Developments in molecular and cellular biochemistry
Najjar, Victor A., ed.: Biological Effects of Glutamic Acid and Its Derivatives, 1981. ISBN 90-6193-841-4
Najjar, Victor A., ed.: Immunologically Active Peptides, 1981. ISBN 90-6193-842-2
Enzyme Induction and Modulation
edited by
v.A. NAJJAR Division of Protein Chemistry Tufts University School of Medicine Boston, Massachusetts, U.S.A.
Reprinted from Molecular and Cellular Biochemistry Volumes 53/54, 1983
1983 MARTINUS NIJHOFF PUBLISHERS a member of the KLUWER ACADEMIC PUBLISHERS GROUP BOSTON / THE HAGUE / DORDRECHT / LANCASTER
Distributors
for the United Stares and Canada: Kluwer Boston, Inc., 190 Old Derby Street, Hingham, MA 02043. USA for aff other countries: Kluwer Academic Publishers Group. Distribution Center, P.O. Box 322, 3300 AH Dordrecht, The Netherlands
Library or CongrtSS Cltlloging in Publication Dltl
LI.,,,, .f CHI"" (. ,.1,,1., I. P.bll"H • • D." Mooi= .=t ry ""d.r title;
(Ile ... l<>p .. nto Ie 1I01uular ""d ~dlul.r blcehu.1.try) Orl,laailT pulol1lb*<l. ..... S3 .. 5. or Nool.eular ..,d
.dlular blochnhtq. 1. taQ"U l_eU"" .... M~lIeo. eo .. ¥I. leetu:-...
2. Llvu e.ll ... _A4~ ........ ..,. •• lee"'...... 3. eUI .l:.l.lur ..... Add. ........ . . """" l.el........ 1. lo.JJI.l" . nel or". II. !!o1«lIIar Uld e.lllll~ hloc.bea1.tq. Itt. Se r in. W'6ol.!'517 1983 m '.Ol925 B:HI13t1
ISBN-13: 978- 1-4613-3881-9 001 : 10. 1007/978- 1-461 3-3879-6
Copyright
e-ISBN- 13: 978- 1-46 13-3879-6
© 1983 by Martinus Nijhorr Publishers, Boston. Soflcover reprint of the hardcover lSI edition 1983 All rights reserved. No part of this publication may be reproduced. stored in a retrieval system, or transmitted in any rorm or by any means. mechanical, photocopying, recording. or otherwise. without the prior written permission of the publishers, Martinus Nijhorr Publishers, 190 Old Derby Street, Hingham. MA 02043. USA.
Contents
Part I 9 T. D. Gelehrter, P. A. Barouski-Miller, P. L. Coleman and B. J. Cwikel: Hormonal regulation of
plasminogen activator in rat hepatoma cells II J. W. Grisham: Cell types in rat liver cultures: their identification and isolation 23 C. Guguen-Guillouzo and A. Guillouzo: Modulation offunctional activities in cultured rat hepatocytes 35 D. F. Haggerty, E. B. Spector, M. Lynch, R. Kern, L. B. Frank and S. D. Cederbaum: Regulation of
expression of genes for enzymes of the mammalian urea cycle in permanent cell-culture lines of hepatic and non-hepatic origin 57
R. Barouki, M.-N. Chobert, J. Finidori, M.-C. Billon and J. Hanoune: The hormonal induction of gamma glutamyltransferase in rat liver and in a hepatoma cell line 77
L. J. Crane and D. L. Miller: Plasma protein induction by isolated hepatocytes 89
Part II III D. K. Granner and J. L. Hargrove: Regulation of the synthesis of tyrosine aminotransferase: the
relationship to mRNATAT 113 J. M. Masserano and N. Weiner: Tyrosine hydroxylase regulation in the central nervous system 129
Part III 153 A. J. Fulco, B. H. Kim, R. S. Matson, L. O. Narhi and R. T. Ruettinger: Nonsubstrate induction of a
soluble bacterial cytochrome P-450 monooxygenase by phenobarbital and its analogs 155 G. Kikuchi and T. Yoshida: Function and induction of the microsomal heme oxygenase 163
Part IV 185 W. L. Miller, D. C. Alexander, J. C. Wu, E. S. Huang, G. K. Whitfield and S. H. Hall: Regulation of
,B-chain mRN A of ovine follicle-stimulating hormone by 17,B-estradiol 187 W. E. G. Muller, A. Bernd and H. C. Schroder: Modulation of poly (A)( +)mRN A-metabolizing and
transporting systems under special consideration of microtubule protein and actin 197 P. H. Pekala and J. Moss: 3T3-Ll preadipocyte differentiation and poly(ADP-ribose) synthetase 221 P. K. Sarkar and S. Chaudhury: Messenger RN A for glutamine synthetase 233
Part V 245 R. Sasaki and H. Chiba: Role and induction of 2,3-bisphosphoglycerate synthase 247 J. G. Cory and A. Sato: Regulation of ribonucleotide reductase activity in mammalian cells 257 M. R. Waterman and R. W. Estabrook: The induction of microsomal electron transport enzymes 267 V. Rubio, H. G. Britton and S. Grisolia: Activation of carbamoyl phosphate synthetase by
cryoprotectants 279
6
G. C. Ness: Regulation of 3-hydroxy-3-methylglutaryl coenzyme A reductase 299 J. W. Porter and T. L. Swenson: Induction of fatty acid synthetase and acetyl-CoA carboxylase by
isolated rat liver cells 307
Preface
In addition to performing its prime function as a vehicle for scientific communications of varied colora­ tions, Molecular and Cellular Biochemistry is again focusing on two subjects which it treats in depth. One of these is a book issue dealing with the transglutaminase reaction. The other is this issue that deals with induction and modulation of enzymes. This is a very broad subject that calls for broader coverage than could be included in one book issue. However, I have elected to include only certain contributions that serve as general examples of the principles involved.
There are six articles on enzyme regulation in hepatocyte culture. These include arginase and argino-succi­ nate synthetase, y-glutamyl transferase and plasminogen activitor. Other regulatory enzymes that are discussed are protein kinases, 2,3-bisphosphoglycerate synthetases, carbamoyl phosphate synthetase, heme oxygenase, cytochrome P-450, tyrosine hydroxylase, fatty acid synthetase, acetyl eoA carboxylase, among others. Also included is the regulation of several enzyme messengers RNAs.
As in the past, subscribers to the journal will receive this double volume as a continuation of regular issues of the journal Molecular and Cellular Biochemistry. This double volume will also be available separately in the form of a book issue to interested readers through Martinus Nijhoff Publishers.
v.A. Najjar
Part I
Molecular and Cellular Biochemistry 53;54,11-21 (1983). © 1983, Martinus 1\ ij hoff Publishers. Boston. Printed in The Netherlands.
Hormonal regulation of plasminogen activator in rat hepatoma cells
Thomas D. Gelehrter, Patricia A. Barouski-Miller, Patrick L. Coleman and Bernard J. Cwikel Departments of Internal Medicine and Human Genetics, University of Michigan Medical School, Ann Arbor, MI48109, U.S.A,
Summary
Plasminogen activators are membrane-associated, arginine-specific serine proteases which convert the inactive plasma zymogen plasminogen to plasmin, an active, broad-spectrum serine protease. Plasmin, the major fibrinolytic enzyme in blood, also participates in a number of physiologic functions involving protein processing and tissue remodelling, and may play an important role in tumor invasion and metastasis. In HTC rat hepatoma cells in tissue culture, glucocorticoids rapidly decrease plasminogen activator (P A) activity. We have shown that this decrease is mediated by induction of a soluble inhibitor of P A activity rather than modulation of the amount of PA. The hormonally-induced inhibitor is a cellular product which specifically inhibits PA but not plasmin. We have isolated variant lines of HTC cells which are selectively resistant to the glucocorticoid inhibition ofP A but retain other glucocorticoid responses. These variants lack the hormonal­ ly-induced inhibitor; P A from these variants is fully sensitive to inhibition by inhibitor from steroid-treated wild-type cells. Cyclic nucleotides dramatically stimulate P A activity in HTC cells in a time- and concentra­ tion-dependent manner. Paradoxically, glucocorticoids further enhance this stimulation. Thus glucocorti­ coids exert two separate and opposite effects on P A activity. The availability of glucocorticoid-resistant variant cell lines, together with the unique regulatory interactions of steroids and cyclic nucleotides, make HTC cells a useful experimental system in which to study the multi hormonal regulation of plasminogen activator.
Introduction
Plasminogen activators (PAs) are membrane­ associated arginine-specific serine proteases found in a variety of tissues (I). P A selectively hydro­ lyses a single Arg-Val bond of the plasma zymo­ gen, plasminogen, to yield the active serine pro­ tease, plasmin, the major fibrinolytic activity in blood (2, Fig. I). Plasmin is a broad-spectrum endopeptidase which can act on a variety of pro­ teins. Because plasminogen is present in plasma in relatively high concentrations (1.5 to 2 ~M, or 0.5% of all plasma proteins), the plasminogen ac­ tivator-plasmin cascade provides considerable po­ tential proteolytic activity (2, 3). Thus generation
of plasmin both amplifies P A activity and broad­ ens the substrate specificity. In addition to plas­ min's well-known role in fibrinolysis, it is also in­ volved in many normal physiologic functions which involve protein processing, cell migration and tissue remodelling (I, 3, 4, Table I). By acting directly on fibrin and directly or indirectly (via activation of procollagenase) on connective tissue matrix (5,6), the plasminogen activator-plasmin cascade may al­ so play an important role in tumor invasion and metastasis (1,3,4,6).
Not surprisingly for an enzyme of such biological importance, plasminogen activator is subject to regulation by a variety of effectors (see 7 for re­ view). Steroid (8-16) and polypeptide hormones
12
Table I. Physiological/ pathological processes mediated by plasminogen activator/ plasmin.
