Isolation and culture of hepatic lipocytes, Kupffer cells, and sinusoidal endothelial cells by density gradient centrifugation with Stractan

ANALYTICAL

BIOCHEMISTRY

161,207-218

(1987)

Isolation and Culture of Hepatic Lipocytes, Kupffer Ceils, and Sinusoidal Endothelial Cells by Density Gradient Centrifugation with Stractan SCOTT L. FRIEDMAN'.' AND F. JOSEPHROLLS Liver Center. Sun Francisco General Hospital and Department ofhfedicine, Unirersitv q/ Caljfhnia, San Francisco. Cal&rnia 94110 Received June 18. 1986 A method for isolating purified populations of hepatic lipocytes. Kupffer cells, and sinusoidal endothelial cells suitable for culture, using density gradient centrifugation on the polysaccharide material Stractan is described. A nonparenchymal cell digest of liver from either normal rats or rats treated with modest doses of vitamin A is layered on a discontinuous gradient of 6, 8, 12, and 20% Stractan; lipocytes are separated efficiently from other nonparenchymal cells and are removed from the top of the gradient. Kupffer cells and sinusoidal endothelial cells, which migrate to denser interfaces in the gradient, are further purified by differential plating and selective trypsinization, respectively. Isolated highly viable lipocytes free of contaminants adhere and spread progressively over several days in primary culture and display both intrinsic vitamin A fluorescence and positive immunostaining for desmin. Lipocytes survive for prolonged periods on plain plastic, and collagen synthesis by these cells remains relatively constant for at least 28 days. Based on serial assay of DNA content, lipocytes in primary culture proliferate, beginning 7 days after plating. Kupffer cells and sinusoidal endothelial cells isolated by Stractan density centrifugation likewise retain their typical morphologic and functional characteristics in culture; the purity of these cell isolates has been confirmed by using specific fluorescent markers. This investigation demonstrates that Stractan density gradient centrifugation is an efficient, sensitive, and reproducible method for isolating pure populations of hepatic nonparenchymal cells. Q 1987 Academic press. Inc. KEY WORDS: lipocytes; Kupffer cells; sinusoidal endothelial cells; Stractan gradient; collagen synthesis; cell culture.

Lipocytes (also known as Ito cells, fat-storing cells, or stellate cells) are hepatic nonparenchymal cells recently found to play active roles in vitamin A storage and hepatic matrix production. In normal liver, lipocytes are situated in the perisinusoidal spaces of the hepatic sinusoid, with extensive foot processes extending into the space of Disse separating hepatocytes from sinusoidal endothelium (1). They are the principal collagen-produc-

ing cells in normal liver (2) and, in addition, are the cell type responsible for storing retinol (Vitamin A), primarily in the form of retinyl esters (3,4). The recognition that hepatic lipocytes perform these important metabolic and synthetic functions has led to a greater need for the isolation and culture of these cells. Previously reported methods for studying nonparenchymal cell function in culture have often failed to specifically identify lipocytes in, or exclude them from, isolates of Kupffer cells or sinusoidal endothelial cells. This suggests that functions of lipocytes in nonparenchymal cell cultures were not clearly identified and, moreover, may have been erroneously attributed to other cell types.

’ Address correspondence to: Scott L. Friedman, M.D., Liver Center Laboratory, Building 40, Room 4102, San Francisco General Hospital, San Francisco, CA 94110. ’ Supported by USPHS Grant IF32 AM 07589-01 and the AGA Research Training Supplement Award. 3 Supported by NIAAA Grant ROl AA 06092. 207

0003-2697187

$3.00

Copyright 0 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.

208

FRIEDMAN

We describe here a useful method for the separation and pure culture of lipocytes, Kupffer cells, and sinusoidal endothelial cells from a single rat liver. The technique avoids the need for centrifugal elutriation, requiring only a single gradient separation using Stractan, a material that is inexpensive, readily prepared, and gives highly reproducible results. MATERIALS

