Retinol and Extracellular Collagen Matrices Modulate Hepatic Ito Cell

THEJOURNALOF BIOLOGICAL CHEMISTRY 01987 by The American Society of Biological Chemists, Inc

Vol. 262, No. 21, Issue of July 25, pp. 10280-10286,1987 Printed in U.S.A.

Retinol and ExtracellularCollagen Matrices Modulate Hepatic Ito Cell Collagen Phenotype and CellularRetinol Binding Protein Levels* (Received for publication, December 15, 1986)

Bernard H. Davis$, Bruce M. Pratt, and JosephA. Madri From the Liver Center, Departments of Medicine and Pathology, Yale University, Schoolof Medicine, New Haven, Connecticut 06510

The hepatic vitamin A-storing Ito cell has been im- (e.g.tissue transglutaminase, collagen, laminin, and keratins) plicated as a causative cell in hepatic fibrogenesis. (4-8). Ito cells may intrinsically have the capacity toproduce Using a modification of a recent method (Friedman, S. significant quantities of collagen matrix (9, 10). In vitro they L., Roll, F. J., Boyles, J., and Bissell, D. M. (1985) produce more collagen (types I, 111, and IV) than any other Proc. Natl. Acad. Sei. U. S. A. 82,8681-8685), rat Ito hepatic cell, and morphologists have often speculated that Ito cells were isolated and passaged in vitro on collagen- cells “transform” into “myofibroblasts” and are responsible coated plastic dishes through cell generation 40-50. for the overproduction of collagen which characterizes the The collagen synthetic phenotype for Ito cells grown broad fibrous bands seen in cirrhosis (9, 11-14). When this on various extracellular matrices was demonstrated by “transformation” occurs, the Ito cell’s vitamin A-laden dropimmunofluorescence and quantitated by competition lets become less prominent and disappear (13). It is unclear enzyme-linked immunosorbent assays. When grown on whether thisloss of vitamin A is causally related to the fibrotic a type I collagen matrix, Ito cells produced type IV > process, but a recent morphologic study demonstrated that type I11 > type I collagen. When grown on a type IV when thehepatic fibrogenic agentcarbontetrachlorideis collagen matrix, the cells produced relatively equal given with vitamin A, hepatic necrosis develops in the rat in amounts of types I and I11 collagen. The absolute the apparentabsence of hepatic fibrosis (15). Ito cell vitamin amounts of type I collagen produced were greaterwhen A status may represent a controlling factor in terms of the cells were grownon type IV versus type I matrix. When cell’s collagen synthetic machinery. lo-’ M retinol was added to cell cultures, there was a During tissueinjury, however, the composition of the extrauniform increase in type I11 collagen regardless of cellular collagen matrix also changes dramatically(16-18). As matrix typebut a decrease in type I collagen when cells matrix effects on cell behavior have been observed in numerwere grown on a type IV matrix and a large increase ous cell systems,it is possible thatIto cell collagen and in type I collagen when cells were grown on a type I vitamin A metabolismmay also be alteredby the large changes collagen matrix. The levels of cellular retinol binding which occur in its surrounding matrix(19-21). protein, a keycytosolic retinol transport protein, were The current work begins to address some of the aforemenquantitated by high performance liquid chromatogra- tioned questions by growing rat hepatic Ito cells on various phy and compared for cells grown on type Iversus type collagen matrices in primary and secondary cultures. This IV collagen matrices. It was found that cells on a type tissue culture system affords the opportunity to study the I matrix contain4.96 f 2.8 times more cellular retinol interactions between collagen and vitaminA metabolism. The binding protein than do cells grown on a type IV ma- accumulated datasuggest that cultured Itocells maintain the trix. capacity toproduce the interstitial-typecollagens I and I11 as In conclusion, Ito cell collagen synthesis may be al- well as the basement membrane collagen type IV through cell tered by underlying extracellular matrix and exoge- generation 30-40. As in numerous othercell systems, Itocells nous retinol. This in vitroculture system should allow respond to changes in the extracellular matrix in a complex the studyof regulatory factors and possible therapeutic fashion (19-21). The current study hasbegun to characterize anti-fibrogenic mediators. this response by noting significant changes in collagen synthetic phenotype and in the cellular response to exogenous retinol whencells are grown on type I (interstitial) uersus type IV (basement membrane) collagen. While the cellular Retinoids in general exert profound effects on cell differ- mechanisms involved inretinolmodulationareunknown, entiation in a variety of systems (1).In uiuo, retinoids are several other studies have implicated a key role for cellular retinol binding protein (CRBP)’ in transporting cytoplasmic largely storedaspalmitatederivativesinthe liver in the hepatic Ito cell, a sinusoidal fat-storing cell in the space of retinol to thenucleus, its presumed siteof action (22). Using Disse (2, 3). It is unclear whether the retinyl palmitate storesa recently described HPLC method of quantitating CRBP levels in tissue culture, the present work has noted thatlevels affect Ito cell protein synthesis. In numerous other systems, retinoids increase or alter the production of various proteins of this protein can also be modified by changes in the extracellular matrix (3, 23). ~

* This work has been supported by United States Public Health Service Grants RO-1-HL 28373, AM-34989,and AM-07356.The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement’’ in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. $ To whom correspondence should be addressed Gastroenterology Section, Dept. of Medicine, Box 400,5841 S. Maryland Ave., University of Chicago, Chicago, IL 60637.