Fibrinolysis Proteolytic processing of cellular and serum proteins
Complement activation Kinin formation Proinsulin conversion
Migration of macrophages during inflammation Tissue remodelling and destruction
Rupture of the ovarian follicle during ovulation Implantation of the mouse embryo Involution of the mammary glandfollowing laclation
Neoplasia Invasiveness lvlelasla/is
(17,18), growth factors (4), cyclic nucleotides (13, 14,17-22), retinoids(22-25), lectins(26) and tumor promoters (4, 22, 25, 27) are all modulators of P A activity in some tissues. Inhibitors of P A and! or plasmin are also important regulators of protease activities (2, 28-33), and in some experimental sys­ tems these inhibitors may also be subject to physio­ logic regulation (10, 11,28,33). The mechanisms by which P A activity is regulated are largely unknown, but could involve regulation of the rates of synthe­ sis or degradation of the P A protein, activation of P A itself or of a P A precursor, or regulation of specific inhibitors of PA.
HTC cells are an established line of rat hepatoma
cells in long-term tissue culture, which provide a favorable experimental system for studying the regulation of plasminogen activators. This line is extremely well characterized with respect to hor­ monal regulation of multiple functions, particular­ ly for the actions of glucocorticoids, insulin, and cyclic nucleotides (34-38). Furthermore, it is possi­ ble to isolate variant HTC cell lines altered in hor­ monal regulation of various properties, and several such lines have been described (9, 39, 40). Over the past several years we have exploited these features of HTC cells to study the hormonal regulation of plasminogen activator and the role ofPA in various cellular functions modulated by hormones. We have described two unique mechanisms of regula­ tion of P A: first, the glucocorticoid induction of a specific inhibitor of plasminogen activator (10, II, 28); and second, a paradoxical effect of glucocorti­ coids on P A regulation in which glucocorticoids alone inhibit P A activity but together with cyclic nucleotides enhance the dramatic stimulation of PA activity by the latter (41).
Materials and methods
Cell culture HTC cells were routinely grown in spinner or
monolayer culture in Minimal Essential Medium (Eagle's) without antibiotics, modified to contain 50 mM tricine, a nonvolatile buffer, 0.5 g/ I sodium bicarbonate, and supplemented with2 mM glutam­ ine and 5% calf and 5% fetal bovine serum. Experi­ ments were performed in a chemically-defined me­ dium identical to the growth medium ~xcept that it lacked serum and was supplemented with neomycin and, where applicable, with 0.1 % bovine serum al­ bumin.
Assays of plasminogen activator P A was routinely assayed in either conditioned
medium (CM) or 0.2% Triton X-100 extracts of cells using either an 125I-fibrinolytic assay (42) or the esterolytic assay (43, 44) developed in this la­ boratory. HTC cells have no demonstrable plas­ minogen-independent fibrinolytic, caseinolytic, or estero lytic activity. Direct addition of dexametha­ sone or cyclic nucleotides to the assay mixture has no effect on PA activity. Inhibitory activity was measured by incubating CM or cell extracts to be
13
tested with either CM or cell extracts of untreated HTC cells (as a source of P A) or urokinase (a human urinary plasminogen activator) for 20 min at 37 0 C or 30 min at 25 0 C prior to assaying P A activity. Inhibitory activity was quantitated by ti­ trating serial dilutions of CM or cell extracts from dexamethasone-treated cells against a fixed amount of UK or HTC cell PA.
Characterization of plasminogen activator in HTC cells
Multiple molecular weight forms of P A were separated by SDS polyacrylamide gel electropho­ resis under nonreducing conditions. Following electrophoresis, SDS was removed from the gels and proteins by exchange with the nonionic deter­ gent Triton X-100, allowing recovery ofPA activity (45). P A activity was then localized and assayed either by elution ofPA from homogenized gel slices and assay of plasminogen-dependent fibrinolysis, by the fibrin-agar underlay method of Granelli-Pi­ perno & Reich (45), or by plasminogen-dependent caseinolytic activity in the gel (46, 47).
Enucleation of HTC cells Anucleate HTC cells (cytoplasts) were prepared
by centrifugation of cells from suspension cultures through a discontinuous Ficoll gradient in the pres­ ence of cytochalasin B. The efficiency of enuclea­ tion was routinely greater than 92%. Cytoplast preparations maintained their membrane integrity for at least 24 hours in culture in the absence of serum (48).
Isolation of variant HTC cells An agar-fibrin overlay technique (9, 49) was used
to identify colonies with plasminogen-dependent fibrinolytic activity. Colonies which expressed P A activity in the presence of dexamethasone (and were thus presumably resistant to the dexametha­ sone inhibition of P A) were isolated through the agar-fibrin overlay and propagated in the absence of dexamethasone. Variant cell lines highly resis­ tant to the dexamethasone inhibition of P A activity were obtained after several cycles of such treatment (9,39).
Materials Tissue culture media and sera were obtained
from Gibco. Dexamethasone was the kind gift of
14
Results
Glucocorticoid regulation of plasminogen activa­ tor
Plasminogen activator actiVIty is found in the membrane fraction of HTC cells from which it can be released by detergents such as Triton X-I 00 (10). Activity is also found in serum-free medium condi­ tioned by HTC cells (9, 41, 43). Analysis of PA activity from both cell extracts and conditioned
94K __ 68K
:3 C D
Fig. 2. Schematic diagram of multiple molecular weight forms of HTC cell plasminogen activator. Monolayer cultures of HTC cells were incubated for 24 hours in serum-free medium without hormones (Lane A). withO.II'M dexamethasone(Lane B), with 3 mM 8-bromo-cAMP; I mM MIBX (Lane C), or with 3 mM 8-bromo-cAMPfI mM MIBX;O.II'M dexamethasone (Lane D). Conditioned media were subjected to SDS polyacrylamide gel electrophoresis and PA activity was localized on a fibrin-agar underlay as described in Materials and methods. Molecular weight markers are phosphorylase B (94K). bovine serum al­ bumin (68K), ovalbumin (43K) and soybean trypsin inhibitor (2IK).
o 4
Time (hrs)
6
Fig. 3. Time course of dexamethasone inhibition of HTC cell plasminogen activator. HTC cells were incubated for the times indicated in serum-free medium containing I I'M dexametha­ sone. Fibrinolytic activity of cell extracts was measured as de­ scribed in Materials and methods and was normalized for the amount or protein in each sample. Each point represents the average of duplicate assays on a single culture. Reproduced from reference 10.
medium on SDS polyacrylamide gels under nonre­ ducing conditions reveals two major molecular weight forms of P A activity of 110000 and 66000 daltons. A minor band of activity is sometimes observed at 33 000 daltons (Fig. 2). There does not appear to be any difference between the molecular weights of the PAs associated with the cell and those released into the medium.
Glucocorticoids rapidly inhibit the activity ofPA in HTC cells. Inhibition is half-maximal after ap­ proximately 90 min and maximal after 4 to 6 hours of incubation (Fig. 3). The magnitude of inhibition is usually 75 to 100%. Inhibition of P A activity is also observed in conditioned medium at later times than in cell extracts. Half-maximal inhibition is achieved at 5 nM dexamethasone, the same concen­ tration that half-maximally induces tyrosine ami­ notransferase and half-maximally inhibits amino acid transport in HTC cells; maximal inhibition is achieved at 10 to 100 nM dexamethasone (39).
Dexamethasone could inhibit PA activity by de­ creasing the amount ofP A protein or by decreasing its activity (possibly by inducing an inhibitor of this protease), or by some combination of these mech­ anisms. When increasing amounts of an extract of dexamethasone-treated cells are mixed with a fixed amount of an extract from control cells (as a source of P A activity), there is a concentration-dependent inhibition of P A activity, demonstrating that the dexamethasone-treated cells contain an inhibitor of
40 ?: 0
~:~ ~ g 30 U .2 "'"0 (,) f! 20 ~n; go . .::'" 10 ~O
*- o
III conditioned medium
Fig. 4. Inhibition of HTC cell plasminogen activator by condi­ tioned medium from dexamethasone-treated HTC cells. HTC cells were incubated for 18 hours with 0.1 I'M dexamethasone (e) or without hormones(O). Increasing amounts of conditioned medium from these cultures were incubated for 30 min at 25°C with 10 I'g of cell extract from untreated HTC cells (as a source of P A) and fibrinolytic activity assayed as described. Each point represents the average of duplicate assays.
P A (10, 11). Cell fractionation experiments have demonstrated that the inhibitor is found primarily in the soluble 100 000 X g supernatant fraction (10), as well as in medium conditioned by dexametha­ sone-treated cells (Fig. 4).
An intact nucleus is required for this hormonal regulation of PA activity. Cytoplasts prepared by
e 100 Control "E
20 e '" 1;;
0 w 5
III conditioned medium
Fig. 5. Inhibition of human urokinase by conditioned medium from dexamethasone-treated HTC cells. HTC cells were incu­ bated for 18 hours with 0.1 I'M dexamethasone (e) or without hormones (0). Increasing amounts of conditioned medium were incubated for 20 min at 37 0 C with 5 milliPloug units of human urokinase and esterolytic activity assayed as described (44). Each point represents the average of duplicate assays.
15
centrifugation through Ficoll gradients in the pres­ ence of cytochalasin B maintain P A activity at lev­ els comparable to or higher than those of intact cells. However, anucleate HTC cells are not respon­ sive to glucocorticoid regulation of P A activity. Dexamethasone does not decrease either intracellu­ lar or extracellular P A activity of the anucleate cells, or induce production of the soluble inhibitor of PA activity (52).
The dexamethasone-induced inhibitor of P A ac­ tivity (PAl) also inhibits the plasminogen-depen­ dent fibrinolytic or estero lytic activity of HeLa cells (Coleman, unpublished work) and of human uro­ kinase (Fig. 5), as well as P A from HTC cells. Thus, PAl inhibits both major immunochemical types of P A, urokinase-like and tissue activator (I). In con­ trast, plasmin is not inhibited by conditioned medi­ um from HTC cells incubated with dexamethasone. The specificity of the inhibitor for plasminogen activation was demonstrated directly by the inhibi­ tion of urokinase-catalyzed activation of 125I-plas­ minogen to 1251-plasmin (28). These results show that the inhibition is not directed against plasmin, but is specific for plasminogen activator.
