Insulin Inhibits Secretin-Induced Ductal Secretion byActivation of PKC Alpha and Inhibition of PKA Activity

Gene D. LeSage,1,2 Luca Marucci,3 Domenico Alvaro,4 Shannon S. Glaser,1 Antonio Benedetti,3 Marco Marzioni,5

Tushar Patel,1 Heather Francis,1 Jo Lynne Phinizy,1 and Gianfranco Alpini1,5,6

Insulin stimulates canalicular bile flow by interaction with hepatocytes. Insulin regulates thefunction of a number of epithelia through activation and membrane translocation of Ca2�-dependent PKC isoforms. No information exists regarding insulin regulation of ductal bilesecretion. The aim of the study was to determine the role and mechanisms of action of insulin inthe regulation of cholangiocyte secretion in BDL rats. We determined the subcellular localizationof insulin receptor in cholangiocytes. We measured the effect of insulin on (1) secretin-stimulatedcAMP levels in cholangiocytes and duct expansion in intrahepatic bile duct units (IBDUs) in theabsence or presence of BAPTA/AM, H7 or rottlerin and (2) bile flow. We evaluated (1) if insulineffects are associated with activation of PKC alpha and (2) if activation of PKC causes inhibitionof secretin-stimulated cAMP levels and PKA activity. We found insulin receptors only in theapical domain of cholangiocytes. Insulin inhibited secretin-induced choleresis and secretin-stimulated cholangiocyte cAMP levels. Insulin inhibited secretin-induced secretion in IBDUswhen applied at the basolateral membrane or microinjected into IBDU lumen. Insulin inhibitoryeffects on cholangiocyte secretion were blocked by BAPTA/AM and H7. Insulin induced activa-tion of PKC alpha, which decreased secretin-stimulated cAMP and PKA activity. In conclusion,insulin inhibited secretin-induced ductal secretion of BDL rats through activation of PKC andinhibition of secretin-stimulated cAMP and PKA activity. In conclusion, insulin counter-regu-lates cholangiocyte secretory processes in the BDL model, which is characterized by cholangio-cyte proliferation. (HEPATOLOGY 2002;36:641-651.)

Secretin stimulates ductal secretion1-8 by interactingwith basolateral receptors expressed by cholangio-cytes.2,3,9 The effect of secretin on cholangiocyte

secretion is mediated by increased intracellular adenosine3�, 5�-monophosphate (cAMP) levels,1-8 activation ofprotein kinase A (PKA),10 and opening of cystic fibrosisconductance transmembrane regulator11 with activationof the Cl�/HCO3

� exchanger2,5,6,12 resulting in bicarbon-ate secretion into bile.1,7,8

Insulin stimulates canalicular bile secretion in dogs,rats, and guinea pigs.13-15 The concept that insulin stim-ulates canalicular bile flow by interaction with hepato-cytes is supported by studies in dogs showing that insulinincreases erythritol clearance.16 No information exists re-garding (1) the functional expression of insulin receptorsin cholangiocytes and (2) the role and mechanism of ac-tion of insulin in the regulation of ductal secretion. Insu-lin regulates the function of a number of epithelia throughactivation and membrane translocation of Ca2�-depen-dent and Ca2�-independent PKC isoforms.17-19 Weposed the following questions: (1) Are functional insulinreceptors expressed by cholangiocytes? (2) Are insulin re-ceptors expressed on the basolateral or apical pole of

Abbreviations: cAMP, adenosine 3�, 5�-monophosphate; PKA, protein kinase A;PKC, protein kinase C; BDL, bile duct ligation; RT-PCR, reverse transcriptionpolymerase chain reaction; GAPDH, glyceraldehyde-3-phosphate dehydrogenase;[Ca2�]i, intracellular Ca2�; BSA, bovine serum albumin; KRH, Krebs RingerHenseleit; IBDU, intrahepatic bile duct units; BAPTA/AM, 1,2-bis(2-amino-phenoxy)-ethane-N,N,N�,N�-tetraacetic acid tetrakis(acetxymethyl ester; H7, 1-(5-Isoquinolinylsulfonyl)-2-methylpiperazine.

From the 1Departments of Internal Medicine, 2Medical Biochemistry and Genetics,and 5Medical Physiology, Scott & White Hospital and The Texas A&M UniversitySystem HSC COM, Temple, TX; 3Department of Gastroenterology, University of An-cona, Ancona, Italy; 4Division of Gastroenterology, University of Rome, “La Sapienza,”Rome, Italy; and 6Central Texas Veterans Health Care System, Temple, TX.

Received April 19, 2002; accepted June 20, 2002.Supported by a grant award to G.A. and G.D.L. from Scott & White Hospital

and Texas A&M University, an NIH grant DK 54208 to G.D.L., an NIH grantDK58411 and VA Merit Award to G.A., a grant from MURST40%(MM06215421/2) progetto nazionale 2000 to D.A., and a grant award“MURSTMM06215421” to the Dept. of Gastroenterology, Ancona, Italy.

Address reprint requests to: Gianfranco Alpini, Ph.D., Associate Professor ofInternal Medicine and Medical Physiology, The Texas A&M University System,HSC COM and Central Texas Veterans HCS, MRB, 702 South West H. K.Dodgen Loop, Temple, TX 76504. E-mail: falpini@mailbox.sw.org orgalpini@tamu.edu; fax: 254-724-5944 and 254-742-7145.

Copyright © 2002 by the American Association for the Study of Liver Diseases.0270-9139/02/3603-0015$35.00/0doi:10.1053/jhep.2002.35537

641

cholangiocytes? (3) Does insulin regulate basal and secre-tin-stimulated ductal secretion? (4) Are insulin effects onductal bile secretion associated with activation and mem-brane translocation of the Ca2�-dependent PKC alpha?and (5) Does activation of PKC (after insulin treatment)lead to decreased secretin-stimulated intracellular cAMPlevels resulting in decreased PKA activity?

Materials and Methods

Animal Model. Male Fischer 344 rats (150-175 g)were purchased from Charles River Laboratories (Wil-mington, MA). Studies were performed in normal and1-week bile duct ligated (BDL) rats. Before experiments,animals were anesthetized with sodium pentobarbital (50mg/kg body weight, intraperitoneally).

Materials. Reagents were purchased from Sigma (St.Louis, MO) unless otherwise specified. The antibody re-acting with the rat insulin receptor20 was a gift of KenSiddle (University of Cambridge, Cambridge, UK).

