Vasopressin Inhibits Type-I Collagen and Albumin Gene Expression in Primary Cultures of Adult Rat Hepatocytes*

The mechanisms that regulate collagen gene expres- sion in hepatic cells are poorly understood. Accelerated Ca2+ fluxes are associated with inhibiting collagen synthesis selectively in human fibroblasts (Flaherty, M., and Chojkier, M. (1986) J. Biol. Chem. 261, 12060- 12065). In suspension cultures of isolated hepatocytes, the Ca2+ agonist vasopressin increases cytosolic levels of free Ca2+ (Thomas, A. P., Marks, J. S., Coil, K. E., and Williamson, J. R. (1983) J. Biol. Chem. 258, 5716-5725). However, whether vasopressin’s inter- actions with plasma membrane VI receptors attenuate hepatic collagen production is unknown. We investi- gated this problem by studying vasopressin’s effects on collagen synthesis and Ca2+ efflux in long-term primary cultures of differentiated and proliferation-com- petent adult rat hepatocytes. Twelve-day-old quiescent cultures were exposed to test substances and labeled with [5-3H]proline. Determinations of radioactivity in collagenase-sensitive and collagenase-resistant pro- teins were used to calculate the relative levels of collagen production. Synthetic [8-arg]vasopressin stimu- lated 46Ca2’ efflux within 1 min and inhibited hepatocyte collagen production within 3 h

5716-5725). However, whether vasopressin's interactions with plasma membrane VI receptors attenuate hepatic collagen production is unknown. We investigated this problem by studying vasopressin's effects on collagen synthesis and Ca2+ efflux in long-term primary cultures of differentiated and proliferation-competent adult rat hepatocytes. Twelve-day-old quiescent cultures were exposed to test substances and labeled with [5-3H]proline. Determinations of radioactivity in collagenase-sensitive and collagenase-resistant proteins were used to calculate the relative levels of collagen production. Synthetic [8-arg]vasopressin stimulated 46Ca2' efflux within 1 min and inhibited hepatocyte collagen production within 3 h by 50%; overall rates of protein synthesis were not affected significantly. In cultures labeled with [35S]methionine, vasopressin also decreased the levels of newly synthesized and secreted albumin, but not fibrinogen, detected in specific immunoprecipitates analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Northern blot analyses using specific [32P]cDNA probes revealed 70% decreases in hybridizable levels of collagen al(1) mRNA in hepatocyte cultures treated with either vasopressin or Ca2+ ionophore A23187; hybridizable levels of albumin mRNA also fell -50% following vasopressin treatment. Vasopressin did not affect collagen production in quiescent cultures of mouse Swiss 3T3, human myofibroblast or rat smooth muscle cells; and hepatocyte collagen production was unaffected by treatment with glucagon or dibutyryl CAMP. Thus, accelerated Ca2+ fluxes induced by vasopressin are associated with decreased produc- tion of hepatocyte collagen and albumin in primary cultures that simulate quiescent adult rat liver.
Hepatic collagen production and collagen content are markedly increased in patients with liver cirrhosis (1, 2). Thus, considerable effort has been devoted toward identifying hepatocellular and biochemical mechanisms that might normally regulate the synthesis and degradation of collagens associated with interstitial and basal laminar elements (3). Since it has been difficult to obtain normal human liver tissue systematically for in vitro study, many investigators have resorted to using short-term primary cultures of adult rat hepatocytes hoping to simplify biochemical and morphological studies of the regulation of collagen production in normal hepatic cells (for review, see Ref. 4). However, conflicting results have been obtained (these various systems employ different culture conditions), and it has been concluded that hepatocytes (5-8), hepatic lipocytes (It0 cells) (9-lo), or hepatic endothelial cells (11) are the principal collagen-producing cells in these cultures. Such differences have complicated interpretation of results of regulatory studies, for example those where corticosteroids inhibited collagen production (12,13).