Because certain cell types can take up serum pro­ tease inhibitors from the serum in medium (31, 33, 53) and release them to serum-free medium condi­ tioned by these cells (31), we investigated the origin of this hormonally-induced inhibitor in HTC cells. SF HTC-HI, a line of cells selected for their ability to-grow in serum-free medium (54), were grown for 76 days (at least 30 generations) in the presence or absence of serum; dexamethasone induced equiva­ lent amounts of inhibitory activity in cells grown under either condition. Furthermore, the inhibitory activity from HTC cells is stable to pH 3 for 2 hours at 37 DC, a treatment which inactivates fibrinolytic inhibitors in serum. These results indicate that the dexamethasone-induced inhibitor is a cellular pro­ duct which differs from serum-derived fibrinolytic inhibitors (28).
The inhibitor is inactivated by boiling and by treatment with pepsin under acidic conditions, sug­ gesting that it is a protein. The P A inhibitory activi­ ty in CM from dexamethasone-treated cells mi­ grates as a single band of approximately 45 000 daltons upon SDS polyacrylamide gel electropho­ resis under nonreducing conditions (Cwikel & Ge­ lehrter, unpublished work). PAl is clearly different from the 38 000 dalton protease inhibitor, protease
16
nexin, described by Low et al. (32) which inacti­ vates both thrombin and urokinase. Conditioned medium from dexamethasone-treated cells, which readily inhibited urokinase, had no effect on thrombin activity, even in the presence of heparin which accelerates the thrombin-protease nexin in­ teraction (Coleman & Gelehrter, unpublished work).
We have begun to investigate the interaction of P Al with various PAs. Preliminary evidence sug­ gests that P AI forms an irreversible covalent com­ plex with urokinase. I n contrast, its interaction with P A from HTC cells can be reversed under conditions of SDS polyacrylamide gel electropho­ resis. Taking advantage of the latter fact, we have asked whether dexamethasone decreases P A activi­ ty directly, in addition to the effects mediated by P AI. Cell extracts and CM from dexamethasone­ treated cells, which have less than 10% of control PA activity when assayed by the fibrin plate assay,
AGAR-FIBRIN OVERLAY TECHNIQUE
PREPARATION OF AGAR-FIBRIN
POUR OYER CELLS p .... TEO OUT ON 601ft.,. CMSH(S
Figure 6. Schematic diagram of the agar-fibrin overlay tech­ nique used to isolate variant HTC cell lines resistant to the glucocorticoid inhibition of PA. Experimental details in Mate­ rials and methods and in references 9 and 39. Reproduced from reference 9.
exhibit PA activity comparable to the controls when measured by the fibrin-agar underlay (Fig. 2) or by direct assay of gel eluates following S OS polyacrylamide gel electrophoresis. These results suggest that PA and PAl dissociate during electro­ phoresis, and that there is no decrease in the amount of PAin cell extracts or medium from dexamethasone-treated cells (Cwikel, Coleman, Barouski-Miller & Gelehrter, unpublished work).
Isolation and characterization oj variant hepatoma cells resistant to hurmunal regulation oJplasmino­ gen activator
Utilizing the agar-fibrin overlay technique (Fig. 6), we have isolated a number of variant HTC cell lines which are highly resistant to the dexametha­ sone inhibition of PA activity (9, 39). These var­ iants are resistant to a concentration of dexametha­ sone I 000 times greater than that necessary to completely inhibit P A activity in wild-type cells (39). The nature of the defect in these cells was defined by mixing experiments analogous to those described above. Dexamethasone-treated variant cells show no PAl activity, whereas the PA from these cells is fully sensitive to inhibitor from gluco­ corticoid-treated wild-type cells. Thus the basis of hormone resistance in the variants appears to be the failure of dexamethasone to induce PAl (10, II).
The growth rate and cloning efficiency of variant and wild-type lines, both on plastic and in soft agar, are indistinguishable. Morphologically there are no consistent differences between the variant and wild­ type cells. Fluctuation analyses support the hy­ pothesis that resistance to dexamethasone arises randomly and is not induced by the hormone. It was not possible to determine the rate at which stable variant cells arise, but only to quantitate the frequency of the first step in this process. The fre­ quency with which colonies from a wild-type popu­ lation form fibrinolytic plaques in the presence of dexamethasone is high (approximately 10-3) and this rate is not altered by treatment with two differ­ ent mutagens: ethylmethane sulfonate and UV light. This observation suggests that mutations are not the primary cause of resistance in this cell line. The karyotypic variability of HTC cells raises the possibility that variants might arise from chromo­ somal segregation events (39).
Biochemical analysis of the variant cell lines
demonstrates that they have a lesion specific for the regulation of plasminogen activator. The hormonal resistance is apparently not due to deficient or de­ fective steroid receptor function since the variants show wild-type induction of tyrosine aminotrans­ ferase. The lesion iIi these variants must therefore be at some step distal to the entry of the hormone­ receptor complex into the nucleus. We have also shown that 6 of 7 variant cell lines tested show wild-type inhibition of amino acid transport by glucocorticoids (one variant is partially resistant to the inhibitory effect of dexamethasone on trans­ port). These findings indicate that there is not a generalized resistance to all membrane-associated dexamethasone responses, but that the cells are selectively resistant to the inhibition of P A (39). Other variant f-!:TC cell lines which are selectively resistant to the dexamethasone induction of tyro­ sine aminotransferase (40) show wild-type inhibi­ tion of P A activity by glucocorticoids (55). These results indicate that the various glucocorticoid-me­ diated responses in HTC cells are independently rather than coordinately regulated (11, 55). The selective resistance of these HTC variants is unique and in contrast to the great majority of glucocorti­ coid-resistant variant cell lines previously described, essentially all of which have been shown to have deficient or defective glucocorticoid receptors (56, 57).
We utilized these variant lines to study the role of plasminogen activator in the hormonal regulation of other membrane properties. HTC cells exhibit increased levels of adhesion to a substrate as well as decreased P A activity when incubated with dexa­ methasone (58). In a variety of cultured cells, adhe­ sion to a substrate requires specific cell surface glycoproteins and intact cytoskeletal elements, and there is evidence that plasmin may affect both of these components of adhesion (59-61). Using the variant HTC cell lines, we tested the hypothesis that dexamethasone induces cell adhesion by decreasing the activity of P A, which in turn allows the accumu­ lation of specific cell surface glycoproteins neces­ sary for adhesion. If this hypothesis were correct there should be little or no dexamethasone inhibi­ tion of adhesion in the variant cell lines. A sensitive quantitative assay was developed which measures the strength of attachment of radioactively-labeled cells to glass scintillation vials following exposure to shearing force (62). We found that dexametha-
17
sone induces the adhesiveness of variant HTC cells to the same extent as that of wild-type cells. This was true when adhesion was measured in serum­ free medium, in serum-containing medium, or in serum-containing medium depleted of, or reconsti­ tuted with, plasminogen. These results indicated that neither P A itself nor plasmin plays a major role in cell adhesion in HTC cells and suggested that the dexamethasone induction of adhesion might oper­ ate through the synthesis of cell surface glycopro­ teins (II , 62).
Cyclic nucleotide regulation of plasminogen activa­ tor activity
Incubation of HTC cells with cAMP deriva­ tives stimulates cell-associated P A activity 8- to 20-fold and extracellular P A activity 30- to I 300- fold. This time- and concentration-dependent in­ crease is enhanced by phosphodiesterase inhibi­ tors such as l-methyl-3-isobutylxanthine (MIBX). Maximal stimulation of P A activity is observed at 3 mM 8-bromo-cAMP and half-maximal stimula­ tion at 0.2 mM. A similar concentration depend­ ence is noted with dibutyryl-cAMP. N6-monobu­ tyryl-cAMP also stimulated P A activity but cAMP itself did not; dibutyryl-cGMP inhibited PA activi­ ty. Increases in PA activity in cells incubated with 8-bromo-cAMP are first detectable at 4 hours in cell extracts and 6 hours in extracellular medium, and are maximal by 8 and 12 hours, respectively (Fig. 7). MIBX increases the level of maximal stim­ ulation, but does not significantly alter the time course (41).
As noted above, dexamethasone decreases P A activity by induction of an inhibitor. Paradoxically, dexamethasone added simultaneously with cAMP derivatives causes a further 4-fold enhancement of the cyclic nucleotide stimulation of P A activity (41). Analysis of the molecular weight forms ofPA under these conditions indicates that the same 110 K and 66 K forms are present in cells treated with cyclic nucleotides or cyclic nucleotides plus dexamethasone as in control and dexametha­ sone-treated cells. In addition a minor 33 K dalton form of PAis sometimes found in cells incubated with cAMP derivatives plus or minus dexametha­ sone (Fig. 2). Dexamethasone also profoundly al­ ters the time course of the cyclic nucleotide en­ hancement of P A activity: increased activity is
18
8.0
.s
12 16 20 24
Time (hours]
Fig. 7. Time course of 8-bromo-cAMP action in the presence and absence of dexamethasone. HTC cells were incubated for 0-24 hours in serum-free medium containingO.1 I'M dexameth­ asone (A), 3 mM 8-bromo-cAMP (0), 3 mM 8-bromo-cAMP/ I mM MIBX (0), or 3 mM 8-bromo-cAMP/ I mM MIBX/ 0.1 I'M dexamethasone ("). (e), Control (no additions). Upper panel: cell-associated PA activity. Lower panel: extracellular PA activity. Inset: expanded ordinate scale to demonstrate the effect of dexamethasone. In these experiments, the fibrinolytic activities of several different amounts of each sample were mea­ sured and the slope of the line (percentage of fibrin solubilized per microgram cellular protein or microliter of extracellular medium) was determined by linear regression analysis. P A activ­ ity is expressed as micrograms fibrin solubilized per microgram of cellular protein or microliter of extracellular medium. Repro­
duced from reference 41.
detected at4 hours in cells incubated with 8-bromo­ cAMP and MIBX but not until 12 hours in cells incubated with dexamethasone as well (41). Induc­ tion of inhibitor by dexamethasone might explain this delay in appearance of the cyclic nucleotide­ stimulated increase in PA activity. Glucocorticoids thus exert two separate and opposite effects on P A activity: induction of an inhibitor and amplification of cyclic nucleotide action. Although permissive
and synergistic effects of dexamethasone on cyclic nucleotide action have been reported previously (38,63), glucocorticoid regulation of P A activity is unique in that the amplification of cyclic nucleotide effects by dexamethasone opposes its regulatory action toward a specific enzyme.