MATERIALSANDMETHODS

Animak-Sprague-Dawley female rats (retired breeders) were routinely used for cell isolation (Camm Research Lab Animals, Wayne NJ).

’

The abbreviations used are: CRBP, cellular retinol binding protein; HPLC, high performance liquid chromatography; PBS, phosphate-buffered saline; ELISA, enzyme-linked immunosorbent assay.

10280

10281

Hepatic It0 Cell Culture Chemicals-Retinol acetate (Sigma) was used for intraperitoneal injections. Trans-retinol (Sigma) and retinoic acid (Sigma) were used for HPLC studies of CRBP. [3H]Retinol (Amersham Corp.) used in HPLC studies of CRBP had a specific activity of 60 Ci/mmol. Cell Isolation-Hepatic Ito cells were isolated using a modification of the method of Friedman et al. (9). Rats received 100,000 IU of retinol acetate intraperitoneallythree times/week for 10-14 days. For cell isolation, portal vein perfusion was begun after heparin administration. Initially, phosphate-buffered saline (PBS) at 37 "C was perfused a t 10 ml/min to adequately blanch the liver. This was followed by 0.1% Pronase (B Grade, Behring Diagnostics and American Hoechst Corp., La Jolla, CA) for 3-4 min andthen 0.03% collagenase (Cooper-4197,Worthington) for 30 min with reperfusion a t 5 ml/min of the latterfrom the inferior vena cava (with continuous 5% C 0 2 instillation to perfusate). The liver was subsequently removed, minced in 0.02% Pronase, and incubated in a shaking 37 "C water bath for 20-30 min with deoxyribonuclease (10 pg/ml, Sigma). The mixture was then centrifuged a t 50 X g for 2 min to remove dead hepatocytes and undigested debris. The supernatant containing hepatic nonparenchymal cells was washed four times in PBS and then layered over a 25% preformed (30,000 X g for 15 min) Percoll (Pharmacia P-L Biochemicals) gradient and centrifuged at 800 X g for 30 min. This gradient produced a top layer of yellowish white oily debris with a band of cells immediately beneath which contained It0 cells. There were Ito cells, identified as lipid droplet-laden cells which demonstrated rapidly fading autofluorescence at 325 nm under the fluorescent microscope, at lower portions of the gradient as well. However,when only the top cell band wasremoved, there was routinely 95% using trypan blue exclusion. The Ito cell band was then washed two times and placed on a 45% unformed Percoll gradient and centrifuged a t 15,000 X g for 20 min. This latter centrifugation removed potential bacterial contamination, leaving the cell band at thetop of the tube. The cells were then washed two times in PBS and resuspended in Dulbecco's minimal essential medium supplemented with 10% fetal bovine serum and plated at 0.7-1.0 X IO6 cells/35-mm tissue culture dish (Falcon Labware, Oxnard, CA). This culture mediumwas routinely used throughout primary and secondary cultures. Cell Culture-Primary cultures were routinely started on Falcon 35-mm plastic dishes (No. 1008)whichwere coated with calf or human type I collagen as previously described (20, 24). In certain select experiments, cells were subcultured on 35-mm Falcon plastic dishes (No. 1008) coated with type IV collagen or fibronectin. The detailed method of this coating has recently been described (25). In general, the primary and passaged cells adhered poorly to uncoated 35-mm tissue culture plastic (Corning No.25000, Corning Glass Works, Corning, NY or Falcon No.3001).However, as previously noted by Friedman et al. (9) with primary Ito cells grown on plastic (Lux, Miles Scientific, Napenrille, IL), passaged Ito cells grown on tissue culture plastic (Corning or Falcon) did produce collagen types I, 111, and IV (see "Results"). When primary cultures reached confluency, the cells were passaged at a 1:4 split ratio. Using this split ratio consistently, an approximation of cell generation could be made. For example, passage 2 would include cells which had undergone approximately eight cell divisions (referred to as eight cell generations). Cells were generally studiedthrough passage 4-5 (or cell generation 40-50). For passaging, cells were released from the corresponding dish with brief exposure to 0.2% trypsin (Sigma), 0.01% EDTA (Sigma). Passaged cells were routinely maintained on Falcon bacteriologic plastic 35-mm dishes or Corning 25-cm2 dishes coated with type I calf collagen (1.0 pg/dish, as determined by ELISA). In some experiments, cell cultures (either primary or secondary) were begun on dishes coated with a type I/type I11 human collagen mixture (1.3 and 1.1 pg/dish, respectively), Engelbreth-Holm-Swarm-produced type IV collagen (1.2 pgldish), Engelbreth-Holm-Swarm-produced laminin (0.63 pg/dish), or fibronectin (0.8 pg/dish). Freshly isolated Ito cells are preloaded with vitamin A, and the available CRBP is probably bound to cellular retinol (radioimmunoassay methodology would be required to measure CRBP levels at this stage). During the 2-3 weeks of primary culture, the cells lose the majority of the vitamin A lipid droplets, and the effects of the high cellular vitamin A content presumably diminish. When retinol effects were studied orwhen CRBP determinations were made, the cells used were either passaged cultures or primary cultures after 3weeks of incuba-