In this report, we approached the problem of the regulation of hepatic collagen production by using a long-term primary culture system of well-differentiated and proliferation-competent adult rat hepatocytes (14). Our work is derived from recent observations linking Ca2+ ionophore and cholecystokinin-accelerated Ca2+ fluxes with selective inhibition of collagen synthesis in cultured human fibroblasts (15), and from reported increases in the levels of cytosolic-free Ca2+ and inositol triphosphate that are induced by vasopressin in suspensions of freshly isolated hepatocytes (16)(17)(18). Since little is known about the effects of vasopressin in the liver, the function(s) of the hepatocyte VI vasopressin receptor, or the role of Ca2+ fluxes in hepatic cellular physiology (19,20), we wondered if vasopressin might attenuate hepatic differentiated function. In this study, we demonstrate that vasopressin decreases Type I collagen and albumin synthesis and the mRNA levels for both of these proteins in long-term hepatocyte cultures.

9583
of medium) into untreated 35-mm Falcon plastic tissue culture dishes and cultured at 37 "C under standard conditions without media changes in humidified 10% C02/air incubators. At the time of plating, arginine-free media (DVMEM)' were supplemented with pretested heat-inactivated 15% dialyzed, heat inactivated FBS, L-ornithine (0.4 mM), and 10 pg each of porcine insulin, inosine, and hydrocortisonesuccinate/ml. Mouse fibroblasts, myofibroblasts, and muscle cells were cultured in standard DVMEM supplemented with 10% FBS.
Radiolabeling of Cells-Quiescent well-differentiated 10-13-day-old primary hepatic cell cultures were used for all biochemical measurements (23,24). The cells were shifted ("zero time") from their spent media into 1-2 ml of fresh arginine-free DVMEM supplemented with 0.2 mM Na+ ascorbate and 0.1 mM L-proline. Depending upon the experiment, test substances, including 15% dialyzed, heat inactivated FBS, vasopressin, A23187, porcine glucagon, or dibutyryl-cyclic AMP also were added at zero time (see legends to figures). For protein synthesis measurements, the cultures were labeled with [5-3H]proline, [35S]methionine, or a mixture of either [5-3H]proline plus [14C]ornithine, or ["C]proline plus [2,3-3H]ornithine starting at 15 min until 3-6 h after the fluid change. Protein synthesis rates were linear during this time. All isotopes were purified prior to use over a Dowex AG50W-X8 ion-exchange column (1 X 45 cm) (25). Swiss 3T3, myofibroblast, and smooth muscle cells were grown to confluence in P-100 culture dishes (Falcon0) at 37 "C in a humidified 5% CO,, 95% air incubator, and fluid changed into 10 ml of standard DVMEM, supplemented with 10% FBS and 0.1 mM L-proline. Labeling conditions for these cultures were similar to those described above.
Measurement of Collagen Production-The production of collagen and noncollagen protein was determined by incubation with purified bacterial collagenase (26) as described previously (15,27). Radioactivity in collagenase-sensitive and collagenase-insensitive labeled proteins was used to calculate the relative rates of collagen production. The levels of radioactivity in collagenase-sensitive proteins released into culture fluids were used to quantify procollagen "secretion" (15,27).
Collagen Purification-Standard purification procedures were followed (27). Soluble proteins were precipitated with 200 mg of [NH,J2S04/ml, pH 8.0, and centrifuged at 23,400 X g for 1 h at 4 'C. The precipitates were collected, dialyzed against 0.5 M acetic acid, and incubated with 10 pg of pepsin/ml at 4 "C for 20 h with stirring. The resulting suspensions were centrifuged at 30,000 X g for 1 h at 4 "C, and the supernatants adjusted to pH 7.5 with 2 N NaOH. The [NH,],SO, precipitation step was repeated twice (for culture samples derived from individual treatments) and pooled precipitates were washed with 70% ethanol and dissolved in a buffer containing 50 mM Tris, pH 7.5, 0.1 M NaCl, 0.3 mM phenylmethylsulfonyl fluoride, and 0.25 mM [-aminocaproic acid. Radioactive collagen from culture fluids or from culture fluids plus cell layers was purified as described above The abbreviations used are: DVMEM, Dulbecco/Vogt's modified Eagles medium; FBS, fetal bovine serum; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; COLL, collagen; ALB, albumin; kb, kilobases. using 1-2 mg of either purified rat collagen (obtained from 12-dayold primary culture fluids) or calf skin collagen (Behring Diagnostics) as carrier.
