Isolation of a nonparenchymal liver cell fraction enriched in cells with biliary epithelial phenotypes

GASTROENTEROLOGY 1989;97:1248-60

Isolation of a Nonparenchymal Liver Cell Fraction Enriched in Cells With Biliary Epithelial Phenotypes GIANFRANCO ALPINI, RENATO LENZI, WEI-RONG ZHAI, MARGARET H. LIU, PHYLLIS A. SLOTT, FIORENZO PARONETTO, and NICOLA TAVOLONI Departments of Medicine (Polly Annenberg Levae Hematology Center) and Pathology, Mount Sinai School of Medicine of the City University of New York and Immunopathology Laboratory, Veterans Administration Medical Center, New York, New York

In the present study we have isolated and purified fractions of nonparenchymal liver cells enriched in biliary epithelial cells. Nonparenchymal liver cells were isolated by collagenase-pronase digestion of the biliary and connective hepatic tissue, which remained undissociated after collagenase perfusion of the liver. Fractionation of the nonparenchymal fractions was then achieved by centrifugal elutriation. Both normal rats and rats with proliferated bile duct-like structures, which were induced either by a l&day bile duct ligation or by feeding 0.1% cY-naphthylisothiocyanate for 28 days, were used in these studies. Using a normal rat liver, the fraction richest in biliary epithelial cells was that obtained at a pump flow rate of 38-40 ml/min. In this fraction 1.8-3.8x lo6 cells per liver were recovered and up to 55% of them were positive for y-glutamyl transpeptidase and cytokeratins 7 and 19,all of which were histochemically or immunohistochemically detected solely in the biliary structures in the intact rat liver. When the nonparenchymal cells were isolated from hyperplastic livers, the number of cells recovered in such a fraction ranged from 12 to 19 x 10’ per liver, and as many as 80%-85% of the cells expressed phenotypes of biliary epithelial cells. These results indicate that (a) by centrifugal elutriation a fraction of nonparenchymal cells enriched in cells with biliary epithelial phenotypes can be obtained from rat liver and (b) the hepatic hyperplasia induced by biliary obstruction or (Ynaphthylisothiocyanate feeding is a useful and valid strategy for improving both the yield and the purity of the isolated biliary epithelial cells. he bile ductules and ducts comprise a system of ramified tubular structures collecting the product of hepatocyte secretion and delivering it to the

T

gallbladder or the intestine. These passageways are lined by simple cuboidal (ductules and small ducts) or columnar (larger ducts) epithelial cells, conventionally referred to as biliary epithelial or bile ductular cells (BDCs). Over the last three decades, extensive research has been carried out to unveil both the structure and function of BDCs and their role in liver disease. From these studies we have learned that pm), and BDCs are relatively large in size (11-14 have a prominent Golgi system, abundant vesicular bodies, little smooth endoplasmic reticulum, and numerous microvilli on their luminal surfaces (l-4). At least in some species, these cells secrete water and electrolytes both spontaneously and under hormonal stimulation (5-7), reabsorb fluid (8,9), and are involved in protein biliary translocation (4,10,11). Finally, we have recognized that the intrahepatic BDCs are the site of injury in various forms of liver disease, particularly when cell-mediated immunity seems to play a pathogenic role [e.g., primary biliary cirrhosis (l2), primary sclerosing cholangitis (13), hepatic graft-versus-host disease (14,15), liver allograft rejection (16,17)], and may be the precursors of chemically induced hepatocellular carcinoma (18-20). Despite these fervent efforts, however, present knowledge of the physiology and pathophysiology of the intrahepatic BDCs is still poor. Obviously, multiple causes underlie this unsatisfactory progress, but the major fault lies in the lack of suitable exper-

Abbreviations used in this paper: ANIT, a-naphtbylisotbiocyanate; BDC, bile ductular cell; CK, cytokeratin; GGT, pglutamyl transpeptidase; G6P, glucose+phosphatase; vWF, von Willebrand factor. 0 1989 by the American Gestroenterological Association 0016.5085/89/$3.50

ISOLATION OF INTRAHEPATIC BILIARY EPITHELIAL CELLS 1249

November 1989

K4.62) against cytokeratin (CK) 19 [molecular weight = imental models. Studies of BDC function have thus 40,000, according to the catalogue by Moll et al. (38)] were far relied on either biliary clearance measurements purchased from Accurate Chemical Co. (Westbury, N.Y.). in vivo, the validity of which has seriously been Monoclonal antibodies (clone 0) against CK 7 (molecular questioned (21), or an isolated preparation of the weight = 54,000] and against vimentin were obtained from extrahepatic bile duct (22-24), whose relevance to Amersham (Arlington Heights, Ill.), and rabbit monospethe intrahepatic biliary epithelium is at best uncercific antihuman von Willebrand factor (vWF) antiserum tain. Morphologic, histochemical, and immunohiswas from Behring Diagnostics (La Jolla, Calif.). Biotinytochemical studies of intact liver tissue have contriblated horse-antimouse immunoglobulin G, biotinylated goat-antirabbit immunoglobulin G, and the avidin-biotinuted significantly to our current understanding of peroxidase complex solution were purchased from Vector the role of BDCs in hepatic function and disease, but Laboratories (Burlingame, Calif.). Other reagents were obneedless to say, these approaches do not lend themtained from Sigma Chemical Co. (St. Louis, MO.), unless selves to unraveling the inner workings of these otherwise indicated. cells. It has thus become increasingly clear that to obtain a better understanding of the biology of these Isolation and Fractionation of cells, we must make use of in vitro models, such as Nonparenchymal Liver Cells isolated BDCs in suspension or in culture. Over the Hepatocytes were isolated by a standard collagelast few years, several investigators have attempted nase perfusion. The undissociated tissue was then used for to isolate (25-33)and even culture (31,32,34,35) the isolation of nonparenchymal liver cells essentially as BDCs or bile duct-like cells from rat liver, yet condescribed by Yaswen et al. (28). Briefly, the undissociated flicting results have been reported with respect to tissue was minced and incubated three times (20 min each BDC purity. Because of our interest in the intrahetime) in Joklik-modified Minimum Essential Medium conpatic biliary epithelium, we have also attempted to taining 0.1% collagenase, 0.1% pronase, and 0.004%0.006% deoxyribonuclease. The cell suspensions were isolate, purify, and characterize a fraction of nonfiltered through a 40-pm nylon mesh, spun down at 300 g parenchymal rat liver cells with biliary epithelial for 5 min, resuspended, and pooled in 15-20 ml of phenotypes. The results of these efforts are the Minimum Essential Medium containing 10% calf serum subject of this report.