tion. This was done to ensure that the vitamin A content of these cells was minimal. Collagens and Antibodies-Isolation and purification of human placental collagen types I and 111, Engelbreth-Holm-Swarm-derived type IV collagen and laminin, human plasma fibronectin, and rat collagen type I and 111 were performed as previously described (24). Affinity-purified antibodies to rat collagen types I and 111, Engelbreth-Holm-Swarm type IV collagen and laminin, and human type V collagen and fibronectin were produced as previously described (20, 24, 26). For measurement of Ito cell collagen production, cells were generally grown to confluence in the presence of ascorbic acid (50 pg/ ml) and P-aminoproprionitrile (100 pg/ml) (added 1-2 times/day). After washing with PBS, cells were then cultured in fresh media for 18-20h. The cellmediawere then collected and stored in frozen aliquots for later analysis. The cell layer was then scraped, sonicated in PBS, and then stirred with 0.05% Triton X-100 Du Pont-New England Nuclear) overnight at 4 "C to solubilize remaining collagen. A portion of the sonicate was frozen separately for later DNA analysis. The cell media and cell layer collagens were assayed using quantitative competition ELISAs as previously described (20, 27). ELISAs were established for collagen types I, 111, IV, and laminin. A representative ELISA for type I rat collagen is shown in Fig. 1. ELISAs were routinely sensitive in the nanogram range, thus permitting serial dilutions of the cell mediaor cell layer collagens to be made. Collagen synthesis was generally expressed relative to micrograms of cell DNA, as determined by the method of Hill and Whatley (28). Immunocytochemistry-For qualitative determination of collagen production, cell cultures were washed several times with PBS and fixed with PBS containing 10% formalin, 0.2% Triton X-100. After several subsequent PBS washes, staining with collagen antibodies was performed as previously described (24). Retinol Modulation-To evaluate the effects of retinol on collagen production, confluent cell cultures grown on either type I or type IV M retinol (solubilized in collagen were incubated with either 100% ethanol) or 0.1% ethanol in fresh tissue culture media containing serum. Preliminary experiments indicated that (a) during a 3-h incubation, a larger percentage of [3H]retinol entered the cells in serum-free media but at least 10% of the label entered the cells in the initial 3 h in the presence of serum and ( b ) a similar percentage of [3H]retinol entered the cells grown on type I collagen or type IV collagen (data not shown). Cells exposed to retinol were then incubated for 18 h and collagen synthetic measurements performed as described above. Therefore, the amount of collagen produced intracellularly or secreted extracellularly was studied in the presence or absence of retinol for 18 h (forconvenience, the datais expressed per 24-h periods). As the production of collagen in tissue culture in general maybe affected by cell density and by cells undergoing logarithmic growth, cells were studied a t confluence in an attempt to attain a stable base line of collagen synthesis. Nevertheless, pulsechase type experiments were not feasible and, therefore, some of the cell layer collagen which was measured in cells treated with retinol was produced prior to theaddition of retinol. As such, the differences between retinol-treated and control group collagen production were potentially larger than those observed. However, as described in Table I, in most instances the majority of the collagen produced was measurable in the cell media. A separate analysis of the cell media l/MEDIA DILUTION 8 4 2 1 100 80

m X z

,' ' ' '

I

-1

60 ~

40

TYPE I RAT COLLAGEN I T 0 CELL

MEDIA 20 0 10

100

1

1000

.

TYPE I RAT COLLAGEN (ng/ml)

FIG. 1. Type I rat collagen ELISA. Representative ELISA of type I rat collagen and serial dilution of Ito cell media. Linear regression-generated slopes are parallel.

10282

Ito

Hepatic

fractions noted a similar pattern of collagen phenotypic modulation after retinol treatment (data notshown). HPLC Analysis of CRBP-To determine the levels of CRBP in various cultures, the method of Bridges et al. (29) was utilized. In brief, cells which had been cultured in the presence of 1O“j M retinol for 3 days were scraped into PBS, centrifuged a t 800 X g, and resuspended in 0.1 M Na2S04,0.02 M Na2HP04, pH6.8. The 3-day incubations were performed in the presence of retinol to attempt to increase CRBP levels as in uiuo CRBP levels appear to respond to dietary retinol availability (30). The cells were then sonicated and centrifuged a t 48,000 X g for 2 h to obtain the cytosolic fraction. A portion of this fraction was stored frozen for later Lowry protein determination(31). The remaining cytosol was incubated in the M [‘Hlretinol in the darkunder N, gas at 4 “C presence of overnight. In some experiments, the cytosolic protein was incubated with the [3H]retinol aswell as a 1,000-fold excess of either cold retinol or cold retinoic acid. These studies were performed to evaluate the reported specificity for CRBP and retinol (29). After incubation, a fraction (0.1 ml) of this cytosolic mixture was analyzed for HPLC by injecting directly onto a Bio-Si1 TSK-250 HPLC gel permeation column (300 X 7.5 mm, Bio-Rad) equipped with a Waters LC spectrophotometer (model 481, Waters Associates, Millipore Corp., Milford, MA) at 280-nm absorbance, and fractions were collected in 0.2ml aliquots afterelution with the same Na2S04/Na2HP04buffer. The fractions were then mixed with Opti-Fluorscintillationmixture (United Technologies and Packard Instrument Co.) and counted (efficiency, 18%). Microscopy-Cell cultures were viewed on an Olympus inverted microscope equipped with phase and Hoffman optics, and photographs were taken with an Olympus PMlO AD camera on Tech-Pan Kodak film. Fluorescent microscopy and photography were performed on a Zeiss microscope as previously described (24). For electron microscopy, cell cultures were fixed directly in 35-mm dishes, viewed on a Philips300 electron microscope, and photographed as previously described (24). Statistical Analysis-Student’s t test was used to evaluate various subgroups.

