Isolation, Cultivation, and Partial Characterization of Microvascular Endothelium Derived from Human Lung William W. Carley, Michael J. Niedbala, and Mary E. Gerritsen Institute for Inflammation and Autoimmunity, Miles Research Center, West Haven, Connecticut

Primary cultures of peripheral lung lobes were grown in a highly supplemented medium. Human lung endothelial cells (HLE) were isolated from the mixed population by FACS. The cells proliferated rapidly and were serially cultivated for at least 16 passages. Both early and late passage cells were positive for the standard endothelial markers, Factor VIII related-antigen (Factor VIII R-Ag), angiotensin-converting enzyme, acetylated low-density lipoprotein labeled with 1,1'-dioctadecyl-l,3,3,3',3'-tetramethyl-indocarbocyanine perchlorate (DiI-Ac-LDL) uptake, and bound the lectin Ulex europaeus agglutinin (UEA). Prostaglandin E 2 was the major cyclooxygenase product of HLE, in contrast to human umbilical vein endothelial cells (HUVE), which synthesized PGI2 in excess of PGE 2 • Factor VIII R-Ag exhibited a diffuse cytoplasmic as well as an extracellular fibrillar distribution in HLE, in contrast to a vesicular (Weibel-Palade body) cytoplasmic distribution in HUVE. The HUVE did demonstrate some extracellular fibrillar Factor VIII R-Ag as well. Urokinase was the predominant plasminogen activator (PA) secreted by HLE, whereas tissue PA was predominant in HUVE cultures. HLE formed tube-like structures within 2 h of plating on a Matrigel matrix whereas HUVE formed larger tube-like structures only after I or more days. The properties described here indicate that human lung microvessel endothelium can be isolated and continuously grown from small tissue segments and express a number of properties that differ from those of HUVE. These studies provide further support for the concept that endothelial cells from different sources can exhibit considerable heterogeneity relating to their phenotypic and biochemical properties.

Introduction The endothelium of the lung microvasculature provides a highly selective conduit for the exchange of gases between the blood and the alveolar airspace. In addition to their functions in gas exchange, the pulmonary capillaries perform additional roles in the filtration of systemic venous blood and

(Received in original form February 26, 1992 and in final form June 25, 1992) Address correspondence to: William Carley, Ph.D., Miles Research Center, 400 Morgan Lane, West Haven, CT 06516. Abbreviations: arachidonic acid, AA; angiotensin-converting enzyme, ACE; acidic fibroblast growth factor, aFGF; bovine aortic endothelial cells, BAEC; bovine serum albumin, BSA; 1,1'-dioctadecyl-l,3,3,3~3'-tetra­ methyl-indocarbocyanine perchlorate-labeled acetylated low-density lipoprotein, DiI-Ac-LDL; dimethyl sulfoxide, DMSO; fluorescent-activated cell sorter, FACS; Factor VIII related-antigen, Factor VIII R-Ag; fetal bovine serum, FBS; fibroblast growth factor, FGF; fluorescein isothiocyanate, FITC; human lung endothelial cells, HLE; human umbilical vein endothelial cell(s), HUVE; interleukin-Io, IL-la; Madin Darby canine kidney epithelial cells, MDCK; ~-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetraw­ lium bromide, MTT; plasminogen activator, PA; plasminogen activator inhibitor-I, PAl-I; phosphate-buffered saline, PBS; prostaglandin, PG; rhodamine, Rh; sodium dodecyl sulfate, SDS; sodium dodecyl sulfate polyacrylamide gel electrophoresis, SDS-PAGE; tumor necrosis factor-a, TNF-a; tissue plasminogen activator, tPA; Ulex europaeus agglutinin, DEA; urokinase type plasminogen activator, uPA. Am. J. Respir. Cell Mol. BioI. Vol. 7. pp. 620-630, 1992

the regulation/clearance of circulating levels of certain vasoactive substances. There are several major features, namely the remarkably low pressures, the extremely thin vascular walls, high capillary density, and the paucity of smooth muscle cells, which distinguish the pulmonary microcirculation from the systemic circulation. The biochemistry and cell biology of the human pulmonary microvascular endothelial cell is an important area of investigation that has been difficult to pursue because of the lack of methods for isolating and cultivating this cell type. Model systems have been described for the culture of pulmonary vascular endothelium from larger blood vessels (1, 2) and microvessels from small animals (3, 4), but methods for the isolation and continuous culture of microvascular/capillary endothelial cells from human lung have not been described. This report documents the successful culture of microvessel endothelium derived from human lung. Early and late passage cultures retain endothelial markers and exhibit several unique features distinguishing them from human umbilical vein endothelial cells (HUVE).