Materials and Methods Animals,

Diets, and Reagents

Male SpragueDawley rats (180-200 g) were obtained from Perfection breeders (Douglasville, Pa.) and housed in a temperature-controlled room (22°C) with alternating 12-h light-dark cycles for at least 1 wk before being used. Experiments were carried out in three groups of animals. Group 1 consisted of control rats fed a standard rat diet. Group 2 included rats in which the common bile duct was ligated to induce proliferation of bile duct-like cells (36). These animals were also fed a standard rat diet and used 3, 7, and 14 days after the induction of biliary obstruction. Group 3 consisted of rats fed a diet containing 0.1% cr-naphthylisothiocyanate (ANIT) for up to 28 days. As in biliary obstruction, ANIT feeding results in proliferation of bile duct-like structures (37). The ANIT diet was prepared by Dyets Inc. (Bethlehem, Pa.). Collagenase was obtained from Boehringer Mannheim Biochemicals (Indianapolis, Ind.); pronase from Calbiochem (La Jolla, Calif.); Minimum Essential Medium and Joklik-modified Minimum Essential Medium from Grand Island Biological Co. (Grand Island, N.Y.); and calf serum from Flow Laboratories (McLean, Va.). Goat-antirat albumin antibodies and sheep-antirat a-fetoprotein antibodies were obtained as FITC conjugates from Cooper Biochemical, Inc. (Malvern, Pa.) and Nordic Immunologic Laboratories (Capistrano Beach, Calif.), respectively. Mouse monoclonal antibodies (Clone OX 1) against rat leukocyte common antigen and mouse monoclonal antibodies (clone

and 0.004%-0.006% deoxyribonuclease. Fractionation of the nonparenchymal liver cell fraction was carried out by centrifugal elutriation using a Beckman J2-21 centrifuge equipped with a JE-6B rotor (Beckman Instruments, Palo Alto, Calif.). The centrifuge chamber was kept at 8”-lO”C, the rotor speed at 2500 rpm, and the pump flow rate was changed progressively from 9 (loading flow rate) to 70 ml/min. Six 50-ml fractions were first collected at a pump flow rate of 11 ml/min to eliminate red blood cells, debris, dead cells, and other contaminants. Four to five SO-ml fractions were then collected at each of the following pump rates: 14, 22, 28, 36, 40, and 44 ml/min. The cells in each fraction were then centrifuged at 500 g for 10 min, and resuspended in an appropriate volume of Minimum Essential Medium containing 10% calf serum. In all steps involved in the isolation and fractionation of nonparenchymal liver cells, nonsiiiconized glassware was used to facilitate adherence of phagocytic cells. Aliquots of the cell suspensions were then used immediately for determining cell concentration and size distribution (Coulter Counter and Channelyzer; Coulter Electronics, Hialeah, Fla.), and for estimating cell viability (trypan blue exclusion). Several cell smears were also prepared from each fraction and air-dried. Some were fixed in glutaraldehyde (1%) for 1-2 min and others were immersed in cold (-20°C) acetone or chloroform for 5-10 min. Smears were used immediately or stored at -70°C. Histologic, Histochemical, and Immunohistochemical Techniques

Liver histology was examined in control rats and in experimental animals killed at various times after bile duct

1250

ALPINI ET AL.

Table

1. Cell Phenotypes Assessed in Liver Sections andlor in Smears of Isolated Parenchymal and Nonparenchymal Liver Cells

Phenotype” G6P Albumin AFP PX vWFC Vimentin LCA GGTd CK-7 CK-19

GASTROENTEROLOGY Vol. 97, No. 5

Cell specificityb Hepatocytes Hepatocytes Fetal heptocytes Kupffer cells; neutrophils; monocytes Vascular endothelial cells Endothelial cells, Kupffer cells, monocytes, fibroblasts Leukocytes Biliary epithelial cells; leukocytes Biliary epithelial cells Biliary epithelial cells

Reference 46 7, 71 7, 60, 72 66. 69 47 63, 73 70 7, 42, 43, 52 45, 54 44,45

L1Expression of glucose-6-phosphatase (G6P), rglutamyl transpeptidase (GGT), cytokeratin-7 (CK-7), CK-19, and vimentin were studied both in liver sections and in smears of isolated cells, whereas detection of endogenous peroxidase (PX], leukocyte common antigen (LCA), von Willebrand factor (vWF), albumin, and cY-fetoprotein (AFP) was limited to smears of parenchymal and nonparenchymal liver cells. b Cell specificity refers only to resident liver-derived cells and other cell types (e.g., blood-borne cells) that may contaminate the nonparenchyma1 liver cell fraction during the isolation procedure. In all instances, specificity refers to normal cells. Erythrocytes and other types of blood cells were recognized by light microscopy (Wright-Giemsa stain). ’ By optical microscopy, only the vascular endothelial cells of the liver display immunoreactivity. At the electron-microscopic level, however, positive stain is observed also in sinusoidal endothelial cells (unpublished observations). d Histochemical activity only.

ligation or the initiation of the ANIT diet. Small liver blocks were obtained from anesthetized rats (pentobarbital sodium, 50 mg/kg body wt, i.p.), fixed in buffered formalin, and processed for hematoxylin and eosin staining by standard procedures. From the same rats, additional liver blocks were obtained, frozen immediately in Z-methylbutane (Eastman Kodak, Rochester, N.Y.) in solid CO,, cut into IO-km-thick sections, air-dried, immersed in cold acetone or chloroform for 5-10 min, and used for histochemical and immunohistochemical analyses (Table 1). Histochemical expression of y-glutamyl transpeptidase (GGT), glucose-6-phosphatase (G6P), and endogenous peroxidase was studied as described by Rutenberg et al. (39), Teutsch (do), and Straus (41), respectively. von Willebrand factor antigen was determined with the avidin-biotinperoxidase complex procedure using rabbit monospecific antihuman vWF antiserum. Briefly, liver sections or isolated cell smears (see Table 1) were immersed first in cold (-20°C) acetone for 5-10 min and, after drying and washing, were incubated for 60 min at 22°C with anti-vWF antiserum (dilution 1:50-250). After washing, sections were incubated for 60 min with biotinylated goat-antirabbit immunoglobulin G (l:lOO), washed, and reincubated for 60 min with the avidin-biotin-peroxidase complex solution. Finally, sections were washed and incubated in the dark for 1-20 min (22°C) with a buffered (pH = 7.6) solution containing Tris (0.05 M), Triton

(0.05%), H,O, (O.Ol%), diaminobenzidine tetrahydrochloride (0.05%), and NiCl, (0.03%). Leukocyte common antigen was immunochemically detected also with’the avidinbiotin-peroxidase complex procedure as described above, and albumin and a-fetoprotein were determined as reported elsewhere (7). For immunohistochemical expression of CK-19 and CK-7, we used mouse monoclonal antibodies again3 CK-19 and CK-7, respectively, and the immunoreactivity was detected with the indirect peroxidase procedure. Liver sections were first immersed in cold chloroform (- 20°C) for 5 min. After drying and washing in phosphate-buffered saline (pH = 7.21, sections were incubated at 2’2°C with anti-CK antibodies for 60 min, washed, and reincubated for 30 min at 22°C with affinity-purified, sheep-peroxidase-conjugated antimouse immunoglobulin G. Thereafter, sections were washed and incubated in the dark for 5-10 min (22°C) with the benzidine solution as described for vWF detection. For vimentin staining, chloroform-treated sections were incubated for 45-60 min with antivimentin monoclonal antibodies, washed, and reincubated for 30 min with biotinylated antimouse immunoglobulin G. After washing, immunoreactivity was detected with the avidin-biotin-peroxidase complex procedure as described above for vWF. When desired, sections were counterstained with Mayer’s hematoxylin. Histochemical and immunohistochemical assays were carried out in liver sections or smears of isolated cells, or both, as described in Table 1. In all immunochemical assays, nonspecific staining was established with nonimmune serum or a mouse monoclonal antibody to an unrelated antigen. Smears of isolated cells were also stained with Wright-Giemsa stain for the identification of blood-borne cells.

