In VitroCell.Dev.BioL27A:312-326.April1991 © 1991TissueCultureAssociation 0883-8364/91 $01.50+0.00
A COMPARISON OF PRIMARY CULTURES OF RAT CEREBRAL MICROVASCULAR E N D O T H E L I A L C E L L S T O RAT AORTIC ENDOTHELIAL CELLS ELLEN L. GORDON, PER E. DANIELSSON, THIEN-SON NGUYEN, AND H. RICHARD WINNl
Department of Neurological Surgery (ZA-86), Universityof Washington School of Medicine, HarborviewMedical Center, 325 Ninth Avenue, Seattle, Washington 98104 (Received 16 July 1990; accepted 5 December 1990)
SUMMARY A method to culture rat cerebral microvascular endothelial cells (RCMECs) was developed and adapted to concurrently obtain cultures of rat aortic endothelial cells (RAECs) without subculturing, cloning, or "weeding." The attachment and growth requirements of endothelial c e l clusters from isolated brain microvessels were first evaluated. RCMECs required fetal bovine serum to attach efficiently. Attachment and growth also depended on the matrix provided (fibronectin laminin >> gelatin > poly-D-lysine ~ Matrigel > hyaluronic acid ~ plastic) and the presence of endothelial cell growth supplement and heparin in the growth medium. Non-endothelial ceils are removed by allowing these cells to attach to a matrix that RCMECs attach to poorly (e.g., poly-D-lysine) and then transferring isolated endothelial cell clusters to fibronectin-coated dishes. These cell cultures, labeled with 1,1 '-dioctadecyl-3,3,3',3 "tetramethyl-indocarboxyamineperchlorate (DiI-Ac-LDL) and analyzed using flow cytometry, were 97.7 + 2.6% (n = 6) pure. By excluding those portions designed to isolate brain microvessels, the method was adapted to obtain RAEC cultures. RAECs do not isolate as clusters and have different morphology in culture, but respond similarly to matrices and growth medium supplements. RCMECs and RAECs have Factor VIII antigen, accumulate DiI-Ac-LDL, contain Weibel-Palade bodies, and have complex junctional structures. The activities of "y-gl.utamyltransferase and alkaline phosphatase were measured as a function of time in culture. RCMECs had higher enzymatic activity than RAECs. In both RCMECs and RAECs enzyme activity decreased with time in culture. The function of endothelial cells is specialized depending on its location. This culture method allows comparison of two endothelial cell cultures obtained using very similar culture conditions, and describes their initial characterization. These cultures may provide a model system to study specialized endothelial ceil functions and endothelial cell differentiation.
Key words: vascular endothelium; culture; rat; aorta; cerebral; matrix.
INTRODUCTION
in culture (25,40). In addition, culture conditions themselves may alter endothelial cell function (2,21). We have developed a method to culture the endothelial cells of microvessels isolated from rat brain (23) and have adapted this method to obtain endothelial cell cultures from rat aorta. These methods allow comparison of two types of endothelial cells which have been isolated from the same animals and cultured under similar conditions. Many protocols report culturing endothelial cells from the cerebrum. These techniques sometimes employ larger animals, which offers obvious advantages in terms of starting material (5,8,37). Protocols using smaller animals (including rat) have also been published (4,11,34). Early reports of cerebral vascular cells in culture describe mixed cultures in which endothelial cells became senescent due to a lack of endothelial cell growth factor or use of matrices (34,35,43). More recent techniques have exploited the availability of growth-enhancing medium supplements and matrices. However, these techniques require large amounts of starting material (4,5,8), and/or "cloning" by mechanical or enzymatic removal of endothelial cell colonies and subsequent expansion of the primary culture (8,11), and/or mechanical removal of non-endothelial cells (weed-
Endothelium, the single cell lining of all blood and lymph vessels, plays a variety of crucial roles in the cardiovascular system. In addition to providing a nonthrombogenic barrier and participating in the control of vascular tone (7), the endothelium of different tissues has further specialized functions. For instance, in the brain the endothelium helps comprise the blood-brain barrier. Brain endothelia are joined by continuous tight junctions, have low permeability and few pinocytic vesicles, and have specialized transport systems and greater mitochondrial content than the endothelium of non-neural tissues (32). By contrast, the endothelium of the aorta, a capacitance vessel, although also forming continuous tight junctions has numerous pinocytic vesicles, high permeability, fewer mitochondria, and, to date, no reported specialized transport systems (32,42). The heterogeneity of endothelial cells is evidenced in vivo in vessels of different types (artery, vein, capillary), tissue origin, and species (19,26,45). Endothelial cells also exhibit heterogeneity
1 To whom requests for reprints should be addressed. 312
313
RAT ENDOTHELIAL CELLS IN CULTURE ing) (39) or transformation of the cells (38) to obtain sufficient numbers of endothelial cells for metabolic studies. Our goal was to develop a primary culture protocol that would reduce the amount of starting material required and yield cultures of sufficient purity so that mechanical removal of non-endothelial cells, cell sorting (44), cloning of colonies, and/or subculturing was unnecessary, Having succeeded in obtaining primary cultures of rat cerebral microvascular endothelial cells (RCMECs) we modified this culture protocol to obtain cultures of rat aortic endothelial cells (RAECs). Relatively fewer publications involving the culture of RAECs have been published (1,29,31). Most RAEC culture protocols employ an explant technique (29), and obtaining cultures of RAECs has been problematic. This is due to the rat aorta's small size and reported difficulties in culturing rat endnthelium in general (14,29). By adapting the isolation protocol for rat cerebral microvessel endothelium to the rat aorta, using the same enzyme solution to dissociate the endothelial cells from the vessel wall, and employing the same growth medium and matrix we have established endothelial cell cultures from the rat aorta. These RAEC cultures, prepared in conjunction with RCMEC cultures, may allow the study of the variability of endothelial cell function in a more controlled manner.
Phosphate buffered saline (PBS-A) = KC1 (2.68 mM), NaCI (136.9 mM), KH~PO4 (1.47 mM), NaHPO4 • 7HzO (8.06 mM), pH 7.2. PBS = PBS-A with MgC12.6H20 (0.49 mM) and CaC12- 2H20 (0.90 raM). Dissecting medium (DM) = M199 with 20 mM HEPES, 20 mM sodium bicarbonate, 1× antibiotic, antimycotic solution, and 50 gg/ml heparin, pH 7.2 with 5% EPDS added unless otherwise indicated. RCMEC growth medium = DMEM with 15% EPDS, 4% FBS, 100 ttg/ml ECGS, 50 #g/ml heparin, 1 mM pyruvate, 2 mM L-glutamine, 1× nonessential amino acids, IX vitamins, 1)< antibiotic, antimycotic solution. Trypsin was diluted to 0.25% with PBS-A. Percoll gradient; percoll (43 ml), 10× M199 buffer (5 ml), 1.5 ml of 1 Mstock HEPES, 0.5 ml antibiotic-antimycoticsolution, 50 ml DM with 10% EPDS. Aliquot to 50 ml Oak Ridge centrifuge tubes. Centrifuged at 25 100 × g in fixed angle rotor (Beckman JA-17) for 70 rain. (Percoll gradients may be stored refrigerated for several weeks.) Aliquots of ECGS, fibronectin, and enzymes are stored as concentrated stock solutions at - 2 0 ° C.