Purification of Cholangiocytes. Pure cholangio-cytes were obtained by immunoaffinity bead purifica-tion.2,4,7,8,12,21,22 Cell number and viability (greater than97%) were assessed by trypan blue exclusion.

Expression of Insulin Receptor. Liver sections andcytospin smears from normal and BDL rats (n � 10) wereincubated with or without the antibody to the insulinreceptor. The reactive sites were detected using a DAKOLSAB kit. Following counterstaining, slides were exam-ined with a microscope (BX 40; Olympus Optical Co.,Tokyo, Japan).

We measured the genetic expression of insulin receptormRNA by reverse transcription polymerase chain reac-tion (RT-PCR) using cholangiocyte poly (A)� mRNAfrom normal and BDL rats. Primers for insulin receptor(sense 5�-AAGTGAGCTATCGGCGATATGGT-3� andantisense 5�-CTCATCCGGAACGTATACGGAAG-3�,fragment length 391 bp) were designed from the sequenceencoding the rat insulin receptor mRNA.23 The com-parability of the RNA was assessed by RT-PCR forglyceraldehyde-3-phosphate dehydrogenase (GAPDH).Primers for GAPDH were designed from the rat GAPDHsequence (sense 5�-GTGACTTCAACAGCAACTCC-CATTC-3� and antisense 5�-GTTATGGGGTCTGG-GATGGAATTGTG-3�, fragment length 294 bp).24

Immunoblots for the insulin receptor were performedin cholangiocytes. Blots were normalized by blotting for�-actin.25 The intensity of the bands was determined byscanning video densitometry using the ChemiImager4000 low light imaging system (Alpha Innotech Corp.,San Leandro, CA).

Intracellular Ca2� Levels. We measured the effect ofinsulin (100 nmol/L) on intracellular Ca2� ([Ca2�]i) lev-els, a second messenger for insulin.18 [Ca2�]i levels weredetermined by a microfluorescent technique26 in cholan-giocytes previously loaded with the fluorescent Ca2� in-dicator fluo-3 (1 micromolar for 10 minutes) prior totreatment with 0.2% bovine serum albumin (BSA) or 100nmol/L insulin in the presence of 0.2% BSA. The fluo-3fluorescence was converted to [Ca2�]i levels employing acalibration kit from Molecular Probes, Inc. (Eugene,OR).26 [Ca2�]i determinations were done in groups of 10to 20 cholangiocytes.

Effect of Insulin on Ductal Secretion: Bile Flow.When steady-state bile flow was reached, normal rats wereinfused with insulin (100 nmol/L for 30 minutes), KrebsRinger Henseleit (KRH, 60 minutes), secretin (100nmol/L for 30 minutes) followed by a final infusion ofKRH for 30 minutes. Since secretin does not change bileflow of normal rats,1,7,12 we did not evaluate the effect ofinsulin on secretin-stimulated bile flow in normal rats.When steady-state bile flow was reached, BDL rats weresubsequently infused with insulin (0.1 to 100 nmol/L for30 minutes), KRH (60 minutes), insulin (0.1 nmol/L to100 nmol/L) � secretin (100 nmol/L) for 30 minutes,KRH for 60 minutes followed by secretin (100 nmol/L)for 30 minutes, and by a final infusion of KRH for 30minutes. Biliary bicarbonate concentration (measured astotal CO2) was determined by a Natelson microgasometerapparatus (Scientific Industries, Bohemia, NY). The lev-els of insulin in bile (before and after administration ofinsulin) were measured by RIA using commercially avail-able kits from Linco Research Inc. (St. Charles, MO).

Effect of Insulin on Ductal Secretion in Intrahe-patic Bile Duct Units and Canalicular Secretion inHepatocytes. We evaluated the effects of insulin on basaland secretin-stimulated ductal secretion in large (sizerange of 50-100 �m with a mean diameter of 70 �m)intrahepatic bile duct units (IBDUs) from BDL rats.Ductal secretion was estimated from the changes in thevolume of IBDUs3,27,28 in two experimental approaches.In the first set of studies, insulin was added to the basolateralmembrane of IBDUs. Studies were performed in the absenceor in the presence of 1,2-bis(2-aminophenoxy)-ethane-N,N,N�,N�-tetraacetic acit tetrakis(acetxymethyl ester)(BAPTA/AM; 5 �mol, an intracellular Ca2� [Ca2�]i chela-tor6,8), or 1-(5-Isoquinolinylsulfonyl)-2-methylpiperazine;(H7; 20 �mol, a Ca2�-dependent PKC alpha inhibitor8).During basolateral application, insulin was added simul-taneously or 30 minutes before the addition of secretin.Since our studies showed insulin receptors to be expressedon the apical (lumenal) membrane of cholangiocytes, inthe second set of studies, insulin was applied at the apical

642 LESAGE ET AL. HEPATOLOGY, September 2002

side of IBDUs by direct microinjection into the lumen.Insulin was injected into IBDU lumen using a micropi-pette with a 5-�m diameter tip.3 Insulin concentration inthe pipette was 200 nmol/L and the injection was finishedwhen the lumen diameter increased 1.4 times the originalsize. Assuming a cylindrical geometry for the IBDU lu-men, the 1.4-increase in IBDU lumen diameter provideda final 100 nmol/L insulin lumen concentration. Secretin,BAPTA/AM, or H7 were added in the bathing solution.We previously showed that BAPTA/AM and PKC inhib-itors do not alter secretin-stimulated ductal secretion inIBDUs.6,10