[%]Thymidine

Histoautoradiography

To establish the origin of the bile duct-like cells proliferating during bile duct obstruction or ANIT feeding, control (sham-operated) rats (n = 2) and rats obstructed for 6-24 h (n = 4) or fed ANIT for 12-72 h (n = 5) were injected intraperitoneally with 1 &i/g body wt of [3H]thymidine (sp act, 2 Ci/mmol; ICN Radiochemicals, Irvine, Calif.). Sixty minutes after the radioactivity injection, rats were killed with an overdose of pentobarbital. The liver was then removed, cut in small blocks, and fixed in buffered formalin. Several sections (4 bk thick) were cut from each block, coated with Kodak NTB-2 emulsion, exposed for 14-28 days, and develqped with Kodak Dektol developer for 2 min at 15°C. After fixing for 5 min with Kodafix, liver sections were washed and stained with hematoxylin and eosin by standard procedures. Reagents for carrying out these studies were all purchased from Eastman Kodak.

Electron

Microscopy

The ultrastructure of biliary epithelial cells was examined in the intact liver, and in the fraction (36-40 ml/min) of elutriated nonparenchymal liver cells most heavily enriched in cells with biliary epithelial phenotypes. Liver tissue was obtained from a control rat, a rat

ISOLATIONOF INTRAHEPATIC BILIARYEPITHELIALCELLS 1251

November 1989

obstructed for 14 days, and a rat kept on the ANIT diet for 28 days. Isolated cells were obtained from 3 separate but similarly treated animals. Small pieces of liver tissue were fixed overnight in 2.5% glutaraldehyde containing 2% paraformaldehyde, post fixed in osmium tetroxide (l%), and embedded in Spurr resin. Nonparenchymal liver cells were isolated and elutriated as already described, and processed for electron microscopic analysis as described for the liver tissue. After elutriation, cells were fixed for 30 min and subjected to low-speed centrifugation between steps. In all instances, ultrathin sections were cut with a LKB ultratome, stained with uranyl acetate and lead citrate, and observed with a Philip 400 electron microscope.

Results Hepatic Histology, Immunochemistry

Histochemistry,

and

Because biliary obstruction and ANIT intoxication are known to induce proliferation of bile duct-like structures (36,37), we have isolated and fractioned nonparenchymal liver cells from control rats and from rats obstructed for 14 days or fed ANIT for 28 days. During the obstruction and ANIT feeding periods, therefore, we have routinely examined hepatic histology and the histochemical or immunohistochemical expression of GGT, G6P, CK-7, and CK-19 to establish the nature of the proliferated bile duct-like cells (Table 1). YGlutamyltranspeptidase biochemical activity can be detected in most biological cells (42), but its histochemical expression in the liver of a normal, mature rat is restricted to the BDCs (7,43). Similarly, both the hepatocytes and BDCs contain various types of intermediate filament proteins (38,44), yet only the BDCs have CK-7 and CK-19 (44,45). On the other hand, G6P is present solely in the hepatocytes and can thus be used to identify parenchymal liver cells (46). In the control livers, hepatic morphology was normal and CK-7, CK-19, and GGT expressions were confined to the bile ductules/ducts (Figure 1). In both the obstructed and ANIT-fed rats, proliferation of bile duct-like structures was the most striking structural abnormality. The hyperplasia was time-related and was massive 14 days after induction of biliary obstruction and 28 days after the animals were fed the ANIT diet. At these respective times, no major morphologic changes in hepatic parenchyma were seen, yet moderate to diffuse tissue infiltration of inflammatory cells was present in both hyperplastic reactions. The proliferated bile duct-like cells were invariably positive for GGT, CK-7, and CK-19 (Figure l), and negative for G6P. Both in the control and hyperplastic livers, only the hepatocytes displayed positive staining for G6P.

Isolation and Fractionation of Nonparenchymal Liver Cells From Normal Rats Using the present technique, the number of nonparenchymal liver cells isolated from a control rat ranged from 2.1 X lOa to 2.9 X lo’, with an average value of 2.23 x 108. Of these, 3%-5% stained strongly for CK-7, CK-19, and GGT, so that the number of recovered BDCs ranged from 0.75 to 1.36 x lo6 per gram of liver (Table 2). Elutriation of the nonparenchymal liver cell population resulted in separating two major cell types, one recovered at low centripetal forces displaying primarily phenotypes of endothelial cells (positive for vimentin and vWF), the other at higher pump flow rates with characteristics of BDCs (positive for GGT and CK-19 and CK-7) (Table 2). The highest enrichment in cells positive for these latter markers was invariably obtained at pump flow rates ranging from 36 to 44 ml/min, regardless of whether the sequential changes in fluid rate were large (‘24 mllmin) or small (12 ml/min). With the sequence adopted routinely and described in Table 2, the highest enrichment (25O-55%) in BDCs was observed at a flow rate of 3640 mllmin (Figure 2). The number of cells recovered in this fraction ranged from 0.22 to 0.45 x 106/g liver, with an average diameter of 12.97 pm. In addition to BDCs, the fraction at 40 ml/min contained mainly (25%40%) mesenchymal cells (vimentin positive), of which vascular endothelial cells (vWF positive, 47) accounted for the major fraction. There were a few leukocytes (3%-8%) and phagocytic cells (endogenous peroxidase positive, 3%7%), and only a minimal (I%-3%) number of hepatocytes (G6P and albumin positive). No erythrocytes or AFP-positive cells were observed. When nonparenchymal liver cells were isolated from two normal rat livers and similarly elutriated, the yield of cells recovered in the 36-40 ml/min fraction was approximately doubled (0.35-0.75 x lo6 cells/g liver), yet the purity of BDCs (25%-60%) was virtually the same as that observed with one rat liver. In the elutriated fractions included in the flow rate range 14-40 mllmin, viability of nonparenchymal cells was always >97%. Isolation and Fractionation of Nonparenchymal Liver Cells From Rats With Bile Ductular-Ductal Cell Hyperplasia When nonparenchymal liver cells were isolated from rats obstructed for 14 days or fed ANIT for 28 days, after which extensive proliferation of bile duct structures was documented histologically, histochemically, and immunohistochemically (Figure

1252

ALPINI ET AL.

