA metabolitesidentified in serum of both sham-operated and HepX animalsremained unconjugated, compared with 49% in the BDLhamster. The increased proportion of the serum LCA, whichwas amidated in the BDL animals, may have contributed tothe observed differences in the extrahepatic distribution ofLCA between the BDL and HepX animals. It is unclear towhat extent the amidated forms of LCA may permeate the cellmembrane, since Hedenborg et al.37 identified glycine- andtaurine-conjugated bile acids in human tissues during extra-hepatic cholestasis. Indeed, as shown by Van Dyke et al.12 asubstantial component of the hepatic uptake of glycine-conjugated bile acids was shown to occur by a non-saturable,sodium-independent process in contrast with that of therespective taurine-conjugated counterpart. Furthermore, dur-ing cholestatic liver disease in man, there is a significantincrease in the excretion of bile acids in urine, in particular

sulfated, glucuronidated, and 1- and 6-hydroxylated metabo-lites.44-45 In the present study, 15 minutes after administra-tion, there was no difference in either the percentage of LCAin the kidney or in the accumulation of LCA in the urine ofthe different experimental models, probably because of itslack of significant biotransformation in the short term. Itwould be expected, however, that effective hepatic modifica-tion of the circulating bile acid such as by sulfation, glucuron-idation, and 6b-hydroxylation would result in enhancedurinary excretion over the long term.

As we previously defined, the partitioning of bile acids inthe bloodstream would be dictated by the comparativedissociation constants of their binding to lipoproteins andalbumin, as well as by the abundance of the respective bileacid/protein.16 In the present study, with one exception, therewas no difference among the experimental groups in thedistribution of LCA among the different lipoprotein andalbumin-containing fractions, 15 minutes after the adminis-tration without a protein carrier. The exception was in theHepX model, in which significantly less LCA was associatedwith the VLDL fraction. This was most likely caused by theinherent lack of VLDL release into the bloodstream. Further-more, the distribution of LCA among the serum proteinfractions was similar to that previously observed in vitro, afterincubation of LCA with whole human plasma.30

During hepatobiliary disorders, there is an increased pro-portion of lipoprotein-bound bile acids,46 which we postu-lated to be responsible for the increased concentrations of bileacids detected in nonhepatic tissues.37 For the first time, thepresent study explores the relative role of LDL and albumin asbile acid carriers targeting LCA to specific organs. When LCAwas administered, bound to either albumin or LDL, there wasno change in the percent dose remaining in blood in eitherthe sham-operated control animals or in the HepX animals, ascompared with its administration in the absence of a proteincarrier. The distribution of LCA among the lipoproteinfractions in the sham animal reflected an enrichment in LCAof the administered protein carrier. In contrast, the distribu-tion of LCA in the HepX animals displayed no difference fromthat in the absence of a protein carrier, 15 minutes afteradministration. This rapid redistribution of LCA reflects therelative affinity of the serum protein fractions for LCA16 andwas probably caused by both a low concentration of serumbile acids as well as a decrease in serum VLDL in this model,in which liver to plasma reflux is impeded.

In both sham and HepX animals, infusion of LCA bound toa protein carrier was associated with less deposition in thelungs. The presence of a protein carrier probably impairs theuptake of LCA during its first pass through the lungs. Thebinding of LCA to LDL significantly enhanced its depositionin kidney and skin in the HepX hamster. It is not clear bywhich mechanism this occurs in the kidney; however, it mayrepresent an increased non-specific association to the tissuein the presence of a lipoprotein carrier. Indeed, we previouslyshowed that the non-specific association of LCA to isolatedhepatocytes was enhanced when a lipoprotein carrier wasused, as compared to an albumin carrier.16 We have postu-lated that this enhanced cellular association in the presence oflipoproteins may represent either lipoprotein binding to itsreceptor or an interaction with liposome binding sites.Likewise, we have previously reported that LDL receptors inthe skin were responsible for greater than 5% to 13% of thetotal LDL uptake in the intact hamster.27 The targeting of LDL

554 CERYAK ET AL. HEPATOLOGY February 1998

to its receptor may facilitate the preferential uptake of LCAfrom the LCA-LDL complex. However, this could be charac-teristic for the skin, as the muscle showed no significantdifference in LCA uptake from the LDL and albumin carriers,despite the fact that, in comparison with skin, it expressesconsiderably more LDL receptor activity in the hamster.27