Materials and Methods Materials Antisera against prostaglandin (PG) E 2 and 6-keto PGF t • were from Advanced Magnetics Inc. (Cambridge, MA). Fluorescein isothiocyanate (FITC)-conjugated anti-Factor

Carley, Niedbala, and Gerritsen: Human Lung Microvessel Endothelium

VIII related-antigen (Factor VIII R-Ag) was from Atlantic Antibodies (Scarborough, ME) and was used at a dilution of 1:100 for the immunohistochemistry studies. Goat antirabbit angiotensin-converting enzyme (ACE) serum was the kind gift of Dr. Richard Soffer (Cornell University Medical College, New York, NY). Goat anti-human Factor VIII R-Ag (used at a dilution of 1:50 for immunohistochemistry studies), murine monoclonal antisera against tissue plasminogen activator (tPA), urokinase type plasminogen activator (uPA), and plasminogen activator inhibitor-l (PAl-I) and plasminogen were from American Diagnostica (Greenwich, CT). 1,1'-dioctadecyl-l,3,3,3:3'-tetramethyl-indocarbocyanine perchlorate-labeled acetylated low-density lipoprotein (DiIAc-LDL) was purchased from Biomedical Technologies Inc. (Staughton, MA). Matrigel, Nu-Serum, insulin, transferrin, and selenium were from Collaborative Research Inc. (Bedford, MA). Gelatin (Bacto) was from Difco Labs (Detroit, MI). Fetal bovine serum (FBS) was purchased from Hyclone (Logan, UT). Arachidonic acid (AA) was from NU-Chek Prep (Elysian, MN). Heparin H-3125, rhodamine (Rhj-conjugated rabbit anti-goat IgG, bovine serum albumin (BSA) fraction V, alkaline phosphatase-conjugated rabbit antimouse IgG, and collagenase type II were obtained from Sigma Chemical Co. (St. Louis, MO). Rh-conjugated Uiex europaeus agglutinin (UEA), biotinylated UEA, and Vectastain kits were from Vector Laboratories (Burlingame, CA). Dimethyl sulfoxide (DMSO) was from Pierce (Rockford, IL). Nitex nylon mesh (100 ~m) was obtained from Tetko (Elmsford, NY). Purified bovine acidic fibroblast growth factor (aFGF) was from R&D Systems (Minneapolis, MN). Recombinant human tumor necrosis factor-a (TNF-a; > 2 X 107 U/mg) and interleukin-lo (lL-lCt; lQ5 U/~g) were from Boehringer Mannheim (Indianapolis, IN). Rh-phalloidin was from Molecular Probes (Eugene, OR). Madin Darby canine kidney epithelial cells.(MDCK) (CCL 34) and human lung fibroblasts (CCD-37Lu) were obtained from the American Type Culture Collection (Rockville, MD). Bovine aortic endothelial cells (BAEC) were provided by Dr. Joseph Madri (Yale University, New Haven, CT). EGM-UV medium used for HUVE propagation was obtained from Clonetics (San Diego, CA). All other cell culture reagents were from GmCO (Grand Island, NY). Endothelial Cell Isolation and Culture Lung tissue was obtained postoperatively from patients undergoing lobectomy for small cell carcinoma. The tissues used for endothelial isolation were from peripheral lobes with no histologic evidence of malignant cells. The outermost layer of cells (0.5 em) was removed with a scalpel to reduce possible mesothelial cell contamination. The remaining tissue was minced in growth medium (RPMI supplemented with 10% FBS, 10% Nu-Serum, 20 ~g/ml heparin, 4 ~l/ml retinal derived growth factor [5, 6], 2 mM glutamine, penicillin [100 U/ml], and streptomycin [100 ~g/ml]). The suspension was centrifuged (800 x g, 5 min), and the pellet resuspended in 0.2 % collagenase and 0.1% BSA in RPMI and incubated with gentle agitation for 30 min at 3rc. The suspension was sheared 20 times by pipetting using a lO-ml pipet, then incubated an additional 30 min. The suspension was filtered through a sterile 100-~m nylon