GASTROENTEROLOGY Vol. 97, No. 5

Figure

l), the percentage

of isolated cells positive for CK-7,

CK-19, and GGT was much greater than that recov-

ered from control rats (Table 3). Elutriation of the nonparenchymal liver cell population isolated from these hyperplastic livers again yielded fractions enriched in endothelial ceils (22-28 ml/min) and in cells with biliary epithelial phenotypes (36-44 ml/min). The fraction richest in cells positive for GGT, CK-7, and CK-19 was that obtained at the pump flow rate of 3640 mUmin, in which 60%85% of the cells displayed phenotypes of biliary epithelial cells (Figure 2 and Table 3). The number of cells recovered in this fraction averaged 16.23 x 10” and 14.07 x 10" for the obstructed and ANIT livers, respectively, values 4-6 times greater than those obtained with a normal liver (Table 3). Most of the cells contaminating the fraction enriched in bile duct-like cells were mesenchymal cells (mainly vWF

1. Immunohistochemical expression of CK-7 in normal rat liver (a] and in two rat livers in which proliferation of bile duct-like structures was induced by either a lb-day bile duct ligation (b) or by feeding ANIT for 28 days (c). In the bile duct-ligated rat (b), liver sections were stained also for CK-19 (d) and GGT (e). Two aspects of these findings need emphasis, First, CK-7 (and CK-19 and GGT as well, reactions not shown) is confined to the biliary structures in the normal rat liver. Second, the bile duct-like cells that proliferate in biliary obstruction and ANIT feeding are also positive for CK-7, CK-19, and GGT. Frozen sections, original magnification X25.

positive), and only a few endogenous peroxidasepositive cells and leukocytes were seen (Table 3). In all instances, cell viability in the 36-40 ml/min fraction ranged from 97% to 100% which was similar to or even higher than that observed before elutriation. Cytogenesis

of Hepatic

Proliferation

To establish the origin of the cells proliferating during biliary obstruction or ANIT feeding, we measured early hepatic incorporation of [3H]thymidine by histoautoradiography in control, obstructed, and ANIT-fed rats. In the control sham-operated animals, ~1% of parenchymal and various nonparenchymal liver cells, including biliary epithelial cells, had nuclei with incorporated thymidine. Twenty-four hours after bile duct ligation, however,

November

1989

ISOLATION OF INTRAHEPATIC BILIARY EPITHELIAL CELLS

1253

Figure 2. Isolated nonparenchymal rat liver cells from the 36-40 ml/mm elutriated fraction, stained for GGT. Cells were isolated from a normal liver [a) and from a liver with proliferated bile ductules and ducts induced by either a 14-day bile duct ligation (b) or ANIT feeding for 28 days (c). The percentage of positive cells is -50 in (a) and 70-85 in (b) and (c). Note that, as staining for GGT ranges from bright-red to light-orange, both black- and gray-stained cells are positive for this enzyme (biliary epithelial cells). Original magnification ~40.

5%10% of bile ductular cells present in ductules and ducts of various size had labeled nuclei (Figure 3A), whereas no change in labeling was observed in

1254

ALPINI ET AL.

GASTROENTEROLOGY Vol. 97, No. 5

Table 3. Characteristics of Nonparenchymal Cells Isolated From Normal and Hyperplastic Rat Livers

Before

and After Elutriation Control rats (n = 6-13)

Nonparenchymal cell pool Cells/liver (x 106) 223.9 + 45.1 Average cell 8.23 + 0.21 diameter (~1 Viability (%I” 97.8 + 1.5 4.1 + 0.2 GGT+ cells (96) CK-7+ cells (%] 4.8 5 0.5 3.7 2 0.2 CK-19+ cells (%] 8.4 + 0.9 CK-19+ cells/liver (X106) Elutriated fraction, 36-40 ml/min 2.84 C 0.45 Cells/liver (X 10’) Average celI 12.97 ? 0.32 diameter @II) 98.1 + 0.8 Viability (I)” 2.2 C 0.6 G6P+ cells (%) 15.3 ? 5.1 vWF+ cells (96) Vimentinf cells (%) 35.2 ? 6.9 3.2 2 1.3 LCA+ cells (%) 4.3 + 2.8 PX+ cells (%) 41.9 ?z 10.3 GGT+ cells (%) 43.0 & 11.2 CK-7+ cells (%) 42.7 + 9.7 CK-19+ cells (%] CK-19+ cells (x10’) 1.22 + 0.23

Obstructed rats (n = 6-12)

ANIT Rata (n = 7-13)

491.7 + 57.9 384.8 + 85.5 9.84 + 0.42 8.53 2 0.21 97.1 28.7 25.9 27.1 132.8

* + + ? 2

1.1 97.4 2.5 27.5 4.7 24.9 3.1 28.9 15.5 103.4

f f ? * f

1.8 3.1 4.2 3.9 18.5

16.23 * 2.84 14.07 + 2.05 12.86 + 0.32 11.32 + Cl.61 98.6 1.3 8.8 15.3 3.9 2.8 73.8 72.4 70.7 11.90

+ r * * + + + t + +

0.9 97.4 0.7 0.9 4.1 9.5 7.1 13.3 0.4 3.7 0.5 3.2 8.9 76.3 9.8 70.1 10.2 72.2 2.02 10.15

+ + ? * * * 2 2 * ?

1.1 0.6 3.7 5.8 0.6 0.6 9.1 9.9 9.3 1.81

CK, cytokeratin; GGT, -y-glutamyl transpeptidase; G8P, glucose6-phosphatase; LCA, leukocyte common antigen; PX, endogenous von Willebrand factor; ANIT, aperoxidase; vWF, naphthylisothiocyanate. Values are mean + SD (n = number of animals) and refer to the cells isolated before elutriation (nonparenchymal cell pool) and to the elutriated fraction (36-M mumin] richest in cells with biliary epithelial phenotypes. The liver weights in control, obstructed, and ANIT rats (rats fed ANIT] were obtained in separate groups of animals of similar body weights and averaged 8.36, 13.78, and 12.65 g, respectively. ’ Cell viability was determined by trypan blue exclusion and refers only to nonparenchymal cells.