Finally, we observed that the uptake of bile acids by humanfibroblasts was more a function of bile acid polarity, i.e.,number of hydroxyl groups, than of absolute hydrophobicity.These findings are supported by those of Van Dyke et al.12 andAldini et al.47 for the nonsaturable, sodium-independentuptake of bile acids in the liver and intestine, respectively.The significance of the present studies in cultured fibroblastslies in the correlation of the rate of bile acid uptake, i.e.,diffusion, with the relative cytotoxicity in a cell which doesnot possess either a bile acid transporter or an efficient systemfor binding, and/or detoxifying these compounds. Indeed, todate, the expression of significant concentrations of bile-acidbinding proteins has been limited to the liver, kidney, andintestine.48 Mitochondrial damage has been implicated in therelative toxicity of CDCA and LCA in rat liver at concentra-tions ranging from 100 to 250 µmol/L and 10 to 100 µmol/L,respectively.49 In the present study, the most nonpolar bileacids exhibited mitochondrial toxicity at concentrations wellbelow their respective critical micellar concentration. Further-more, what is remarkable about the bile acid–induced cyto-toxic effects is that they were observed in less than 1 hour andbefore any changes in plasma membrane integrity associatedwith detergent effects could be observed. Therefore, en-hanced uptake of hydrophobic bile acids, in an acute fashionand by certain tissues, could conceivably result in pathologi-cal consequences for the targeted organs.

This, we feel, underlines the relevance of the present study,in that under conditions of impaired hepatic transport of bileacids, an elevation of hydrophobic bile acids in the systemiccirculation, as well as changes in the serum protein composi-tion, could preferentially target bile acid uptake in thosetissues that we identify, namely skin, colon, muscle, heart,brain, adrenal glands, and thymus. Under these conditions,very low concentrations of bile acids are sufficient (i.e., 1 and50-100 µmol/L of LCA and CDCA, respectively) to inhibitmitochondrial dehydrogenase activity. Thus, even under thecondition in which the bile acid may be present for only ashort period of time, it may be sufficient to alter mitochon-drial metabolic events. This is clearly significant underconditions of certain hepatobiliary disorders without com-plete obstruction of bile flow, in which serum LCA concentra-tions of 1 to 4 µmol/L have been measured.5-6 Of equalimportance is the likelihood that, under conditions of im-paired liver function and decreased bile secretion, such as inpatients with portal cirrhosis3 and in infants with extrahe-patic biliary atresia and neonatal hepatitis,50 serum concentra-tions of CDCA may also reach levels of 20 to 100 µmol/L,which would result in significant tissue accumulation of thisbile acid. In addition to the potentially toxic effects of certainbile acids, we previously showed that, at extracellular concen-trations as low as 10 µmol/L, dihydroxy bile acids, includingCDCA, maximally inhibit the production of adenosine 38,58-cyclic monophosphate, in the presence of an appropriatestimulus in tissues outside of the enterohepatic circulation.14

This underlines the importance of potential pathophysiologi-cal mechanisms by which a significant increase in particularsystemic bile acids may not only result in cytotoxicity, but

may also lead to changes in the response of extrahepatictissues to hormones and cytokines, which are regulated bysecond messengers.

Acknowledgment: The authors wish to thank Bryan Ra-batic for his skillful technical assistance; and Frances Noonan,Ph.D., for the gift of human fibroblasts, as well as the use ofher cell culture facilities.

REFERENCES

1. Angelin B, Bjorkhem I, Einarsson K, Ewerth S. Hepatic uptake of bileacids in man. Fasting and postprandial concentrations of individual bileacids in portal venous and systemic blood serum. J Clin Invest1982;70:724-731.

2. Setchell KDR, Lawson AM, Blackstock EJ, Murphy A. Diurnal changes inserum unconjugated bile acids in normal man. Gut 1982;23:637-642.

3. Makino I, Nakagawa S, Mashimo K. Conjugated and unconjugatedserum bile acid levels in patients with hepatobiliary diseases. Gastroen-terology 1969;56:1033-1039.

4. Lewis B, Panveliwalla D, Tabaqchali S, Wootton IDP. Serum-bile acids inthe stagnant-loop syndrome. Lancet 1969;1:219-220.

5. Bartholomew TC, Summerfield JA, Billing BH, Lawson AM, SetchellKDR. Bile acid profiles of human serum and skin interstitial fluid andtheir relationship to pruritis studied by gas chromatography-massspectrometry. Clin Sci 1982;63:65-73.

6. Poupon RE, Chretien Y, Poupon R, Paumgartner G. Serum bile acids inprimary biliary cirrhosis; effect of ursodeoxycholic acid therapy. HEPATOL-OGY 1993;17:599-604.

7. Ananthanarayanan M, Bucuvalas JC, Shneider BL, Sippel CJ, Suchy FJ.An ontogenically regulated 48-kDa protein is a component of theNa1-bile acid cotransporter of rat liver. Am J Physiol 1991;261:G810-G817.

8. Shneider BL, Dawson PA, Christie DM, Hardikar W, Wong MH, SuchyFJ. Cloning and molecular characterization of the ontogeny of a rat ilealsodium-dependent bile acid transporter. J Clin Invest 1995;95:745-754.