MATERIALSANDMETHODS
Matrices
Tissue Culture Solutions
Tissue Culture Supplies All reagents were tissue culture or analytical grade. Medium 199 (M199) (modified); 100× antibiotic, antimycotic solution (10 000 U/ml penicillin, 10 000 #g/ml streptomycin, 2.5 #g/ml Fungizone); Dulbecco's modification of Eagle's medium (DMEM) without L-glutamine, with 1 g/liter dextrose; sodium pyruvate (100 mM); 100× nonessential amino acids for minimum essential medium (Eagle); 100× vitamins for minimum essential medium Eagle (modified); trypsin (2.5%), all from Flow Lab. Inc., McLean, VA. Sodium bicarbonate; heparin (porcine intestinal mucosa, sodium salt grade II); dextran (clinical grade, av. molecular weight 79 100); 4,6-diamidino-2-pheuyhndole (DAPI); poly-D-lysine (hydrobromide); hydrocortisone; albumin (bovine fraction V); hyaluronic acid (grade I, from human umbilical cord), all from Sigma, St. Louis, MO. 4-(2-Hydroxyethyl)-l-piperazine-ethanesulfonicacid (HEPES, ultra-pure grade); protease (neutrale, dispase); collagenase-dispase from Achromobacter iophagus /BaciUus polymyxa; Collagenase from Clostridium histolyticuum; fibronectin (human), all from Boehringer Mannheim Biochemicals, Indianapolis, IN. Endothelial mitogen (ECGS); fibronectin (human); 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarboxyamine perchlorate (DiI-Ac-LDL); Anti-ghal fibrillary acidic protein (GFAP) (rabbit) polyclonal antiserum, all from Biomedical Technologies Inc., Stoughton, MA. Percoll from Pharmacia, Inc., Pleasant Hill, CA. Rabbit X human Factor VIII antigen from Accurate Chemical & Scientific Corp., San Diego, CA. Second antibody fluorescein isothiocyanate (FITC) APA to rabbit IGG from ICN Immunobiologicals, Lisle, IN. Oak Ridge Centrifuge tubes, 50 ml polycarbonate, Nalgene, Rochester, NY. Nitex nylon mesh from Tetko, Inc., Elmsford, NY. Endothelial cell growth supplement (CR-ECGS), laminin and Matrigel from Collaborative Research, Bedford, MA. Custom processed equine plasma derived serum (EPDS), prepared by Pel-Freez, Rogers, AK. Fetal bovine serum (FBS), from Hyclone, Logan, UT. Gelatin (calf skin), from Eastman Kodak Co., Rochester, NY.
Fibronectin was diluted for use with DMEM, and dishes were coated with 1 to 5 gg/cm 2 for at least 30 min at 37" C. The fibronectin solution is aspirated just before the addition of cells. Poly-o-lysine solution is diluted to 20 #g/ml with sterile HzO and incubated in culture dishes (2.5 ttg/cm ~ for 5 rain), the solution is then aspirated, the dishes rinsed twice with sterile H20, and air dried overnight in a cell culture hood. Gelatin was dissolved in water and sterilized by autoclaving. Dishes were coated for at least 1 h at 37 ° C with either 1% gelatin in H20 or 2% gelatin in medium (2.36 g gelatin in 100 ml H20 is autoclaved and 10 ml of 10X medium, 20 ml serum and 3.5 ml of 7.5% bicarbonate solution are added). Before plating, the gelatin is aspirated and the dish rinsed with DM. Laminin was diluted to 10 #g/ml in growth medium and applied to the culture dish 45 min before plating cells.
Cell Lines A bovine aortic endothelial cell (BAEC) line was generously provided by Dr. John Harlan (University of Washington, Seattle), and maintained in the same culture medium as RCMEC. C6 ghoma cells were obtained from the American Type Culture Collection (ATCC, CCL 107, Rockville, MD), and maintained in RCMEC culture medium (without ECGS). A human glioma line, UW28. was used as a positive control for GFAP immunochemistry experiments. UW28 (provided by Dr. Francis Ali-Osman, University of Washington, Seattle) was maintained in 10% FBS in DMEM with 1× antibioticantimycotic.
Routine Cell Culture Methods Albumin blocks the adherence of microvessel preparations to glassware (8), therefore all glassware was rinsed using DM (with 5% EPDS) before use. Cells were counted using a hemacytometer (0.25% trypsin, 20 to 30 rain, 37 ° C, with further dissociation by repeated pipetting).
314
GORDON ET AL.
Viability was assessed by exclusion of trypan blue dye. Protein was determined by the method of Bradford (6). Alkaline phosphatase (AKP) and 7-glutamyhransferase ('y-glutamyl transpeptidase, 7-GT) activities were measured using ReagentSet assay kit no. 396460 and FastChem 7-GT kit no. 158208, Boehringer Mannheim Diagnostics, Houston, TX. Cellular uptake of DiI-Ac-LDL was accomplished as described by Voyta et al. (44). Cells were visualized with a Nikon Diaphot phase contrast inverted microscope with EPI fluorescence attachments, using standard rhodamine excitation-emission filters. For flow cytometry, cultures were analyzed on a ORTHO 5 0 + + Flow Cytometer. C6 glioma cells were used as the negative control and BAEC as the positive control for setting gates. For Factor VIII staining cells were fixed in methanol at - 2 0 ° C for 10 min, rehydrated with PBS for 10 min and incubated with rabbit anti-human Factor VIII (1:100 dilution) for 1 h at 37 ° C. The second antibody was goat anti-rabbit IGG with FITC label (1:32 dilution, for 1 h at 37 ° C). BAEC and C6 glioma cells were used as positive and negative controls, respectively. For GFAP staining, cells were treated as for Factor VIII staining except the primary antibody was used at a 1:200 dilution. UW28 glioma cell line and BAEC were used as positive and negative controls, respectively. Cultures grown on Lab-Tech slides coated with fibroneetin were used to test for mycoplasma contamination. Cells were fixed with 100% methanol, air dried, and stained with 1 #g/ml DAPI, then observed at 385 nm on Zeiss EPI fluorescent microscope for nonnuclear DNA.
Electron Microscopy Transmission electron microscopy(TEM). Cells in 35-mm culture dishes or tissue samples in microfuge tubes were rinsed with 0.1 M cacodylate buffer, pH 7.4, and fixed for 15 min in 2% p-formaldehyde/0.5% glutaraldehyde (wt/vol in cacodylate buffer), then for 4 b in 1% p-formaldehyde/1.5% (wt/vol in eacodylate buffer) at 4 ° C. Postfixation was carried out in 1% osmium tetraoxide (vol/vol) in cacodylate buffer with 0.75% potassium ferrocyanide (wt/vol). The samples were dehydrated with ethanol and embedded in MEDCAST with 2% DMP-30. Samples prepared in tissue culture dishes were removed using liquid nitrogen and remounted for thin sectioning. Thin sections were stained with uranyl acetate and lead citrate and examined with a JEOL 100B electron microscope. Scanning electron microscopy (SEM). Tissue samples were fixed for 15 min in 2% p-formaldehyde/0.5% glutaraldehyde (wt/ vol) in 0.1 Mcacodylate buffer, pH 7.4, at 4 ° C, then for 3 h in 3% glutaraldehyde in cacodylate buffer. Postfixation was carried out in 1% osmium tetraoxide (vol/vol in cacodylate buffer) for 20 min. Pelco tissue carriers lined with 5-#m nylon mesh were used for dehydration with ethanol and Freon exchange. Tissue was mounted on stubs coated with colloidal silver paste, sputter coated with palladium/gold alloy, and viewed on a JEOL 840A scanning electron microscope. For electron microscopy, chemicals were obtained from Sigma, and microscopy supplies from Ted Pella, Inc., Redding, CA. High PerformanceLiquid Chromatography(HPLC)Analysis HPLC analysis of adenine nucleotides was an adaptation of the method of Geisbuhler et al. (18). Briefly, nucleotides are extracted
PRIMARY CULTURE OF RAT AORTIC AND CEREBRAL MICROVESSEL ENDOTHELIAL CELLS
ISOLATE TISSUE*
/
\
ISOLATE CEREBRALHEMISPHERES
ISOLATETHORACIC AORTA
MINCE
REFRIt ;ERATE
¢
HOMOGENIZE FILTER THROUGH 149 M|CRON MESH INCUBATE WITH PROTEASE ISOLATE VESSELS USING CENTRIFUGATIONDExTRANx~THROUGH
CUT INTO 1-2 MM RINGS /
INCUBATE WITH COLLAGENASE/DISPASE
L
PERCOLL GRADIENT CENTRIFUGATION t ALIQUOT TO 24 WELL DISH AND COLLECTTHOSE WELLSCONTAINING ENDOTHELIAL CELL CLUSTERS FILTER RESUSPEND IN FEED MIX AND PLATE TO POLY-D-LYSINE OR GELATIN COATED DISHES INCUBATE 4 HOURS TRANSFER TO FIBRONECTINCOATED DISHES "THE PREPARATIONIS WASHEDUSINGDM AND CENTRIFUGAl'ION(800 X G) BETWEENEACH STEP
FIG. 1. Flow chart of RCMEC and RAEC primary culture protocols.