621

mesh, and the filtrate collected and pelleted by centrifugation (800 X g, 5 min). The pellet was resuspended in 6 ml growth medium and transferred to a T-25 flask precoated with 1.5% gelatin in phosphate-buffered saline (PBS). Numerous colonies of cells were evident within 2 days, and confluence was attained in 7 days. The medium was removed and the cells were subcultured to three T-25 flasks using Ix trypsin-EDTA. An additional 1:3 (1 flask:3 flasks) split was performed before fluorescent-activated cell sorting (FACS), and aliquots of mixed cell populations (second passage) were frozen in 90% FBS, 10% DMSO for later comparisons and isolations. HUVE were purchased from Clonetics. The identities of the cells used were confirmed by staining for Factor VIII R-Ag and DiI-Ac-LDL uptake. Greater than 99% of the HUVE used in these studies were positive for these endothelial specific markers. HUVE were grown in EGM-UV medium supplemented with 10% FBS. Cells were subcultivated and split at a 1:3 ratio using trypsin. HUVE used in the present studies were from the second to eighth passage from primary culture. Human Lung Fibroblasts Human lung fibroblasts (CCD-37Lu) were used as a control for immunohistochemistry studies. These cells contained 0% positive DiI-Ac-LDL cells when analyzed by FACS (not shown). The cells were grown in growth medium and subcultivated at a 1:3 ratio using trypsin-EDTA. CCD-37Lu used in the present studies were from the fourth to sixth passage from primary culture. Endothelial Cell Isolation by FACS Preconfluent mixed cultures (1 T-75 flask/sort) of cells were incubated with 5 ~glml DiI-Ac-LDL (7) in growth medium for 4 h at 37°C, washed, trypsinized, and resuspended in 1 ml growth medium. Cells preferentially incorporating DilAc-LDL were separated from other cell types by FACS on a FACstar Plus flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA). An argon-ion laser emitting 200 mW at 488 om generated forward and orthogonal light scatter signals used to select intact cells and provided an excitation source for the DiI-Ac-LDL. DiI-Ac-LDL fluorescence was measured through a bandpass filter with a 42 nm bandwidth centered at 585 om and amplified logarithmically. The positive sorted cells (the top 0.5%, approximately 3 X 104 cells) were grown to confluence under the same conditions as the initial isolates and were used for subsequent experiments. The usual time for obtaining sorted microvessel human lung endothelial cells (HLE) from the initial primary isolate was 3 to 4 wk. Endothelial cells were identified using several markers, i.e., Factor VIII R-Ag, ACE, DiI-Ac-LDL uptake, and binding of UEA as previously described (6, 8-10). For the determination of UEA staining in human lung, sections were blocked with 0.3% hydrogen peroxide in methanol for 20 min at room temperature, washed in PBS, and incubated 45 min in 1% BSA in PBS at 3rC. Sections were then incubated with 10 ~glml biotinylated UEA. UEA staining was visualized using a streptavidin-horseradish peroxidase based kit (Vectastain kit; Vector Laboratories) following the manufacturer's directions. Controls included sec-