9. Amelsberg A, Schteingart, CD, Ton-Nu, H-T, Hofmann AF. Carrier-mediated jejunal absorption of conjugated bile acids in the guinea pig.Gastroenterology 1996;110:1098-1106

10. Cabral DJ, Small DM, Lilly HS, Hamilton JA. Transbilayer movement ofbile acids in model membranes. Biochemistry 1987;26:1801-1804.

11. Donovan JM, Jackson AA. Rapid translocation of fully ionized bile salts(BS) depends on BS concentration, hydrophobicity and membranecholesterol content [Abstract]. Gastroenterology. 1996;110:1184A.

12. Van Dyke R, Stephens JE, Scharschmidt BF. Bile acid transport incultured rat hepatocytes. Am J Physiol 1982;243:G484-G492.

13. Aldini R, Roda A, Montagnani M, Cerre C, Pellicciari R, Roda E.Relationship between structure and intestinal absorption of bile acidswith a steroid or side chain-modification. Steroids 1996;61:590-597.

14. Bouscarel B, Ceryak S, Gettys TW, Fromm H, Noonan F. Alteration ofcAMP-mediated hormonal responsiveness by bile acids in cells ofnonhepatic origin. Am J Physiol 1995;268:G908-G916.

15. Rudman D, Kendall FE. Bile acid content of human serum. II. Thebinding of cholanic acids by human plasma proteins. J Clin Invest1957;36:538-542.

16. Ceryak S, Bouscarel B, Fromm H. Comparative binding of bile acids toserum lipoproteins and albumin. J Lipid Res 1993;34:1661-1674.

17. Hofmann AF, Small DM. Detergent properties of bile salts: correlationwith physiological function. Ann Rev Med 1967:18:333-376.

18. Scholmerich J, Becher M-S, Schmidt K, Schubert R, Kremer B, FeldhausS, Gerok W. Influence of hydroxylation and conjugation of bile salts ontheir membrane-damaging properties-studies on isolated hepatocytesand lipid membrane vesicles. HEPATOLOGY 1984;4:661-666.

19. Velardi ALM, Groen AK, Oude Elferink RPJ, Van der Meer R, PalascianoG, Tytgat GNJ. Cell type-dependent effect of phospholipid and choles-terol on bile salt cytotoxicity. Gastroenterology 1991;101:457-464.

20. Spivey JR, Bronk SF, Gores GJ. Glycochenodeoxycholate-induced lethalhepatocellular injury in rat hepatocytes. Role of ATP depletion andcytosolic free calcium. J Clin Invest 1993;92:17-24.

21. Sokol RJ, Winklhofer-Roob BM, Deveraux MW, McKim JM. Generationof hydroperoxides in isolated rat hepatocytes and hepatic mitochondriaexposed to hydrophobic bile acids. Gastroenterology 1995;109:1249-1256.

22. Bouscarel B, Fromm H, Nussbaum R. Ursodeoxycholic acid mobilizesintracellular calcium and activates phosphorylase a in isolated hepato-cytes. Am J Physiol 1993;264:G243-G251.

HEPATOLOGY Vol. 27, No. 2, 1998 CERYAK ET AL. 555

23. Bouscarel B, Gettys TW, Fromm H, Dubner H. Ursodeoxycholic acidinhibits glucagon-induced cAMP formation in hamster hepatocytes: arole for protein kinase C. Am J Physiol 1995; 268:G300-G310.

24. Matsuzaki Y, Bouscarel B, Le M, Ceryak S, Gettys TW, Shoda J, Fromm H.Effect of cholestasis on regulation of cAMP synthesis by glucagon andbile acids in isolated hepatocytes. Am J Physiol 1997;273:G164-G174.

25. Redgrave TG. Formation of cholesteryl ester-rich particulate lipidduring metabolism of chylomicrons. J Clin Invest 1970;49:465-471.

26. Bouscarel B, Fromm H, Ceryak S, Cassidy MM. Ursodeoxycholic acidincreases low-density lipoprotein binding, uptake and degradation inisolated hamster hepatocytes. Biochem J 1991;280:589-598.

27. Malavolti M, Fromm H, Ceryak S, Roberts IM. Modulation of lowdensity lipoprotein receptor activity by bile acids: differential effects ofchenodeoxycholic and ursodeoxycholic acids in the hamster. J Lipid Res1987;28:1281-1295.

28. Malavolti M, Fromm H, Setchell KDR, Albert MB, Cohen B, Ceryak S.Formation, absorption, and biotransformation of D 6-lithocholenic acidin humans. Am J Physiol 1993;264:G163-G171.

29. Cass OW, Cowen AE, Hofmann AF, Coffin SB. Thin-layer chromato-graphic separation of sulfated and nonsulfated lithocholic acids andtheir glycine and taurine conjugates. J Lipid Res 1975;16:159-160.