Hepatocytes were isolated as described.29 Hepatocyteswere resuspended in culture medium at 1 � 105/mL.One-milliliter aliquots were cultured in cover slip bottomculture plates (Becton Dickinson, Lincoln Park, NJ).Cells were incubated for 4 hours in 5% CO2/air at 37°C.The cells were washed with phosphate-buffered saline andthe culture media was replaced with KRH, pH 7.4. Theplates were transferred to the stage of a Nikon TE300inverted microscope (Melville, NY). Hepatocyte coupletswith a clearly defined central lumen were identified usinga �40 objective. Images of the couplets were obtainedthrough a �63 objective in a focal plane in which theluminal diameter was greatest. Images were obtained be-fore and 15 minutes after the addition of 100 nmol/Linsulin using a CCD camera (Photometrics Sensys, Tus-con, AZ). Images were stored on a computer workstationand analyzed using the integrated morphometric analysisfunctions of the Metamorph imaging system (UniversalImaging Corp., West Chester, PA). As changes in themaximal canalicular luminal area correlate with the can-alicular volume, the percentage change in luminal areaprovides an index of canalicular bile secretion.29

cAMP Levels. Cholangiocytes (1 � 105) from normalrats were incubated for 5 minutes at room temperaturewith (1) 0.2% BSA (basal), (2) 100 nmol/L secretin with0.2% BSA, (3) 100 nmol/L insulin with 0.2% BSA, or (4)insulin plus secretin (at 100 nmol/L) with 0.2% BSA.Cholangiocytes (1 � 105) from BDL rats were incubatedfor 5 minutes at room temperature with (1) 0.2% BSA(basal), (2) 100 nmol/L secretin with 0.2% BSA, (3) 100nmol/L insulin with 0.2% BSA, or (4) insulin (0.1 to 100nmol/L) plus 100 nmol/L secretin with 0.2% BSA.Cholangiocytes from BDL rats were incubated for 5 min-utes at room temperature with 100 nmol/L insulin plus100 nmol/L secretin in the absence or presence ofBAPTA/AM, H7, or rottlerin with 0.2% BSA. We deter-mined the effects of BAPTA/AM, H7, and rottlerin oncAMP levels. cAMP levels were measured by RIA kits(Amersham Life Science, Inc., Arlington Heights, IL).

Are Insulin Effects on Secretin-Stimulated DuctalSecretion Associated With Activation of PKC alpha?Cholangiocytes were stimulated for 90 minutes8,17 at37°C with (1) 0.2% BSA for 90 minutes or (2) 100nmol/L insulin with 0.2% BSA and analyzed for PKCalpha by immunoblots.8 Blots were normalized by blot-ting for �-actin.25 The intensity of the bands was deter-mined by scanning video densitometry.

To determine if the effects of insulin on secretin-stim-ulated ductal secretion are associated with cellular redis-tribution of PKC alpha, we evaluated the intracellulardistribution of PKC alpha in liver sections, and in a tri-ton-soluble (containing cytoplasm � membrane) and atriton-insoluble (containing cytoskeleton) cholangiocytefraction from BDL rats infused via a jugular vein withKRH (control) or insulin (100 nmol/L) for 2 hours. Liversections were incubated with or without the mouse PKCalpha antibody (1:20 dilution in phosphate-buffered sa-line) and the reactive sites were detected using a DAKOLSAB kit according to the manufacturer’s instructions.Triton-soluble and triton-insoluble fractions were ob-tained from purified cholangiocytes.8,17 Immunoblots forPKC alpha in the triton-soluble and triton-insoluble frac-tion of cholangiocytes were performed as described.8

Does Insulin Activation of PKC Induce a Decreasein Secretin-Stimulated PKA Activity? We evaluated ifactivation of PKC (induced by insulin treatment) causesinhibition of secretin-stimulated PKA activity. For mea-surement of PKC activity, cholangiocytes (5 � 106) fromBDL rats were stimulated at 37°C for 90 minutes with (1)0.2% BSA or (2) insulin (100 nmol/L) with 0.2% BSA.For the determination of PKA activity, cholangiocytes(5 � 106) from BDL rats were treated at 37°C with (1)0.2% BSA (90 minutes), (2) secretin (30 minutes at 100nmol/L) with 0.2% BSA, (3) insulin (90 minutes at 100nmol/L) with 0.2% BSA, (4) insulin (90 minutes at 100nmol/L) before stimulation with secretin (30 minutes at100 nmol/L) with 0.2% BSA, or (5) BAPTA/AM, H7, orrottlerin (20 minutes each) before treatment with insulin(90 minutes at 100 nmol/L) followed by secretin (30 min-utes at 100 nmol/L) stimulation with 0.2% BSA. Cholan-giocyte samples were analyzed with the PepTag assay fornonradioactive detection of PKC and PKA from Promega(Madison, WI). Phosphorylated and nonphosphorylatedpeptide bands were visualized on a 0.8% agarose gel.Phosphorylated peptide bands were quantitated by scan-ning video densitometry. Densitometric values were nor-malized on the basis of cell number.

Statistical Analysis. All data are expressed as mean �SEM. The differences between groups were analyzed byStudent’s t test when two groups were analyzed.

HEPATOLOGY, Vol. 36, No. 3, 2002 LESAGE ET AL. 643

Results

Expression of Insulin Receptors. Immunohisto-chemistry of liver sections shows positive staining for theinsulin receptor in the apical domain of bile ducts fromnormal and BDL rats (Fig. 1A). The negative control isshown in Fig. 1A. Immunohistochemistry of cytospin

smears shows positive staining for insulin receptor incholangiocytes from normal and BDL rats (Fig. 1B). ByRT-PCR and immunoblot analysis, insulin receptor wasdetected in cholangiocytes from normal and BDL rats(Fig. 1C). Following BDL, we found an increase inthe expression of the insulin receptor in cholangiocytes(Fig. 1C).

Fig. 1. Immunohistochemistry for insulin receptor in (A) sections and (B) pure cholangiocyte smears from normal and BDL rats. (A) A specificreaction appears in the apical region of cholangiocytes. (B) The expression of insulin receptor is higher in purified cholangiocyte smears from BDLrats compared with normal cholangiocytes. The negative control section and cell smears from BDL rats incubated without the primary antibody showabsence of specific staining for the insulin receptor. Original magnification �150 (sections) and �250 (cell smears). (C) Immunoblots (normalizedby immunoblots for �-actin, the internal control) and RT-PCR (normalized by immunoblots for GAPDH, the internal control) for � insulin receptor inpure cholangiocytes isolated from normal and BDL rats. By RT-PCR and immunoblot analysis, insulin receptor (at both the message and protein levels)was detected in normal cholangiocytes and up-regulated in cholangiocytes from BDL rats. Data are mean � SE of 3 experiments. *P � .05 vs. thecorresponding value of normal cholangiocytes.

644 LESAGE ET AL. HEPATOLOGY, September 2002

Effect of Insulin on [Ca2�]i. Insulin (added at timeshown by the arrow of Fig. 2) caused a persistent increasein [Ca2�]i in purified single cholangiocytes (Fig. 2). Thetracing of Fig. 2 comes from 10 to 20 cholangiocytes fromBDL rats stimulated with BSA (basal value) or 100nmol/L insulin.