622

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

tions incubated in the absence of lectin or in the absence of lectin and streptavidin-horseradish peroxidase but in the presence of the developing reagents. Cell Growth Analyses The 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay for cell growth was used (11). Briefly, 100 p.l of MTT (5 mg/ml) in growth medium was added to each well of a 96-well plate and incubated for 4 h at 37°C. After aspiration of the culture medium, the blue precipitate was dissolved in 150 p.l of 0.04 N HCI in isopropanol and absorption at 595 nm determined on a Molecular Devices (Menlo Park, CA) microtiter plate reader. Tube Formation Confluent HUVE, CCD-37Lu, MDCK, and HLE were removed from their respective growth media and grown for 48 h in RPMIIIO% FBS, then trypsinized, and cells (5 x 104) plated in RPMIIIO% FBS on Matrigel (0.6 ml/35-mm dish). After various periods, the cells were examined by phase microscopy and fixed in 3.7% formalin in PBS for 30 min. Actin was stained by incubation overnight with 100 ng/ml Rh-phalloidin, and photographs were taken using Hoffman modulation optics or Rh fluorescence illumination on a Leitz Fluorovert FS microscope. ACE Activity Intact monolayers (1 X 106 cells/15 ml medium) were incubated 20 min with 100 ng/ml human angiotensin I (Peninsula Laboratories, Belmont, CA). Angiotensin II formation was monitored using a radioimmunoassay kit for angiotensin II (Peninsula Laboratories). Under the conditions of the assay, there was no cross-reactivity of angiotensin I with the angiotensin II assay. Controls included incubations in the absence of cells and incubations in the presence of the ACE inhibitor captopril (Sigma) at a concentration of 50 p.M. Prostaglandin Assays HUVE and HLE were grown to confluence in 24-well culture dishes and were fed with RPMIIIO% FBS 48 h before testing. The culture media were removed, and the cells incubated with or without TNF-a in RPMIIO.l% BSA for 4 h. After removal of these media, the cells were washed 3 times with I ml of Dulbecco's PBS with Ca2 + and Mg2 + , then incubated 15 min with PBS with or without 3 p.M AA. The incubation media were removed and stored at -20°C until assayed by radioimmunoassays. Radioimmunoassays for PGE 2 and 6-keto PGF 1a (the stable hydrolysis product of PGI 2) were performed as described previously (12, 13). Determination of Plasminogen Activator (PA) Immunologic quantitation ofPA and PAI-Iin appropriate dilutions of 25x concentrated cell-free conditioned medium was carried out using'a micro-ELISA assay as previously described (14). This assay utilized murine monoclonal antibodies directed against tPA, uPA, and PAI-I in conjunction with a secondary detection system using rabbit anti-mouse alkaline phosphatase conjugate. Hydrolysis of p-nitrophenol phosphate was monitored at 405 nm with an automated plate reader, and values were compared with standard tPA, uPA,

and PAI-I antigen curves under conditions of linearity. The substitution of nonimmune mouse IgG for the primary antibody served as a negative control. This micro-ELISA assay detects tPA and uPA antigen irrespective of their molecular form, i.e., latent versus active or enzyme/inhibitor complexes. Zymographic analysis of PApresent in 25 X concentrated cell free-conditioned medium was performed on a 10% polyacrylamide gel containing 0.1% sodium dodecyl sulfate (SDS) polymerized in the presence of plasminogen and casein as previously described (15). The caseinolytic activity observed was shown to represent true PA activity since caseinolytic activity in replicate gels polymerized in the absence of plasminogen was not present.

Results Primary Culture A number of cell morphologies were evident in the initial cell isolate (HL), including phenotypes characteristic of epithelium, fibroblasts, and macrophages. At confluence, cells were multilayered, and no one particular phenotype appeared dominant. Incubation of mixed cell cultures with DiIAc-LDL revealed a limited population of labeled cells (Figure I). Two cell types are known to incorporate significant levels of DiI-Ac-LDL in a 4-h period, namely macrophages and endothelial cells (7, 16). Stein and Stein (16) reported that macrophage uptake exceeded that of endothelial cells by 3- to 6-fold. Based on the difference in cell morphology and fluorescence intensity, it is likely that the polygonal cells (see Figure Ib) incorporating the Dil-Ac-LDL were macrophages and the elongated labeled cells were endothelia. FACS Results The specific uptake of Dil-Ac-LDL by endothelial cells was used as the basis for segregation of cell populations. The FACS profiles of different cell types incubated with Dil-AcLDL under the same conditions described above for the lung isolates are given in Figure 2a. The epithelial cells, MDCK, exemplify the profile for the negative control, i.e., cells not incorporating the Dil-Ac-LDL to any significant extent, and are clearly distinguished from the profiles of the two endothelial populations (positive controls), HUVE and BAEC. The lung isolates (human lung culture, passage 3, HLp3) exhibited a profile similar to the MDCK, although a small percentage of cells exhibited fluorescence above the negative control (Figure 2b). Additionally, a few cells were observed I log unit beyond the positive controls (not shown). When these highly fluorescent cells were collected, no cell growth was observed although the cells attached readily to the substrate. However, when the top 0.5 % of cells were collected (excluding the highly fluorescent cells), seeded to gelatincoated 35-mm dishes, and grown in growth medium, the resulting populations readily attached, proliferated, and attained confluence in I wk. The cells were subcultured and used for cell identification, cell proliferation, and further FACS analyses. As shown in Figure 2c, the sorted cells showed a right (positive) shift in the FACS profile that was maintained as many as 16 passages after the initial sort. The FACS sort of initial isolates was repeated two additional

Carley, Niedbala, and Gerritsen: Human Lung Microvessel Endothelium

a

623

a.