from cells or tissues by perchloric acid/EGTA treatment, and neutralized extracts then analyzed by HPLC using a Whatman Partisil SAX cartridge system (4.6 × 125 ram) with an anion exchanger (AX) guard column, and a 5-era precolumn of silica Pre-Column Gel (Whatman Intern'l, Ltd., Maidstone, England). The detector was set at 254 nm, and flow maintained at 1.5 ml/min. The HPLC program was as follows: Buffer A was run isocratically for 6 rain followed by a linear gradient to 90% buffer B from 6 to 40 min, and 40 to 43 min isocraticaUy with 90% buffer B (buffer A = 0.005 M KH2PO4, low-absorbance grade, EM Science, Cherry Hill, NJ; buffer B = 0.600 M KH2POa, both pH 4.5). Chromatographic peaks were identified and quantified by comparison to known standards.
Primary Culture Methods An outline of our final protocols for culture of RCMEC and RAEC appears in Fig. 1. We developed culture conditions for RCMEC first, assuming that the cerebral cells would require a more complex method and be more fastidious in their growth requirements than the RAEC (16). Subsequently, the method employed to culture RCMEC was adapted to RAEC. During the course of developing tissue culture methods, different sets of homogenizers, lots of enzymes, or concentrations of enzymes resulted in different amounts of tissue dissociation. The success of the culture was criti-
RAT ENDOTHELIAL CELLS IN CULTURE
315
TABLE 1 EVALUATION OF MICROVESSEL ISOLATION Sample
Energy Charge (I/2 ADP + ATP)/(ATP + ADP + AMP)
Brain" Brain homogenate Microvessels after dextran Cell clusters after percoll
0.915 0.278 0.700 0.865
+ + + +
0.004 0.051 0.060 0.049
(4)b (7) (5) (3)
3t-GT Aclivily, U/rag protein
AKP Activily, U/rag protein
-0.0053 + 0.0010 (5) 0.0591 + 0.0170 (5) 0.0943 + 0.0203 (3)
-0.0579 + 0.0110 (5) 0.4263 + 0.0719 (5) 1.5695 + 0.6993 (3)
By freeze-blow technique (48). b Values reported as mean _+ SD [n), where n is the number of preparations studied. Each preparation was analyzed in triplicate and the average value used. °
cally dependent on adjusting these factors to achieve maximum yield and purity of the cells plated out. In the protocol description, ranges of treatment are listed to reflect these adjustments.
Primary Culture of RCMECs Male Sprague-Dawley rats (6 rats of 35 to 50 g) were killed by the anesthetic halothane, and their brains removed. The isolated brains were placed in 70% ethanol (2 to 4 min) and then rinsed in DM. The thoracic aortae are also removed at this time, placed in DM, and refrigerated. The cerebral hemispheres are dissected away from the brain stem (yielding about ~ 1.2 g of tissue per rat), cut into small pieces, and rinsed several times in DM. The cerebral hemispheres are next homogenized [Wheaton-Dounce, 40 ml, 2 to 6 times with pestle B (clearance = 0.0025 to 0.0035 in.) followed by 2 to 10 times with pestle A (clearance = 0.001 to 0.003 in.)]. The amount of homogenization is dependent on the particular homogenizer. After homogenization the tissue is filtered through a 149-#m nylon mesh to remove large vessels, and then washed with DM containing no serum. (Placed in a 50-ml tissue culture tube and centrifuged at 800 Xg for 5 to 10 min, the superuatant is then aspirated and the pellet suspended again.) The homogenate is further dissociated using protease [0.005 to 0.05% protease (wt/vol) in 30 ml of DM containing no serum, in a 125-ml screw-top Ehrlenmeyer flask, 37 ° C, shaking water bath for 1.0 to 5.0 hi. After protease treatment the microvessels are separated from neurons and their supporting ghal cells by centrifugation through dextran (15% wt/vol in DM). The protease-treated homogenate is washed in DM, and the pellet suspended again by vigorous vortexing in the dextran solution (high setting for at least 3 rain) and then centrifuged (4 ° C, 4500 × g for 10 min, using a swinging bucket rotor, Beckman JS 5.2). The fat pad and dextran are collected to a new centrifuge tube and vortexed and centrifuged once more. The pellets from the two dextran spins are collected and washed in DM. The microvessels are further dissociated using collagenase/dispase [0.035 to 0.050% collagenase/dispase (wt/vol) in 10 ml DM, 125-ml screw-topped Ehrlenmeyer flask, 37 ° C, shaking water bath, for 6 to 18 h]. The collagenase/dispase-treated microvessels are centrifuged and suspended in 1 ml DM, layered across the top of a percoll gradient and centrifuged to separate cells (1650 ×g, 10 rain, 4 ° C). The percoll gradient was divided into 1.5-ml aliquots in a 24-well plate and the portions of the gradient containing endothelial cells identified. These wells were collected and the cells washed in DM. (percoll is toxic.) The endothelial ceils were present in approximately the middle third of the gradient, or a density range of ~ 1.04 to 1.07 g/ml. Finally, the ceils were suspended in culture medium for plating. The
best results were achieved using culture medium composed of 15% EPDS, 4% FBS, 1X antibiotic, antimycotic solution, 100 ~tg/ml endothelial mitogen, 50 ~g/ml heparin, 1 X pyruvate, 1X glutamine, 1X nonessential amino acids, 1X vitamins in DMEM (RCMEC growth medium). Cells plated directly to fibronectincoated tissue culture dishes (5/.tg/cm 2) sometimes gave good resuits (cultures > 90% purity as determined by Dil-Ac-LDL). However, on many occasions this was not the case (cultures of 50 to 60% purity as determined by DiI-Ac-LDL). Even greater success was achieved by first filtering through a double layer of 10-#m mesh and placing the cell-cluster isolates in uncoated dishes or dishes coated with poly-D-lysine or gelatin and allowing non-endothelial cells to attach. After 4 h, unattached cell clusters were transferred to fibronectin-coated dishes. Resulting primary cultures of RCMECs were of very high purity. The cells were rinsed and fed the day after their plating, and reached confluence in 5 to 8 days. Yield varies, but typically six rats yield 15 to 20 million cells, covering 120 to 160 cm 2.