AND METHODS

A. Materials

Male Sprague-Dawley rats (350-450 g), purchased from Bantin and Kingman, were fed ad libitum with standard rodent chow (Purina Co., vitamin A content 15 IU/g). Retinyl acetate powder was purchased from Sigma (St, Louis, MO) and dissolved into corn oil (Mazola) by sonication. Implantable retinyl acetate pellets ( 100 mg/pellet, 100 mg = 290,000 IU) were purchased from Collaborative Research (Bethesda, MD). Formalinfixed Staphylococcusaureus were purchased from Zymed Laboratories (Burlingame, CA), polyclonal antibodies to desmin from Accurate Scientific (Westbury, NY), Pronase (77.5 P.U.K.) and deoxyribonuclease (DNase) from Calbiochem (La Jolla, CA), collagenase from Cooper Biomedical (Malvern, PA), and EDTA from Sigma. Trypsin was purchased from Difco Laboratories (Detroit, MI). Trypsin/EDTA solution was prepared by dissolving 1.0 g trypsin, 40 mg EDTA, and 200 mg glucose in 200 ml L- 15 salts and then centrifuging the mixture at 4°C for 20 min at 12,000g. The supematant was collected on ice, and 0.4 ml phenol red (0.5%) was added; the 0.5% Trypsin/0.02% EDTA stock solution was filter-sterilized and stored at -20°C. Working solutions were made by dilution with L- 15 salts. Stractan (arabinogalactan), grade II, was purchased as a crude powder from Sigma Chemical Co. It was prepared exactly according to the method of Corash et al. (5): 450 g of Stractan

AND ROLL

powder was dissolved into 400 ml of distilled water by gentle stirring in a 37°C water bath. The crude Stractan solution was deionized sequentially in 900 g each of Amberlite 1R- 120 cation-exchange and 1RA-4 10 anion-exchange resins (Sigma) by mixing the Stractan directly with each resin for 30 min at 4°C and then removing the resins from the solution by filtration on a Buchner funnel containing Pyrex wool. The density of the deionized Stractan was assessed by measuring the weight of a Stractan aliquot relative to an equal volume of water, and by using this information the concentration of Stractan (Cs) was extrapolated from a density vs concentration graph of sucrose, with Stractan having the same refractive index as sucrose (w/w). The osmolarity was determined by freezing point depression (Wescor osmometer, Logan, UT) of Stractan diluted 2: 1 with buffered saline with glucose (BSG),4 (BSG, pH 7.4, osmolarity 291 mOsmo1 per liter; NaCl, 8.1 g; NaHPO,, 1.22 g; NaH2P04 2H20). The true osmolarity was then calculated according to the formula True osmolarity = (OsM 2: 1 dilution) X(2+AW)-2XOsMBSG AW

7

where AW (available water) = 1 - 0.64 X CS (concentration Stractan). The pH of the deionized Stractan was adjusted to 7.4, and the osmolarity was increased using the following additions: for each 100 ml available water-l 16 mg MgClz - 6H,O and 200 mg glucose; for each 100 ml Stractan solution11.1 ml 0.15 M potassium phosphate solution, pH 7.4 (&HPO,, 5.226 g; KH2P04, 1.02 g in 250 ml deionized water), and 3.0 g 4 Abbreviations used: BSG, buffered saline with cose; DiI, 3,3’-dioctadecyl indocarbocyanine; DMEM, Ham’s modification of Dulbecco’s minimal sential medium; MEM-E, Eagle’s minimal essential dium.

gluH/ esme-

LIF’OCYTE

ISOLATION

BY

STRACTAN

bovine serum albumin (Sigma, Catalog No. 8022 or 4503). The pH was readjusted to 7.4, and the concentration and true osmolarity of the Stractan were redetermined. Final adjustment of the osmolarity to 290-295 mOsmol per liter was accomplished by stepwise addition of dry NaCl ( 100 mg NaCl/ 100 ml AW = 31.5 mOsmo1 per liter). The stock solution of Stractan (usually 35-40%) was stored at -20°C and then thawed and diluted with BSG as required into 35-ml working aliquots of 68, 12, and 20% (v/v). Working aliquots were filter-sterilized through a 0.45~pm filter and stored in sterile plastic tubes at -20°C between experiments. Each batch of stock Stractan solution provided enough material for approximately 50 experiments. Sterile tissue culture dishes were purchased from Nunclon (Denmark) and Miles Scientific (Naperville, IL). Medium 199 was supplied by GIBCO (with Hanks’ salts as 10X powder, Catalog No. 330-l 18 l), and calf and horse sera were purchased from Flow Laboratories (McLean, VA). In addition to serum, Medium 199 used for plating contained the following supplements: NaHCO,, 3.3 mg/lOO ml; glucose, 1 mg/ml; penicillin (Pfizer), 100 U/ml; U-40 regular insulin (Squibb/Nova), 4 mu/ml; corticosterone (Sigma), 10m6 M, from 1O-2 M stock solution in dimethyl sulfoxide (Sigma, grade I); and L-ascorbic acid (Sigma), 50 pg/ml. Sterile filters were from Nalgene (Rochester, NY). The isotope [2,3,4,5-3H]proline (sp act 100 Ci/mmol, 3.7 TBq/mmol) was purchased from Amersham (Arlington Heights, IL). B. Methods