Determination of Radiolabeled Albumin and Fibrinogen-Twelveday-old primary hepatocyte cultures were incubated with [35S]methionine in fresh methionine-free plating media under standard conditions and culture fluids harvested and centrifuged 3-4 h later (23,31). In some experiments, albumin was purified from these fluids using an Affi-Gel Blue column as described previously (25). Radiolabeled albumin and fibrinogen secreted into culture fluids were recovered by immunoprecipitation using rabbit anti-rat albumin (Cappel) or rabbit anti-human fibrinogen (DAKO) antibodies, respectively. Immune complexes were precipitated with Protein-A (25,32), washed, and centrifuged four times, and analyzed by SDS-PAGE and autoradiography.
Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis-SDS-PAGE and gel staining with Coomassie Brilliant Blue were performed according to Laemmli (33) as described previously (27). When necessary, gels were dried under a vacuum and exposed to X-Omat film (Eastman) at -70 "C. Intensifying screens or flashed film (fluorogram (30)) were used as needed.
Hepatocyte Contribution to Hepatic Collagen Production-This calculation was made with a previously validated method used for in vivo studies of hepatic collagen synthesis in intact rats (25,32). After primary rat hepatic cell cultures were incubated with the isotope mixture containing [3H]proline and [14C]ornithine, the relative contribution of hepatocytes to the total collagen production in these cultures was determined by comparing the ratio of [3H]proline/[14C] arginine in hepatic collagen to the same ratio in albumin. We calculated the predicted [3H]proline radioactivity in hepatic collagen produced by hepatocytes, as reported previously (25).
The relative contribution of hepatocytes and nonparenchymal cells (NPC) to hepatic collagen production was calculated as reported previously (25).

Hepatic [3H]ProCoLL
The NPC collagen production was determined as follows: Amino Acid Analysis-Low molecular weight fractions of cell culture fluid supernatant (obtained following fractionation with 66% ethanol) were evaporated to dryness, dissolved in distilled water, and chromatographed over Bio-Gel P2. The retained material was eluted and the resulting fractions used for amino acid analysis. Proteins precipitated in 66% ethanol or proteins recovered from purification or immunoprecipitation procedures described above were hydrolyzed in 6 N HCl at 120 'C for 3 h and evaporated to dryness. Amino acid analyses were performed as described previously (25,32). Radioactivity was measured in aqueous scintillant using a Mark 111 counter; automatic corrections were applied for spillover of I4C radioactivity into the 3H channel and for respective counting efficiencies.
Determination of 45Cuz+ Efflux-Twelve-day-old primary hepato-cyte cultures were incubated for 24 h with spent plating media (-1.5 ml/dish) in the presence of 33 pCi of ''CaCI, (15,37). The fluids were aspirated, the cultures were washed three times with 2 ml of fresh plating media, and shifted (time 0) into 2 ml of similar media without or supplemented with 100 nM [8-arg]vasopressin. Fifty pl of culture fluids were sampled from cultures in both groups at varying times thereafter. Northern Blotting-Primary hepatocyte cultures were generated and incubated, as described above, with or without test substances. Cells were processed as described previously (27), and total cellular RNA isolated by standard procedures (38). For treatment of cell cultures with Pronase, culture fluids were removed and cells washed twice with 50 mM Tris, 0.15 M NaC1, pH 7.5. Pronase (0.25% w/v) was incubated in 0.15 M NaCl at 37 "C for 1 h, and 1 ml was added to each dish. Cells were treated with the Pronase solution for 7 min, and then washed with 50 mM Tris, 0.15 M NaCl, pH 7.5, at 4 "C. The resulting RNA was chromatographed over oligo(dT) columns (30), and the recovered poly(A)+ RNA was used for Northern blotting as described previously (27, 39). Autoradiograms were made with intensifying screens and quantified by a scanning laser densitometer interfaced with an integrator.