Primary Culture of RAEC As mentioned above, our strategy for culture of RAECs was to use culture conditions as close to those of RCMECs as possible (Fig. 1). While the brain homogenate was incubating in pro/ease, the isolated rat aortae were processed for cell culture. The periadventitial fat was removed, and the aortae washed with DM and cut into rings 1 to 2 mm wide. The aorta rings were transferred with a wide-mouth pipette to a 15-ml centrifuge tube and rinsed several times in DM with 5% EPDS. The rings were not incubated with protease (as are the RCMECs) because this would digest the aorta wall, releasing non-endothelial cells, nor is it necessary to use dextran. The endothelial cells were released from the rings by incubation in collagenase/dispase (10 ml of 0.035% wt/vol in DM in a 50-ml, screw-topped Ehrlenmeyer flask, 37 ° C shaking water bath, 3 to 4 h). After collagenase/dispase treatment the rings were transferred to a 15-ml centrifuge tube. The tube was filled to the top by rinsing the Ehrlenmeyer flask with additional DM. It is critical at this step to vortex on high for at least 2 min to separate the endothelial cells from the vessel wall. The rings and endothelial cells were washed, suspended in DM and vortexed again. After centrifugation the rings and endothelial cells were suspended in 2 ml of DM. This was layered onto the top of a percoll gradient using a wide-mouth pipette. The culture protocol proceeds as with RCMECs, except that the percoll is aliquoted using a wide-mouth pipette to recover the aortic rings. Non-endothelial cells were found mostly in the top third of the percoll gradient, whereas the endothelial cells and aortic rings were in the next half, and the final sixth of the gradient was mostly
316
GORDON ET AL.
red blood cells. Cells and rings were plated out as with RCMECs. When cells are rinsed and fed the next day, it is important to remove all pieces of aorta as they will give rise to cells (either smooth muscle cells or fibroblasts) which migrate from the vessel wall onto the tissue culture dish. RESULTS The development of the culture techniques proceeded on three fronts. It was necessary to determine the appropriate conditions for the isolation of endothelial cell clusters, the selective attachment of endothelial cells, and the selective growth of endothelial cells. The success of different culture protocols was initially evaluated by observing cell morphology and uptake of DiI-Ac-LDL. Endothelial cells (and macrophage cells) take up DiI-Ac-LDL by a receptor mediated process, and the DiI fluorescence probe accumulates in lysosomal membranes (44). DiI-Ac-LDL is therefore a convenient marker because cells may be identified without fixation and staining steps. The purity of the cell cultures could be evaluated any time after the cells had attached simply by adding DiI-Ac-LDL to the culture medium and observing the cells, using fluorescence microscopy.
Isolation of RCMECs We followed the vessel isolation protocol using HPLC analysis to determine cellular energy charge (1/2ADP + ATP)/(ATP + ADP + AMP) (3), and enzymatic assays for alkaline phosphatase and 3'-GT, proposed markers for microvascular and brain endothelium (22,49). Phase microscopy, TEM, and SEM were used to help assess the purity, viability, and ultrastructure of the freshly isolated vessels. Table 1 lists the cellular energy charge at different points in the microvessel isolation process along with "y-GT and AKP enzyme activities. These data reflect the damage necessary to isolate the cells and the day-to-day variability associated with techniques of this type. The microvessels are still metabolically viable (76% of initial energy charge) and accumulate radiolabeled adenosine into their adenine nucleotide pools after both homogenization and protease treatment (24). After the cerebral hemispheres were homogenized and passed through the 149-#m nylon mesh, samples were prepared for SEM (Fig. 2 a) and TEM (Fig. 2 B-D) analysis. Phase contrast microscopy and SEM analysis confirmed that the majority of vessels present after the homogenate was filtered (149-#m mesh) were less than ~ 6 0 #m in diameter. TEM analysis (Fig. 2 B-D) of the filtered homogenate revealed that the majority of vessels were capillary vessels and small arterioles and venules. Vessels with one or two layers of smooth muscle cells were occasionally seen, but never any larger vessels, (SEM analysis of the material collected on the 149#m nylon mesh detected many larger vessels.) TEM analysis (Fig. 2 B-D) illustrates that rat brain microvessels are heavily invested by pericytes, with both endothelium and pericytes being surrounded by basement membrane. Astrocytic foot processes then surround and enclose the basement membrane. In larger microvessels, smooth muscle cells replace pericytes. The TEM analysis shows that although the isolated microvessels seem quite pure by SEM and phase contrast microscopy, the structure of the microvessels makes the isolation of endothelial cells ditficult. After the protease and dextran step of the protocol, the vessels
start to dissociate from the surrounding tissue. Next, collagenase/ dispase treatment was used to digest the basement membrane and release endothelial cells from the other cellular components of the microvessels. Several protease and collagenase preparations were analyzed using SEM, and it was found that within a preparation and from preparation to preparation the extent of enzymatic digestion varied widely. The preparation consisted of red blood cells, single cells, debris, and clusters of rounded cells. In a few experiments, albumin (25%) was used in place of dextran (47). A higher yield of microvessels was obtained with albumin. However, subsequent cell plating and growth were worse. Different types of collagenase treatment were also investigated. Many investigators have tried Ca++/Mg ++-free buffers for collagenase incubation. Using collagenase (CIostridiumhistolyticuum) or collagenase/ dispase we found that viability was decreased in Ca++/Mg++-free buffers. In fact, we found addition of 5% serum (EPDS) to every step of the protocol (except the protease treatment) resulted in better yield and viability of cells. Initially, the different fractions of the percoll gradient were washed and plated to fibronectin or gelatin-coated dishes, and the growth from each fraction evaluated by morphology and uptake of DiI-Ac-LDL. Figure 3 A-B illustrates this approach, showing some morphologies seen in primary cultures and the positive DiI-Ac-LDL labeling of some cell colonies. During initial outgrowth the morphology of endothelial cell colonies were similar to either the "classic cobblestone" morphology typical of BAECs or, more often, an elongated swirling pattern described in many reports of cultured mierovascular endothelial ceils. As the cultures matured and the cells in the colonies multiplied and cell density increased, most endothelial cells became elongated and assumed a swirling pattern (Fig. 3 C-D). We observed that the greatest number of endothelial cell colonies were derived from very refractive clusters of cells in the middle third of the percoll gradient (density 1.04 to 1.07 g/ml). The majority of these clusters exclude trypan blue dye (>90%). Many single cells are present in this portion of the gradient as well, but their viability is generally low (30 cells) the cultures derived from them are mixed. The task therefore was to adjust the isolation procedure (homogenization, protease treatment, and collagenase treatment) to maximize the yield of these clusters. Enzymes are purchased in large lots, and once isolation conditions are established, they are prealiquoted and stored frozen ( - 2 0 ° C). If the yield at the percoll step decreases, homogenization, enzyme concentration, or incubation times are decreased. Conversely, if large clusters of cells are obtained, homogenization, enzyme concentration, or incubation times are increased.