Labeling of Nonparenchymal Cells Using Specific Fluorescent Markers In animals exceeding 450 g, lipocytes were easily recognized and adequately separated from other nonparenchymal liver cells without pretreatment with vitamin A. Smaller animals required pretreatment with vitamin

GRADIENT

CENTRIFUGATION

209

A to increase lipocyte buoyancy and facilitate their recognition. This was accomplished either by multiple subcutaneous injections of retinyl acetate (total dose 500-750,000 IU) suspended in olive oil (60,000 U/ml) or by subcutaneous implantation of retinyl acetate pellets (total dose = 590,000 IU) 2 to 3 weeks prior to cell isolation. In other studies lipocytes in culture were fixed with 0.5% paraformaldehyde for 1 h, then incubated with either normal rabbit serum or rabbit anti-human desmin at 1:20 dilution, followed by rhodamine-conjugated goat anti-rabbit IgG (6). Endothelial cells were identified by intravenous injection of acetoacetylated low density lipoprotein tagged with the fluorescent marker, 3,3’-dioctadecyl indocarbocyanine (DiI), 10 min prior to liver perfusion. The fluorescent marker, DiI, is metabolically inert and remains detectable exclusively within sinusoidal endothelial cells for several days in culture (7). It is specific for sinusoidal, and not vascular, endothelium (8). Kupffer cells were identified in vitro by their ability to ingest formalin-fixed, FITC-labeled Staphvlococcus aureus. Fluorescein-conjugated staphylococci were prepared by the conjugation procedure of Wells et al. (9). For labeling Kupffer cells in culture, cells were washed with L- 15 salts, incubated for 10 min in Medium 199 containing 10% fetal calf serum and 0.25% bovine serum albumin, and then incubated with staphylococci in the same medium for 5 min at 37°C (40 pg dried staphylococci per 1 ml cell medium). Cells were then extensively washed with fresh medium. Uptake of S. aureus specifically identifies Kupffer cells and correlates with histochemical staining for endogenous peroxidase (J. Trudell and F. J. Roll, unpublished observation). Isolated cells, labeled by these methods, were identified by fluorescence microscopy. Lipocytes displayed fading green-blue fluorescence under epi-illumination with ultraviolet light ( 10). Endothelial cells labeled with

210

FRIEDMAN

AND ROLL

DiI were fluorescent in the rhodamine range, and Kupffer cells labeled with S. aureus displayed stable fluorescence in the fluorescein range.

6% H202, filter-sterilized catalase (0.1 mg/ml) and then sterile L- 15 salts. Gradients were centrifuged at 20,000 rpm for 30 min at 25°C in a Beckman SW40 rotor.

Liver Digestion and Nonparenchymal Cell Separation

Cell Culture

Animals were anesthetized with ether and treated with 0.1 cc heparin ( 1000 u/ml) via direct inferior vena cava injection immediately before perfusion. The liver was perfused in situ at 37°C at a flow rate of 10 cc/min, first with 75 cc L- 15 salts, followed by 100 cc of 0.2% Pronase in Ham’s modification of Dulbecco’s minimal essential medium (H/ DMEM), and then via recirculation with 0.0 15% collagenase in H/DMEM for approximately 25 min (11) or until the hepatic parenchyma beneath the capsule appeared liquified. Following perfusion the liver was excised, opened with scissors, placed in a siliconized flask containing 0.02% Pronase and 10 pg/ml DNase in 100 cc H/DMEM, and agitated in a 37°C rotary shaking both for 30 min at 275 rpm. The resulting cell suspension was filtered through gauze and washed three times at 500g for 7 min in 75 ml Eagle’s minimal essential medium (MEM-E) without calcium or magnesium, with 10 pg/ml DNase. Nonfiltered tissue was dispersed further by magnetic stirring at room temperature in a siliconized flask containing 50 cc of MEM-E for 15 min. This material was refiltered through gauze, washed at 5008 for 7 min in 50 ml MEM-E containing 10 pg/ml DNase, and pooled with the remainder of the cell suspension. The combined digest was suspended in 25 cc MEM-E containing 10 rg/ml DNase and distributed evenly across four sterile Stractan gradients containing 1.5 cc of 6, 8, 12, and 20% solutions (densities of 1.053, 1.058, 1.080, and 1.111, respectively, calculated relative to an equal volume of water). Gradients were poured in Beckman Ultra-Clear 13-ml centrifuge tubes that had been sterilized by washing sequentially with