In variant HTC cells selectively resistant to the glucocorticoid inhibition of P A activity, dexameth­ asone still enhances the stimulation of P A activity by cyclic nucleotides (Fig. 8). Furthermore, in con­ trast to its effect in wild-type cells, dexamethasone does not alter the time course of 8-bromo-cAMP stimulation of P A activity in variant cells; en­ hancement is first observed at 6 hours and is max­ imal by 12 hours incubation. These results appear to dissociate the glucocorticoid induction of inhibi­ tor from its enhancement of cyclic nucleotide stim­ ulation of P A activity.
The steroid specificity of the glucocorticoid en­ hancement effect appears to be similar to that for glucocorticoid inhibition of P A activity and amino
!: :~ 20 ti co 0 'ij 10 ~
'iii B '0 0 4 8 12 16
*' Variant
~ u::
o Time, hr
Fig. 8. Time course of8-bromo-cAMP stimulation of PA activi­ ty in the presence or absence of dexamethasone in wild-type and variant cells. Wild-type and variant HTC cells were incubated for 0-24 hours in serum-free medium containing: no additions (e), 0.1 I'M dexamethasone (A), 3 mM 8-bromo-cAMP/ I mM MIBX (0), orO.II'M dexamethasonej3 mM 8-bromo-cAMP/ I mM MIBX ("). Cell-associated P A activity (in 2.5 micrograms cellular protein of wild-type cells or I microgram cellular protein of variant cells) was measured on "'I-fibrin plates as described in Materials and methods, and expressed as a percentage of total radioactivity released.
acid transport (64), as well as induction of tyrosine aminotransferase (65). The optimal inducers, or full agonists, dexamethasone and cortisol, show a sim­ ilar concentration dependence curve for all of these phenomena. In each case, dexamethasone was ten times more potent than cortisol. The partial ago­ nists, II f3-hydroxyprogesterone and deoxycortico­ sterone, cause submaximal enhancement of the cyc­ lic nucleotide stimulation of P A activity and do so only at higher steroid concentrations. Tetrahydro­ cortisol, which does not interact with the glucocor­ ticoid receptor, fails to enhance cyclic nucleotide stimulation of P A activity. The glucocorticoid an­ tagonist, 17a-methyltestosterone, which has no ef­ fect on the enhancement of cyclic nucleotide stimu­ lation by itself, blocks the enhancement by dexa­ methasone in a concentration-dependent manner. These observations suggest that the steroidal en­ hancement of cyclic nucleotide stimulation of P A activity is mediated by the same glucocorticoid re­ ceptor mechanism which mediates the induction of transaminase, and the inhibition of amino acid transport and PA activity (7, Barouski-Miller & Gelehrter, unpublished work).
Incubation of HTC cells with cAMP derivatives also alters cell morphology, causing cell elongation and extension of processes followed by flattening of the cells. Plasminogen activator has been reported to alter cell morphology in several lines either di­ rectly (66) or by production of plasmin (67). We have shown, however, that the cyclic nucleotide effects on cell morphology are not caused by the stimulation of P A activity and can be dissociated from them. The morphologic changes appear with­ in 30 to 60 min incubation with cyclic AMP derivatives, long before any detectable changes in either intracellular or extracellular P A are appar­ ent. Upon removal of cyclic nucleotides from the medium, cell morphology returns to normal within two to four hours, a time at which P A activity is still significantly elevated. Furthermore, when protein synthesis is blocked by cycloheximide, the cyclic nucleotide stimulation of P A activity is completely blocked; however, induction of mor­ phologic changes still occurs. Analogous to the si­ tuation described above for glucocorticoid regula­ tion of P A and cell adhesion, these results suggest independent regulation by cyclic nucleotides of P A activity and cell morphology (7, Barouski-Miller & Gelehrter, unpublished work).
19
Discussion
HTC cells provide a useful experimental model for studying the hormonal regulation of plasmino­ gen activator and the role of PAin several cellular functions modulated by hormones. In addition to various well-characterized hormonal responses in these cells, variant cell lines which are resistant to specific hormone-mediated functions have been is­ olated (9,39,40,55,68). Our studies have revealed two unique regulatory mechanisms: glucocorticoids inhibit P A activity by inducing a soluble inhibitor rather than by regulating the amount of enzyme (10, II, 28); and glucocorticoids together with cy­ clic nucleotides paradoxically enhance the dramat­ ic stimulation of P A activity by cyclic nucleotides (41).
Further investigation of this system should yield interesting information about mechanisms of hor­ monal regulation of this important protease. The ability to study the multiple molecular weight forms ofPA should allow studies on the hormonal regula­ tion of the expression of these forms. We can inves­ tigate whether various hormones cause differential regulation of these forms of plasminogen activator and whether they affect interconversion of these forms. The isolation and characterization of the dexamethasone-induced inhibitor (PAl) and the preparation of specific antibodies to it should allow studies on the hormonal regulation of PAl at a molecular level. Finally, the paradoxical effects of glucocorticoids on PAin this system provide a unique opportunity to study the nature of glucocor­ ticoid-cyclic nucleotide interactions.
Acknowledgements
This work was supported by Grant CA 22729 from the National Cancer Institute. P.A.B-M. was supported by Predoctoral Training Grant G M 97544 from the National Institutes of Health. We thank Ms Judy Worley for secretarial assistance, and Denis Lee for helping create Fig. I.
References
I. Christman, J. K., Silverstein, S. C. and Acs, G .. 1977. In: Proteinases in Mammalian Cells and Tissues. (Barrett, A. J., ed.), pp. 91-149, New York: North Holland Publishing Co.
20
2. Lijnen, H. R. and Collen, D., 1982. Seminars in Thrombosis and Hemostasis 8: 2·10.
3. Reich, E., 1978. In: Biological Markers of Neoplasia: Basic and Applied Aspects. (Ruddon, R. W., Jr., ed.), pp. 491-500, New York: Elsevier-North Holland.
4. Weinstein, I. B., Wigler, M., Yamasaki, H. et aI., 1978. In: Biologlcal Markers of Neoplasia: Basic and Applied As­ pects. (Ruddon, R. W., Jr., ed.), pp. 451-471, New York: Elsevier-N orth Holland.
5. Werb, Z., Mainardi, C. L."Vater, C. A. and Harris, E. D., Jr., 1977. N. Engl. J. Med. 296: 1017-1023.
6. Quigley, J. P., 1979. In: Surfaces of Normal and Malignant Cells. (Hynes, R.O., ed.), pp. 247-285, Chichester: John Wiley.
7. Miller, P. A. Barouski-, 1982. Ph.D. thesis, University of Michigan.
8. Wigler, M., Ford, J. P. and Weinstein, I. B., 1975. In: Pro­ teases and Biological Control. (Reich, E., Rifkin, D. B. and Shaw, E., eds.), pp. 849-856, New York: Cold Spring Harbor Laboratory.
9. Carlson, S. A. and Gelehrter, T. D., 1977. J. Supramolecular Structure 6: 325-331.
10. Seifert, S. C. and Gelehrter, T. D., 1978. Proc. Natl. Acad. Sci. U.S.A. 75: 6130-6133.
II. Gelehrter, T. D., Seifert, S. C. and Fredin, B. L., 1979. Cold Spring Harbor Conf. Cell Prolif. 6: 259-267.
12. Laishes, B. A., Roberts, E. and Burrowes, c., 1976. Bio­ chern. Biophys. Res. Commun. 72: 462-471.
13. Vassali, 1.-D., Hamilton, J. and Reich, E., 1976. Cell 8: 271-281.
14. Granelli-Piperno, A., Vassalli, J.-D. and Reich, E., 1977. J. Exp. Med. 146: 1693-1706.
15. Roblin, R. and Young, P. L., 1980, Cancer Research 40: 2706-2713.
16. Werb, Z., 1978. J, Exp. Med. 147: 1695-1712. 17. Beers, W. H., Strickland, S. and Reich, E., 1975. Cell 6:
387-394. 18. LaCroix, M. and Fritz, I. B., 1982. Molec. and Cell. Endo­
crinol. 26: 247-258. 19. Laug, W. E., Jones, P. A., Nye, C. A. and Benedict, W. F.,
1976. Biochem. Biophys. Res. Commun. 68: 114-119. 20. Rosen, N., Piscitello, J., Schneck, J. et aI., 1979. J. Cell.
Physiol. 98: 125-136. 21. Rosen, N., Schneck, J., Bloom, B. R. and Rosen, O. M.,
1978. J. Cyclic Nucleotide Research 5: 345-358. 22. Wilson, E. L. and Reich, E., 1979. Cancer Research 39:
1579-1586. 23. Strickland, S. and Mahdavi, V., 1978. Cell 15: 393-403. 24. Schroder, E. W., Chou, I.-N. and Black, P. H., 1980. Cancer
Research 40: 3089-3094. 25. Miskin, R., Easton, T. G. and Reich, E., 1978. Cell 15:
1301-1312. 26. Mochan, E., 1979. Biochim. Biophys. Acta 558: 273-278. 27. Wigler, M. and Weinstein, I. B., 1976. Nature 259: 232-233. 28. Coleman, P. L., Barouski, P. A. and Gelehrter, T. D., 1982.
J. BioI. Chern. 257: 4260-4267. 29. Loskutoff, D. J. and Edgington, T. S., 1977. Proc. Natl.
Acad. Sci. U.S.A. 74: 3903-3907. 30. Roblin, R. 0., Young, P. L. and Bell, T. E., 1978. Biochem.
Biophys. Res. Commun. 82: 165-172.