Cell Culture apparent cell diameter.Thisappearanceismaintained throughout the2-3 weeks of primary cultureas well as during secondary cultures (through passage 10). These cells never assumed the characteristic “cobblestone” appearance of sinusoidal endothelial cells(32). This also served as further corroboration of their distinction from endothelial cells or Kupffer cells. As stated under “Materials and Methods,” these Ito cells were negative for staining withlysozyme or diacetylated low density lipoprotein, presumed markersof the other two cell types. Endothelial cells retain their capacity to stain with diacetylatedlow density lipoprotein in passaged cultures. Freshly isolated Ito cells, passaged Ito cells, and Ito cells in frozen liver sections stain for desmin (Dako Corp.) (data not shown). The ultrastructural morphology of passaged Ito cells is characterizedby prominent stressfibers, dilated prominent rough endoplasmicreticulum,and a paucity of lysosomes (data not shown). These characteristics are similar to primary culture Ito cells except for the absence of lipid droplets (9). Using Hoffman and phase contrast optics, the cells appear identical when grown on types I,111, or IV collagen, fibronecAs stated, the endoplasmic reticulum did tin,orlaminin. appear prominent forcells grown on a type I collagen matrix. It is possible that the endoplasmic reticulum appeared differently for cells grown on a type IV collagen matrix, but this was not studied.

Collagen Phenotype Qualitative-Passaged Ito cells were cultured on a variety of matrices (on 35-mm plastic dishes)or tissue culture plastic and then stained forcollagen types I, 111, IV, V, laminin, and fibronectin (see “Materials and Methods” above). The cells were simultaneously fixed and permeabilized as stated to permit maximum staining of predominantly intracellular proRESULTS teins. The amounts of extracellular collagen and lamininwere Cellular Morphology detected using ELISA methodology (see below). As shown in Fig. 3, the cells appear to produce collagen types I, 111, and As shownin Fig. 2 A , freshlyisolatedIto cellsgenerally appeared as slightly roundedcells with very prominent cyto- IV and lesser quantities of laminin. On tissue cultureplastic, plasmic lipid-like droplets. These droplets displayed the char- there was moderate intracellular staining of types I and I11 as collagen and weaker staining for type IV collagen and laminin. acteristicvitamin A autofluorescence(datanotshown) previously described by Friedman et al. (9). During the sub- On a type I collagen matrix ora fibronectin matrix, therewas sequent 2 weeks in culture, these dropletsbecome less prom- diffuse intracellular and extracellular staining of type I collagen, moderate intracellular type I11 collagen staining, modinent and cannot be detected clearly in secondary cultures. After 3 days of culture, themorphology of these cells changes est intracellular typeIV collagen staining, and weak laminin dramatically, as shown in Fig. 2B. The cells flatten, spread, staining. On a type IV collagen matrix, there was moderate and assume a stellate-like appearance with 2-4 X increase in intracellular staining of types I and I11 collagen and weak laminin staining (type IV collagen staining was not done as dishes were coated withhomologous type IV collagen). Fibronectin staining was also prominent (data not shown). There was no type Vcollagen detected. The relative intensity of collagen staining on the various matrices mayreflect the relative ratios of intracellular and secretedcollagen. Furthermore, some of the apparent extracellular collagen staining appears to vary between matrices perhaps due to differences in adherenceof various collagens for the underlying extracellular matrices (ie. type I collagen adherence to fibronectin). Therefore, estimates of relative collagen synthetic ratios on various matrices require a more quantitative approach. Quantitatiue-Relative collagen synthesis was quantitated &,. using ELISAmethodology as described under “Materials and 1‘. Methods.”Cell cultures studied included primarycultures FIG. 2. Ito cell morphology. A , Ito cell 8 h after isolation on after 2-3 weeks incultureas well as secondarycultures type I collagen-coated dish. Note cytoplasm engorged with lipid-like through passage 5. There was a similar quantitative collagen droplets; Hoffman optics were used. Bar = 50 pm. B, Ito cell 3 days phenotypic pattern for primary or secondary cultured cells. after isolation on type I collagen-coated dish. Note prominent peribetween these nuclear aggregates (analogous to type 111 collagen staining shown in Therefore, the ELISA data do not distinguish Fig. 3 ) as well as apparent increase in size compared to Fig. 3A. Phase two culture types. Fig. 4 summarizes the collagen phenotype for Ito cells grown on type I collagen. The data represent the contrast optics were used. Bar = 50 pm.