Lipocytes. Lipocytes were recovered at the interface between medium and 6% Stractan, washed in 25 cc MEM-E at 500g for 7 min, and suspended in Medium 199 supplemented with calf and horse sera at 10% each (v/v). The concentration was adjusted with an automated cell counter (Sysmex, TOA Medical Electronics, Japan) to l-2 X lo6 cells/ml, and cells were plated in 35-mm tissue culture dishes. Plating efficiency, morphology, and survival were comparable on Nunclon vs Miles plastic dishes. The medium was replaced 18-24 h after plating, and every 24 h thereafter. Kup& cells. Cells present in the 8- 12% Stractan interface following centrifugation were collected with a sterile pipette, washed in 25 cc MEM-E at 500g for 7 min, suspended in Medium 199 supplemented with 20% serum, and plated in 60-mm uncoated plastic tissue culture dishes at a concentration of 4-5 X lo6 cells/ml. After 20 min, nonadherent cells and residual serum were washed off the monolayer with L-l 5 salts and then replaced with fresh medium containing 20% serum (12). Twenty-four hours later the culture was incubated for 3 min at 37°C with L- 15 salts containing 0.5% trypsin and 0.02% EDTA. EDTA/trypsinized cells were removed by gently aspirating the monolayer with a sterile pipet and the culture was continued by addition of fresh medium containing 20% serum. Medium was subsequently replaced every 24 h. Endothelial cells. Cells (mixed endothelial cells, Kupffer cells, and erythrocytes) from the 12-20% Stractan interface were removed with a sterile pipet, washed in 25 cc MEM-E at 500g for 7 min, and plated in 60-mm plastic dishes at a concentration of 3-4 X lo6

LIPOCYTE

ISOLATION

BY

STRACTAN

cells/ml. Medium was replaced after 24 h to remove nonadherent cells and erythrocytes; after an additional 24 h, cells were washed twice with L-l 5 salts and incubated with 0.125% trypsin/0.005% EDTA for 3 min. The monolayer was gently aspirated with a sterile pipet, and the trypsinized cells were resuspended in medium containing 20% serum and replated into uncoated plastic tissue culture dishes. Medium was subsequently replaced every 24 h. Viability of cultured lipocytes, sinusoidal endothelial, and Kupffer cells was determined by their ability to exclude trypan blue ( 13). Quantitation of Collagen Production Total lipocyte collagen and noncollagen protein synthesis was assessed by measuring the incorporation of [3H]proline into peptide-bound [3H]hydroxyproline and [3H]proline, respectively (14). Cells were incubated in Medium 199 for 24 h with [2,3,4,53H]proline (50 &i/ml; 1 Ci = 37 GBq) and fresh ascorbate (50 ~g/mIf~ Cells and media were harvested separately into 0.5 M acetic acid containing 10 mM EDTA and 8 mM iV-ethylmaleimide at 4°C and then dialyzed against distilled water until dialysate counts were minimal. Dialyzed samples were hydrolyzed in 6 M HCl and chromatographed on a Beckman 119 amino acid analyzer equipped with a stream splitter. Fractions in Aquasol were counted on a Tri-Carb liquid scintillation spectrometer (Packard Instrument Co., Downers Grove, IL) with an efficiency of 33%. An aliquot of the original cell fraction was analyzed for DNA content by the fluorimetric method of Labarca and Paigen (15), and final data were expressed as disintegrations per minute of tritiated hydroxyproline or proline per microgram of DNA, per 24 h. Morphologic