The plasmid pHF677 (40) containing cDNA encoding the human al(1) collagen gene was provided by F. Ramirez (SUNY Health Science Center, Brooklyn, NY). The plasmid LK 280 containing cDNA encoding the rat / 3 actin gene was provided by L. Kedes (Stanford University, Palo Alto, CA). The plasmid malb2 containing cDNA encoding the mouse albumin gene was provided by S. Tilghman (Princeton University, Princeton, NJ) (41). Plasmid DNA was purified by alkali lysis and CsCl gradient centrifugation as described previously (39). The cDNA inserts were purified by digestion with the appropriate restriction endonucleases (EcoRI for pHF677, Hind111 for malb2, and PuuII for LK 280), electrophoresis on a low melting point agarose gel, and elution on an anion exchange column (Elutip). The cDNA fragments were radiolabeled with [a-"PIdCTP using the random primer synthesis method (42). The labeled cDNA probes were separated from dCTP by centrifugation through a G-50 spin column. The specific activity of the cDNA probes were approx- Statistical Analysis-Unless noted, results were expressed as means f S.E. Student's t test was used to evaluate the difference of the means between groups, acceptingp < 0.05 as significant (43).

Collagen Production in Primary Hepatocyte
Cultures-Collagen production was measured in 12-day-old proliferationcompetent cultures composed of hepatocytes (-80%) and nonparenchymal cells (-20%). Such cultures display a broad spectrum of hepatocyte-specific functions (44). During a 24h interval following fluid change, the relative level of collagen to total protein production was 2-3% (see control experiments in Table I). Within this time frame, about 70% of newly synthesized procollagen was secreted; the relative level of collagen to total protein in the culture fluids was 4-6%. More than 95% of the preexisting and newly synthesized collagen had the electrophoretic characteristics of al(1) and a2(I) chains as shown in stained reducing SDS-gels (Fig. lA) following fluorography (Fig. 1B). Gel chromatography of this purified collagen demonstrated (Fig. 2) that the preexisting (major peak of proteins eluting at 90 ml) and newly synthesized material, which coeluted with authentic Type-I collagen, were also composed solely of al(1) and a2(I) subunits (Fig. 2, inset). The relative contributions of hepatocytes and nonparenchymal cells to collagen production were estimated in 12-dayold cultures using two different approaches. First, a method was employed that involves amino acid analysis of newly synthesized proteins in cultures dual-labeled with [3H]proline and ["Clornithine. The results are shown in Fig. 3.
The rationale underlying this approach was based upon the presence in hepatocytes and absence in nonparenchymal cells of arginine biosynthetic capacity (principally due to the hepatocyte-specific function, ornithine transcarbamylase [EC  25,45,46)) in these cultures. When labeled proline and ornithine are presented to hepatic cells all of the cells incorporate proline into proteins, but only hepatocytes generate proteins containing labeled arginine derived from ornithine via the urea cycle (21, 25,45,46). Four h after the fluid change, total secreted proteins (panel A ) including albumin (panel B ) , contained -80% of the 14C radioactivity as arginine. Most of the 14C radioactivity remaining in the culture fluid amino acids (5.3 x lo5 dpm) was found in proline and in other related amino acids (94.8%). Negligible amounts of 14C radioactivity remaining in the culture fluid as amino acids were identified as arginine (0.6%) and ornithine (4.6%). These observations indicated that the cellular uptake of ["C] ornithine was rapid and nearly quantitative; and, that among [ "Clornithine's metabolites, proline alone was significantly exported into the culture medium.
Forty-five percent of the 14C in collagen purified from these cultures was found as ["Clarginine (Fig. 3C). The collagen [I4C]arginine radioactivity was -70% of the level expected were collagen produced exclusively by hepatocytes (25). In contrast, when human fibroblast cultures were dual-labeled under similar conditions there was, as expected (45), a negligible amount of "C radioactivity present in arginine in hydrolyzed fibroblast proteins (Fig. 3 0 ) . Since [14C]arginine produced by hepatocytes is unlikely to have been transferred to nonparenchymal cells (47), these experiments suggested, based upon previous analyses (251, that more than two-thirds of the newly synthesized collagen in this in vitro system was produced by hepatocytes. Evidence supporting this conclusion was obtained from a second type of experiment, in which Pronase was used to selectively destroy cultured hepatocytes (48,49) FIG. 2. Subunit structure of highly purified collagen synthesized by primary cultures of adult rat hepatocytes. Preexisting and newly synthesized collagens were obtained and purified from media of 12-day-old cultures as described in Fig. 1 (panel B ) .