Effect of Insulin on Bile Secretion. Secretin did notalter bile flow (62.99 � 6.34 vs. 59.64 � 7.57 [basalvalue] �L/min/kg body weight), biliary bicarbonate con-centration (25.83 � 0.12 vs. 24.60 � 0.18 [basal value]mEq/L) or secretion (1.61 � 0.32 vs. 1.41 � 0.16 [basalvalue] �Eq/min/kg body weight) in normal rats. Parallelto other studies,13-15 insulin (100 nmol/L) increased bileflow in normal rats (71.71 � 3.63 vs. 60.64 � 3.89 [basalvalue] �L/min/kg body weight) presumably by selectiveinteraction with hepatocytes. Consistent with previousstudies,13 insulin-induced choleresis in normal rats wasnot associated with an increase in biliary bicarbonate con-centration (34.06 � 0.38 vs. 34.99 � 0.002 [basal value]mEq/L) or secretion (2.45 � 0.14 vs. 2.12 � 0.19 [basalvalue] �Eq/min/kg body weight). In BDL rats, secretinincreased bile flow, and bicarbonate concentration andsecretion (Fig. 3A-C). Similar to normal rats, insulin in-creased bile flow in BDL rats (Fig. 3A). Insulin did notincrease biliary bicarbonate concentration of BDL rats(Fig. 3B) but induced a slight increase in bicarbonatesecretion of BDL rats associated with an increase in basalbile flow (Fig. 3C). When infused together with secretin(100 nmol/L), insulin (only at the doses of 10 and 100nmol/L) inhibited secretin-stimulated bile flow and bicar-bonate concentration and secretion in BDL rats (Fig. 3A-C). In normal rats, the concentration of insulin in bile was29.99 � 5.56 pmol/L and was increased by intravenousadministration of 100 nmol/L insulin (45.42 � 5.89pmol/L). In BDL rats, the biliary concentration of insulinwas 31.69 � 7.65 pmol/L and increased during intrave-nous infusion of insulin (59.76 � 6.26 pmol/L).

Effect of Insulin on Ductal Secretion in IBDUs andCanalicular Secretion in Hepatocytes. An increase inductal secretion in IBDUs from BDL rats was observedafter stimulation with secretin (Table 1). Insulin alone(applied at the basolateral or apical membrane) did notalter basal IBDU lumen size (data not shown). Wheninsulin was superfused (basolateral application) simulta-neously with secretin, no modification of secretin-in-duced lumen expansion in IBDUs from BDL rats wasfound (Table 1). When IBDUs were pretreated withinsulin (added in the incubation medium, basolateralapplication) for 30 minutes, secretin-stimulated lumenexpansion was inhibited (Table 1). Since the inhibitoryeffects of insulin were only observed when insulin wasadded 30 minutes before secretin, and insulin receptor isexpressed on the apical membrane, we consider the re-quirement for insulin to cross cholangiocytes to reach theinsulin receptor on the apical membrane. We tested thishypothesis by demonstrating that insulin, when directlyinjected into IBDU lumen, inhibits the lumen expansioninduced by superfusion of IBDUs with secretin (Table 1).The inhibitory effect of insulin on secretin-increased duc-tal lumen expansion was abolished by pretreating IBDUswith BAPTA/AM or H7 (Table 1). BAPTA/AM or H7did not alter ductal lumen size (not shown).

We assessed the effect of insulin on canalicular secre-tion in hepatocyte couplets. Incubation with insulin for15 minutes increased the canalicular luminal area by9.1 � 1.1% (n � 11). Similar increases were seen inlumen perimeter (7.6 � 0.9%), but not in couplet area. Incontrol experiments (n � 11), incubation without insulinfor 15 minutes induced no measurable changes in cana-licular luminal area, lumen perimeter, and couplet area.Changes in the canalicular luminal area can be used as anindex of canalicular bile secretion. These data indicatethat insulin increases canalicular luminal secretion of he-patocyte couplets and that the choleretic effects of insulinare consistent with a hepatocyte origin.

cAMP Levels. Secretin increased cAMP levels in nor-mal cholangiocytes (Fig. 4A). Insulin alone did not alterbasal or secretin-stimulated cAMP levels in normalcholangiocytes (Fig. 4A). Insulin did not alter basal cAMPlevels, but inhibited (in a dose-dependent fashion) secre-tin-stimulated cAMP levels of cholangiocytes from BDLrats (Fig. 4B). Insulin inhibition of secretin-stimulatedcAMP synthesis was ablated by BAPTA/AM and H7 butnot rottlerin (Fig. 4C). BAPTA/AM, H7, or rottlerin didnot alter basal or secretin-stimulated cAMP levels (notshown).

Are Insulin Effects on Cholangiocyte Secretion As-sociated With Activation of PKC Alpha? PKC alphawas expressed by cholangiocytes (Fig. 5). Insulin in-

Fig. 2. Effect of insulin (100 nmol/L) on [Ca2�]i of cholangiocytesfrom BDL rats. [Ca2�]i was determined by a microfluorescent techniquein cholangiocytes previously loaded with the fluorescent Ca2� indicatorfluo-3 (1 �mol for 10 minutes). Insulin (added at time shown by arrow)caused a persistent increase in [Ca2�]i in purified cholangiocytes. Thetracing of Fig. 2 comes from a group of 10 to 20 cholangiocytes.Representative of 3 experiments.

HEPATOLOGY, Vol. 36, No. 3, 2002 LESAGE ET AL. 645

creased PKC alpha protein expression in cholangiocytesfrom BDL rats (Fig. 5). In sections from BDL rats, PKCalpha was localized mainly in the cytoplasm of cholangio-cytes (Fig. 6A). Following in vivo infusion of insulin toBDL rats, the specific staining for PKC alpha appearsstrongly distributed along the apical and nuclear mem-brane as well as in the cytoplasm of cholangiocytes (Fig.6B). A section incubated without the primary antibodyshowed absence of specific staining for PKC alpha incholangiocytes (Fig. 6C). In vivo infusion of insulin toBDL rats induces cytoskeleton to membrane distributionof PKC alpha, which was evidenced by increased PKCalpha protein expression in the triton-soluble fraction

compared with the triton-soluble fraction of BDL ratsinfused with KRH (Fig. 6D). We found a correspondingloss of PKC alpha protein expression in the triton-insol-uble fraction of cholangiocytes compared with the triton-insoluble fraction of BDL rats infused for 2 hours withKRH (Fig. 6D).