Methods

The appearance of lipocytes under highpower fluorescence and phase-contrast mi-

GRADIENT

CENTRIFLJGATION

211

croscopy was monitored and photographed with an Olympus BH research microscope (Olympus Corp. of America) equipped for epi-illumination with UG5 exciter and 530 nm barrier filters. Low-power photomicrographs were made with a Nikon Diaphot inverted microscope equipped for phase-contrast microscopy and Panatomic X film (ASA 32). For fluorescence micrographs, exposure times with Ektachrome film (ASA 400) at f 2.8 ranged from 2 to 8 s. RESULTS Composition and Yield of the Stractan Gradient

Centrifugation of hepatic nonparenchyma1 cells on Stractan yielded four distinct populations of cells whose composition was reproducible, as assessed using cell-specific fluorescent markers (Table 1). The total nonparenchymal cell yield obtained following enzymatic digestion was 2.23 X lo8 cells (mean data, n = 3), not including the pellet which consisted of only a few surviving hepatocytes. The lipocyte yield is greater than results reported previously ( 16). In addition, the total nonparenchymal cell yield, gradient composition, and cell viability were similar between untreated rats and those injected with vitamin A. The most critical aspect of the isolation procedure was the quality of the in situ liver digestion. Optimal yields were obtained when virtually all hepatocytes were destroyed by Pronase digestion and the liver appeared liquid. The activity of the Pronase varied with storage time and from batch to batch. The Pronase concentration for each batch was titrated to the point at which little or no hepatocytes were present in the mixed nonparenchymal cell digest; perfusion time and the volume of Pronase solution were kept constant. Concentrations in excess of this endpoint led to reduced yields of lipocytes and Kupffer cells, relative to sinusoidal endothelial cells, suggesting that the latter population is relatively insensitive to Pronase.

212

FRIEDMAN

AND ROLL

TABLE 1 YIELDANDCOMPOSITIONOFNONPARENCHYMALCELLSISOLATEDBY

Pregradient

Postgradient

Mixed nonparenchymal cell isolate Total No. of cells X 10m6: 223.0 f 115.0 Composition (mean %) Lipocytes Hepatocytes Kupffer cells Endothelial cells Other”

STRAC~ANGRADIENTCENTRIFUGATION

Medium-6%

Stractan layer

Total No. of cells X 10m6:12.2 f 1.7 (mean f SD.) 3.9 3.0 24.9 36.8 31.4

Composition (mean %) Lipocytes Hepatocytes Kupffer cells Endothelial cells

98 2 0 0

6-8% Stractan layer Total No. of cells X IO-? 14.4 * 3.6 Composition (mean %) Lipocytes Hepatocytes Kupffer cells Endothelial cells

11.4 26.1 19.0 43.5

8-12% Stractan layer Total No. of cells X lo-? 105.5 f 19.3 Composition (mean %) Lipocytes Hepatocytes Kupffer cells Endothelial cells

0 0 44.7 55.3

12-20% Stractan layer Total No. of cells X IO-? 93.7 * 9.2 Composition (mean %) Lipocytes Hepatocytes Kupffer cells Endothelial cells Other”

0 4.4 12.9 25.9 56.8

Note. The composition of each gradient layer is shown above. Cells were identified using specific fluorescent labels as described under Methods. The yield and composition data represent the mean + standard deviation of three experiments. DThis population consists mainly of erythrocytes, although some nucleated cells were also present, including leukocytes and bile duct epithelial cells.

Cells from three of the four gradient interfaces were used for culture. The top layer (medium-6% Stractan) almost exclusively contained lipocytes, although some nonvi-

able hepatocytes were occasionally present. This layer also contained substantial debris if the mixed nonparenchymal cells were inadequately washed prior to gradient separation.

LIPOCYTE

ISOLATION

BY STRACTAN

The second layer (6-8% Stractan) was rich in lipocytes but also contained other nonparenchymal cells and intact nonviable hepatocytes, which interfered with lipocyte adherence in culture. Therefore, cells from this gradient layer were not used for cell culture studies. The lower two fractions, at the 8- 12% and 12-20% Stractan interfaces, contained mixed populations of Kupffer and endothelial cells and were usually free of lipocytes although lipocytes were occasionally present at the 8-12% interface. The 12-20% fraction also contained erythrocytes and nucleated nonliver cells, either leukocytes or bile duct epithelial cells. The plating efficiency of cells from the individual gradient layers was greater for the lower fractions. Lipocytes from the medium-6% Stractan interface had a plating efficiency of 59.5 -t 8.9% (n = 4) when the number of attached cells was measured 24 h after plating. The plating efficiency of cells from the 8-l 2% Stractan interface, from which Kupffer cells were obtained, was 80.2 t- 9.4% (n = 4) and that of the 12-20% Stractan interface, from which endothelial cells were obtained, was 80.9 f 3.2% (n = 4). Characteristics

of Cultured

Lipocytes

Lipocytes were recovered from the top of the Stractan gradient (Table 2). The yield, TABLE