Collagens were fractionated by gel chromatography as described under "Experimental Procedures." Conductivity and absorbance were determined for each fraction. The arrow indicates the elution position of collagen Type I (90 ml). Proteins eluting in this peak were dialyzed, precipitated with 66% ethanol, and dissolved in 0.2 N NaCl, 1 mM phenylmethylsulfonyl fluoride, 50 mM Tris, pH 7.5. The resulting material was analyzed by SDS-PAGE and gel staining (inset) as described in Fig. 1.  Hydrolyzates of precipitated proteins (0.2 N sodium citrate, 0.1% Brij 35, pH 2.2) were obtained as described under "Experimental Procedures," applied to a 1 X 45-cm AG-50 column, and eluted (1 ml/min) at 50 "C with 120 ml of 0.7 N sodium citrate, 0.1% Brij 35, pH 8.6, and then with 0.2 N NaOH. Four-ml fractions were collected. Elution of authentic proline, ornithine, and arginine standards is indicated (arrows). Recoveries of standards of radioactive amino acids on this column were quantitative. Radioactivity was measured by liquid scintillation spectroscopy. B, newly synthesized albumin secreted into the medium during the 24-h labeling interval was purified from 12day-old hepatocyte cultures by Affi-Gel Blue chromatography as described under "Experimental Procedures." Hydrolyzates were obtained and analyzed as described in A . C, newly synthesized collagen secreted into the medium during the 24-h labeling interval was purified from 12-day-old hepatocyte cultures using carrier rat collagen from 12-day-old primary rat hepatocyte cultures as described under "Experimental Procedures." Hydrolyzates were obtained and ana- Northern blots. The cultured cells contained, before Pronase treatment, 5.5and 2.2-kb species of collagen d(1) and albumin mRNA, respectively (Fig.  4, lane l ) . After Pronase Bound RNA was hybridized to RzP-labeled cDNA plasmids (human collagen al(1) and mouse albumin) as described under "Experimental Procedures." The filters were exposed to x-ray film at -70 "C with an intensifying screen. Lanes 1-3 show RNA from untreated cells (undiluted, diluted 1:16, diluted 1:64, respectively); RNA from Pronase-treated cultures, obtained as described under "Experimental Procedures," is shown in lane 4. treatment, neither collagen mRNA nor albumin mRNA were detected (Fig. 4, lane 4 ) , although quantitative yields (-20% of the total RNA obtained from untreated cultures (10-40 pg RNA/106 cells)) of intact 18 S and 23 S RNA were recovered together with a strong B actin signal, the internal cDNA hybridization control used in the Northern blot studies (not shown). The lack of albumin mRNA signal (Fig. 4, lane 4 ) indicated effective Pronase digestion of hepatocytes, since albumin mRNA was negligible or not detectable in dilutions of total RNA recovered from untreated cultures (%6 and Vm), (Fig. 4, lanes 2 and 3, respectively).
Thus, the results from both dual-labeling and RNA blotting indicated that hepatocytes generated the bulk of the Type I collagen under the conditions of long-term culture that were used in these experiments.
Effects of Vasopressin on Protein Synthesis and 4sCa2+ Efflux-Vasopressin inhibited collagen production in a dosedependent fashion in 12-day-old primary hepatocyte cultures (Table I). Significant inhibitory effects were observed at 10 nM peptide. This response was obtained with either synthetic pharmaceutical preparations, a 99% pure octapeptide, or synthetic [8-arg]vasopressin. The average maximal inhibitory effect was about 50% (IDso -80-100 nM); only small (10-20%) and generally insignificant effects on noncollagen protein production were observed ( Table I)   b.e Determined as described in Table I. vasopressin. primary hepatocyte cultures, Table I1 shows that vasopressin did not attenuate collagen production in confluent cultures of Swiss mouse 3T3 fibroblasts, human myofibroblasts, or rat smooth muscle cells.