Attachment and Growth of RCMECs The isolation procedure enriches for endothelial cells but other cell types are also present. Fibroblasts, pericytes, smooth muscle, and glial cells are all possible contaminants when the "purified" cell preparations are plated out. It was necessary to choose conditions that selected for attachment and growth of RCMECs and if possible inhibited attachment and growth of non-endothelial cell types. Once the percoll fractions yielding the greatest numbers of endothelial cell colonies were identified, the fractions were pooled and
RAT ENDOTHELIAL CELLS IN CULTURE
317
FIG. 2. SEM and TEM photomicrographs of brain microvessels. Obtained by dextran centrifugation of fihered (149-#m mesh) brain homogenate. Note the close association of the cells of the vessel wall. This is responsible for the difficulty of isolating pure endothelial cell preparations from the rat brain. A, SEM, bar = 10 /.tm. B, TEM, X10 000 and (C and D). Enlargement of B, )
A COMPARISON OF PRIMARY CULTURES OF RAT CEREBRAL MICROVASCULAR E N D O T H E L I A L C E L L S T O RAT AORTIC ENDOTHELIAL CELLS ELLEN L. GORDON, PER E. DANIELSSON, THIEN-SON NGUYEN, AND H. RICHARD WINNl
Department of Neurological Surgery (ZA-86), Universityof Washington School of Medicine, HarborviewMedical Center, 325 Ninth Avenue, Seattle, Washington 98104 (Received 16 July 1990; accepted 5 December 1990)
SUMMARY A method to culture rat cerebral microvascular endothelial cells (RCMECs) was developed and adapted to concurrently obtain cultures of rat aortic endothelial cells (RAECs) without subculturing, cloning, or "weeding." The attachment and growth requirements of endothelial c e l clusters from isolated brain microvessels were first evaluated. RCMECs required fetal bovine serum to attach efficiently. Attachment and growth also depended on the matrix provided (fibronectin laminin >> gelatin > poly-D-lysine ~ Matrigel > hyaluronic acid ~ plastic) and the presence of endothelial cell growth supplement and heparin in the growth medium. Non-endothelial ceils are removed by allowing these cells to attach to a matrix that RCMECs attach to poorly (e.g., poly-D-lysine) and then transferring isolated endothelial cell clusters to fibronectin-coated dishes. These cell cultures, labeled with 1,1 '-dioctadecyl-3,3,3',3 "tetramethyl-indocarboxyamineperchlorate (DiI-Ac-LDL) and analyzed using flow cytometry, were 97.7 + 2.6% (n = 6) pure. By excluding those portions designed to isolate brain microvessels, the method was adapted to obtain RAEC cultures. RAECs do not isolate as clusters and have different morphology in culture, but respond similarly to matrices and growth medium supplements. RCMECs and RAECs have Factor VIII antigen, accumulate DiI-Ac-LDL, contain Weibel-Palade bodies, and have complex junctional structures. The activities of "y-gl.utamyltransferase and alkaline phosphatase were measured as a function of time in culture. RCMECs had higher enzymatic activity than RAECs. In both RCMECs and RAECs enzyme activity decreased with time in culture. The function of endothelial cells is specialized depending on its location. This culture method allows comparison of two endothelial cell cultures obtained using very similar culture conditions, and describes their initial characterization. These cultures may provide a model system to study specialized endothelial ceil functions and endothelial cell differentiation.
Key words: vascular endothelium; culture; rat; aorta; cerebral; matrix.
INTRODUCTION
in culture (25,40). In addition, culture conditions themselves may alter endothelial cell function (2,21). We have developed a method to culture the endothelial cells of microvessels isolated from rat brain (23) and have adapted this method to obtain endothelial cell cultures from rat aorta. These methods allow comparison of two types of endothelial cells which have been isolated from the same animals and cultured under similar conditions. Many protocols report culturing endothelial cells from the cerebrum. These techniques sometimes employ larger animals, which offers obvious advantages in terms of starting material (5,8,37). Protocols using smaller animals (including rat) have also been published (4,11,34). Early reports of cerebral vascular cells in culture describe mixed cultures in which endothelial cells became senescent due to a lack of endothelial cell growth factor or use of matrices (34,35,43). More recent techniques have exploited the availability of growth-enhancing medium supplements and matrices. However, these techniques require large amounts of starting material (4,5,8), and/or "cloning" by mechanical or enzymatic removal of endothelial cell colonies and subsequent expansion of the primary culture (8,11), and/or mechanical removal of non-endothelial cells (weed-
Endothelium, the single cell lining of all blood and lymph vessels, plays a variety of crucial roles in the cardiovascular system. In addition to providing a nonthrombogenic barrier and participating in the control of vascular tone (7), the endothelium of different tissues has further specialized functions. For instance, in the brain the endothelium helps comprise the blood-brain barrier. Brain endothelia are joined by continuous tight junctions, have low permeability and few pinocytic vesicles, and have specialized transport systems and greater mitochondrial content than the endothelium of non-neural tissues (32). By contrast, the endothelium of the aorta, a capacitance vessel, although also forming continuous tight junctions has numerous pinocytic vesicles, high permeability, fewer mitochondria, and, to date, no reported specialized transport systems (32,42). The heterogeneity of endothelial cells is evidenced in vivo in vessels of different types (artery, vein, capillary), tissue origin, and species (19,26,45). Endothelial cells also exhibit heterogeneity
1 To whom requests for reprints should be addressed. 312
313
RAT ENDOTHELIAL CELLS IN CULTURE ing) (39) or transformation of the cells (38) to obtain sufficient numbers of endothelial cells for metabolic studies. Our goal was to develop a primary culture protocol that would reduce the amount of starting material required and yield cultures of sufficient purity so that mechanical removal of non-endothelial cells, cell sorting (44), cloning of colonies, and/or subculturing was unnecessary, Having succeeded in obtaining primary cultures of rat cerebral microvascular endothelial cells (RCMECs) we modified this culture protocol to obtain cultures of rat aortic endothelial cells (RAECs). Relatively fewer publications involving the culture of RAECs have been published (1,29,31). Most RAEC culture protocols employ an explant technique (29), and obtaining cultures of RAECs has been problematic. This is due to the rat aorta's small size and reported difficulties in culturing rat endnthelium in general (14,29). By adapting the isolation protocol for rat cerebral microvessel endothelium to the rat aorta, using the same enzyme solution to dissociate the endothelial cells from the vessel wall, and employing the same growth medium and matrix we have established endothelial cell cultures from the rat aorta. These RAEC cultures, prepared in conjunction with RCMEC cultures, may allow the study of the variability of endothelial cell function in a more controlled manner.
Phosphate buffered saline (PBS-A) = KC1 (2.68 mM), NaCI (136.9 mM), KH~PO4 (1.47 mM), NaHPO4 • 7HzO (8.06 mM), pH 7.2. PBS = PBS-A with MgC12.6H20 (0.49 mM) and CaC12- 2H20 (0.90 raM). Dissecting medium (DM) = M199 with 20 mM HEPES, 20 mM sodium bicarbonate, 1× antibiotic, antimycotic solution, and 50 gg/ml heparin, pH 7.2 with 5% EPDS added unless otherwise indicated. RCMEC growth medium = DMEM with 15% EPDS, 4% FBS, 100 ttg/ml ECGS, 50 #g/ml heparin, 1 mM pyruvate, 2 mM L-glutamine, 1× nonessential amino acids, IX vitamins, 1)< antibiotic, antimycotic solution. Trypsin was diluted to 0.25% with PBS-A. Percoll gradient; percoll (43 ml), 10× M199 buffer (5 ml), 1.5 ml of 1 Mstock HEPES, 0.5 ml antibiotic-antimycoticsolution, 50 ml DM with 10% EPDS. Aliquot to 50 ml Oak Ridge centrifuge tubes. Centrifuged at 25 100 × g in fixed angle rotor (Beckman JA-17) for 70 rain. (Percoll gradients may be stored refrigerated for several weeks.) Aliquots of ECGS, fibronectin, and enzymes are stored as concentrated stock solutions at - 2 0 ° C.