other types of liver cells. Similarly, 48 h after the animals were placed on the ANIT diet (the earliest time at which increased deoxyribonucleic acid synthesis was observed) the biliary epithelial cells (2%5%) contained grains in their nuclei (Figure 3B). These results indicate that in both hyperplastic reactions, the proliferating cells originate from preexisting bile ductular-ductal epithelial cells. Ultrastructure

of Biliary Epithelial

Cells

To further establish the characteristics of biliary epithelial cells after isolation and purification, particularly when obtained from bile duct-ligated or ANIT-fed rats, the ultrastructure of these cells was studied both in the intact liver and in elutriated

fractions (36-40 ml/min) of nonparenchymal liver cells. In the normal rat liver, only one or two bile ductules were seen in the portal tracts. The epithelium of the ductules was arranged around a lumen and displayed a round to oval nucleus with serrated edges. Whereas the apical portion of the cytoplasm was devoid of organelles and displayed microvilli on the surface, scattered organelles were observed around the nucleus (Figure 4A). The junctional complex, including desmosomes, tight junction, and interdigitation of adjoining cell lateral membranes, was prominent. Numerous bile ductules, frequently without a well-defined lumen, were observed in the portal tracts of the rat obstructed for 14 days and of the rat fed ANIT for 28 days. Morphologically, these ductules were similar to those observed in the normal rat, except that an increase in number of organelles (especially mitochondria, Golgi, and rough enobserved in the doplasmic reticulum) was proliferated liver. Electron microscopy of nonparenchymal liver cells isolated after a 14-day biliary obstruction or a 28-day ANIT feeding revealed that, by far, most of the cells present in the elutriated 38-40 ml/min fraction had ultrastructural characteristics of biliary epithelial cells. Liver endothelial cells were the major contaminants and only a few macrophages, lymphocytes, and neutrophils were seen in such a fraction. Hepatocytes were rare. When parenchymal liver cells were isolated from a normal rat liver, the contaminating cells were more numerous on a percentage basis, yet the biliary epithelial cells accounted again for the largest population (-50%) present in such an elutriated fraction. Ductular cells were in most instances completely dissociated and displayed numerous microvilli on their surface. When in close approximation, the cells displayed a prominent junctional complex and a periluminal arrangement (Figure 4B). The morphologic appearance of biliary epithelial cells isolated from the control, obstructed, and ANIT-fed rats was virtually the same as that observed in the respective tissues except that, in the isolated state, the cytoplasm of bile ductular epithelial cells contained vesicles of various sizes (Figure 4). These vesicular structures have already been found in isolated biliary epithelial cells (30) and are most likely the result of anoxia caused by the isolation or purification procedures, or both.

Discussion The isolation and purification of BDCs entails three major technical problems. First of all, BDCs are surrounded by bundles of connective tissue and, because of their collecting nature, are tightly sealed

November

Figure

1989

ISOLATION OF INTRAHEPATIC BILIARY EPITHELIAL CELLS

1255

3. Histoautoradiographs of liver sections from a bile duct-ligated rat (A) and from a rat fed ANIT (B). Animals were injected intraperitoneally with -1 &i/g [3H]thymidine 24 h after bile duct ligation or 48 h after being placed on the ANIT diet, and killed 90-120 min after the injection. Note that in both livers, the bile duct cells were labeled (arrows). Hematoxylin and eosin, original magnification X 250.

to each other. These characteristics make it difficult to isolate BDCs by a standard collagenase procedure. Second, BDCs are similar in size to other nonparenchymal liver cells and are outnumbered by the latter. As a result, purification methods that exploit differences in cell size are bound to provide relatively impure fractions. Finally, the biology of BDCs is poorly understood so that their identification in the isolated condition must be based on cellular markers, the validity of which has as yet to be conclusively established. In the present studies, we have isolated BDCs by

collagenase-pronase digestion of the biliary and connective tissue, which remains undissociated after collagenase perfusion of the liver. This has been the procedure of choice in previous reports (26-32) and, even though pronase has at times been omitted (25,26,29,33) or replaced by ethyleneglycol-bis(p aminoethylether)-N,N’-tetraacetic acid or trypsin, or both (30,32), we have not encountered sufficient difficulties with such a technique as to warrant any modification. As to the purification of BDCs, we have also used a standard procedure, namely, centrifugal elutriation of the nonparenchymal liver cell isolate.

1256

ALPINIETAL.

GASTROENTEROLOGYVol. 97,No.5

Figure 4. Electron microscopy of biliary spithelial cells in a normal rat liver (A) and after isolation and purification (36-40 ml/min elutriated fraction) from a rat obstructed for 14 days (B). A. Two biliary epithelial cells are seen around a lumen (L) showing microvilli (small arrowhead), nuclei (N) with serrated borders, small mitochondria (M), sparse rough endoplasic reticulum (R), prominent tight junctions (large arrowheads) and desmosomes (arrow), and abundant cytoskeleton filaments (thin arrow). B. Two isolated biliary epithelial cells in close approximation, displaying the same structural characteristics as in A, including large nucleus, scanty organelles, microvilli, and junctional complex. Vesicles (V) of varying sizes were seen in the cytoplasm of biliary epithelial cells after isolation and elutriation, but not in the intact liver. As in the intact tissue (A), note the presence of desmosomes (arrow within rectangular area) shown at higher power in the inset (upper left corner). Sections stained with uranyl acetate and lead citrate (X 24,000). Bars represent 1 pm.

Previous studies dealing with BDC purification have utilized various methods, including counterflow elutriation (28,48), centrifugation on Percoll(29-31,33), Ficoll (25),or metrizamide (25-27)gradients, and

free-flow electrophoresis in combination with centrifugation on Percoll gradients (49). All of these procedures, however, exploit virtually the same principle for cell separation, that is, differences in

November

1989

ISOLATION

cell sedimentation velocity. Thus, as a centrifugal elutriator is available in our laboratory, we have adopted this procedure to fractionate the nonparenchymal cell population. However, we have devoted a major effort to identifying the various types of cells present in the isolate and in the elutriated fractions. This is because previous reports have provided somewhat conflicting data with respect to BDC purity, and the failure to establish reliable techniques for unambiguous BDC recognition may have contributed to the disparate results. Furthermore, no rigorous identification of the cells contaminating the BDC-enriched fraction has previously been attempted. Our experience, here and elsewhere (7), as well as that of others (2629,33,49), strongly suggests that GGT cytochemistry is a valid approach to establish BDC purity as, in the intact rat liver, this enzyme is histochemically localized solely on the biliary epithelial structures (7,42,43). It has been reported, however, that hepatocytes (50,51), leukocytes and platelets (52), and even endothelial cells (53) may stain for GGT under certain conditions. Thus, to substantiate the GGT results, we have studied the immunocytochemical expression of CK-19 and CK-7, which have been shown previously (44,45,54) and in this study as well to be confined to the biliary ductules and ducts. Consistent with the tissue results, CK-19 and CK-7 immunocytochemistry yielded approximately the same BDC purity as that estimated with GGT cytochemistry. Naturally, the restriction of GGT and CK-19 and CK-7 to the intact biliary structures does not prove their validity for the identification of BDCs in the isolated state. During their isolation or purification, or both, BDCs may undergo structural changes that could compromise the expression of these markers. However, this is very unlikely and, even if such changes occurred, they would result solely in underestimating BDC purity. Thus, in absence of an indisputable marker of BDCs in the isolated condition, CK-19 and CK-7 immunocytochemistry and GGT cytochemistry appear to be the methodologies of choice for in vitro identification of BDCs. Based on the CK-19, CK-7, and GGT data, our experiments indicate that, by centrifugal elutriation, a fraction of nonparenchymal liver cells containing 25%-55% of BDCs can be isolated from a normal rat liver. At these purities, the yield ranges from 0.23 to 0.45 x lo6 cells/g liver. These results are similar to those reported previously by some other groups who have claimed a BDC enrichment ranging from 20% to 67% at a yield of 1.5-3.8 X lo6 cells/liver (25,27,28, 48). On the other hand, our BDC enrichment is more modest than that described in some recent reports in which, using a normal rat liver, a 68%-87% BDC