Vasopressin's inhibitory effects on cultured hepatocyte protein synthesis were selective, but not entirely specific, as the levels of newly synthesized and secreted albumin were also decreased 50% after 4 h of exposure to 100 nM peptide (Fig.  6A) whereas the levels of newly synthesized and secreted fibrinogen, another major hepatocyte-specific plasma protein, were unaffected (Fig. 6B).
Effects of A23187, Glucagon, and Dibuty&cAMP on Collugen Production-calcium ionophore A23187 (0.6 PM) decreased the production of newly synthesized collagen in primary hepatocyte cultures (Table 111). Similar results were obtained with cultured smooth muscle cells and myofibroblasts (data not shown) and with human fibroblasts (15). The ionophore's inhibitory effects were detected after 2 h of incubation and persisted for at least 6 h after contact with cells; full attenuation required exposure up to 24 h (not shown). Unlike vasopressin, the ionophore exerted more of a nonspe-

B )
were recovered from culture fluids specifically by immunoprecipitation using rabbit anti-rat albumin or rabbit anti-human fibrinogen antibodies, respectively. Immune complexes were precipitated with Protein-A, washed, and centrifuged. Immunoprecipitates were pretreated with 100 mM dithiothreitol and analyzed by SDS-PAGE and the resulting gels subjected to autoradiography with intensifying screens as described under "Experimental Procedures." The specificity of the immunoprecipitation reactions was indicated by the failure to immunoprecipitate similar radiolabeled bands when culture fluids were incubated under similar conditions but with preimmune sera (data not shown). *.' Determined as described in Table I.  Twelve-day-old primary rat hepatocyte cultures were incubated for 3 h with no additions or in the presence of glucagon or CAMP, as described under "Experimental Procedures." b.c Determined as described in Table I. cific "biphasic" effect, particularly at 2 h, when a significant reduction in noncollagen protein was observed that was somewhat dissipated by 6 h. In contrast to vasopressin and A23187, neither 10-100 nM glucagon (a CAMP-generating dose range in this system (53)) nor 1 PM dibutyryl CAMP affected collagen production significantly in 12-day-old hepatocyte cultures (Table IV).

Effects of Vasopressin on Collagen and Albumin mRNA
LeueLs-Twelve-day-old hepatocyte cultures were incubated with 100 nM vasopressin; RNA was recovered from the cultures 3-4 h later and analyzed for hybridization to different "P-labeled cDNA probes. (No morphological differences were observed in either treated or mock-treated cultures prior to isolation of the RNA which, in both cases, was obtained in similar yield and free of degradation as analyzed on ethidium bromide stained gels.) Northern blots revealed (Fig. 7) that vasopressin-treated cells contained levels of 2.2-kb albumin and 5.5-kb collagen d ( 1 ) mRNA that, in comparison to mocktreated cells, were reduced by 50 and 70%, respectively. Similar attenuation also was seen in the level of a second putative al(1) mRNA species of -6 kb ( Fig. 7) and, again, for all three mRNA species in cultures treated with A23187 (data not shown). In contrast, the hybridization of "'P-labeled probes to 2-kb p actin mRNA was similar in its intensity when analyzed on Northern blots of RNA from vasopressin-treated and mock-treated cultures (data not shown).