31. Rohrlich, S. T. and Rifkin, D. B., 1981. J. Cell. Physiol. 109: 1-15.
32. Low, D. A., Baker, J. B., Koonce, W. C. and Cunningham, D. D., 1981. Proc. Natl. Acad. Sci. U.S.A. 78: 2340-2344.
33. Finlay, T. H., Katz,J., Rasums, A. etal., 1981. Endocrinol­ ogy 108: 2129-2136.
34. Thompson, E. B., 1979. In: Glucocorticoid Hormone Ac­ tion. (Baxter, J. D. and Rousseau, G. G., eds.), pp. 203-217, Heidelberg: S pringer-Verlag.
35. Higgins, S. J., Baxter, J. D. and Rousseau, G. G., 1979. In: Glucocorticoid Hormone Action. (Baxter, J. D. and Rous­ seau, G. G., eds.), pp. 135-160, Heidelberg: Springer-Verlag.
36. Gelehrter, T. D., 1979. In: Glucocorticoid Hormone Action. (Baxter, J. D. and Rousseau, G. G., eds.), pp. 561-574, Hei­ delberg: S pringer-Verlag.
37. Gelehrter, T. D., 1979. In: Glucocorticoid Hormone Action. (Baxter, J. D. and Rousseau, G. G., eds.), pp. 583-591, Hei­ delberg: S pringer-Verlag.
38. Granner, D. K., 1979. In: Glucocorticoid Hormone Action. (Baxter, J. D. and Rousseau, G. G., eds.), pp. 593-611, Hei­ delberg: S pringer-Verlag.
39. Seifert, S. C. and Gelehrter, T. D., 1979. J. Cell. Physiol. 99: 333-342.
40. Thompson, E. B., Aviv, D. and Lippman, M. E., 1977. En­ docrinology 100: 406-419.
41. Barouski-Miller, P. A. and Gelehrter, T. D., 1982. Proc. Natl. Acad. Sci. U.S.A. 79: 2319-2322.
42. Strickland, S. and Beers, W. H., 1976. J. BioI. Chern. 251: 5694-5702.
43. Coleman, P. L. and Green, G. D. J., 1981. Annals, N.Y. Acad. Sci. 370: 617-626.
44. Coleman, P. L. and Green, G. D. J., 1981. Meth. Enzymol. 80: 408-414.
45. Granelli-Piperno, A. and Reich, E., 1978. J. Exp. Med. 148: 223-234.
46. Huessen, C. and Dowdle, E. B., 1980. Anal. Biochem. 102: 196202.
47. Miskin, R. and Soreq, H., 1982. Anal. Biochem. 118: 252-258.
48. McDonald, R. A. and Gelehrter, T. D., 1981. J. Cell BioI. 88: 536-542.
49. Jones, P., Benedict, W., Strickland, S. et aI., 1975. Cell 5: 323-329.
50. Deutsch, D. G. and Mertz, E. T., 1970. Science 170: 1095-1096.
51. Gilbert, L. R. and Wachsman, J. T., 1976. Anal. Biochem. 72: 480-484.
52. Barouski, P. A. and Gelehrter, T. D., 1980. Biochem. Bio­ phys. Res. Commun. 96: 1540-1546.
53. Van Leuven, F., Cassiman, J-J. and Van den Berghe, H., 1979. J. BioI. Chern. 254: 5155-5160.
54. Thompson, E. B., Anderson, C. U. and Lippman, M. E., 1975. J. Cell. Physiol. 86: 403-412.
55. Thompson, E. B., Granner, D. K., Gelehrter, T. D. et aI., 1979. Molec. and Cell. Endocrinol. 15: 135--150.
56. Pfahl, M. R., Kelleher, R. J. and Bourgeois, S., 1978. Molec. and Cell. Endocrinol. 10: 193-207.
57. Yamamoto, K. R., Gehring, U., Stampfer, M. R. and Sibley, C. H., 1976. Rec. Prog. Hormone Res. 32: 3-23.
58. Ballard, P. and Tomkins, G. M., 1970. J. Cell BioI. 47: 222-234.
59. Culp, L., 1978. Curf. Top. Membf. Transp. 11: 327-395. 60. Pollack, R. and Rifkin, D., 1975. Cell 6: 495-506. 61. Weber, 1. M., 1975. Cell 5: 253-261. 62. Fredin, B. L., Seifert, S. C. and Gelehrter, T. D., 1979. Na­
ture 277: 312-313. 63. Rousseau, G. G., 1977. Eur. J. Biochem. 76: 309- 316. 64. Gelehrter, T. D. and McDonald, R. A., 1981. Endocrinology
109: 476-482.
21
65. Samuels, H. H. and Tomkins, G. M., 1970.1. Mol. BioI. 52: 57-74.
66. Quigley, J. P., 1979. Cell 17: 131-141. 67. Urquhart, c., Whur, P., Gordon, M. et aI., 1978. Exptl. Cell
Res. 113: 31-38. 68. Grove, 1. R. and Ringold, G. M., 1981. Proc. Natl. Acad.
Sci. U.S.A. 78: 4349-4353.
Received 24 September 1982.
Molecular and Cellular Biochemistry 53/54, 23-33 (1983), © 1983, Martinus Nijhoff Publishers, Boston. Printed in The Netherlands.
Cell types in rat liver cultures: their identification and isolation
J. W. Grisham Department of Pathology. University of North Carolina. School of Medicine, Chapel Hill, NC 27514. U.S.A.
Abstract
This paper reviews the various types of cells in the liver in vivo and in hepatic cellular suspensions produced by perfusion ofthe liver with collagenase solutions. Methods to identify and isolate different types of hepatic cells are discussed. In vitro culture of various types ofliver cells is reviewed and the identification of cultured cells is considered.
I. Introduction
The major aim of this paper is to consider in broad outline methods and strategies for isolating and culturing the several types of cells that compose the liver. A secondary aim is to discuss the problems and possibilities of identifying specific types of cells in cultures of liver tissue in order to relate the cultured cells to cells in the liver in vivo.
Culture of liver tissue and of cells isolated from liver affords the investigator the opportunity to examine hepatic function at the cellular level, un­ impeded by those constraints that follow from the fact that the liver in vivo is a complex tissue com­ posed of several types of cells and unaffected by the influences of cells and cellular products from other parts of the body. Although much progress has been made in the culture of liver and in the disper­ sion of liver tissue into popUlations of single cells for culture, much less effort has been directed to­ ward the separation of the mixed population that comprises a suspension of total liver cells into sub­ popUlations containing only one type of cell. Many investigators seem to assume that an enzymatic suspension ofliver parenchyma contains a virtually pure population of hepatocytes, although such an assumption can readily be shown to be incorrect by
using relatively simple cytological or tissue culture techniques. An unquestioning attitude about the cellular composition of an enzymatic suspension of liver cells has already led to possibly incorrect con­ clusions concerning the primacy of the hepatocyte in diverse areas of hepatic function.
A related pro blem pertains to the identity of cells in long-term cell cultures derived from dispersed liver cells. When a suspension of liver cells is main­ tained in culture for several weeks, hepatocytes are replaced by small cells with clear cytoplasm, some of them epithelial in character, which can often be subcultured. Such long-term epithelial cultures de­ rived from liver have been used for a variety of experimental investigations. Although they bear no morphological resemblance to authentic hepato­ cytes, these continually culturable hepatic epithelial cells are frequently termed hepatocytes. Since there is strong evidence that cultured hepatic epithelial cells are derived from cells in the original suspen­ sion other than mature hepatocytes, the designa­ tion of these cells as hepatocytes is misleading. If data from the analysis of hepatic cells in long-term culture are to be applied to an interpretation of hepatic function in vivo, then identification of the cell of origin in the liver is of considerable impor­ tance.
24
A variety of techniques for culturing liver are now available, including organ and explant cul­ tures of intact liver, primary culture of total hepatic cellular suspensions or of partially purified subfrac­ tions of the whole suspension, and secondary (prop­ agable) cell cultures. Many valuable studies have already been carried out using these techniques. Recent reviews show that primary cultures of dispersed hepatic cells are suitable preparations to analyze many facets of hepatic function (1,2), as well as to evaluate the metabolism of chemicals and drugs and to ascertain the cytotoxicity of their me­ tabolites (3-5). The phenomenon of carcinogenesis in epithelial cells (as contrasted to mesenchymal cells) was first examined in propagable hepatic epithelial cells (reviewed in 3). Perhaps the greatest handicap to the accurate interpretation of the re­ sults of studies employing cultured liver cells stems from the combination of not knowing the exact composition ofthe mixed cellular population one is studying and/ or not being able to identify the cell one is studying in vitro with a cell type present in the liver in vivo. This paper focuses on these problems and, although it presents few answers, aims to stim­ ulate studies that will yield solutions.
II. Types of cells in the liver
The liver is a cytologically and structurally com­ plex tissue (6, 7). In the normal liver, hepatocytes predominate in both number and volume, but the liver contains several additional types of cells and these non parenchymal cells are responsible for many important hepatic functions. As a point of departure for discussing the identification and iso­ lation of different types of hepatic cells, it is useful to consider liver tissue in terms of its major structu­ ral compartments-portal tracts and lobular paren­ chyma. Portal tracts comprise a dense collagenous matrix containing afferent blood vessels (both arte­ rial, portal venous, and capillary), secondary and larger order bile ducts, nerves, and lymphatic ves­ sels (6, 7). The spaces in the collagenous matrix also contain a variable population of cells, including fibroblasts, hematopoietic stem cells, and variable numbers of leukocytes, including neutrophiles, lymphocytes, plasma cells, mast cells, macrophages, and eosinophiles. Included in the tissue of the por­ tal tracts are epithelial cells of bile ducts, endothe-
lial cells of blood and lymphatic vessels, smooth muscle cells of arteries and veins, cells that form nerves, and a variety of mesenchymal cells, includ­ ing fibroblasts and inflammatory cells. The lobular parenchyma is histologically less complex than por­ tal tracts, consisting predominantly of interposed plates of hepatocytes and sinusoids, and of the ef­ ferent blood vessels into which the sinusoids drain (6,7). The lobular parenchyma is supported by only a light investment of collagen fibers, which form a network around efferent vessels and along the inter­ face between sinusoids and plates of hepatocytes. As in the whole liver, hepatocytes are the most numerous cellular component of the lobular paren­ chymal compartment, and they comprise the major volume fraction of this compartment (8, 9). How­ ever, sinusoidal endothelial cells and Kupffer cells, although small in size, are also numerous constitu­ ents of the lobular parenchymal compartment (9, 10); fat storing (Ito) cells (II) and so-called pit cells (12) are also represented in the lobular parenchymal compartment of the normal liver, residing in the tissue space between hepatocytic plates and sinus­ oids. Epithelial cells of the terminal bile ducts (usu­ ally called bile ductules) and fibroblasts and smooth muscle cells in the walls of efferent hepatic veins are frequently overlooked cellular constituents of the lobular parenchyma. Bile ductules connect the bile canaliculi of hepatocytic plates to bile ducts in por­ tal tracts, providing a pathway for the outflow of bile (13, 14). Terminal ends of bile ductules are connected by attachment complexes to hepatocytes at the portal ends of hepatocytic plates; bile duc­ tules protrude for short distances into the lobular parenchymal compartment without a heavy in­ vestment of connective tissue (13, 14).