OF INTRAHEPATIC

BILIARY

EPITHELIAL

CELLS

1257

purity has been claimed (30,31,33,55). As the procedures for the isolation and purification of BDCs used in these previous studies (30,31,33,55) are somewhat different from those adopted here, it is possible that technical differences underlie the divergent results. In one of these reports, however, BDC purity was established primarily on morphologic criteria (31). In others, GGT cytochemistry (33) or immunofluorescence staining for prekeratin (30,31,55), or both, was carried out, yet the claimed BDC enrichment was not objectively documented. The modest enrichment and relatively low yield of isolated BDCs obtained with a normal rat liver cast some doubt on the suitability of this preparation for in vitro studies of BDC function. Thus, we have also isolated nonparenchymal liver cells from rats obstructed for 14 days or fed ANIT for 28 days, in which extensive proliferation of bile duct-like cells was induced. Counterflow elutriation of the cells obtained from these hyperplastic livers resulted not only in improving the enrichment in cells with biliary epithelial phenotypes, but also the yield. For instance, in the 36-40 ml/min fraction, which was the richest in bile duct-like cells, 60%-85% of the nonparenchymal cells isolated from obstructed and ANIT livers were positive for CK-19, CK-7, and GGT. And, the cell yield in such a fraction was 4-6 times greater than that recovered from a normal liver. These findings are consistent with those previously reported by Grant and Billing (25), Sirica and Cihla (29), and Ledda et al. (27), who also obtained much purer preparations of bile duct-like cells when isolated from obstructed or ANIT-fed rats. As both the yield and the enrichment of bile duct-like cells isolated from obstructed or ANIT-fed rats would be rather satisfactory for the conduct of in vitro studies of BDC biology, our results raise the fundamental question as to whether the cells isolated from these hyperplastic livers are suitable for this purpose. Some reservation about their applicability arises from the observations that (a) metaplasia of hepatocytes into ductular structures has been and documented under certain conditions (5658), (b) the initial stage of chemical hepatocarcinogenesis is associated with proliferation of a cell type (“oval” cell) that expresses phenotypes of both BDCs and immature or transformed hepatocytes (18-20,28,59, 60). Both in the obstructed and ANIT-fed rats, however, the hyperplastic cells were positive for GGT and, unlike oval cells, were negative for G6P, albumin, and cY-fetoprotein. Furthermore, Yaswen et al. (61) have reported that the bile duct-like cells proliferating during chronic biliary obstruction display genetic characteristics of biliary epithelial cells. We have recently shown that in both biliary obstruction (7,62) and ANIT feeding (62), the hyperplastic

1258

ALPINI ET AL.

ductules and ducts secrete a bicarbonate-rich fluid both under spontaneous conditions and during secretin-induced choleresis, and this is a secretory property of normal biliary epithelial cells (5,7). Finally, we have demonstrated here that the proliferated cells originate from preexisting biliary epithelial cells and retain their morphology and other phenotypic traits. For instance, they are positive for CK-19 and CK-7, and there is strong evidence that these intermediate filament proteins are highly conserved during cell differentiation and transformation (63,64). Our immunocytochemical findings need to be stressed as they have shown for the first time that, in the rat, both the normal and hyperplastic BDCs express CK-19 and CK-7. Altogether, therefore, these observations provide persuasive evidence that the proliferated bile duct-like cells associated with either biliary obstruction or ANIT feeding are the progeny of intrahepatic BDCs and retain their characteristics. Hence, both hyperplastic models lend themselves to increasing the number of intrahepatic BDCs so as to allow their isolation at a relatively high yield and purity. However, whether the isolated BDCs obtained from either a normal or a proliferated rat liver are suitable for in vitro studies of BDC function is not entirely known at this time. The present finding that the cells contaminating the elutriated fraction most enriched in BDCs consisted primarily of mesenchymal cells and included only 0%-Z% hepatocytes, together with the preliminary observation that BDCs in short-term culture display endocytotic activity (65,66), suggests their potential applicability to studies of BDC transport function. On the other hand, it has been demonstrated here and elsewhere (67) that BDCs in vivo have very low, if any, physiologic replicative activity, so that the BDCs isolated under proliferative conditions should be applied with caution to the study of BDC growth and its regulation. In conclusion, the present study has demonstrated that, by centrifugal elutriation, a fraction of nonparenchymal cells (1.8-3.8 X lO’/liver) containing up to 55% BDCs can be obtained from a normal rat liver. If the nonparenchymal cells are isolated from a rat liver rendered hyperplastic by biliary obstruction or ANIT feeding, the yield of the bile duct-like cells ranges from 12 to 19 x lo6cells/liver and the purity can be as high as 85%. As the hyperplasia associated with either bile duct ligation or ANIT intoxication results in a population of cells that originate from and retain the characteristics of biliary epithelial cells, our results strongly suggest that both models of hyperplasia lend themselves to the isolation of BDCs at a yield and purity satisfactory for in vitro studies of BDC physiology and pathophysiology.