DISCUSSION
Developmental, proliferative, and functional processes in animal cells are markedly dependent on extracellular matrices in contact with the cell surface (54). Recent reports of proliferative and differentiated properties of short-term proliferation competent primary cultures of adult rat hepatocytes, cultured on plastic substrata coated with differing combinations of matrix components, lend further support to this biological principle (55,56). As major components of various tissue matrices, collagenous proteins are being scrutinized for their structural properties and cellular attachment sites (57); the molecular mechanisms that regulate collagen production also are under intense investigation (58). In hepatic tissue, where collagen overproduction is associated intimately with liver cirrhosis, controversy prevails over which hepatic cell(s) synthesize and secrete collagen, the chemical type(s) of collagen produced, and the regulatory mechanisms involved. In this report, using well-differentiated long-term proliferationcompetent primary cultures of adult rat hepatocytes as a model system, we provide biochemical evidence from dual [3H]proline/['4C]ornithine labeling studies showing that hepatocytes synthesize and secrete Type I collagen. We also have found that vasopressin, an osmoregulatory nonapeptide (59) rapidly reduces the synthesis of both albumin and Type I collagen together with coordinate reductions of -50 and 70%, respectively, in the mRNA levels for both of these proteins. These observations suggest that vasopressin regulates hepatocyte gene expression.
Several findings reinforce this conclusion. First, in addition to the pharmaceutical preparations employed, the synthetic analog [8-arg]vasopressin (CYFQNCPRG) also reduced collagen and albumin synthesis and the mRNA levels of each of these proteins in these cultures. These results eliminate the possibility that nonspecific contaminants in the pharmaceutical preparations acted as the inducing agents. The in vitro requirement of supraphysiological levels of vasopressin (its basal plasma levels range between 2-7 PM, its plasma halflife ranges 17-35 min, and -33% of its clearance in rats is performed by the liver (60) has not yet been explained, but it is unlikely to be the consequence of long-term hepatocyte culture conditions since nanomolar levels of [8-arg]vasopressin also were required for the activation of the sodium pump and 45Ca2+ fluxes in suspension cultures of freshly isolated adult rat hepatocytes (17,19). Second, although it is arguable that vasopressin's affects are indirect via the peptide's interactions with nonparenchymal cell subpopulations present in hepatocyte cultures, this seems unlikely as vasopressin failed to elicit detectable effects on Type I collagen production in three different kinds of nonhepatic cell cultures, including smooth muscle cells, Swiss 3T3 fibroblasts and myofibroblasts, all of which harbor vasopressin receptors (50-52). Furthermore, the Pronase digestion of hepatocytes in primary cultures eliminated all of the hybridizable albumin and Type I collagen mRNA molecules detected in Northern blots, indicating that a Pronase-sensitive cell population was responsible for both albumin and most of the collagen production in these cultures. Third, vasopressin's ability to inhibit albumin synthesis and attenuate the levels of albumin mRNA reflected a specific effect on hepatocytes, as the synthesis of albumin in these cultures and in vivo is solely due to expression of this gene in hepatocytes. The failure of vasopressin to inhibit the synthesis of fibrinogen, another uniquely hepatocyte-specific function, further indicates that vasopressin's effects were selectively pleiotropic. These findings, plus morphological observations showing no obvious cellular alterations in vasopressin-treated cultures, imply that such effects on hepatocyte function did not arise from generalized cytotoxicity.
The use of the dual-labeling method in these in vitro studies, to identify the hepatocellular origin of a ubiquitous proteinlike Type I collagen, has been validated earlier by in vivo studies using adult rats (25); observations made with longterm cultures are entirely consistent with in vivo findings. Moreover, direct evidence from studies employing in situ hybridization (61) and immunohistochemistry combined either with light microscopic (62) or with electronmicroscopic (63, 64) analyses of liver tissue sections also indicated previously that hepatocytes synthesize Type I collagen in vivo.
Similar conclusions were reached from similar studies using short-term primary cultures of adult rat hepatocytes (65,66). Taken together, the combined results clearly indicate that hepatocytes synthesize Type I collagen in vivo and in vitro.