Not only is there heterogeneity of cell types in the lobular parenchyma, but hepatocytes located in dif­ ferent parts of the hepatocellular plates differ in form and function (15, 16). In general, structure and function of hepatocytes follow gradients along the distance of hepatocellular plates from the vicini­ ty of portal tracts (Zone I) to the vicinity of termi­ nal hepatic veins (Zone 3). In adult rats, hepato­ cytes of Zone I are diploid, whereas cells of Zone 3 are polyploid. Zone I hepatocytes are smaller and contain fewer but larger mitochondria than do hep­ atocytes of Zone 3. A larger fraction of the typical hepatocyte in Zone I is occupied by the Golgi appa­ ratus than is true of the Zone 3 hepatocyte, whereas
the opposite situation obtains in the instance of the smooth endoplasmic reticulum. Functionally, hep­ atocytes of Zone I appear to be predominantly involved in gluconeogenesis, while cells in Zone 3 are mainly involved in glycolysis. Analogous varia­ tions in protein and lipid metabolism, drug metabo­ lism, and bile formation may also typify hepato­ cytes located in different parts of the lobular parenchymal compartment.
The cellular composition of the liver described here typifies the normal liver of the adult rat. N or­ mal livers of some other species and pathologically altered livers of all species may show a heavier investment of collagenous connective tissue in the lobular parenchyma and different distributions and arrangements of various types of cells. In addition, cells in the normal livers of the developing animal may vary considerably from those described, espe­ cially by containing a large population of hemato­ poietic cells. Techniques that allow the separation of cells from normal livers may not apply to devel­ oping or pathologically altered livers.
III. Features of different hepatic cells that allow selective recovery
The remainder of this discussion focuses on the cells of the lobular parenchymal compartment, since it is these cells that are predominantly released by limited perfusion of the liver with collagenase solutions, the method used most frequently to dis­ perse liver cells. Each of the types of cells that make up the tissue of the lobular parenchymal compart­ ment are morphologically and functionally distinc­ tive and can be accurately identified in vivo by their characteristic morphologies, synthetic capacities, enzymatic features, metabolic responses to specific challenge, capacity to store metabolites, locations of receptors and receptor-mediated internalization ofligands, and ability to phagocytically ingest large particulates. This section emphasizes some features of various hepatic cells that facilitate their selective recovery from an enzymatic suspension of cells of the whole liver and their identification in long-term cultures. The differential features of the various types of hepatic cells that are presently useful in selectively recovering them from a mixed popula­ tion are size, density, surface membrane receptors and enzymes, and differences in the content of sub-
25
stances that can be tagged with fluorochromes. For purposes of identification of different types of cells, these features plus the histochemical demonstra­ tion of selected enzymes are useful.
Hepatocytes in the adult rat are large cells that vary from about 15 /-Lm to nearly 25 /-Lm in diameter and have densities that vary from about 1.100 to 1.140 gm/ ml (17). In livers of immature rats and in livers that are the sites of pathological oval cell reactions, smaller hepatocytes are also found (18). Isolated hepatocytes are spherical and their sur­ faces are completely covered by densely numerous microvilli (17, 19). Ultrastructurally, hepatocytes contain a rich endowment of organelles, including large aggregates of rough and smooth endoplasmic reticulum, prominent Golgi apparatuses, many large mitochondria, numerous Iysosomes, and per­ oxisomes (17, 19). Hepatocytes from well fed ani­ mals contain large stores of glycogen. Isolated hep­ atocytes possess the functional ability to synthesize several serum proteins, including albumin, trans­ ferrin, and coagulation factors (1,2). They metabol­ ize bilirubin and possess a broad capability to me­ tabolize exogenous chemicals by mixed-function mono-oxidation (3-5). Numerous membrane re­ ceptors have been identified on hepatocytes (20). Hepatocyte-specific monoclonal antibodies have been developed (21).
Isolated Kupffer cells are 8-12 /-Lm in diameter (22) and stellate in shape with a kidney-shaped nucleus (23). Their peak density has been measured at about 1.076 gm/ ml (22). The surface membranes of isolated Kuppfer cells form complex folds and microvilli (23) and internally these cells are distin­ guished by a large number of phagosomes (22). They are actively phagocytic in vitro, avidly ac­ cumulating relatively large (0.8 /-Lm) particles (22-24). The surface membranes of Kupffer cells contain Fe and C3 receptors (23,24), receptors for mannose and N-acetyl glucosamine (25-26), and they bind avidly to glass (23, 24), as do other mac­ rophages. They contain high levels of peroxidase (22, 23), acid phosphatase (and other lysosomal enzymes) (22, 23), and glucose-6-phosphate dehy­ drogenase (27).
Isolated endothelial cells are smaller than Kupffer cells, measuring 4-6 /-Lm in diameter (28). Their surfaces are more or less smooth, and when isolated they are round and have a high nuclear cytoplasmic ratio (22, 26). Ultrastructurally, the cytoplasm of
26
freshly isolated cells has a lace-like appearance, which is interpreted to represent the remnants of in-situ sieve plates (22). Their density is about 1.060 to 1.080 gm/ ml (22). Endothelial cells are said to synthesize type IV collagen (29). Whether or not sinusoidal endothelial cells can synthesize and store coagulation factor VIII, as do endothelial cells of other vessels, is now a matter of dispute. Factor VIII has been localized immunologically by fluo­ rescence microscopy to sinusoidal endothelial cells in sections of liver tissue (29), but recent studies on isolated endothelial cells are said to demonstrate that these cells lack Factor VIII (28). Sinusoidal endothelial cells contain receptors for glycopro­ teins that have a terminal mannose or N-acetyl glucosamine residue (26).
Bile ductular epithelial cells are 6-12 Mm in di­ ameter( 18) and their surface membrane is occupied by microvilli (13, 14, 31). These cells are round and they contain a round nucleus (18). Their cytoplas­ mic organelles are not distinctive (18). The density of bile ductular cells ranges from 1.075 to 1.1 00 gm/ ml with a peak of about 1.095 gm/ ml (18,32). Antibodies have been developed against isolated biliary epithelial cells and shown to bind to bile ductules in vivo (32). Bile ductular epithelial cells contain large amounts of ),-glutamyl transpepti­ dase (18,32) and leucine aminopeptidase (32).
F at-storing cells can be detected in isolated popu­ lations of normal hepatic cells by the presence of lipid droplets (11) or vitamin A fluorescence (33). Presence oflipid should make them much less dense than cells that do not contain lipid, but the density of these cells apparently has not been measured. Although these cells show bright vitamin A fluores­ cence in vivo, attempts to distinguish a subpopula­ tion of isolated nonparenchymal cells that contains a high content of vitamin A (34) or lipid droplets (35) have not succeeded.
A potentially useful feature for identifying hepat­ ic cells and, perhaps, for rapidly separating them as multicellular clumps is their tendency to form ag­ gregates of similar cells when maintained in suspen­ sion cultures. Many investigators have observed the reaggregation of dispersed hepatocytes in culture with the formation of intercellular attachment complexes (17, 19). Other studics have demonstrat­ ed that other hepatic cells, including biliary ductu­ lar cells, endothelial cells, and Kupffer cells, reag­ gregate in suspension culture (36).
IV. Dispersion of liver cells
The development of methods to enzymatically separate liver tissue into suspensions of viable sin­ gle cells has been a signal technological accomp­ lishment (37,38). U sing now standard techniques of liver perfusion with solutions of collagenase, it is possible routinely to separate the lobular paren­ chyma into suspensions of single cells that are im­ permeable to trypan blue and that retain for at least short periods a high level offunctional integrity (39, 40). It is possible to harvest selectively the cells of the lobular parenchyma by controlled collagenase perfusion. The structures of the larger portal tracts and of the larger hepatic veins are only poorly released from their dense stromal matrix by the usual techniques of collagenase perfusion, and they are left behind in the undigested residue. In con­ trast, cells of the lobular parenchyma are more readily dispersed since they lack a dense investment of collagenous connective tissue. Recent studies from several laboratories have shown that it is pos­ sible to harvest the cells ofthe lobular parenchymal compartment virtually quantitatively (28). By ap­ plying selective proteolysis with pronase or trypsin, hepatocytes can be removed without rendering the nonparenchymal cells permeable to trypan blue (22, 23, 31). However, strong proteases must be used cautiously because they may damage surface membrane receptors, impeding further separation of cells or analysis of function.
The nonparenchymal cell popUlation remaining after removal of all hepatocytes from of a whole liver cell suspension contains K upffer cells, en­ dothelial cells, bile ductular epithelial cells, Ito cells, and various cells from blood (31, 35). By detecting cells that are histochemically positive for peroxidase or that can ingest 0.8 Mm latex spheres, about 15 to 20% of the nonparenchymal cell popu­ lation can be shown to represent Kupffer cells (22, 23). The remaining cells are frequently termed en­ dothelial cells (22), but the subpopulation is heterog­ enous. Histochemical detection of )'-glutamyl transpeptidase-positive cells shows that 5-15% of the nonparenchymal popUlation, excluding K upffer cells, represents biliary epithelial cells (Grisham, unpublished work).