Hepatic It0 Cell Culture

10283

FIG. 3. Ito cell collagen immunofluorescence. Collagen staining of culturedIto cells (passage 3 ) onvarying matrices: A-D, on plastic; E-H, on type I collagen matrix; I-L, onfibronectin matrix; "0, on type IV collagen matrix. Antibody staining: A, E, I, M, anti-type I collagen; B, F, J , N , anti-type 111 collagen; C, G, K, anti-type IV collagen; D, H , L, 0, anti-laminin. Bar = 50 pm.

*

?! 1200 a

z

0 1000

. m 3

m

800 600,

2

0

a

400

2

200

0 0

0

TYPE I

TYPE 1 1 1

TYPE IV

TYPE OF COLLAGEN SYNTHESIS

FIG. 4. Ito cell collagen synthesis (on type I collagen matrix). Collagen syntheticphenotype for cultured Ito cells. Data represents sum of cell layer + cell media collagen (& S.E.; n = 6).

mean (fS.E.) from six separate Ito cell cultures. The total amounts (cell layer + cell media) of each collagen subtype are shown. Using a laminin ELISA sensitive to 10 ng, there was nodetectable in laminin apparent secondary Ito cell culture media not (data shown). The amounts mean (fS.E.) of types I, 111, and IV collagens were 129 f 61 (ng/pg of DNA/24 h), 545 f 82, and 1181 f 290, respectively. The differences between the various subtypes were statistically significant: I uersus 111,p < 0.002; I uersus IV, p < 0.005; and I11 uersus IV, p < 0.06. When cells were grown type on IV collagen, the relative amounts of types I and I11 collagen were 2172 f 1811 and 963 f 420, respectively (mean f S.E. for three separate Ito cell cultures). Fig. 5 shows the relative amounts of type I collagen produced when cells were grown on a type I versus IV matrix. There was a trend toward more type I Wnthesis while cells were grown on a type I v matrix ( p < 0.1). There were relativelyequal amounts of type 111 collagen produced on either underlying matrix. Table I summarizes

It0

10284

Hepatic

UNDERLYING MATRIX TYPE I TYPE IV

3000 P

Cell Culture

W

UNDERLYING MATRIX HTYPE I

$

H T Y P E IV

U X

0 w 1

: 4

TYPE I

TYPE 111


FIG. 5. Collagen synthesis uersus extracellularmatrix. Comparison of types I and I11 collagen synthesis by Ito cells grown on type I collagen matrix ( n = 6) or type IV collagen matrix ( n = 3).

TABLE I Relative It0 cell collagen secretion

TYPE I

TYPE Ill


FIG. 6. Retinol modulation of collagen synthesis. Confluent M retinol for 16 h and cultures of Ito cells were exposed to collagen synthesis was measured (see “Materialsand Methods”). Data are expressedas the relative changes in synthesis for cells exposedto retinol versus control (mean of three to six separate cultures).

dioimmunoassay (2). This is a relatively high concentration of this protein and is thought toreflect the normal Ito cell’s function of processing vitamin A (2). Furthermore, it has been I IV suggested that CRBP may function as a critical protein in media collagen/cell layer collagen transporting retinol from the cytoplasm to the nucleus, its Type I, n = 6 1.09 f 0.3 0.7 f 0.1 3.0 f 1.0 presumed site of action (22). It was of interest then to determine therelative amounts of this protein in cultured Ito cells. Type IV, n = 3 0.3 f 0.3 1.4 f 0.9 ND“ Since the aforementioned data suggested that retinol modua Not done. lation of type I collagen synthesis varied according to the underlying collagen matrix, it was of interest to determine the relative secretion of the various collagen subtypes uersus the effect of the underlying collagen matrix on CRBP levels. cell layer collagens. The observation that themajority of the The measurement of CRBP was performed using a HPLC type IV collagen was secreted into the medium explains in method recentlydescribed by Bridges et al.(29). The method’s part the relatively weak stainingpatternnotedin Fig.3. relative specificity is shown in a representative assay in Fig. While therelative ratios of collagen subtypes varied, the total 7. When [3H]retinol is incubated (see “Materials and Methamount of collagen produced (1500-3000 ng/pg of DNA/24 ods”) with cytosolic protein, two distinct fractions are idenh) is similar to the previously reported amount of collagen tified (Fig. 7 A ) .The lower molecular weight species approxiproduced by primary Ito cell cultures grown on plastic (9). mates the reported CRBP molecular weight of 14,600 (34,35, 40). A second molecular weight species in the 150,000 range Modulation of Collagen Production by Retinol is also identified. When excess cold retinol is added in addition of the labeled retinoid Since Ito cells normally handle vitamin A in uiuo, in was of to the [3H]retinol, the apparent binding interest to determine the effect of retinol oncollagen produc- and the 16-kDa protein is competitively inhibited (Fig. 7 B ) . M retinol concentration used to study the effect In addition, the binding of the labeled retinoid and the 150tion. The may represent a physiologic or a pharmacologic concentration. kDa protein is inhibited as well. When excess cold retinoic While serum vitamin A levels are known, it is unclear what acid isaddedinadditiontothe[3H]retinol,thereisno concentration of vitamin A stimulates the normal Itocellular inhibition of binding of label with the 16-kDa protein.Howmilieu. It is possible that Itocells are exposed to higher dietary ever, there is inhibition of label binding with the 150-kDa retinol concentrationsin portal blood (2, 3). Furthermore, Ito protein. Fig. 8 summarizes the levels of the 16-kDa retinol cells may actually receive vitamin A compounds from the binding protein to Itocells at varying lengths of culture. As previously reported by Bridges et al. (29) using a retina tissue neighboring hepatocytes which could also increase the imt o demediate pericellular vitamin A concentration (2, 3). The ef- culture system, this retinol binding protein appears fects of retinol administration are summarized in Fig. 6 (data crease with cell passages. The levels of the protein for cell represents the meanof three to six separate Ito cell cultures). passages 4, 5, and 8 were 2.1 (ng/pg of cytosolic protein), 1.8, There was a dramatic difference in the type I collagen syn- and 0.3, respectively (for cells grown on type I collagen) as thetic response to retinol when thecells were grown on type shown in Fig. 8. A similar pattern was noted for cells subculI uersus type IV collagen. When cells were grown on type I tured on type IV collagen (data not shown). However, there collagen, retinol administration resulted in a 433% increase was generally less CRBP in cells grown on a type IV matrix in type I collagen whereas when cells were grown on type IV uersus a type I matrix as shown in Fig. 9. collagen, retinol treatment resulted ina 49% reduction in the DISCUSSION amount of type I collagen. Retinol treatment resulted in a 190 and 160% increaseintype I11 collagen synthesis forcells The current study has demonstrated that vitamin A-storing grown on typeI or typeIV collagen, respectively.The collagen Ito cells can be isolated from rat liver with a high degree of synthetic response includes the summation of cell media and purity without the use of centrifugal elutriation and subsecell layer collagen normalized for pg of DNA/24 h. quently be maintained in primary and secondary cultures. Due to the general poor adherence of the passaged cells on HPLC Measurement of CRBP tissue culture plastic, it appears that long-term secondary by growth on an underlyingcollagen Freshly isolated Ito cells reportedly contain approximately culture is best supported 6 pg of CRBP/pg of cytosolic protein as determined by ra- matrix. These cells maintain a stellate-like appearance after Underlying collagen matrix