Does Activation of PKC Induce a Decrease in Se-cretin-Stimulated PKA Activity? Densitometric analy-sis of phosphorylated peptide bands shows that insulinincreased PKC activity (Fig. 7A). We found that secretinincreases cAMP-dependent PKA activity and that insulininhibits secretin-induced increases in PKA activity (Fig.7B). Insulin had no effect on basal PKA activity (Fig. 7B).

Fig. 3. Effect of insulin on basal and secretin-induced bile flow andbicarbonate secretion in BDL rats (A-C). When infused together withsecretin (100 nmol/L), insulin (only at the doses of 10 and 100 nmol/L)inhibited (A) secretin-stimulated bile flow and both (B) bicarbonateconcentration and (C) secretion in BDL rat. *P � .05 vs. correspondingbasal value. #P � .05 when compared with secretin-induced increasesin bile flow or bicarbonate secretion. Data are mean � SEM of at least6 rats.

646 LESAGE ET AL. HEPATOLOGY, September 2002

The inhibitory effect of insulin on secretin-stimulatedPKA activity was ablated by BAPTA/AM and H7 but notrottlerin (Fig. 7B).

DiscussionThis study relates to the inhibitory effect of insulin on

secretin-induced ductal secretion of BDL rats. Insulin in-hibits (1) secretin-stimulated cAMP levels in cholangio-cytes, (2) secretin-induced increases in secretion inIBDUs, and (3) secretin-stimulated choleresis by interact-ing with apically located insulin receptors on cholangio-cytes through transduction pathways involving [Ca2�]i,PKC, and PKA. Insulin inhibitory effects on secretin-

stimulated ductal secretion are associated with increasedactivity and membrane redistribution of PKC alpha, in-creased [Ca2�]i, inhibition of secretin-stimulated cAMPlevels, and PKA activity with subsequent decreased secre-tin-stimulated choleresis.

Table 1. Secretin-Stimulated Lumen Expansion(% of Basal, Unstimulated)

Secretin plus basolateral application of insulinControl (secretin alone) 27.2 � 4.8Insulin (simultaneous with insulin)* 28.4 � 5.7Insulin pretreatment* 13.7 � 4.9†BAPTA/AM plus insulin pretreatment* 21.3 � 4.3H7 plus insulin pretreatment* 25.9 � 6.1

Secretin plus apical application of insulin bymicroinjection into IBDU lumen

Control (KRB microinjected into lumen) 31.2 � 4.7Insulin (microinjected into lumen) 15.6 � 4.1†BAPTA/AM (added basolaterally) plus insulin

(microinjected into lumen) 22.9 � 4.3H7 (added basolaterally) plus insulin

(microinjected into lumen) 25.6 � 3.2

*In studies where insulin was added to the basolateral aspect of IBDUs, insulinwas added either at the same time as secretin (simultaneous) or added 30minutes prior to secretin (pretreatment).

†P � .05 compared with control.

Fig. 4. Measurement of basal and hormone-regulated cAMP levels in pure cholangiocytes isolated from (A) normal and (B and C) BDL rats. (Aand B) Cholangiocytes from normal and BDL rats were stimulated for 5 minutes at 22°C with (1) 0.2% BSA, (2) 100 nmol/L secretin with 0.2%BSA, (3) 100 nmol/L insulin with 0.2% BSA, or (4) insulin (0.1-100 nmol/L) plus 100 nmol/L secretin with 0.2% BSA. (C) Cholangiocytes werestimulated for 5 minutes at 22°C with (1) 0.2% BSA, (2) 100 nmol/L secretin with 0.2% BSA, (3) 100 nmol/L insulin with 0.2% BSA, or (4) insulinplus secretin (both at 100 nmol/L) in the presence or absence of BAPTA/AM, H7, or rottlerin with 0.2% BSA. *P � .05 vs. basal values. #P � .05vs. secretin-stimulated cAMP levels. Data are mean � SEM of at least 14 experiments.

Fig. 5. Immunoblot for PKC alpha in cholangiocytes from BDL ratstreated in vitro with 0.2% BSA or insulin (100 nmol/L) with 0.2% BSA for90 minutes8,17 at 37°C. Insulin increased the expression of PKC alphaprotein in cholangiocytes. *P � .05 vs. its corresponding basal value.Data are mean � SEM of at least 3 experiments.

HEPATOLOGY, Vol. 36, No. 3, 2002 LESAGE ET AL. 647

The effects of insulin on epithelial secretion are unde-fined.30,31 Insulin inhibits secretin-stimulated pancreaticbicarbonate secretion in dogs30 and increases pancreaticsecretion stimulated by secretin and cholecystokinin inhumans.31 In the liver, insulin increases bile flow in sev-eral species.13-16,32,33 The choleretic effect of insulin isdirect and not mediated by changes in the plasma levels ofhormones (e.g., glucagon and somatostatin) or glu-cose.15,32,34 Although the mechanism of insulin choleresisis undefined, studies showed that (1) insulin acts at thebile canaliculus since insulin-induced choleresis is associ-ated with increased erythritol clearance16 and (2) insulinincreases the bile salt-independent fraction of bile flowbut it does not alter bile salt secretion.16,32,34 We used theBDL rat because this model (characterized by up-regula-tion of the secretin receptor4,9 and by secretin-inducedcholeresis,1,7 absent in normal rats1,7,12) allows for a betterevaluation of ductal secretion.

In agreement with previous studies,16,34 insulin in-duces a similar increase in total bile flow in normal andBDL rats. These data suggest that the anatomical sites ofinsulin choleresis are hepatocytes rather than cholangio-cytes. The concept that insulin stimulates canalicular se-cretion by interaction with hepatocytes is supported bystudies in dogs showing that insulin increases erythritolclearance.16 This concept was directly confirmed by theexperiments (1) in isolated hepatocyte couplets showingthat insulin increases canalicular luminal area and (2) inIBDUs where insulin did not alter duct volume. Never-theless, rat cholangiocytes express insulin receptors andrespond to insulin with a sustained increase of [Ca2�]i.Furthermore, we found that insulin receptors are locatedin the apical domain of cholangiocytes and that their ex-pression is increased in cholangiocytes from BDL rats.Similar to cholangiocytes, insulin receptors were found inthe apical membrane of colonocytes.35 Insulin is taken up