-II,)

U

Log Fluorescence

b

II,)

U

Log Fluorescence

c.

Figure 1. DiI-Ac-LDL staining of mixed culture of human lung cells. (a) Phase contrast of HLp2 cells. (b) Same field as panel a, visualized under fluorescent optics for rhodamine. Two different cell morphologies incorporate DiI-Ac-LDL, one rounded and one elongated. The rounded cell incorporates more DiI-Ac-LDL. Bar = 10 p.m.

times, with essentially identical results. A low-power phase micrograph of the sorted HLE passage 3 (HLEp3) is shown in Figure 3. The incorporation of DiI-Ac-LDL into HLE is depicted in Figure 3c, and shows the typical perinuclear vesicular location characteristic for endothelial cells (7). Although the uptake of the DiI-Ac-LDL by the HLE was not as great as that seen for the HUVE and BAEC, a number of investigators have described variability in the degree of DiIAc-LDL incorporation by cell density and endothelial cell origin (6, 17). Other Endothelial Markers Immunohistochemical studies using FITC-labeled anti-human Factor VIII R-Ag demonstrated that the HLE exhibited diffuse cytoplasmic, and additionally, an intense extracellu-

-

....HLEp16

II,)

U

Log

Fl~orescence

Figure 2. FACS analyses of cell populations. All cells were incubated with DiI-Ac-LDL as described in MATERIALS AND METHODS, trypsinized, and suspensions analyzed by FACS. (a) Negative controls (Madin Darby canine kidney [MDCK] cells) and positive controls (human umbilical vein endothelial [HUVE] and bovine aortic endothelial [BAEC] cells). (b) FACSprofile of unsorted lung cells (HLp3) after isolation, passage 3. Arrow indicates population that was sorted, 0.5 % of total population. (c) FACSprofile of sorted human lung endothelial cells (HLE), passages 3. and 16 after sorting, demonstrating a significant shift in fluorescence intensity from the unsorted cells and maintenance of this phenotype for at least 13 additional passages.

lar fibrillar, staining pattern (Figures 4a and 4b). In contrast to HUVE (Figures 4c and 4d), there was no evidence for intracellular vesicular staining, characteristic of the WeibelPalade bodies. The HUVE also demonstrated some extracellular fibrillar staining (Figure 4d). The matrix association of

624

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

a

b

c

the HLE fibrillar stain was demonstrated by repetition of the staining procedures in nonpermeabilized cells, which revealed the same fibrillar staining pattern (not shown). The FITC-conjugated anti-Factor VIII R-Ag antibody did not stain human lung fibroblasts (CCD-37Lu: Figures 4e and 4f), nor MDCK (not shown). Essentially identical staining patterns to those shown in Figure 4 were obtained with both polyclonal and monoclonal anti-human Factor VIII R-Ag from different suppliers (ICN, Costa Mesa, CA and Accurate Antibodies, Westbury, NY), in conjunction with the appropriate Rh-conjugated secondary antibody (not shown). The HLE maintained the matrix- and cell-associated staining for Factor VIII R-Ag for at least eight passages after cell sorting (not shown). There was no detectable staining of gelatin-coated dishes incubated with growth medium or of MDCK (not shown). The presence of active ACE in the HLE was obtained from experiments monitoring the conversion of angiotensin I to angiotensin II and the inhibition of this conversion by the specific ACE inhibitor captopril (50 J.IoM) (Table 1). There was no conversion of angiotensin I to angiotensin II in the absence of cells. Immunohistochemical localization of ACE demonstrated a typical homogeneous surface staining pattern similar to that reported previously in cultured rabbit pulmonary microvessel endothelium (3, 18) (not shown). Additionally, the goat anti-rabbit ACE antibody immunoprecipitated a single band in pSS]methionineprelabeled HLE with a molecular weight of 170 kD (not shown). Purified ACE has a molecular weight of 155 to 170 kD; the variability in the estimate is a result of the high carbohydrate content of the protein (19). The HLE maintained expression of ACE immunoreactivity for at least eight passages after sorting (not shown). The HLE also labeled with Rh-conjugated UEA. The UEA lectin selectively stained endothelial cells in thin section of formalin-fixed lung, indicating that this lectin is a useful, selective marker for HLE (Figures Sa to 5d). UEA staining of HLE was maintained for at least seven passages after sorting (not shown). Tube Formation Tube-like (also known as "capillary-like") structures are characteristically formed by endothelial cells cultured on an appropriate extracellular matrix such as Matrigel (20-22). Application of HUVE or HLE suspensions to preformed Matrigel matrices resulted in the morphologic differentiation of these cells into tube-like structures (Figures 6a to 6d). There were notable differences in both the composition and the time course of tube formation between the HUVE and