Collagen type produced 111

10285

Hepatic It0 Cell Culture 20000

5

A.

2.0

-n

n=3

1

8

3

8

s 10000

H-RETINOL

Q

2m

1.0

m

a 0

45

V

I

m

35

5575

65

0.0

Underlying Collagen Matrix

6.

FIG. 9. CRBP levels versus underlying collagen matrix. CRBP levels were compared for cells grown on either a type I or type IV collagen matrix. The data represent the mean of threeseparate cultures ranging from cell passage 4 to 8 (+ S.E.).

*

z

0

H-RETINOL

+

s

EXCESS COLD RETINOL

10000

20000

4 'q. 'r TK H-RETINOL

+

s

EXCESS COLD RETlNOlC ACID

0

45

55

35

65

75

FX #

FIG. 7. HPLC measurement of retinol binding proteins. Representative analysis of Ito cell cytosol exposed to lo" M [3H] retinol at 4 "C overnight and then injected directly onto a Bio-Si1 TSK-250 HPLC gel permeation column (300 X 7.5 mm). Fractions were collected in 0.2-ml aliquots after elution with 0.1 M Na2S04,0.02 M Na2HP04,pH 6.8. M , standards: IgG (158,000);ovalbumin (44,000); myoglobin (17,000).The asterisk depicts the approximate position for cellular retinol binding protein ( M , 14,600). A, [3H]retinol; B, [3H] retinol and 1000 X excess cold retinol; C, [3H]retinol and 1000 X excess cold retinoic acid. FX, Fraction number.

0 '

I

4

I

5 6 CELL PASSAGE #

I

I

7

S

FIG. 8. Cellular retinol binding protein levels during Ito cell culture. CRBP levels were quantitated by HPLC determination of 16-kDa retinol binding protein (see Fig. 7) for Ito cells grown on type I collagen matrix.

2-3 days in culture on a type I collagen matrix and continue to display a distinctive morphology through approximately 40-50 cell generations. Potential fibroblast overgrowth was not observed as it would be readily discernible due to the characteristic morphology. The use of relatively low cell split ratios (i.e. 1:2/1:4) appeared to be important as cell viability was poor when higher split ratios (i.e. 1%)were attempted (data not shown). This study has demonstrated that cultured Ito cells maintain the capacity to produce significant quantitiesof interstitial collagen types I and I11 as well as basement membrane collagen type IV. Furthermore, this collagen synthetic capacity is well maintained at least through cell passage 5. These findings are in contrast to studies of fibroblasts in general, the cell type which has often been suggested to evolve from Ito cells during hepaticfibrosis (13).Unlike cultured Itocells, fibroblasts generally produce mainly type I collagen, and this synthetic capacity declines rapidly during culture (36). It is particularly notable that these three collagen subtypes are produced by Ito cells i n vitro, asrecentimmunoelectron microscopic studies of human liver have demonstrated that Ito cells in vivo produce collagen types I, 111, and IV as well (10). As dramatic changes in extracellular matrix composition occur during liver injury,Ito cellsmight respondtothe changes in their extracellular matrix (16, 18). For example, the capillarization of sinusoids which has been reported to occur in alcohol-induced clinical liver injury would theoretically result in a basement membrane forming in the space of Disse immediately adjacent to the Ito cell (14). The current study began t o approach this theoretical possibility by comparing thecollagen phenotype of cells grown on type I (interstitial) versus type IV (basement membrane) collagen. There was a trend toward greater amounts of type I collagen production when the cells were grown on a type IV collagen matrix. The amounts of type I11 collagen produced were the same regardless of the type of underlying collagen matrix. The relative type I/type I11 ratio then was altered by the two different matrices; on a type I matrix this ratio was d . 0 whereas ona type IV matrix this ratio approximated1.0. It is interesting to note that a shift from predominantly type I11 collagen to type I collagen has often been noted with i n vivo models of pulmonary and hepatic fibrosis (16-18). Ito cells (or other hepatic cells) may respond to subtle shifts in their surrounding milieu with significant changes in their collagen synthetic machinery. It is conceivable that basement membrane proteins (i.e. type IV collagen, laminin, or proteoglycans) are involved in transmitting this message. The previ-