GASTROENTEROLOGY Vol. 97, No. 5

References 1. Steiner JW, Carruthers JS. Studies on the fine structure of the terminal branches of the biliary tree. Am J Path01 1961; 38:639-61. 2. Schaffner F, Popper H. Electron microscopic studies of normal and proliferated bile ductules. Am J Path01 1961;38: 393-410. 3. Grisham JW, Hartroft WS. Morphologic identification by electron microscopy of oval cells in experimental degeneration. Lab Invest 1961;10:317-32. 4. Sternlieb I, Quintana N. Biliary proteins and ductular ultrastructure. Hepatology 1985;5:139-43. 5. Wheeler HO, Mancusi-Ungaro PL. Role of bile ducts during secretin choleresis in dogs. Am J Physiol 1966;210:1153-9. 6. Forker EL. Two sites of bile formation as determined by mannitol and erythritol clearance in the guinea pig. J Clin Invest 1967;46:1189-95. 7. Alpini G, Lenzi R, Sarkozi L, Tavoloni N. Biliary physiology in rats with bile ductular cell hyperplasia. Evidence for a secretory function of proliferated bile ductules. J Clin Invest 1988;81:569-78. 8. Wheeler HO, Ramos OL. Determinants of the flow and composition of bile in the unanesthetized dog during constant infusion of sodium taurocholate. J Clin Invest 1960;39:16170. 9. Tavoloni N. Role of ductular bile water reabsorption in canine bile secretion. J Lab Clin Med 1985;106:154-61. 10. LaRusso NF. Proteins in bile: how they get there and what they do. Am J Physiol 1984;247:G199-205. 11. Tomana M, Kulhavy R, Mestecky J. Receptor-mediated binding and uptake of immunoglobulin A by human liver. Gastroenterology 1988;94:762-70. 12. Kaplan MM. Primary biliary cirrhosis. N Engl J Med 1987; 316:521-8. 13. Nakanuma Y, Hirai N, Kono N, Ohta G. Histological and ultrastructural examination of the intrahepatic biliary tree in primary sclerosing cholangitis. Liver 1986;6:317-25. 14. Miglio F, Pignatelli M, Mazzeo V, et al. Expression of major histocompatibility complex class II antigens on bile duct epithelium in patients with hepatic graft-versus-host disease after bone marrow transplantation. J Hepatol 1987;5:182-9. 15. Sale GE, Storb R, Kolb H. Histopathology of hepatic acute graft versus host disease in the dog: a double blind study confirms the specificity of small bile duct lesions. Transplantation 1978;26:103-6. 16. Vierling JM, Fennel1 RI-I. Histopathology of early and late human hepatic allograft rejection: evidence of progressive destruction of interlobular bile ducts. Hepatology 1985;5: 1076-82. 17. Ludwig J, Wiesner RH, Batts KP, Perkins JD, Krom RAF. The acute vanishing bile duct syndrome (acute irreversible rejection] after orthotopic liver transplantation. Hepatology 1987; 7:476-83. 18. Braun L, Goyette M, Yaswen P, Thompson NL, Faust0 N. Growth in culture and tumorigenicity after transfection with the ras oncogene of liver epithelial cells from carcinogentreated rats. Cancer Res 1987;47:4116-24. 19. Germain L, Noel M, Gourdeau H, Marceau N. Promotion of growth and differentiation of rat ductular oval cells in primary culture. Cancer Res 1988;48:368-78. 20. Sell S, Leffert HL. An evaluation of cellular lineages in the pathogenesis of experimental hepatocellular carcinoma. Hepatology 1982;2:77-86. 21. Tavoloni N. Biliary clearance of inert carbohydrates. Expectations and reality. Gastroenterology 1988;94:217-28.

November

1989

ISOLATION OF INTRAHEPATIC BILIARY EPITHELIAL CELLS

J. Secretory function of the rabbit common 22. Chenderovitch bile duct. Am J Physiol 1972;223:695-706. 23. Strasberg SM, Petrunka CN, Ilson RG. The contribution of the extrahepatic bile ducts to bile formation. Can J Physiol Pharmacol 1976;54:757-63. 24. Tavoloni N, Wyssbrod HR, Jones MJT. Permeability characteristics of the guinea pig biliary apparatus. Hepatology 1986; 6:1369-81. 25. Grant AG, Billing BH. The isolation and characterization of a bile ductule cell population from normal and bile-duct ligated rat livers. Br J Exp Path01 1977;58:301-10. 26. Sells MA, Katyal SL, Shinozuka H, Estes LW, Sell S, Lombardi B. Isolation of oval cells and transitional cells from the livers of rats fed the carcinogen nL-ethionine. JNCI 1981; 66:355-62. 27. Ledda GM, Sells MA, Yokoyama S, Lombardi B. Metabolic properties of isolated rat liver cell preparations enriched in epithelial cells other than hepatocytes. Int J Cancer 1983; 31:231-7. 28. Yaswen P, Hayner NT, Faust0 N. Isolation of oval cells by centrifugal elutriation and comparison with other cell types purified from normal and preneoplastic livers. Cancer Res 1984;44:324-31. 29. Sirica AE, Cihla HP. Isolation and partial characterization of oval and hyperplastic bile ductular cell-enriched populations from the livers of carcinogen and noncarcinogen-treated rats. Cancer Res 1984;44:3454-66. 30. Gall JAM, Bhathal PS. The isolation of intrahepatic biliary epithelial cells from normal rat livers. Cell Biol Int Rep 1985;9:315-22. 31. Kumar V, Jordan TW. Isolation and culture of biliary epitheha1 cells from the biliary tract fraction of normal rats. Liver 1986;6:369-78. 32. Gall JAM, Bhathal P.S. Isolation and culture of intrahepatic bile ducts and its application in assessing putative inducers of biliary epithelial cell hyperplasia. Br J Exp Path01 1987; 68:501-10. 33. Kenick S, Lowry RP, Forbes RDC. Sterile isolation of rat hepatic parenchymal and bile ductular cells. Transplant Proc 1987;19:3061-3. 34. Sirica AE, Sattler CA, Cihla HP. Characterization of a primary bile ductular cell culture from the livers of rats during extrahepatic cholestasis. Am J Path01 1985;120:67-78. 35. Schier F, Schier C. Cell cultures of bile duct epithelium and the pathogenesis of biliary atresia. Eur J Pediatr 1987;146: 27-30. 36. Steiner JW, Carruthers JS, Kalifat SR. The ductular cell reaction of rat liver in extrahepatic cholestasis. I. Proliferated biliary epithelial cells. Exp Mol Path01 1962;1:162-85. 37. Lopez M, Mazzanti L. Experimental investigations on anaphthylisothiocyanate as a hyperplastic agent of the biliary ducts in the rat. J Path01 Bacterial 1955;69:243-50. 38. Moll R, Franke WW, Schiller DL. The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors, and cultured cells. Cell 1982;31:11-24. 39. Rutenberg AM, Kim H, Fishbein JW, Hanker JS, Wasserkrug HL, Seligman AM. Histochemical and ultrastructural demonstration of r_glutamyltranspeptidase activity. J Histochem Cytochem 1969;17:517-26. 40. Teutsch HF. Improved method for the histochemical demonstration of glucose-6-phosphatase activity. A methodological study. Histochemistry 1978;57:107-17. 41. Straus W. Imidazole increases the sensitivity of the cytochemical reaction for peroxidase with diaminobenzidine at a neutral pH. J Histochem Cytochem 1982;30:491-3. 42. Tate SS, Meister A. y_Glutamyltranspeptidase: catalytic,

43.

44.

45.

46

47

48.

49.

50.

51.