30. Malavolti M, Fromm H, Ceryak S, Shehan KL. Interaction of potentiallytoxic bile acids with human plasma proteins: binding of lithocholic(3a-hydroxy-5b-cholanic-24-oic) acid to lipoproteins and albumin.Lipids 1989;24:673-676.

31. Bouscarel B, Nussbaum R, Dubner H, Fromm H. The role of sodium inthe uptake of ursodeoxycholic acid in isolated hamster hepatocytes.HEPATOLOGY 1995;21:145-154.

32. Denizot F, Lang R. Rapid colorimetric assay for cell growth and survival.Modifications to the tetrazolium dye procedure giving improved sensitiv-ity and reliability. J Immunol Methods 1986;89:271-277.

33. Ng PY, Hofmann AF. Tissue distribution of [114C]cholylglycine in ratsand hamsters with a bile fistula or bile duct ligation. Proc Soc Exp BiolMed 1977;154:134-137.

34. Cowen AE, Korman MG, Hofmann AF, Thomas PJ. Metabolism oflithocholate in healthy man. III. Plasma disappearance of radioactivityafter intravenous injection of labeled lithocholate and its derivatives.Gastroenterology 1975;69:77-82.

35. Oh SY, Dupont J. Identification and quantitation of cholanoic acids inhepatic and extra-hepatic tissues of the rat. Lipids 1975;10:340-347.

36. Little JM, Zimniak P, Radominska A, Lehman P, Lester R. Urinaryexcretion of lithocholic acid and its conjugates by the bile duct-ligatedrat. HEPATOLOGY 1991;14:690-695.

37. Hedenborg G, Norlander A, Norman A. Bile acid conjugates present in

tissues during extrahepatic cholestasis. Scand J Clin Lab Invest 1986;46:539-544.

38. Wang SD, Parsson H, Andersson R, Soltesz V, Johansson K, Bengmark S.Bacterial translocation, intestinal ultrastructure and cell membranepermeability early after major liver resection in the rat. Br J Surg1994;81:579-84.

39. Cucchiaro G, Meyers WC, Young SL, Branum GD, Quarfordt S. Bile acidtransport in the anhepatic rat. Gastroenterology 1992;103;1034-1040.

40. Spigelman MK, Zappulla RA, Malis LI, Holland JF, Goldsmith SJ,Goldberg JD. Intracarotid dehydrocholate infusion: a new method forprolonged reversible blood-brain barrier disruption. Neurosurgery 1983;12:606-612.

41. Wahler JB, Swain MG, Carson R, Bergasa NV, Jones EA. Blood-brainbarrier permeability is markedly decreased in cholestasis in the rat.HEPATOLOGY 1993;17:1103-1108.

42. Luxon BA, Weisiger RA. Sex differences in intracellular fatty acidtransport: role of cytoplasmic binding proteins. Am J Physiol 1993;265:G831-G841.

43. Podevin P, Calmus Y, Bonnefis MT, Veyrunes C, Chereau C, Poupon R.Effect of cholestasis and bile acids on interferon-induced 28,58-adenylatesynthetase and NK activities. Gastroenterology 1995;108:1192-1198.

44. Makino I, Hashimoto H, Shinozaki K, Yoshino K, Nakagawa S. Sulfatedand nonsulfated bile acids in urine, serum, and bile of patients withhepatobiliary diseases. Gastroenterology 1975;68:545-553.

45. Shoda J, Tanaka N, Osuga T, Matsuura K, Miyazaki H. Altered bile acidmetabolism in liver disease: concurrent occurrence of C-1 and C-6hydroxylated bile acid metabolites and their preferential excretion intourine. J Lipid Res 1990;31:249-259.

46. Hedenborg G, Norlander A, Norman A, Svenson A. Bile acid transport byserum lipoproteins in patients with extrahepatic cholestasis. Scand JClin Lab Invest 1986;46:745-750.

47. Aldini R, Montagnani M, Roda A, Hrelia A, Biagi PL, Roda E. Intestinalabsorption of bile acids in the rabbit: different transport rates in jejunumand ileum. Gastroenterology 1996;110:459-468.

48. Stolz A, Takikawa H, Ookhtens M, Kaplowitz N. The role of cytoplasmicproteins in hepatic bile acid transport. Annu Rev Physiol 1989;51:161-76.

49. Krahenbuhl S, Talos C, Fischer S, Reichen J. Toxicity of bile acids on theelectron transport chain of isolated rat liver mitochondria. HEPATOLOGY

1994;19:471-479.50. Javitt NB, Keating JP, Grand RJ, Harris RC. Serum bile acid patterns in

neonatal hepatitis and extrahepatic biliary atresia. J Pediatrics 1977;90:736-739.

556 CERYAK ET AL. HEPATOLOGY February 1998