10286

Hepatic It0 Cell Culture

ously mentioned “transformation” of Ito cells during injury interplay between collagen and vitaminA metabolism as well may in fact partly representa response to thatmessage. as to identify endogenous and exogenous soluble mediators If matrix-driven changes in theregulation of collagen syn- which may ultimately have therapeutic implications. thesis indeed occur, one might predict that the cell’s response Acknowledgments-We would like to thank Dr. Heinz Furthmayr to external stimuli might also vary. The cell’s ability to alter for his interest in the project and the use of his HPLCapparatus. We its response would have obvious importance in maintaining would also like to thank Dr. C. D. Bridges for his advice and preprint intracellular and tissuehomeostasis. As vitamin A processing for the CRBP HPLC analysis. appears to be amajor Ito cell function in uiuo, it was of REFERENCES interest to determine( a ) whether collagen synthesis could be 1. Roberts, A. B., and S orn, M. B. (1984) in The Reinoids (Sporn, M., modulated by exposure to vitaminA and ( b ) whether a vitaRoberts, A. B., and Ebodman, D. S., eds) Vol. 2, pp. 210-276, Academic min A response was altered by the extracellular matrix. RePress, Orlando, FL 2. Blomhoff, R., Rasmussen, M., Nilsson, A,, Norum, K. R., Berg, T., Blaner, tinoids had previously been shown to modulatecollagen synW. S., Kato, M., Mertz, J. R., Goodman, D. S., Eriksson, U., and Peterson, P. A. (1985) J. Biol. Chem. 260, 13560-13565 thetic enzymes in other cell systems (37). It was found that R., Norum, K. R., andBerg, T. (1985) J. Biol. Chem. 260, the Itocell’s response to retinol was to some extent dependent 3. Blomhoff, 13571-13575 4. Fuchs, E., and Green, H. (1981) Cell 25,617-625 upon the underlying matrix. When cells were grown on a type 5. Murtaugh, M. P., Dennison, O., Stein, J. P., and Davies, P. J. A. (1986) J. I collagen matrix, retinol exposure resulted in a significant Exp. Med. 163,1325-1330 6. Wang, S.-Y., and Gudas, L. J. (1983) Proc. Natl. Acad. Sei. U. S. A. 80, increase in type I and type I11 collagen synthesis (type IV 5880-5884 ~ ~ . . collagen synthesis increased aswell (data not shown)).How7. Wang, S.-Y., and Gudas, L. J. (1984) J. Biol. Chem. 259,5899-5906 8. Wu, R., and Wu, M. M.J. (1986) J. Cell. Physiol. 127, 73-82 ever, when cellswere subcultured ona type IV matrix, retinol 9. Friedman. S. L.. Roll. F. J.. Bodes. J.. and Bissell. D. M. (1985) . . Proc. Natl. resulted in a decrease in type I collagen production and an Acad. Sei. U. .S. A.’82,868118685 10. Clement, B., Grimaud, J. A,, Campion, J. P., Deugnier, Y., and Guillouzo, increase in type I11 production. The mechanism of retinol A. (1986) He tology 6,225-234 modulation is unclear. Studies of skin fibroblasts and retinol 11. Ballardini, G., g p o s t i , S. D., Bianchi, F. B., De Giorgi, L. B., Faccani, A,, Biolchini, L., Busachi, C. A,, and Pisi, E.(1983) Liuer 3, 58-63 had suggested an effect at the transcriptional level (38, 39). 12. Kent, G., Gay, S., Inouye, T., Bahu, R., Minick, 0. T., and Popper, H. (1976) Proc. Noti. A d . Sei. U. S. A. 73,3719-3722 Other workers have emphasized the role of cellular retinol 13. Mak, K. M., Leo, M. A,, and Lieber, C. S. (1984) Gastroenterology 87, 188binding protein,a 14.6-kDa protein, in intracellular transport 200 14. Minato, Y., Hasumura, Y., and Takeucbi, J. (1983) Hepatology 3, 559-566 of retinol (22). 15. Senoo, H., and Wake, K. (1985) L a b . Inuest. 5 2 , 182-194 CRBP, which is present inrelatively higher concentrations 16. Murata, K., Kudo, M., Onuma, F., and Motoyama, T. (1984) Hepatogastroenterology 31,158-161 in liver, lung, skin, and retina, has been shown to be immu- 17. Raghu,,G., Striker, L. J., Hudson, L. D., and Striker, G. E.