Fig. 6. (A-C) Immunohistochem-istry for PKC alpha in liver sectionsfrom BDL rats infused for 2 hoursthrough a jugular vein with 0.2%BSA or insulin (100 nmol/L) with0.2% BSA. (A) In control section,specific staining for PKC alpha isdetected mainly in the cytoplasm ofcholangiocytes. (B) In sections fromBDL rats infused with insulin, PKCalpha appears strongly distributedalong the apical and nuclear mem-brane as well as in the cytoplasm ofcholangiocytes. (C) A control liversection incubated without the pri-mary antibody shows absence ofspecific staining for PKC alpha. Orig-inal magnification �150 (controland insulin) and �50 (negative).(D) Measurement of PKC alpha pro-tein expression in a triton-solubleand a triton-insoluble cholangiocytefraction obtained from 1-week BDLrats infused via a jugular vein withKRH (control) or insulin for 2 hours.In vivo infusion of insulin to BDL ratsinduces cytoskeleton to membranedistribution of PKC alpha, which wasevidenced by increased PKC alphaprotein expression in the triton-sol-uble fraction as compared with thetriton-soluble fraction of BDL ratsinfused for 2 hours with KRH. Wefound a corresponding loss of PKCalpha protein expression in the tri-ton-insoluble fraction of cholangio-cytes as compared with the triton-insoluble fraction of BDL ratsinfused for 2 hours with KRH. *P �.05 vs. the triton-soluble fraction ofBDL rats infused for 2 hours withKRH.

648 LESAGE ET AL. HEPATOLOGY, September 2002

by hepatocytes and cross hepatocytes by transcytosis to beexcreted into bile.36 Signaling via interaction with recep-tors on the apical membrane has been demonstrated forATP regulation of Cl� channels in cholangiocytes.37 Thelocalization of insulin receptors on the apical domain sug-gests that insulin acts from the lumenal side of cholangio-cytes, a concept supported by the finding that insulinbiliary concentration increased during infusion of thishormone to normal and BDL rats and by the direct dem-onstration that insulin applied at the apical pole of IBDUsinhibits secretin-stimulated expansion of lumen space. In-hibition of secretin-stimulated fluid secretion could bealso caused by increased glucose reabsorption from theapical pole of bile ducts induced by insulin. Although thishypothesis cannot be ruled out from our experiments,studies have shown38 that the glucose transporter identi-fied in cholangiocytes is not regulated by insulin. Thesestudies show that the facilitative glucose transporter(GLUT1) is expressed on the basolateral membraneof cholangiocytes and the Na�-glucose cotransporter(SGLT1) is expressed on the apical membrane of cholan-giocytes. Together, these glucose transporters have beenproposed to mediate glucose uptake from bile.38 It is un-likely, however, that insulin modulates ductal bile forma-tion through changes in these glucose transporters sinceinsulin regulates only GLUT4 in the glucose transporterfamily.39

The inhibitory effects of insulin on secretin-stimulatedductal secretion occur through activation of [Ca2�]i andPKC. Within the different isoforms of the PKC super-family, we exclude the involvement of PKC delta, becauseits specific inhibitor rottlerin failed to influence the inhib-itory effect of insulin on secretin-stimulated cAMP levelsand ductal lumen expansion. Consistent with activationof Ca2�-dependent PKC alpha in insulin inhibition ofsecretin-stimulated ductal secretion, we showed8 that ac-tivation of PKC alpha (by gastrin) inhibits secretin-stim-ulated cAMP levels. Similar counter-regulatory effects ofPKC agonists on cAMP-dependent Cl�/HCO3

� activityhas been observed in rat hepatocytes.40 We observed thata PKC inhibitor, H7, and a [Ca2�]i chelator, BAPTA/AM,6,8 prevented insulin effects on secretin-stimulatedcAMP levels in cholangiocytes and secretin-induced duc-tal secretion in IBDUs.

We evaluated whether activation and cholangiocytemembrane translocation of PKC alpha (induced by insu-lin) leads to inhibition of secretin-stimulated PKA ac-tivity. The interaction of secretin with its own receptorleads to increased cholangiocyte cAMP and PKA activ-ity,2,3,6-8,10,12,22 opening of CFTR11 with subsequent ac-tivation of the Cl�/HCO3

� exchanger.2,6,12 Thus, the

Fig. 7. Measurement of (A) PKC and (B) PKA activity. PKA and PKCassays were performed according to the manufacturer’s instructionsusing PepTag Assay Protein Kinase Kits. Data are mean � SEM of 6experiments.

HEPATOLOGY, Vol. 36, No. 3, 2002 LESAGE ET AL. 649

elevation of intracellular cAMP levels and the recruit-ment of PKA play a central role in secretin-stimulatedbicarbonate-rich choleresis.7,8,10,22 We found that insulininhibits both secretin-stimulated cAMP levels and PKAactivity but does not alter basal cAMP levels and PKAactivity, which supports the concept that insulin inhibitssecretin-stimulated choleresis but not basal ductal secre-tion. We found that insulin decreased secretin-stimulatedPKA activity in cholangiocytes, and that insulin inhibi-tion of PKA activity was ablated by H7 or BAPTA/AM.

These studies showed that insulin increased [Ca2�]i

and that increased [Ca2�]i was required for insulin inhi-bition of secretin-stimulated cAMP levels and ductal se-cretion. Consistent with our findings, studies showed18

that insulin effects occur through increased [Ca2�]i. Thecross talk between [Ca2�]i and cAMP signalling pathwaysis complex and dependent on the species, cell types, andthe different agonists/antagonists involved.41-45 Modula-tion of agonist-induced cAMP levels by [Ca2�]i may oc-cur at the level of receptors, G proteins, adenylate cyclase,or phosphodiesterase.41-43 There are at least 11 differentisoforms of adenylate cyclases with different sensitivity toand regulation by [Ca2�]i, PKC, PKA, and G proteins.41

The adenylyl cyclase isoform 8 is stimulated by Ca2�-calmodulin pathways and by this mechanism, muscarinicagents increase agonist-induced cAMP release in differentcells.41 We described6 a similar mechanism in cholangio-cytes, where acetylcholine increased secretin-inducedcAMP release via calmodulin-, and calcineurin-depen-dent but PKC-independent pathways. In other cells, ad-enylyl cyclase isoforms 5 and 6 are inhibited by Ca2�

agonists through a calmodulin-independent mechanism,which leads to decreased cAMP release.41-46

Insulin inhibits secretin-stimulated ductal secretionthrough activation and membrane translocation of PKCalpha, which leads to decreased secretin-stimulated cAMPlevels and PKA activity. The presence of an apically lo-cated insulin receptor on cholangiocytes suggests the pos-sibility of paracrine regulation of ductal secretion, withinsulin release from hepatocytes modulating downstreamductal secretion. Insulin may have a dual and counterregulatory role on bile secretion in hepatocytes andcholangiocytes, respectively, in the BDL hyperplastic ratmodel, which is characterized by cholangiocyte prolifera-tion. Since changes in bile flow and biliary lipid secretionhave been described in experimental and human diabetes(the latter being characterized by increased risk of gall-stone disease),47,48 we also propose that insulin-depen-dent alteration of ductal secretion may be an additionalmechanism determining pathologic changes of bile com-position in the course of diabetes.