TABLE 1

Angiotensin-converting enzyme activity in HLE* Angiotensin II Formed/20 Min/10 6 Cells

Figure 3. Morphology. of sorted HLE. (a) Low-magnification

Group

phasemicrograph of sortedHLEp3 cells(bar = 40 /lm). (b) Highmagnification phase micrograph ofHLEp3 incubated with DiI-AcLDL as detailed in MATERIALS AND METHODS. (c) Same field as panel b visualized under fluorescent optics for rhodamine (bar =

HLE cells HLE cells

10 /lm).

(ng)

+ captopril

5.67 ± 0.45 2.11 ± 0.31

* Confiuent HLE were washed 3 times with 15 m1 serum-free RPMI,,.then incubated for 20 min at 37°C in RPMI (no serum) with 1.5 p.g angiotensin I in a total volume of 15 m1 in the presence or absence of 50 p.M captopril. Angiotensin II formation was determined by radioimmunoassay. Data are expressed as the mean of duplicate determinations ± range of values. Cross-reactivity of the substrate (angiotensin I) was < 1 %.

Figure 4. Endothelial cell markers. (a) HLEp3 cells, phase microscopic view. (b) Same field as panel a, stained for Factor VIII related-antigen as described in MATERIALS AND METHODS (bar = 28 ttm) . (c) HUVEp8, phase micrograph. (d) Same field as panel c, stained for Factor VIII related-antigen as described in MATERIALS AND METHODS (bar = 20 ttm). (e) CCD-37Lu, phase micrograph. (f) Same field as panel e, stained for Factor VIII related-antigen as described in MATERIALS AND METHODS (bar = 28 ttm).

Figure 5. UEA staining in human lung and HLEp3 cells. (a) Low-power, bright-field illumination of thin section of human lung reacted with streptavidin-horseradish peroxidase as detailed in MATERIALS AND METHODS. (b) Low-power view of human lung stained with biotinylated UEA followed by streptavidin-horseradish peroxidase (bar = 80 ttm). (c) High-power view of human lung stained with UEA demonstrating selective staining of endothelial cells (bar = 8 ttm) . (d) Fluorescent micrograph of HLE stained with rhodamine- UEA (bar = 10 ttm).

o

626

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

b

d

c

HLE. The tube structures formed by the HLE were narrower, appeared to be formed by one or two cells in close longitudinal apposition, and were oriented in three-dimensional arrays throughout the matrix. In contrast, the structures formed by the HUVE were multicellular, the arrays two dimensional, and the individual cells comprising the tubes exhibited a cobblestone-like morphology. The HLE rapidly associated in the Matrigel and formed elongated strand-like structures, and clear areas indicated rapid digestion of the Matrigel. In contrast, the HUVE remained as single-cell isolates for many hours after seeding, and only after one or more days could multicellular aggregates, matrix degradation, and tube-like structures be discerned. MDCK or CCD37Lu cells plated under identical conditions remained as isolated, individual cells for 24 to 48 h, and after I wk attained confluence with no evidence of tube formation (not shown). The HLE maintained the ability to form tubes in culture for at least 10 passages after cell sorting (not shown). HLE Growth Dependence on Fibroblast Growth Factor (FGF) FGFs are known to be potent mitogens for vascular endothelial cells, and the use ofFGF has allowed the establishment of clonal endothelial cell populations from a number of vascular beds (23, 24). The growth medium used in the isolation of the HLE contains two sources of FGF, i.e., NuSerum (which contains endothelial cell growth supplement) and retinal derived growth factor (which is a rich source of FGF). As shown in Figure 7, HLE (passage 4 after sorting) cultured in the presence of purified aFGF (100 ng/ml) ex-