(1985) Am. Reu. nolocalized in theliver primarily in the Ito cell (41,42). Using Resprr. Dts. 131, 281-289 18. Rojkind, M., Giambrone, M. A,, and Biempica, L. (1979) Gastroenterology radioimmunoassay methodology, recent studies have shown 76, 710-719 that Ito cells quantitatively contain thehighest CRBP levels 19. Carey, D. J., Todd, M. S., and Rafferty, C.M. (1986) J. Cell Biol. 102, 2254-2263 as compared to other hepatic cells (2). The current study 20. Madri, J. A,, and Williams, S. K. (1983) J . Cell Biol. 97, 153-165 measured CRBP levels and noted that( a ) levels decline with 21. Madri, J. A., and Pratt, B. M. (1986) J. Histochem. Cytochem. 34,85-91 Chytil, F. (1985) Lab. Inuest. 52,465-467 prolonged cell passage and ( b ) the relative levels are higher 22. 23. Allegretto, E. A,, Kelly, M. A,, Donaldson, C. A., Levine, N., Pike, J. W., and Haussler, M. R. (1983) Bioehem. Biophys. Res. Commun. 116, 75when cells are grown ontype I collagen uersus type IV 81 collagen. This former observationwas previously observed in 24. Pratt, B. M., and Madri, J. A. (1985) Lab. Inuest. 52,650-656 25. Form, D. M., Pratt, B. M., and Madri, J. A. (1986) Lab. Inuest. 5 5 , 521a retina tissue culture system (29). It is interesting to note wn that the CRBPlevel of 2.1 ng/pg of cytosolic protein for cell 26. Madri, J. A., Roll, F. J., Furthmayr, H., and Foidart, J.”. (1980) J. Cell Biol. 86,682-687 passage 4 is in general agreement with a previous study of 27. Davis, B. H., and Madri, J. A. (1987) Am. J. Pathol. 126,119-129 freshly isolated Ito cells (2).Thisotherstudy utilized a 28. Hill, B. T., and Whatley, S. (1975) FEES Lett. 5 6 , 20-23 29. Bridges, C. D. B., Oka, M. S., Fong, S.-L., Liou, G. I., and Alvarez, R. A. radioimmunoassay and noted approximately 6 ng of CRBP/ (1987) Neurochem. Int., in press fig of cytosolic protein (2). The apparent differences in CRBP 30. Kato, M., Blaner, W. S., Mertz, J. R., Das, K., Kato, K., and Goodman, D. S. (1985) J. Biol. Chem. 260,4832-4838 levels between the studiesmay be accountedfor by the differ- 31. Lowry 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Swl.’ Chem. 193,265-275 ent methodologies as well as the use of fresh uersus cultured 32. Irving, M. G., Roll, F. J., Huang, S., and Bissell, D. M. (1984) Gastroentercells. ology 87, 1233-1247 Taylor, C. R. (1978) Arch. Pathol. Lab. Med. 102,113-121 In conclusion, the current work implies that in the Itocell 33. 34. Ong, D. E. (1985) Nutr. Reu. 43,225-232 culture system, there may be an intimate relationship between 35. Sundelin, J., Anundi, H., Tra irdh, L., Eriksson, U., Lind, P., Ronne, H., P. A,, and Rask, &. (1985) J . Biol. Chem. 260, 6488-6493 vitamin A and collagen metabolism. Vitamin A clearly has 36. vonPeterson, der mark K. and von der mark H. (1977) J. Cell Bzol. 73, 736-747 37. Roguska, M. k.,And Gudas, L. J. (1685) J. Biol. Chem. 260, 13893-13896 dramatic effects on collagen synthetic phenotype, and the R. P., Meeker, C. A,, Oikarinen, H., Oikarinen,A. I., and Uitto, J. extracellular collagen matrix appears to alter both collagen 38. Abergel, (1985) Arch. Dermatol. 121,632-635 and vitamin A metabolism. The different collagen matrices 39. Oikarinen, H., Oikarinen, A. I., Tan, E. M. L., Abergel, R. P., Meeker, C. A.. Chu. M. L., ProckoD, . . D. J., and Uitto. J. (1985) J. Clm. Inuest. 75, affect collagen synthetic phenotype but also alter thecell’s Ito 1545-1553 Eriksson, U., Das, K., Busch, C., Nordlinder, H., Rask, L., Sundelin, J., 40. response to retinol as well as the levels of CRBP. Future Sallstrom, J., and Peterson, P. A. (1984) J. Biol. Chem. 259, 13464studies will need to determine whether the underlying matrix 13470 41. Kato, M., Kato, K., and Goodman, D. S. (1985) Lab. Inuest. 52, 475-484 also alters the Ito cell’s intracellular handling of retinol. Ito 42. Porter, S. B., Fraker, L. D., Chytil, F., and Ong, D. E. (1983) Proc. Natl. Acad. Sei. U. S. A . 80,6586-6590 cell culture appears to bea promising system to evaluate the ~~~~

’