MATERIALSANDMETHODS
Matrices
Tissue Culture Solutions
Tissue Culture Supplies All reagents were tissue culture or analytical grade. Medium 199 (M199) (modified); 100× antibiotic, antimycotic solution (10 000 U/ml penicillin, 10 000 #g/ml streptomycin, 2.5 #g/ml Fungizone); Dulbecco's modification of Eagle's medium (DMEM) without L-glutamine, with 1 g/liter dextrose; sodium pyruvate (100 mM); 100× nonessential amino acids for minimum essential medium (Eagle); 100× vitamins for minimum essential medium Eagle (modified); trypsin (2.5%), all from Flow Lab. Inc., McLean, VA. Sodium bicarbonate; heparin (porcine intestinal mucosa, sodium salt grade II); dextran (clinical grade, av. molecular weight 79 100); 4,6-diamidino-2-pheuyhndole (DAPI); poly-D-lysine (hydrobromide); hydrocortisone; albumin (bovine fraction V); hyaluronic acid (grade I, from human umbilical cord), all from Sigma, St. Louis, MO. 4-(2-Hydroxyethyl)-l-piperazine-ethanesulfonicacid (HEPES, ultra-pure grade); protease (neutrale, dispase); collagenase-dispase from Achromobacter iophagus /BaciUus polymyxa; Collagenase from Clostridium histolyticuum; fibronectin (human), all from Boehringer Mannheim Biochemicals, Indianapolis, IN. Endothelial mitogen (ECGS); fibronectin (human); 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarboxyamine perchlorate (DiI-Ac-LDL); Anti-ghal fibrillary acidic protein (GFAP) (rabbit) polyclonal antiserum, all from Biomedical Technologies Inc., Stoughton, MA. Percoll from Pharmacia, Inc., Pleasant Hill, CA. Rabbit X human Factor VIII antigen from Accurate Chemical & Scientific Corp., San Diego, CA. Second antibody fluorescein isothiocyanate (FITC) APA to rabbit IGG from ICN Immunobiologicals, Lisle, IN. Oak Ridge Centrifuge tubes, 50 ml polycarbonate, Nalgene, Rochester, NY. Nitex nylon mesh from Tetko, Inc., Elmsford, NY. Endothelial cell growth supplement (CR-ECGS), laminin and Matrigel from Collaborative Research, Bedford, MA. Custom processed equine plasma derived serum (EPDS), prepared by Pel-Freez, Rogers, AK. Fetal bovine serum (FBS), from Hyclone, Logan, UT. Gelatin (calf skin), from Eastman Kodak Co., Rochester, NY.
Fibronectin was diluted for use with DMEM, and dishes were coated with 1 to 5 gg/cm 2 for at least 30 min at 37" C. The fibronectin solution is aspirated just before the addition of cells. Poly-o-lysine solution is diluted to 20 #g/ml with sterile HzO and incubated in culture dishes (2.5 ttg/cm ~ for 5 rain), the solution is then aspirated, the dishes rinsed twice with sterile H20, and air dried overnight in a cell culture hood. Gelatin was dissolved in water and sterilized by autoclaving. Dishes were coated for at least 1 h at 37 ° C with either 1% gelatin in H20 or 2% gelatin in medium (2.36 g gelatin in 100 ml H20 is autoclaved and 10 ml of 10X medium, 20 ml serum and 3.5 ml of 7.5% bicarbonate solution are added). Before plating, the gelatin is aspirated and the dish rinsed with DM. Laminin was diluted to 10 #g/ml in growth medium and applied to the culture dish 45 min before plating cells.
Cell Lines A bovine aortic endothelial cell (BAEC) line was generously provided by Dr. John Harlan (University of Washington, Seattle), and maintained in the same culture medium as RCMEC. C6 ghoma cells were obtained from the American Type Culture Collection (ATCC, CCL 107, Rockville, MD), and maintained in RCMEC culture medium (without ECGS). A human glioma line, UW28. was used as a positive control for GFAP immunochemistry experiments. UW28 (provided by Dr. Francis Ali-Osman, University of Washington, Seattle) was maintained in 10% FBS in DMEM with 1× antibioticantimycotic.
Routine Cell Culture Methods Albumin blocks the adherence of microvessel preparations to glassware (8), therefore all glassware was rinsed using DM (with 5% EPDS) before use. Cells were counted using a hemacytometer (0.25% trypsin, 20 to 30 rain, 37 ° C, with further dissociation by repeated pipetting).
314
GORDON ET AL.
Viability was assessed by exclusion of trypan blue dye. Protein was determined by the method of Bradford (6). Alkaline phosphatase (AKP) and 7-glutamyhransferase ('y-glutamyl transpeptidase, 7-GT) activities were measured using ReagentSet assay kit no. 396460 and FastChem 7-GT kit no. 158208, Boehringer Mannheim Diagnostics, Houston, TX. Cellular uptake of DiI-Ac-LDL was accomplished as described by Voyta et al. (44). Cells were visualized with a Nikon Diaphot phase contrast inverted microscope with EPI fluorescence attachments, using standard rhodamine excitation-emission filters. For flow cytometry, cultures were analyzed on a ORTHO 5 0 + + Flow Cytometer. C6 glioma cells were used as the negative control and BAEC as the positive control for setting gates. For Factor VIII staining cells were fixed in methanol at - 2 0 ° C for 10 min, rehydrated with PBS for 10 min and incubated with rabbit anti-human Factor VIII (1:100 dilution) for 1 h at 37 ° C. The second antibody was goat anti-rabbit IGG with FITC label (1:32 dilution, for 1 h at 37 ° C). BAEC and C6 glioma cells were used as positive and negative controls, respectively. For GFAP staining, cells were treated as for Factor VIII staining except the primary antibody was used at a 1:200 dilution. UW28 glioma cell line and BAEC were used as positive and negative controls, respectively. Cultures grown on Lab-Tech slides coated with fibroneetin were used to test for mycoplasma contamination. Cells were fixed with 100% methanol, air dried, and stained with 1 #g/ml DAPI, then observed at 385 nm on Zeiss EPI fluorescent microscope for nonnuclear DNA.
Electron Microscopy Transmission electron microscopy(TEM). Cells in 35-mm culture dishes or tissue samples in microfuge tubes were rinsed with 0.1 M cacodylate buffer, pH 7.4, and fixed for 15 min in 2% p-formaldehyde/0.5% glutaraldehyde (wt/vol in cacodylate buffer), then for 4 b in 1% p-formaldehyde/1.5% (wt/vol in eacodylate buffer) at 4 ° C. Postfixation was carried out in 1% osmium tetraoxide (vol/vol) in cacodylate buffer with 0.75% potassium ferrocyanide (wt/vol). The samples were dehydrated with ethanol and embedded in MEDCAST with 2% DMP-30. Samples prepared in tissue culture dishes were removed using liquid nitrogen and remounted for thin sectioning. Thin sections were stained with uranyl acetate and lead citrate and examined with a JEOL 100B electron microscope. Scanning electron microscopy (SEM). Tissue samples were fixed for 15 min in 2% p-formaldehyde/0.5% glutaraldehyde (wt/ vol) in 0.1 Mcacodylate buffer, pH 7.4, at 4 ° C, then for 3 h in 3% glutaraldehyde in cacodylate buffer. Postfixation was carried out in 1% osmium tetraoxide (vol/vol in cacodylate buffer) for 20 min. Pelco tissue carriers lined with 5-#m nylon mesh were used for dehydration with ethanol and Freon exchange. Tissue was mounted on stubs coated with colloidal silver paste, sputter coated with palladium/gold alloy, and viewed on a JEOL 840A scanning electron microscope. For electron microscopy, chemicals were obtained from Sigma, and microscopy supplies from Ted Pella, Inc., Redding, CA. High PerformanceLiquid Chromatography(HPLC)Analysis HPLC analysis of adenine nucleotides was an adaptation of the method of Geisbuhler et al. (18). Briefly, nucleotides are extracted
PRIMARY CULTURE OF RAT AORTIC AND CEREBRAL MICROVESSEL ENDOTHELIAL CELLS
ISOLATE TISSUE*
/
\
ISOLATE CEREBRALHEMISPHERES
ISOLATETHORACIC AORTA
MINCE
REFRIt ;ERATE
¢
HOMOGENIZE FILTER THROUGH 149 M|CRON MESH INCUBATE WITH PROTEASE ISOLATE VESSELS USING CENTRIFUGATIONDExTRANx~THROUGH
CUT INTO 1-2 MM RINGS /
INCUBATE WITH COLLAGENASE/DISPASE
L
PERCOLL GRADIENT CENTRIFUGATION t ALIQUOT TO 24 WELL DISH AND COLLECTTHOSE WELLSCONTAINING ENDOTHELIAL CELL CLUSTERS FILTER RESUSPEND IN FEED MIX AND PLATE TO POLY-D-LYSINE OR GELATIN COATED DISHES INCUBATE 4 HOURS TRANSFER TO FIBRONECTINCOATED DISHES "THE PREPARATIONIS WASHEDUSINGDM AND CENTRIFUGAl'ION(800 X G) BETWEENEACH STEP
FIG. 1. Flow chart of RCMEC and RAEC primary culture protocols.