V. Isolation of specific types of celIs from suspen­ sions
Several techniques are now available by which to separate the suspension of single cells of the hepatic parenchymal compartment into more or less hom­ ogenous populations. Separation techniques have generally been directed toward the separation of hepatocytes from non parenchymal cells, the sub­ fractionation of nonparenchymal cells into the var­ ious types of cells, and the subfractionation of hep­ atocytes into different classes. It is obvious that any separation technique must be nontoxic if cells are to be used for short-term functional studies or if they are to be propagated in culture. Tissue culture also requires that sterility be maintained during the sep­ aration procedures.
The separation methods most widely used today employ sedimentation of cells in gravitational fields. Separation of cells by sedimentation techniques is determined by differences in the size and/ or density of cells in the mixed cellular suspension. Isopycnic sedimentation utilizes density gradients and rela­ tively high centrifugal forces to band cells at a level of equal density in the gradient. Attempts have been made to separate hepatocytes and nonparenchymal cells by isopycnic sedimentation using gradients of Ficoll (41-43), Percoll (44), or Metrizamide (45, 46). In a few investigations, the objective has been to subfractionate hepatocytes (47) or nonparen­ chymal cells (22, 32). Velocity sedimentation de­ pends mainly on differences in cell size for separa­ tion and utilizes low centrifugal forces, in most instances for short periods of time. Commonly em­ ployed methods of velocity sedimentation are unit gravity sedimentation, isokinetic sedimentation, and elutriation. U nit gravity sedimentation has been widely used as a means to partially separate hepatocytes from nonparenchymal cells (48-50). Isokinetic sedimentation in albumin (5 I) or Ficoll (47,52) has been used to separate populations of hepatocytes with varying purity. Similar methods have been used to partially subfractionate nonpar­ enchymal cells (32). Iso kinetic sedimentation on Ficoll in a reorienting zonal rotor has been applied to the separation of altered hepatocytes and non­ parenchymal cells from livers of rats treated with the hepatocarcinogen diethylnitrosamine (53). A combination of velocity and isopycnic sedimenta­ tion methods, employing sedimentation at unit
27
gravity to remove most of the hepatocytes and sed­ imentation on a density gradient ofMetrizamide to separate non parenchymal cells, was used to isolate a subfraction of nonparenchymal cells, termed oval cells, from livers of rats that had been treated with the hepatocarcinogen ethionine(l8). Reverse phase elutriation has been used to effectively separate hepatocytes and nonparenchymal cells (54, 55), to subfractionate nonparenchymal cells (56), and to subfractionate hepatocytes (57). Separation of he­ patic cell suspensions has also been attempted by sedimentation on discontinuous gradients of Ficoll (32, 58) or by partitioning on a phasic system of dextran and polyethylene glycol (59), but these methods generally have not yielded satisfactory separation of mixed hepatic cells. Interesting at­ tempts to make different categories of hepatic cells more separable by stimulating the accumulation of selected cytoplasmic organelles or storage products have been reported. Selective hypertrophy of hep­ atocyte smooth endoplasmic reticulum and stimu­ lation of glycogen storage has been used to modify the size and density range of hepatocytes and there­ by influence their separability ( 17,60). Kupffer cells have been induced to phagocytize various materials with a similar objective in mind (22). Attempts to selectively remove K upffer cells that have ingested iron by passing hepatic suspensions through a magnetic field have shown this imaginative method to be inefficient (61).
In general, velocity sedimentation methods ap­ pear to more effectively separate hepatic cells than do isopycnic sedimentation methods. Separation of the much larger hepatocytes from small nonparen­ chymal cells is fairly efficient with velocity sedimen­ tation methods. However, clean separation of non­ parenchymal cells of nearly the same size and density is more difficult. Even though high degrees of purity of separation have been claimed for non­ parenchymal cells in some instances (22), the meth­ ods employed to distinguish different types of cells were inadequate to detect many types of cells in the nonparenchymal population. Most studies have clearly distinguished only the Kupffer cell subpopu­ lation and then assumed that the remainder of the nonparenchymal population was composed of en­ dothelial cells. 0 bviously other types of cells are known to be present and some of them can be identified by applying appropriate methods. Based on ploidy distribution, elutriation seems clearly to
28
be superior to isopycnic sedimentation on Ficoll gradients to subfractionate hepatocytes (17). Unit gravity sedimentation enjoys the advantage of sim­ plicity, since it does not require a centrifuge, and allows partial separation of the large hepatocytes from the small nonparenchymal cells. Since it af­ fords a cheap and easy way to enrich the fraction of hepatocytes in the hepatic cell suspension that is cultured, a unit gravity sedimentation step is incor­ porated in many protocols for establishing hepato­ cyte primary cultures, but complete separation is not achieved. Another disadvantage of unit gravity sedimentation is its slowness. Both velocity and isopycnic sedimentation methods require expensive centrifuges and rotors, but these techniques are rapid and can quickly separate large numbers of cells. Separation of the large, dense hepatocytes from small, usually less dense nonparenchymal cells by these methods appears to be quite good if not absolute. However, it seems unlikely that sin­ gle-step sedimentation techniques will ever be able to yield absolutely precise separations of different types of cells, especially cells in the nonparenchy­ mal population, which differ little in size and densi­ ty. Complete separation of such cells that are close­ ly similar in size and density will require methods that take advantage of distinct, nonoverlapping dif­ ferences in some property such as a surface enzyme or other antigen. The latter methods may usefully start with populations of cells that are partially separated by sedimentation methods.
More specific methods to purify various subpop­ ulations of hepatocytes and nonparenchymal cells are under investigation in many laboratories. These methods attempt to take advantage of differences in the surface properties of cells, in the content of proteins (including enzymes) and other antigens, and in the content of other macromolecules to yield more precise separation of cells. A particularly promising general method is the use of antibodies to selected cellular antigens to facilitate cellular separation by secondary means. Differential at­ tachment and release of cells from surfaces, either through nonspecific or poorly understood mechan­ isms or through specific binding of cellular recep­ tors to immobilized ligand, offer the possibility of precise separation of some types of cells in hepatic cell suspensions. Various cells in the nonparen­ chymal cell population may be enriched selectively by nonspecific attachment to surfaces of glass or
plastic. Attachment of Kupffer cells to glass or plastic surfaces allows these cells to be highly puri­ fied from other nonparenchymal cells (23). Specific attachment of cells by receptors to ligands immobil­ ized on surfaces is a potentially powerful method that has not been widely applied to the separation of different types of hepatic cells. Receptor-ligand combinations that may be used are limited only by the requirement that they be differentially localized to some ofthe various types of hepatic cells. Recep­ tors for sugar residues of glycoproteins provide a good example of the general attributes of a poten­ tially useful system. Hepatocytes have receptors for oligosaccharide chains that terminate in galactase residues, whereas sinusoidal endothelial cells and Kupffer cells contain specific receptors for complex sugars that terminate in N-acetyl glucosamine or mannose residues (26). Endothelial cells are sixfold more active in internalizing the latter residues than are Kupffer cells (26). Other types of hepatic cells have not been discerned to bind these sugars, al­ though this possibility needs more rigorous study. Hepatocytes bind avidly to polyacrylamide sur­ faces containing galactose via a Ca++-requiring reaction, and the bound cells may be released (62). If other hepatic cells lack receptors for galactose and man nose, it should be possible to separate hep­ atocytes and sinusoidal cells (endothelial and Kupffer cells) from other hepatic cells with consid­ erable precision by passing hepatic cell suspensions sequentially over surfaces to which galactose and man nose are bound. The number of receptors on hepatic cells is vast, and it is possible that other receptors are located differentially on various types of hepatic cells. For example, hepatocytes that con­ tain concanavalin-A binding sites have been isolat­ ed by binding to the ligand (63). Other cells of the nonparenchymal population may have cell specific receptors. Epithelial cells of terminal biliary ducts selectively bind certain steroids (64). For cells to be effectively separated by receptor-ligand binding they must be dispersed by techniques that preserve the integrity of receptor molecules. Proteases, in­ cluding pronase and those proteases that contami­ nate the relatively crude collagenase preparations used to disperse hepatic tissue, may damage or destroy receptors and prevent specific binding to ligands.
Fluorescence activated cell sorting may be used to separate cells that can be specifically tagged with
a fluorescent material. Hepatic cells have been sort­ ed into different population groups based on ploidy by using a fluorescent compound that binds stoi­ chiometrically to DNA (65). This technique separ­ ates diploid cells (all nonparenchymal cells and a fraction of the hepatocytes) from polyploid cells of other classes (several fractions of hepatocytes). Fluorescently tagged antibodies to cellular antigens that are located on the surface membrane of cells may also be used to separate these cells by fluores­ cence activated cell sorting. Epithelial cells from biliary ductules have been sorted after reacting them with a specific antibody to the membrane enzyme 'Y-glutamyl transpeptidase (66). An anti­ body to the hepatocyte galactose receptor protein, which has been isolated (67), might be used to sep­ arate hepatocytes. Antibodies, monoclonal and po­ Iyclonal, that bind specifically to unknown antigens of hepatocytes or bile duct epithelium have been reported (21, 32). Antibodies directed toward these and other hepatic cells could provide the basis for precise fractionation and separation of hepatic cells by fluorescence activated cell sorting. Fluorescence activated cell sorting is limited by the relatively small number of cells that can be economically separated by this method. Although sufficient cells can be readily sorted to establish tissue cultures, it is not economical today to use this technique to iso­ late cells for biochemical studies.
Separation of cells by electrophoresis has been advocated for many years, but apparently has not been recently used to attempt to separate hepatic cells.
A frequently overlooked and simple method to isolate specific cells is to plate suspensions as single cells and to selectively subculture colonies that have the desired morphological and functional proper­ ties (68). Selective culture media and conditions (biomatrix) show promise of allowing the propaga­ tion of specific types of cells, although these meth­ ods are still poorly defined.