Other investigators found that contamination with hepatic lipocytes ("1t0" cells) are responsible for the increased collagen production in primary hepatic cell cultures (from -2.5 to -9% relative rates) under some culture conditions (10). These short-term cultures, however, display sharp differences in the behavior of the recovered cell populations that eventually survive during the first week postplating. This is not surprising since fundamental differences exist between the two primary systems with respect to the conditions of hepatocyte isolation, plating, and culturing. In the short-term cultures, although special measures are used to "purify" the initial cell suspension recovered following tissue dispersion and to coat the plastic culture dishes with Type I collagen, the plated hepatocytes eventually deteriorate and are overgrown by nonparenchymal contaminants. Moreover, in our studies, 12-dayold hepatocyte cultures produced collagen at rates that were similar (2% of total proteins) to those observed in "noncontaminated 8-day-old hepatocyte cultures following centrifugal elutriation and gradient separation (10). In addition, the proportion of [14C]arginine recovered in newly synthesized collagen was that expected if 70% of the collagen were produced by hepatocytes. We recognize that a potential pitfall of the dual isotopic labeling approach is that a fraction of the [14C]arginine produced by hepatocytes might be transported to nonparenchymal cells and incorporated into their proteins. Were this to occur to any significant extent, it would clearly complicate the interpretation of an hepatocellular origin of synthesis of an ubiquitous protein-like Type I collagen. However, since the amount of [14C]arginine present in the media after the 4-h incubation period used in the current studies was only 0.003% of the starting dpm ((3000/1.1 x 10' dpm [14C]ornithine)), it would seem unlikely that any significant amount of extracellular [14C]arginine (relative to ['Hlproline) could have entered the precursor pool of protein synthesis in nonparenchymal cells (where further reductions in arginine's specific activity would have occurred) and have been detected in newly synthesized protein (see also Ref. 42). Direct cell-tocell transfer of [I4C]arginine from hepatocytes to nonparenchymal cells also would seem unlikely in the absence of electronmicroscopically detectable intercellular junctions between hepatocytes and nonparenchymal cells in vitro.' Further experiments are needed to delineate the mechanism(s) and the causal relationships, if any, by which vasopressin lowers the synthesis and mRNA levels of Type I collagen and albumin in cultured hepatocytes. By itself, CAMP, although implicated earlier in the regulation of collagen synthesis in certain fibroblast systems (67), is apparently not involved, since neither dibutyryl-CAMP nor glucagon (a CAMP-generating peptide in these cultured hepatocytes (53)) affected collagen and albumin production.
The available evidence suggests involvement of one or more Ca'+-dependent processes, possibly including activation of protein kinase C (68): (i) vasopressin accelerates 4sCa2+ fluxes in this and in other adult rat hepatocyte systems (16,17), (ii) vasopressin's effects are reportedly additive with phorbol esters that rapidly activate hepatocyte Na+-K+ [ATPase] and =Rb+ flux (68), and (iii) calcium ionophore A23187 also inhibits collagen and albumin synthesis in these primary cultures and collagen synthesis in cultured fibroblasts (15).
Although the ionophore concentration that was used in the present studies bordered upon a level that suppresses initia-K. Dempo and H. L. Leffert, unpublished observations. tion of hepatocyte DNA synthesis in this system (76), its nonselective inhibitory effects on hepatocyte protein synthesis dissipated with time. Lastly, recent work indicates that collagenase (69) and albumin gene expression (70, 71) are regulated by a complex network of nuclear factors that bind specifically to regions of untranslated 5"control sequences in both genes. Whether vasopressin affects some of these transacting factors remains to be determined.
It is reasonable to ask if the inhibitory effects of vasopressin observed in this hepatocyte culture system are physiologically relevant in uiuo. While these cultures have demonstrated their physiological relevance in many instances, particularly when used to probe into the mechanisms controlling liver regeneration (21, 31, 44, 49), thus far there is no evidence that liver collagen and albumin gene expression might be regulated by vasopressin. On the other hand, there is mounting evidence that various stressful conditions clearly alter hepatocyte gene expression. For example, malnutrition (30), inflammation (72), and exposure to tumor necrosis factor (Y (73) inhibit hepatic albumin synthesis and alter the distribution of hepatocyte production of many of the acute-phase proteins (74). Further studies of hepatic vasopressin levels and receptor function in various physiological states and in various animal models, like vasopressin-deficient Brattleboro rats (75), combined with primary culture studies of hepatocytes from such animals, may provide new insights into the regulation of hepatic gene expression and protein synthesis in normal and diseased liver.