GRADIENT

appearance, and behavior in culture of lipocytes isolated from normal rats (greater than 450 g) were comparable to those obtained from smaller animals (less than 400 g) pretreated with vitamin A. Lipocytes appeared under phase microscopy when first isolated as round phase-dense cells, containing refractile droplets (Fig. 1A). Cells adhered and spread relatively slowly in primary culture. Spreading was complete only after 3 days of incubation; after 7 days, the culture appeared as a dense homogenous population of stellate cells (Fig. IB). Lipocytes in primary culture retained many of the ultrastructural characteristics that characterize these cells in viva These include abundant vitamin-A-containing droplets and prominent rough endoplasmic reticulum (2). All cells displayed characteristic rapidly fading vitamin A fluorescence when examined under ultraviolet epi-illumination. Desmin, an intermediate filament specific for lipocytes among liver sinusoidal cells, was used as a confirmatory marker of lipocyte purity ( 17). Greater than 99% of the cells in primary lipocyte cultures displayed desmin immunofluorescence when stained with antiserum to this protein (Fig. 2). The stability of lipocytes in primary culture was studied both morphologically and biochemically. The cells retained their characteristic phase and fluorescent appearance for at least 28 days. Over the same period, they secreted collagen at a steady rate 2

YIELD, PURITY, AND VIABILITYOF~SOLATED LIP~CYTES,KUPFFERCELLS, SINUSOIDALENDOTHELIALCELLS Yield per animal (X IO6 cells) Lipocytes Kupffer cells Endothelial cells Note. The three cell populations automated cell counter. Viability experiments.

12.2 k 1.7 22.1 f 1.2 35.0 f 2.9

213

CENTRIFUGATION

Purity

(%)

Viability

under Data

(W)

96.0 i 0.1 96.9 f 5.4 96.8 f 3.1

>99 95.4 f 2.7 90.4 k 3.6

were purified and identified as detailed was assessed by trypan blue exclusion.

AND

Methods represent

and counted using an mean k SD of three

214

FRIEDMAN

AND

ROLL

FIG. 1. (A) Phase photo micrograph of lipocytes in primary culture 24 h after plating onto plain plastic. Most cells appear as dark round clusters, while a few cells have begun to spread. (Bar equals 100 pm.) (B) Lipocytes in primary culture for 7 days. There is a uniform population of well-spread stellate-shaped cells, all of which demonstrated rapidly fading vitamin A fluorescence (Bar equals 100 pm.)

LIPOCYTE

ISOLATION

BY STRACTAN

GRADIENT

CENTRIFUGATION

215

FIG. 2. lmmunofluorescent photo micrograph of lipocytes in primary culture for 7 days, stained with antiserum to desmin (see Methods). The antiserum recognizes cytoplasmic filaments in virtually all cells. whereas a control culture stained with normal rabbit IgG (not shown) was negative. (Bar equals 25 pm.)

(Fig. 3), and the ratio of collagen-to-total protein synthesis remained relatively constant (Fig. 3). Serial DNA assays were used as an index of cell number and showed an increase in DNA content per plate, beginning 7 days after plating and reaching a peak at 14 days (Fig. 4). In preliminary experiments, DNA content correlated closely with cell number (r = 0.9934). Lipocytes were passaged by treating the cells with 0.5% trypsin/0.02% EDTA for 10 min and by resuspending and plating in medium containing 20% serum. Secondary cultures retained intracellular vitamin A fluorescence (not shown). Kupffer Cell Culture

Kupffer cells were the predominant cell type sedimenting to the 8- 12% Stractan interface with centrifugation (Table 1). Cells

from this fraction were plated on plain plastic, from which purified cultures of Kupffer cells were obtained by selectively removing contaminants (primarily sinusoidal endothelial cells) with trypsin. The remaining adherent population contained greater than 95% Kupffer cells, as assessed by phase-contrast morphology and phagocytosis of fluoresceinated staphylococci (Table 2). Kupffer cells purified in this manner and maintained in medium containing 20% serum remained adherent and phagocytic in primary culture for at least 10 days. Endothelial Cell Culture