References1. Alpini G, Lenzi R, Sarkozi L, Tavoloni N. Biliary physiology in rats with

bile ductular cell hyperplasia. Evidence for a secretory function of prolif-erated bile ductules. J Clin Invest 1988;81:569-578.

2. Alpini G, Roberts SK, Kuntz SM, Ueno Y, Gubba S, Podila P, LeSage G,LaRusso NF. Morphological, molecular and functional heterogeneity ofcholangiocytes from normal rat liver. Gastroenterology 1996;110:1636-1643.

3. Alpini G, Glaser S, Robertson W, Rodgers R, Phinizy JL, Lasater J, LeSageGD. Large but not small intrahepatic bile duct units are involved in secre-tin-regulated ductal bile secretion in normal rat liver. Am J Physiol 1997;272:G1064-G1074.

4. Alpini G, Glaser SS, Ueno Y, Pham L, Podila PV, Caligiuri A, LeSage G,LaRusso NF. Heterogeneity of the proliferative capacity of rat cholangio-cytes after bile duct ligation. Am J Physiol 1998;274:G767-G775.

5. Alvaro D, Cho WKC, Mennone A, Boyer JL. Effect of secretin on intra-cellular pH regulation in isolated rat bile duct epithelial cells. J Clin Invest1993;92:1314-1325.

6. Alvaro D, Alpini G, Jezequel AM, Bassotti C, Francis C, Fraioli F, RomeoR, et al. Role and mechanisms of acetylcholine in the regulation of cholan-giocyte secretory functions. J Clin Invest 1997;100:1349-1362.

7. Glaser S, Rodgers R, Phinizy JL, Robertson W, Lasater J, Caligiuri A,Tretjak Z, et al. Gastrin inhibits secretin-induced ductal secretion by in-teraction with specific receptors on rat cholangiocytes. Am J Physiol 1997;273:G1061-1070.

8. Glaser S, Benedetti A, Marucci L, Alvaro D, Baiocchi L, Kanno N, Cali-giuri A, et al. Gastrin inhibits cholangiocyte growth in bile duct ligated ratsby interaction with CCK-B/gastrin receptors via IP3-, Ca2�-, and PKCalpha-dependent mechanisms. HEPATOLOGY 2000;32:17-25.

9. Alpini G, Ulrich II C, Phillips J, Pham L, Miller L, LaRusso N. Upregu-lation of secretin receptor gene expression in rat cholangiocytes after bileduct ligation. Am J Physiol 1994;266:G922-G928.

10. Alvaro D, Mennone A, Boyer JL. Role of kinases and phosphatases in theregulation of fluid secretion and Cl-/HCO3

- exchange in cholangiocytes.Am J Physiol 1997;273:G303-G313.

11. Fitz JG, Basavappa S, McGill J, Melhus O, Cohn JA. Regulation of mem-brane chloride currents in rat bile duct epithelial cells. J Clin Invest 1993;91:319-328.

12. LeSage G, Glaser S, Gubba S, Robertson WE, Phinizy JL, Lasater J, Rodg-ers R, et al. Regrowth of the rat biliary tree after 70% partial hepatectomyis coupled to increased secretin-induced ductal bile secretion. Gastroenter-ology 1996;111:1633-1644.

13. Jones RS. Effect of insulin on canalicular bile formation. Am J Physiol1976;23:40-43.

14. Meyers WC, Wellman C, Quarfordt SH, Jones RS. Insulin or glucagoncholeresis in the isolated perfused guinea pig liver. Proc Soc Exp Biol Med1983;173:56-62.

15. Thomsen OO, Larsen JA. The effect of glucagon, dibutyrylic cyclic AMPand insulin on bile production in the intact rat and the perfused rat liver.Acta Physiol Scand 1981;111:23-30.

16. Garberoglio CA, Richter H, Henarejos A, Moossa AR, Baker AL. Pharma-cological and physiological doses of insulin and determinants of bile flow indogs. Am J Physiol 1983:G157-G163.

17. Andre F, Rigot V, Remacle-Bonnet M, Luis J, Pommier G, Marvaldi J.Protein kinase C-� and-� are involved in insulin-like growth factor-I mi-gration of colonic epithelial cells. Gastroenterology 1999;116:64-77.

18. Komatsu M, Noda M, Sharp GW. Nutrient augmentation of Ca2�-de-pendent and Ca2�-independent pathways in stimulus-coupling to insulinsecretion can be distinguished by their guanosine triphosphate require-ments: studies on rat pancreatic islets. Endocrinol 1998;139:1172-1183.

19. Reks SE, Smith PH, Messina JL, Weinstock RS. Translocation of PKCdelta by insulin in a rat hepatoma cell line. Endocrine 1998:161-167.

20. Ganderton RH, Stanley KK, Field CE, Coghlan MP, Soos MA, Siddle K.A monoclonal anti-peptide antibody reacting with the insulin receptor

650 LESAGE ET AL. HEPATOLOGY, September 2002

beta-subunit. Characterization of the antibody and its epitope and use inimmunoaffinity purification of intact receptors. Biochem J 1992;288:195-205.

21. Ishii M, Vroman B, LaRusso NF. Isolation and morphological character-ization of bile duct epithelial cells from normal rat liver. Gastroenterology1989;97:1236-1247.

22. LeSage G, Alvaro D, Benedetti A, Glaser S, Marucci L, Eisel W, CaligiuriA, et al. Cholinergic system modulates growth, apoptosis and secretion ofcholangiocytes from bile duct ligated rats. Gastroenterology 1999;117:191-199.