from cells or tissues by perchloric acid/EGTA treatment, and neutralized extracts then analyzed by HPLC using a Whatman Partisil SAX cartridge system (4.6 × 125 ram) with an anion exchanger (AX) guard column, and a 5-era precolumn of silica Pre-Column Gel (Whatman Intern'l, Ltd., Maidstone, England). The detector was set at 254 nm, and flow maintained at 1.5 ml/min. The HPLC program was as follows: Buffer A was run isocratically for 6 rain followed by a linear gradient to 90% buffer B from 6 to 40 min, and 40 to 43 min isocraticaUy with 90% buffer B (buffer A = 0.005 M KH2PO4, low-absorbance grade, EM Science, Cherry Hill, NJ; buffer B = 0.600 M KH2POa, both pH 4.5). Chromatographic peaks were identified and quantified by comparison to known standards.
Primary Culture Methods An outline of our final protocols for culture of RCMEC and RAEC appears in Fig. 1. We developed culture conditions for RCMEC first, assuming that the cerebral cells would require a more complex method and be more fastidious in their growth requirements than the RAEC (16). Subsequently, the method employed to culture RCMEC was adapted to RAEC. During the course of developing tissue culture methods, different sets of homogenizers, lots of enzymes, or concentrations of enzymes resulted in different amounts of tissue dissociation. The success of the culture was criti-
RAT ENDOTHELIAL CELLS IN CULTURE
315
TABLE 1 EVALUATION OF MICROVESSEL ISOLATION Sample
Energy Charge (I/2 ADP + ATP)/(ATP + ADP + AMP)
Brain" Brain homogenate Microvessels after dextran Cell clusters after percoll
0.915 0.278 0.700 0.865
+ + + +
0.004 0.051 0.060 0.049
(4)b (7) (5) (3)
3t-GT Aclivily, U/rag protein
AKP Activily, U/rag protein
-0.0053 + 0.0010 (5) 0.0591 + 0.0170 (5) 0.0943 + 0.0203 (3)
-0.0579 + 0.0110 (5) 0.4263 + 0.0719 (5) 1.5695 + 0.6993 (3)
By freeze-blow technique (48). b Values reported as mean _+ SD [n), where n is the number of preparations studied. Each preparation was analyzed in triplicate and the average value used. °
cally dependent on adjusting these factors to achieve maximum yield and purity of the cells plated out. In the protocol description, ranges of treatment are listed to reflect these adjustments.
Primary Culture of RCMECs Male Sprague-Dawley rats (6 rats of 35 to 50 g) were killed by the anesthetic halothane, and their brains removed. The isolated brains were placed in 70% ethanol (2 to 4 min) and then rinsed in DM. The thoracic aortae are also removed at this time, placed in DM, and refrigerated. The cerebral hemispheres are dissected away from the brain stem (yielding about ~ 1.2 g of tissue per rat), cut into small pieces, and rinsed several times in DM. The cerebral hemispheres are next homogenized [Wheaton-Dounce, 40 ml, 2 to 6 times with pestle B (clearance = 0.0025 to 0.0035 in.) followed by 2 to 10 times with pestle A (clearance = 0.001 to 0.003 in.)]. The amount of homogenization is dependent on the particular homogenizer. After homogenization the tissue is filtered through a 149-#m nylon mesh to remove large vessels, and then washed with DM containing no serum. (Placed in a 50-ml tissue culture tube and centrifuged at 800 Xg for 5 to 10 min, the superuatant is then aspirated and the pellet suspended again.) The homogenate is further dissociated using protease [0.005 to 0.05% protease (wt/vol) in 30 ml of DM containing no serum, in a 125-ml screw-top Ehrlenmeyer flask, 37 ° C, shaking water bath for 1.0 to 5.0 hi. After protease treatment the microvessels are separated from neurons and their supporting ghal cells by centrifugation through dextran (15% wt/vol in DM). The protease-treated homogenate is washed in DM, and the pellet suspended again by vigorous vortexing in the dextran solution (high setting for at least 3 rain) and then centrifuged (4 ° C, 4500 × g for 10 min, using a swinging bucket rotor, Beckman JS 5.2). The fat pad and dextran are collected to a new centrifuge tube and vortexed and centrifuged once more. The pellets from the two dextran spins are collected and washed in DM. The microvessels are further dissociated using collagenase/dispase [0.035 to 0.050% collagenase/dispase (wt/vol) in 10 ml DM, 125-ml screw-topped Ehrlenmeyer flask, 37 ° C, shaking water bath, for 6 to 18 h]. The collagenase/dispase-treated microvessels are centrifuged and suspended in 1 ml DM, layered across the top of a percoll gradient and centrifuged to separate cells (1650 ×g, 10 rain, 4 ° C). The percoll gradient was divided into 1.5-ml aliquots in a 24-well plate and the portions of the gradient containing endothelial cells identified. These wells were collected and the cells washed in DM. (percoll is toxic.) The endothelial ceils were present in approximately the middle third of the gradient, or a density range of ~ 1.04 to 1.07 g/ml. Finally, the ceils were suspended in culture medium for plating. The
best results were achieved using culture medium composed of 15% EPDS, 4% FBS, 1X antibiotic, antimycotic solution, 100 ~tg/ml endothelial mitogen, 50 ~g/ml heparin, 1 X pyruvate, 1X glutamine, 1X nonessential amino acids, 1X vitamins in DMEM (RCMEC growth medium). Cells plated directly to fibronectincoated tissue culture dishes (5/.tg/cm 2) sometimes gave good resuits (cultures > 90% purity as determined by Dil-Ac-LDL). However, on many occasions this was not the case (cultures of 50 to 60% purity as determined by DiI-Ac-LDL). Even greater success was achieved by first filtering through a double layer of 10-#m mesh and placing the cell-cluster isolates in uncoated dishes or dishes coated with poly-D-lysine or gelatin and allowing non-endothelial cells to attach. After 4 h, unattached cell clusters were transferred to fibronectin-coated dishes. Resulting primary cultures of RCMECs were of very high purity. The cells were rinsed and fed the day after their plating, and reached confluence in 5 to 8 days. Yield varies, but typically six rats yield 15 to 20 million cells, covering 120 to 160 cm 2.