VI. Long-term culture of isolated cells
Hepatocytes can be maintained in primary cul­ ture for several weeks, although their major func­ tions deteriorate markedly during this period of time (69). Changes in culture media allow the long­ term expression of some functional properties by
29
hepatocytes in primary cultures (70) but the hepat­ ocytes cannot be subcultured. The major impedi­ ment to subculturing hepatocytes has been their inability to divide in culture to form daughter cells. Recent reports indicate that under appropriate conditions hepatocytes may proliferate in culture (71, 72). Coupling of culture conditions that allow hepatocytes to cycle with those that facilitate the maintenance of critical functions may allow the eventual development of propagable (subcultur­ able) populations of hepatocytes and the estab­ lishment of clonally derived cultures of hepatocytes from single cells.
Several investigators have established long-term, propagable diploid cultures of liver cells that have so-called epithelial, fibroblastic, or macrophagic patterns of growth in vitro (reviewed in 3). When mass cultures are established from crude isolates of liver cells, it is impossible to precisely identify the origin of the cultured cells from a specific type of cell present in the intact liver tissue. However, the use of a pure population of isolated cells or the establishment of clonal culture populations from single isolated cells as starting material has allowed the clear demonstration that Kupffer cells and pre­ sumed bile ductular epithelial cells can be main­ tained in culture in vitro for relatively long periods with the maintenance of major functional capabili­ ties. Preliminary evidence suggests that endothelial cells may also be maintained in culture in pure populations. However, propagable cultures of Kupffer cells or endothelial cells that have been rigorously identified before cultures were estab­ lished apparently have not been reported.
Kupfter cells have been grown in liquid medium after isolation from a crude nonparenchymal cell population by adherence to glass or plastic (23). After plating, Kupffer cells flatten but retain a stel­ late shape (23). When Kupffer cells proliferate, they form clusters typical of other macrophages, but confluent sheets of cells are not formed (73). Cul­ tured cells are distinguished by their phagocytic capability, by the presence of Fe receptors, and by the high specific activity of lysosomal enzymes and peroxidase (23, 73). Kupffer cells have been main­ tained in culture for up to several weeks (73). Re­ cently a cell line isolated from liver several decades ago was shown to have functional characteristics of macrophages, presumably Kupffer cells (74).
Presumptive endothelial cells flatten and form
30
continuous sheets after attachment (Grisham, un­ published work). Ultrastructurally, they are simple and they lack distinctive morphologic features. Fen­ estrations have not been reported in endothelial cells that have been maintained in culture. Endothe­ lial cells fo{m type IV collagen in culture, but this does not distinguish them precisely from bile ductu­ lar cells (29). A recent study suggests that cultured endothelial cells do not contain Factor VIII (29). Lacking a specific marker for endothelial cells, such as Factor VIII, it is difficult at present to identify these cells once they have been cultured.
After plating, small epithelial cells that are not hepatocytes and that are presumed to be bile ductu­ lar epithelial cells flatten and attach to form a con­ tinuous sheet of cells joined by attachment com­ plexes (68). Because the origin of these cells from bile ductular epithelium is unproved, these cultured cells are best termed, noncommittally, hepatic epithelial cells. Cultured hepatic epithelial cells contain relatively high levels of I'-glutamyl trans­ peptidase (Grisham, unpublished work). They syn­ thesize collagen of types I, III, and IV (75) and they ha ve levels of receptors for epidermal growth factor (Earp and Grisham, unpublished work) and insulin (76) that are similar to the receptor levels found on isolated hepatocytes. Many propagable hepatic epithelial lines and strains possess a few hepatocyte­ like functions, but no single line possesses a large combination of functions reminiscent of mature hepatocytes. Cells from some individual lines and strains may synthesize and secrete one or more serum proteins, including albumin, transferrin, (X­
fetoprotein, and fibrinogen (reviewed in 3). Greatly increased fractions of biliary ductular cells, having properties similar to cells from normal livers, can be isolated from livers that are sites of oval cell reac­ tions (IS).
VII. Identification of cells in long-term cultures
The principles of identification of cells in long­ term cultures are similar to the principles of identi­ fication of cells in the liver in vivo or in freshly isolated suspensions; one can identify cells in long­ term cultures if it can be shown that they have properties that place them in one of the major types of cells in the liver in vivo and that exclude them from other types. Although it is not possible at
present to uneq uivocally identify all types of cells in long-term cultures, as we gain further detailed in­ formation on the differences in characteristics of the various types of liver cells such identification should become possible. Antibodies to specific types of hepatic cells would provide potentially the most precise and accurate method to identify cells in long-term culture.
At present, perhaps the most difficult hindrance to the precise and facile identification of cells in long-term cultures is the possible phenotypic alter­ ation (dedifferentiation) of cells under conditions of culture so that they no longer possess the identi­ fying characteristics ofthe cell of origin in vivo (77). Although recent observations on the macrophagic characteristics of a long established line of liver cells suggests that culture-associated changes in mor­ phology and function may not prevent identifica­ tion (74), opinions differ. This is a point of conten­ tion as regards the origin of lines of continuously propagable hepatic epithelial cells. Recognizable hepatocytes deteriorate structurally and functional­ ly during the first 7 -10 days in culture, and colonies of small, morphologically simple epithelial cells appear. It has been suggested that these colonies of simple epithelial cells arise from the altered hepato­ cytes (77). The analysis of this problem has been approached directly in two ways: by determining the effect of proteolytic destruction of hepatocytes by trypsin or pronase on formation of epithelial colonies in culture and by assessing the ability of different types of single liver cells to proliferate to form epithelial clones in vitro. The number of epithelial colonies formed when standard inocu­ lums of hepatic cell suspension are plated does not decrease even though all discernable hepatocytes have been destroyed (6S). Establishment of colonies by primary cloning from single cells isolated from collagenase-dispersed liver cell suspensions demon­ strated that hepatocytes never formed clones, while clones developed from small nonparenchymal cells (6S). Nonparenchymal clonogenic cells were6-S!lm in longest diameter and had scant, nongranular cytoplasm (6S). These results indicate that hepatic epithelial colonies can originate from a cell or cells in the non parenchymal cell popUlation and exclude the required involvement of mature or differentiated hepatocytes. Although these studies demonstrate that hepatocytes do not produce epithelial clones under the conditions of these experiments, they do
not exclude the possibility that mature hepatocytes may phenotypically modulate or retrodifferentiate under the artificial conditions of culture and yield epithelial clones that grow in long-term culture. Further studies are needed to corroborate or ex­ clude this possibility.
VIII. Suggested nomenclature for liver cells in long­ term culture
To avoid ambiguity, cells in long-term culture should not be given the same name as a type of cell in the liver unless they can be shown unequivocally to be derived from that £ell. Single cell cloning appears at present to be the only certain way to prove that a cultured cell population arises from a specific type of cell. Even this circumstance requires that the single cell be precisely identified if the cultured population is to be accurately traced to a specific type of cell in vivo. Unless the cell in ques­ tion possesses unique features that may be dis­ cerned without rendering it unculturable, it may not be possible to provide sufficiently strong evi­ dence to enable one to accurately trace a cultured cell population to a particular type of cell in the liver. Culturing of populations of cells, even popu­ lations that are greatly enriched in a single type of cell, cannot be used as unequivocal proof of the origin of the cultured population. No method of fractionation of single cell suspensions of liver yet has the precision required to exclude the presence of small numbers of different types of cells which may grow preferentially in vitro and yield a subcul­ turable population. Cell sorting on the basis of a unique cellular characteristic (i.e., a feature not shared by other hepatic cells) may eventually allow the establishment of pure cultures from specific types of hepatic cells. More experience is needed.
Cultured cells whose origin is not rigorously es­ tablished cIonogenically, but whose functional at­ tributes suggest their origin from a particular type of cell in the liver, should be designated as being like that particular cell, e.g., hepatocyte-like, Kupffer cell-like, etc. Cultured cells whose origin clonogeni­ cally is not certain and whose functional properties are unknown should be given a less committal de­ signation. For example, cultured cells that possess epithelial characteristics but lack a history or func­ tional qualities that would limit their origin from
31
either hepatocytes or bile ductular cells should be designated as hepatic epithelial cells. Epithelial cells cultured from liver are the source of perhaps the greatest terminological ambiguity. Although such cells are often loosely referred to as hepatocytes in the literature, there is no firm evidence at present that they may be derived from hepatocytes while it has been demonstrated that they can be derived from a small, non parenchymal cell that may be derived from the epithelium of bile ductules or some other source. The situation is made even more ambiguous because cultured hepatic epithelial cells that are rigorously shown not to be derived from hepatocytes may demonstrate limited hepatocyte­ like functional characteristics in vitro. Although there is suggestive evidence that bile ductular epithelial cells and hepatocytes may undergo phe­ notypic interchanges under some normal and patho­ logical conditions in vivo, and in tissue culture, these potential interconversions are as yet obscure. Until such potential cellular interchanges are un­ equivocally documented and their mechanisms bet­ ter discerned, it is best not to refer to any hepatic epithelial cell in long-term culture as a hepatocyte. To do so leads at this time to ambiguity and confu­ sion.
Some hepatic cells may grow in culture as coher­ ent sheets, but cannot be determined unequivocally to be either epithelial or endothelial. Such cultured cells should be given some no~specific designation that indicates their origin from liver and, perhaps, describes a major attribute, but is noncommittal about their derivation from a specific type of cell in the liver. The designation hepatic clear cells can be used for this purpose.
IX. Conclusions
The investigation of cultured hepatic cells has already led to many new insights on the cellular functions of the liver and improved the accuracy and precision of previous observations from in vivo studies. In order to maximize the utility of the observations made, investigators who study hepatic cells in vitro need to be as precise as possible about the types of hepatic cells that are being examined in vitro and about the need not to confuse or obscure the functions of different types of hepatic cells by oversimplifying their correlation in vivo and in vi-
32
tro. The liver is a cytologically complex tissue that should not be analyzed as if the hepatocyte were the sole type of cell present, and, indeed, hepatocytes themselves cannot be treated as a homogenous population. Separation and culture of hepatic cells has already facilitated the development of new knowledge on the contributions to integrated he­ patic functions by different types of hepatic cells, and additional insights on cellular functions, can be confidently predicted.
Studies on cultured epithelial cells from liver - both he