1259

structural, and functional aspects. Mol Cell Biochem 1981; 39:357-68. Ronchi G, Desmet J. Histochemical study of rglutamyl transpeptidase in experimental intrahepatic and extrahepatic cholestasis. Beitr Path01 1973;150:316-21. Franke WW, Denk H, Kalt R, Schmid E. Biochemical and immunological identification of cytokeratin proteins in hepatocytes of mammalian liver tissue. Exp Cell Res 1981;131: 299-318. Van Eyken P, Sciot R, Von Damme B, De Wolf-Peeters C, Desmet VJ. Keratin immunochemistry in normal human liver. Cytokeratin pattern of hepatocytes, bile ducts and acinar gradient. Virchows Arch [A] 1987;412:63-72. Chiquoine AD. The distribution of glucose-6-phosphatase in the liver and kidney of the mouse. J Histochem Cytochem 1953;1:429-35. Lenzi R, Alpini G, Rand JH, Tavoloni N. Immunolocalization of von Willebrand factor in rat liver vascular endothelial cells (abstr). Hepatology 1986:6:1191. Ishii M, Vroman BT, Kost LJ, La Russo NF. Isolation and morphological characterization of bile duct epithelial cells from rat liver (abstr). J Cell Biol 1987;105:161A\. Miller SB, Saccomani G, Pretlow TP, et al. Purification of cells from livers of carcinogen-treated rats by free-flow electrophoresis. Cancer Res 1983;43:4176-9. Kitagawa T, Imai F, Sato K. Re-evaluation of y-glutamyl transpeptidase activity in periportal hepatocytes of rats with age. Jpn J Cancer Res 1980;71:362-6. Russell JJ, Staffeldt EF, Wright BJ, Prapuolenis A, Carnes BA, Peraino C. Effect of rat strain, diet composition, and phenobarbital on hepatic y-glutamyl transpeptidase histochemistry and on the induction of altered hepatocyte foci and hepatic tumors by diethylnitrosamine. Cancer Res 1987;47:

1130-4. 52. Khalaf MR. Hayhoe

FGJ. Cytochemistry of y-glutamyltransEerase in haemic cells and malignancies. Histochem J 1987; 19:385-95. 53. Dropulic B, Masters CL. Culture of mouse brain capillary endothelial cell livers that express factor VIII, y-glutamyltranspeptidase, and form junctional complexes in vitro. In Vitro Cell Dev Biol 1987;23:775-81. 54. Ramaekers F, Huysmans A, Schaart G, Moesker 0, Vooijs P. Tissue distribution of keratin 7 as monitored by a monoclonal antibody. Exp Cell Res 1987;170:235-49. 55. Parola M, Cheeseman KH, Biocca ME, Dianzani MU, Slater TF. Isolation and characterization of biliary epithelial cells from normal rat liver. J Hepatol 1988;6:175-86. 56. Nakanuma Y, Ohta G. Immunohistochemical study of bile ductular proliferation in various hepatobiliary diseases. Liver 1986;6:205-11. 57. Vanstapel MJ, Gatter KC, DeWolf-Peeters C, Mason DY, Desmet VJ. New sites of human S-100 immunoreactivity detected with monoclonal antibodies. Am J Clin Path01 1986;85:160-8. 58. Butron Vila M, Haot J, Desmet VJ. Cholestatic features in focal nodular hyperplasia of the liver. Liver 1984;4:387-95. 59. Plenat F, Brawn L, Faust0 N. Demonstration of glucose6-phosphatase and peroxisomal catalase activity by ultrastructural cytochemistry in oval cells from livers of carcinogen-treated rats. Am J Path01 1988;130:91-102, 60 Petropoulos CJ, Yaswen P, Panzica M, Faust0 N. Cell lineages in liver carcinogenesis: possible clues from studies of the distribution of a-fetoprotein RNA sequences in cell populations isolated from normal, regenerating, and preneoplastic rat livers. Cancer Res 1985;45:5762-8. 61 Yaswen P, Goyette M, Shank PR, Faust0 N. Expression of c-Ki-ras, c-Ha-ras, and c-myc in specific cell types during hepatocarcinogenesis. Mol Cell Biol 1985;5:780-6.

1260

ALPINI ET AL.

62. Alpini G, Lenzi R, Zhai WR, et al. Bile secretory function of intrahepatic biliary epithelium in the rat. Am J Physiol 1989;257:G124-33. 63. Osborn M, Weber K. Intermediate filaments: cell-type-specific markers in differentiation and pathology. Ceil 1982;31:303-6. 64. Cooper D, Schenner A, Sun TT. Clarification of human epithelia and their neoplasms using monoclonal antibodies to keratins: strategies, applications, and limitations. Lab Invest 1985;52:243-56. 65. Ishii M, Vroman BT, La Russo NF. Morphologic evidence of receptor-mediated endocytosis of epidermal growth factor by isolated bile duct epithelial cells (abstr). Gastroenterology 1988;94:A550. 66. Ishii M, Vroman B, La Russo NF. Fluid-phase endocytosis by isolated intrahepatic bile duct epithelial cells (abstr). Hepatology 1988;8:1251. 67. Wu PC, Lily MA, Gibson JB, Hirai H, Tsukada Y. Serum alpha-fetoprotein in rats after ligation of the common bile duct: relation to ductular cell (oval cell] proliferation. J Path01 1981;133:61-74. 68. Wisse E. Observations on the fine structure and peroxidase cytochemistry of normal rat liver Kupffer cells. J Ultrastruct Mol Struct Res 1974;46:393-426. stain using benzi69. Kaplow LS. Simplified myeloperoxidase dine dihydrochloride. Blood 1965;26:215-9. 70. Sunderland CA, McMaster WR, Williams AF. Purification with monoclonal antibody of a predominant leukocyte-

GASTROENTEROLOGY Vol. 97, No. 5

common antigen and glycoprotein from rat thymocytes. Eur J Immunol 1979;9:155-9. localization 71. Shroyer KR, Nakane PK. Immunohistochemical of albumin and in situ hybridization of albumin mRNA. Cell Biochem Funct 1987;5:195-210. 72. Gitlin D. Normal biology of a-fetoprotein. Ann NY Acad Sci 1975;259:7-16. 73. Nagle RB. Intermediate filaments: a review of the basic biology. Am J Surg Path01 1988;12(Suppl 1):4-16.

Received May 5, 1988. Accepted March 27, 1989. Address requests for reprints to: N. Tavoloni, Ph.D., Division of Hematology, Atran Building, Box 1679, Mount Sinai Medical Center, 190th Street and Madison Avenue, New York, New York 10029. This work was supported by grant HD 17556 from the National Institute of Child Health and Human Development. Portions of this study were presented at the Digestive Disease Week meeting held in San Francisco, California on May 18-21, 1986, and at the American Association for the Study of Liver Diseases meeting held in Chicago, Illinois on November 4-5, 1986, and have been published in abstract form (Gastroenterology 1986;90:1707, Hepatology 1986;6:1192). The authors thank Mary Barrett for her invaluable assistance in preparing the manuscript.