Endothelial cells comprised the major liver-derived cell population migrating to the I2-20% Stractan interface (Table l), with lipocytes usually appearing in only the two most buoyant layers of the Stractan density

216

FRIEDMAN

0

,,+ 7

6

l4 TIME

AND ROLL

n IN CULTURE

n

2s

(DAYS)

FIG. 3. Graph demonstrating long-term collagen production by lipocytes and relative amount of collagen synthesized, as a percentage of total protein. Lipocytes in primary culture were harvested at the time points indicated and collagen synthesis was assessedby measuring the incorporation of [3H]proline into collagen as [‘Hlhydroxyproline (solid line). Collagen synthesis at each time point represents mean -CSD of 3-4 experiments. except on day 7, where n = 2. The relative collagen synthesis (dotted line) is based on the values for [‘Hlproline incorporation at each time point and is derived using a modification of the formula of Diegelman, as reported previously (2). gradient. Endothelial cells were efficiently separated from other cell types by trypsinization 48 h after initial plating, erythrocytes and lymphocytes were removed when the medium was changed at 24 h, and Kupffer cells remained adherent to the culture plate even after exposure to trypsin (Table 2). Sinusoidal endothelial cells were identifiable by their pavement-like morphology (7), by their fluorescence after injection of DiIAcAcLDL in vivo, and by their inability to ingest fluorescently labeled S. aureus in culture. DISCUSSION

Stractan gradient centrifugation provides a reproducible method of separating nonparenchymal liver cells, with several advantages over previously reported techniques. Other investigators have utilized Stractan to separate subpopulations of erythrocytes (5) and

platelets (18) and we have chosen the material, a polysaccharide of 30,000 Da, because it is isosmolar, simple to prepare, economical, and biologically inert (5). From a stock solution of approximately 40%, gradients can be tailored to optimize the yield of the cell populations under study. Centrifugation through Stractan has no detectable effect on the viability or attachment and spreading of Kupffer and endothelial cells, in comparison with the behavior of cells isolated using other gradient materials (19). By combining the gradient technique with differential trypsinization, individual cultures of three nonparenchymal cell types can be achieved without centrifugal elutriation, a time-consuming procedure that requires expensive equipment and an additional level of technical support. The use of unambiguous fluorescent markers for each nonparenchymal cell documents the homogeneity of the respective cultures.

LIPOCYTE

ISOLATION

BY STRACTAN

26 1 241

bL $E!‘4-1 1210-, 5

7

9

14

TIME IN CULTURE

17

21

(DAYiS)

FIG. 4. DNA content of cultured lipocytes in longterm primary culture. Equal numbers of suspended cells were plated in plastic culture plates, maintained in serum-containing medium, and harvested for assay of DNA content at the time points indicated. Each data point represents the mean value for two separate experiments.

Methods for isolation and culture of lipocytes have only recently been described ( 11, 20) and most studies report a cell purity of 7040%. Lipocytes isolated by Stractan density gradient centrifugation are virtually free of contaminating cell types. It is noteworthy that only smaller animals required pretreatment with vitamin A to facilitate lipocyte recognition and isolation. Enhanced yields of lipocytes from older (and therefore larger) rats has also been recognized by others (20). Lipocytes are the main storage site for vitamin A in the liver (3), even in animals with hypervitaminosis A (2 I), and preloading smaller animals with this vitamin appears to enhance lipocyte buoyancy comparable to that in larger animals, so that lipocytes are completely separated from other nonparenchymal cell types on a density gradient. The vitamin A present in lipocytes enables them to be distinguished from Kupffer and endothelial cells by phase-contrast or fluorescence microscopy. Lipocytes containing lit-