Primary Culture of RAEC As mentioned above, our strategy for culture of RAECs was to use culture conditions as close to those of RCMECs as possible (Fig. 1). While the brain homogenate was incubating in pro/ease, the isolated rat aortae were processed for cell culture. The periadventitial fat was removed, and the aortae washed with DM and cut into rings 1 to 2 mm wide. The aorta rings were transferred with a wide-mouth pipette to a 15-ml centrifuge tube and rinsed several times in DM with 5% EPDS. The rings were not incubated with protease (as are the RCMECs) because this would digest the aorta wall, releasing non-endothelial cells, nor is it necessary to use dextran. The endothelial cells were released from the rings by incubation in collagenase/dispase (10 ml of 0.035% wt/vol in DM in a 50-ml, screw-topped Ehrlenmeyer flask, 37 ° C shaking water bath, 3 to 4 h). After collagenase/dispase treatment the rings were transferred to a 15-ml centrifuge tube. The tube was filled to the top by rinsing the Ehrlenmeyer flask with additional DM. It is critical at this step to vortex on high for at least 2 min to separate the endothelial cells from the vessel wall. The rings and endothelial cells were washed, suspended in DM and vortexed again. After centrifugation the rings and endothelial cells were suspended in 2 ml of DM. This was layered onto the top of a percoll gradient using a wide-mouth pipette. The culture protocol proceeds as with RCMECs, except that the percoll is aliquoted using a wide-mouth pipette to recover the aortic rings. Non-endothelial cells were found mostly in the top third of the percoll gradient, whereas the endothelial cells and aortic rings were in the next half, and the final sixth of the gradient was mostly
316
GORDON ET AL.
red blood cells. Cells and rings were plated out as with RCMECs. When cells are rinsed and fed the next day, it is important to remove all pieces of aorta as they will give rise to cells (either smooth muscle cells or fibroblasts) which migrate from the vessel wall onto the tissue culture dish. RESULTS The development of the culture techniques proceeded on three fronts. It was necessary to determine the appropriate conditions for the isolation of endothelial cell clusters, the selective attachment of endothelial cells, and the selective growth of endothelial cells. The success of different culture protocols was initially evaluated by observing cell morphology and uptake of DiI-Ac-LDL. Endothelial cells (and macrophage cells) take up DiI-Ac-LDL by a receptor mediated process, and the DiI fluorescence probe accumulates in lysosomal membranes (44). DiI-Ac-LDL is therefore a convenient marker because cells may be identified without fixation and staining steps. The purity of the cell cultures could be evaluated any time after the cells had attached simply by adding DiI-Ac-LDL to the culture medium and observing the cells, using fluorescence microscopy.
Isolation of RCMECs We followed the vessel isolation protocol using HPLC analysis to determine cellular energy charge (1/2ADP + ATP)/(ATP + ADP + AMP) (3), and enzymatic assays for alkaline phosphatase and 3'-GT, proposed markers for microvascular and brain endothelium (22,49). Phase microscopy, TEM, and SEM were used to help assess the purity, viability, and ultrastructure of the freshly isolated vessels. Table 1 lists the cellular energy charge at different points in the microvessel isolation process along with "y-GT and AKP enzyme activities. These data reflect the damage necessary to isolate the cells and the day-to-day variability associated with techniques of this type. The microvessels are still metabolically viable (76% of initial energy charge) and accumulate radiolabeled adenosine into their adenine nucleotide pools after both homogenization and protease treatment (24). After the cerebral hemispheres were homogenized and passed through the 149-#m nylon mesh, samples were prepared for SEM (Fig. 2 a) and TEM (Fig. 2 B-D) analysis. Phase contrast microscopy and SEM analysis confirmed that the majority of vessels present after the homogenate was filtered (149-#m mesh) were less than ~ 6 0 #m in diameter. TEM analysis (Fig. 2 B-D) of the filtered homogenate revealed that the majority of vessels were capillary vessels and small arterioles and venules. Vessels with one or two layers of smooth muscle cells were occasionally seen, but never any larger vessels, (SEM analysis of the material collected on the 149#m nylon mesh detected many larger vessels.) TEM analysis (Fig. 2 B-D) illustrates that rat brain microvessels are heavily invested by pericytes, with both endothelium and pericytes being surrounded by basement membrane. Astrocytic foot processes then surround and enclose the basement membrane. In larger microvessels, smooth muscle cells replace pericytes. The TEM analysis shows that although the isolated microvessels seem quite pure by SEM and phase contrast microscopy, the structure of the microvessels makes the isolation of endothelial cells ditficult. After the protease and dextran step of the protocol, the vessels
start to dissociate from the surrounding tissue. Next, collagenase/ dispase treatment was used to digest the basement membrane and release endothelial cells from the other cellular components of the microvessels. Several protease and collagenase preparations were analyzed using SEM, and it was found that within a preparation and from preparation to preparation the extent of enzymatic digestion varied widely. The preparation consisted of red blood cells, single cells, debris, and clusters of rounded cells. In a few experiments, albumin (25%) was used in place of dextran (47). A higher yield of microvessels was obtained with albumin. However, subsequent cell plating and growth were worse. Different types of collagenase treatment were also investigated. Many investigators have tried Ca++/Mg ++-free buffers for collagenase incubation. Using collagenase (CIostridiumhistolyticuum) or collagenase/ dispase we found that viability was decreased in Ca++/Mg++-free buffers. In fact, we found addition of 5% serum (EPDS) to every step of the protocol (except the protease treatment) resulted in better yield and viability of cells. Initially, the different fractions of the percoll gradient were washed and plated to fibronectin or gelatin-coated dishes, and the growth from each fraction evaluated by morphology and uptake of DiI-Ac-LDL. Figure 3 A-B illustrates this approach, showing some morphologies seen in primary cultures and the positive DiI-Ac-LDL labeling of some cell colonies. During initial outgrowth the morphology of endothelial cell colonies were similar to either the "classic cobblestone" morphology typical of BAECs or, more often, an elongated swirling pattern described in many reports of cultured mierovascular endothelial ceils. As the cultures matured and the cells in the colonies multiplied and cell density increased, most endothelial cells became elongated and assumed a swirling pattern (Fig. 3 C-D). We observed that the greatest number of endothelial cell colonies were derived from very refractive clusters of cells in the middle third of the percoll gradient (density 1.04 to 1.07 g/ml). The majority of these clusters exclude trypan blue dye (>90%). Many single cells are present in this portion of the gradient as well, but their viability is generally low (30 cells) the cultures derived from them are mixed. The task therefore was to adjust the isolation procedure (homogenization, protease treatment, and collagenase treatment) to maximize the yield of these clusters. Enzymes are purchased in large lots, and once isolation conditions are established, they are prealiquoted and stored frozen ( - 2 0 ° C). If the yield at the percoll step decreases, homogenization, enzyme concentration, or incubation times are decreased. Conversely, if large clusters of cells are obtained, homogenization, enzyme concentration, or incubation times are increased.
Attachment and Growth of RCMECs The isolation procedure enriches for endothelial cells but other cell types are also present. Fibroblasts, pericytes, smooth muscle, and glial cells are all possible contaminants when the "purified" cell preparations are plated out. It was necessary to choose conditions that selected for attachment and growth of RCMECs and if possible inhibited attachment and growth of non-endothelial cell types. Once the percoll fractions yielding the greatest numbers of endothelial cell colonies were identified, the fractions were pooled and
RAT ENDOTHELIAL CELLS IN CULTURE
317
FIG. 2. SEM and TEM photomicrographs of brain microvessels. Obtained by dextran centrifugation of fihered (149-#m mesh) brain homogenate. Note the close association of the cells of the vessel wall. This is responsible for the difficulty of isolating pure endothelial cell preparations from the rat brain. A, SEM, bar = 10 /.tm. B, TEM, X10 000 and (C and D). Enlargement of B, )