23. Goldstein BJ, Dudley AL. The rat insulin receptor: primary structure andconservation of tissue-specific alternative messenger RNA splicing. MolEndocrinol 1990;4:235-244.

24. Fort P, Marty L, Piechaczyk M, el Sabrouty S, Dani C, Jeanteur P, Blan-chard JM. Various rat adult tissues express only one major mRNA speciesfrom the glyceraldehyde-3-phosphate-dehydrogenase multigenic family.Nucleic Acid Res 1985;13:1431-1442.

25. Alpini G, Ueno Y, Glaser S, Phinizy JL, Francis H, LeSage G. Bile acidfeeding stimulates proliferative activity of both small and large cholangio-cytes through activation and membrane translocation of protein kinase Calpha. HEPATOLOGY 2001;34:868-876.

26. Hovis JG, Meyer T, Teasdale RM, Albrecht BN, Yorek MA, LoweWLJ. Intracellular calcium regulates insulin-like growth factor-I messengerribonucleic acid levels. Endocrinol 1993;132:1931-1938.

27. Mennone A, Alvaro D, Cho W, Boyer JL. Isolation of small polarized bileduct units. Proc Natl Acad Sci U S A 1995;92:6527-6531.

28. Roberts S, Kuntz S, Gores G, LaRusso N. Regulation of bicarbonate-dependent ductular secretion assessed by lumenal micropuncture of iso-lated rodent intrahepatic bile ducts. Proc Natl Acad Sci U S A 1993;90:9080-9084.

29. Trauner M, Mennone A, Gigliozzi A, Fraioli F, Boyer JL. Nitric oxide andguanosine 3’,5’-cyclic monophosphate stimulate bile secretion in isolatedrat hepatocyte couplets, but not in isolated bile duct units. HEPATOLOGY

1998;28:1621-1628.30. Berry SM, Fink AS. Insulin inhibits secretin-stimulated pancreatic bicar-

bonate output by a dose-dependent neurally mediated mechanism. Am JPhysiol 1996;270:G163-G170.

31. Kim C, Kim K, Lee H, Song C, Ryu H, Hyun J. Potentiation of cholecys-tokinin and secretin-induced pancreatic exocrine secretion by endogenousinsulin in humans. Pancreas 1999;18:410-414.

32. Thomsen OO, Larsen JA. Interaction of insulin, glucagon, and DBcAMPon bile acid-independent bile production in the rat. Scand J Gastroenterol1982;17:687-693.

33. Thomsen OO, Larsen JA, Orskov H. Insulin-induced choleresis in relationto insulin concentrations in plasma and bile in the cat. Scand J Gastroen-terol 1982;17:297-303.

34. Jones RS, Putnam W, Andersen DK, Hanks JB, Lebovitz HE. Insulin’seffect on bile flow and lipid excretion during euglycemia and hypoglyce-mia. Dig Dis Sci 1984;29:33-39.

35. Pillion DJ, Ganapathy V, Leibach FH. Identification of insulin receptorson the mucosal surface of colon epithelial cells. J Biol Chem 1985;260:5244-5247.

36. Renston RH, Maloney DG, Jones AL, Hradek GT, Wong KY, GoldfineID. Bile secretory apparatus: evidence for a vesicular transport mechanismfor proteins in the rat, using horseradish peroxidase and [125I]insulin. Gas-troenterol 1980;78:1373-1388.

37. Roman RM, Feranchak AP, Salter KD, Wang Y, Fitz JG. EndogenousATP release regulates Cl- secretion in cultured human and rat biliary epi-thelial cells. Am J Physiol 1999;276:G1391-400.

38. Lazaridis KN, Pham L, Vroman B, de Groen PC, LaRusso NF. Kinetic andmolecular identification of sodium-dependent glucose transporter in nor-mal rat cholangiocytes. Am J Physiol 1997;272:G1168-G1174.

39. Bryant NJ, Govers R, James DE. Regulated transport of the glucose trans-porter glut4. Nat Rev Mol Cell Biol 2002;3:267-277.

40. Benedetti A, Strazzabosco M, Ng OC, Boyer JL. Regulation of Activityand Apical Targeting of the Cl-/HCO3

- Exchanger in Rat Hepatocytes.Proc Natl Acad Sci U S A 1994;91:792-796.

41. Watson EL, Jacobson KL, Singh JC, Idzerda R, Ott SM, DiJulio DH,Wong ST, et al. The type 8 adenylyl cyclase is critical for Ca2� stimulationof cAMP accumulation in mouse parotid acini. J Biol Chem 2000;275:14691-14699.

42. Kawabe J, Iwami G, Ebina T, Ohno S, Katada T, Ueda Y, Homcy CJ, et al.Differential activation of adenylyl cyclase by protein kinase C isoenzymes.J Biol Chem 1994;269:16554-16558.

43. Geoffroy V, Fouque F, Nivet V, Clot JP, Lugnier C, Desbuquois B, BenelliC. Activation of a cGMP-stimulated cAMP phosphodiesterase by proteinkinase C in a liver Golgi-endosomal fraction. Eur J Biochem 1999;259:892-900.

44. Chiono M, Mahey R, Tate G, Cooper DM. Capacitative Ca2� entryexclusively inhibits cAMP synthesis in C6-2B glioma cells. Evidence thatphysiologically evoked Ca2� entry regulates Ca(2�)-inhibitable adenylylcyclase in non-excitable cells. J Biol Chem 1995;270:1149-1155.

45. Hirst RA, Lambert DG. Adenylyl cyclase in SH-SY5Y human neuroblas-toma cells is regulated by intra- and extracellular calcium. Biochem Phar-macol 1995;26:1633-1640.

46. Yu HJ, Ma H, Green RD. Calcium entry via L-type calcium channels actsas a negative regulator of adenylyl cyclase activity and cyclic AMP levels incardiac myocytes. Mol Pharmacol 1993;44:689-693.

47. Gonzalez J, Fevery J. Spontaneously diabetic biobreeding rats and impair-ment of bile acid-independent bile flow and increased biliary bilirubin,calcium and lipid secretion. HEPATOLOGY 1992;16:426-432.

48. De Santis A, Attili AF, Ginanni Corradini S, Scafato E, Cantagalli A, DeLuca C, Pinto G, et al. Gallstones and diabetes: a case-control study in afree-living population sample. HEPATOLOGY 1997;25:787-790.

HEPATOLOGY, Vol. 36, No. 3, 2002 LESAGE ET AL. 651