GRADIENT

CENTRIFUGATION

217

tle or no vitamin A lack phase-dense vesicles and may be overlooked as contaminants of either parenchymal(22) or nonparenchymal (unpublished data) liver cell cultures. In separate studies we have examined whether vitamin A administration alters lipocyte function in culture; we find no relationship between the dose of vitamin A or hepatic vitamin A content and collagen production of lipocytes in culture (23). By microscopic inspection and an increase in DNA content it is apparent that lipocytes proliferate in culture as previously reported (20). In addition, our finding that trypsin/EDTA may be used to passage the cells confirms the report of De Leeuw et al. (20). We are currently examining the phenotype of passaged lipocytes. The yield, morphology, and survival of Kupffer and sinusoidal endothelial cells isolated by Stractan centrifugation are comparable to data previously reported using other methods. In particular, Kupffer cells remain phagocytic and adherent to the plastic substratum. The sinusoidal endothelial cells purified from the 12-20% Stractan interface likewise are typical in appearance; they retain intracellular fluorescence when labeled with DiI and secrete measurable amounts of matrix proteins (2,22). In conclusion, Stractan density centrifugation is a reproducible and accurate technique for isolating lipocytes, Kupffer cells, and endothelial cells from a nonparenchymal digest of a single normal rat liver, involving low cost and avoiding the need for centrifugal elutriation. ACKNOWLEDGMENTS The authors gratefully acknowledge the technical assistance of Mark Malamud and Alan Sato and the typing and graphics skills of Heather Mates.

REFERENCES 1. Wake, K. (1980) Inc. Rev. Cytol. 66, 303-353. 2. Friedman, S. L., Roll, F. J., Boyles, J., and Bissell,

218

3. 4.

5. 6.

7. 8. 9.

FRIEDMAN

D. M. (1985) Proc. Natl. Acad. Sci. USA 82, 8681-8685. Blomhoff, R., Holte, K., Naess, L., and Berg, T. (1984) Exp. Cell Res. 150, 186-193. Blomhoff, R., Rasmussen, M., Nilsson, A., Norum, K. R., Bert, T., Blaner, W. S., Kato, M., Mertz, J. R., Goodman, D. S., Eriksson, U., and Peterson, P. A. (1985) J. Biol. Chem. 260, 1356013565. Corash, L. M., Piomelli, S., Chen, H. C., and Gross, E.(l974)J. Lab. Clin. Med. 84, 147-151. Roll, F. J., and Madri, J. A. (1982) in Immunochemistry of the Extracellular Matrix, Methods and Application (Furthmayr, H., Ed.), Vol II, pp. 49-88, CRC Press, New York. Irving, M. G., Roll, F. J., Huang, S., and Bissell, D. M. (1984) Gastroenterology 87, 1233-1247. Pitas, R. E., Boyles, J., Mahley, R. W., and Bissell, D. M. (1985) J. Cell Biol. 100, 103-I 17. Wells, A. F., Miller, C. E., and Nadel, M. V. (1966) Appl. Microbial.

14, 27 1-275.

10. Popper, H. (1941) Arch. Pathol. 31, 1l-32. 11. Knook, D. L., Seffelaar, A. M., and de Leeuw, A. M. (1982) Exp. Cell Res. 139,468-47 1. 12. Smedsrod, B., and Pertoft, H. (1985) J. Leukocyte Biol. 38, 2 13-230.

AND ROLL 13. Girardi, A. J., McMichael, H., and Henle, W. A. (1956) Virology 2, 532-544. 14. Bienkowski, R. S., Cowan, M. J., MacDonald, J. A., and Crystal, R. G. (1978) J. Biol. Chem. 253, 4356-4363. 15. Labarca, C., and Paigen, K. (1980) Anal. Biochem. 102,344-352.

16. Knook, D. L., Blansjaar, N., and Sleyster, E. Ch. (1977) Exp. Cell Rex 109, 317-329. 17. Yokoi, Y., Namihisa, T., Kuroda, H., Komatsu, I., Miyazaki, A., Watanabe, A., and Usui, K. (1984) Hepatology

4,709-7

14.

18. Corash, L., Tan, H., and Gralnick, H. R. (1977) Blood

49,71-87.

19. Pertoft, H.. Rubin, K., Kjellen, L., Laurent, T. C., and Klingeborn, B. (1977) Exp. Cell Rex 110, 449-457. 20. DeLeeuw, A. M., McCarthy, S. P., Geerts, A., and Knook, D. ( 1984) Hepatology 4, 392-403. 21. Earnest, D. L., Hendricks, H. F. J., Brouwer, A., de Leeuw, A. M., Wennekers, H. M., and Knook, D. M. (1986) Gastroenterology 90, 1723. 22. Maher, J. J., Friedman, S. L., Roll, F. J., and Bissell, D. M. (1985) Hepatology 5, 1038. 23. Friedman. S. L., and Blaner, W. S. (1986) Hepatology 6, 1136.