Figure 6. Tube formation by and actin distribution in HUVE (a and b) and HLE (c and d) cells. In panels a and c the tubes are visualized under Hoffman modulation optics, and in panels b and d the cells, stained with rhodamine-phalloidin, are viewed under fluorescent optics for rhodamine (bar = 68 ttm).

hibited a more rapid growth rate than in FBS alone. HLE cultured in the absence of aFGF were larger and stopped replicating before attaining confluence 3 to 5 days after plating. Parallel experiments with growth medium yielded growth

0.30

It)

0.20

01

It)

ci

o

0.10

- 0 - 10%FBS --0- 10 % FBS + aFGF

0.00 +-~-r--~---.-~-,--~.,.....~---{ 20 o 40 60 80 100 Time (Hrs.)

Figure 7. Growth dependence of HLE cells on aFGF. HLE cells were seeded to 48-well plates (5 x 103 cells/well initial density) and cultured for the indicated periods of time in RPMI supplemented with 10 % FBS with or without 200 ng/ml aFGF. At the indicated time periods, MTT uptake was determined by the method of Mosmann (11), allowing estimation of relative viable cell number. An 0.D. above 0.25 is representative of a confluent monolayer.

Carley, Niedbala, and Gerritsen: Human Lung Microvessel Endothelium

627

TABLE 2

Prostaglandin synthesis by HLE and HUVE* Prostaglandin Release in HLE (ng/well)

Prostaglandin Release in HUVE (ng/well)

6-keto POF,"

POE,

6-keto PGF,.

POE,

0.02 ± 0.01

0.07 ± 0.02

< 0.02

< 0.02

0.20 ± 0.1 0.30 ± 0.01§

1.52 ± 0.06 4.74 ± 1.09§

0.66 ± 0.07 1.25 ± 0.08§

0.37 ± 0.05 0.89 ± 0.05§

Incubation conditions 15 mint PBS

4-h treatment + AA, 3 IlMt Medium Medium + TNF-a

* All data are expressed as ng/well (mean ± SEM, n = 4). Each well contained approximately 10' cells. t HLE or HUVE were incubated with PBS for 15 min at 37"C. Levels of both prostaglandins in PBS not incubated with cells were below the level of detection « 1 pg). t HLE or HUVE were incubated in RPMIIlO% FBS with or without (media) TFN-a (10 ng/ml). After the 4-h treatment with or without TNF-a, the cells were washed 3 times with I ml PBS then incubated 15 min with 3 I"M AA. § Significantly different from cells similarly incubated but not treated with TNF-a, Student's r-test for unpaired samples, P < 0.05.

curves identical to those shown for RPMII1O% FBS supplemented with purified aFGF (not shown). HLE propagated poorly in the EGM-UV medium used for the HUVE cells, yielding growth curves similar to that shown for RPMI supplemented with 10% FBS (not shown). Prostaglandin Assays PGE, is the major cyclooxygenase product of eicosanoid metabolism in microvessel endothelial cells from a number of species, including human (10, 25, 26), in contrast to large vessel endothelial cells which produce PGI, as the major product (27). As shown in Table 2, both cell types released relatively little of either prostanoid under unstimulated conditions. To compare the relative proportions of PGE, and POI, produced by HLE and HUVE; the cells were incubated for 4 h in RPMI, 0.1% BSA with or without 10 ng/ml TNF-a, washed, and then stimulated with exogenous AA (3 jLM). As shown in Table 1, control HLE synthesized PGE, in excess of PGI, (ratio of 7.6:1), whereas the control HUVE synthesized less PGE, and POI, (ratio of 0.56:1). In the experiments shown, the HUVE were passage 4 after primary culture and HLE were passage 4 after sorting. Similar ratios of PGE, to PGI, were observed in HLE passages 3 through 8 after sorting (not shown). Pretreatment of HUVE and HLE with TNF-a resulted in a 2- to 3-fold increase in the synthesis of both prostaglandins by both cell types upon addition of exogenous AA compared with cultures incubated in media alone. Similar observations were obtained when the HUVE or HLE were pretreated with IL-la (1 U/ml) under the same experimental conditions (data not shown). Plasminogen Activator Studies Extracellular PA and PAI-I associated with 24-h cell-free conditioned medium derived from either HUVE or HLE was determined using SDS-polyacrylamide gel electrophoresis (PAGE) PA zymography and micro-ELISA assays. For these studies, endothelial cells were removed from their respective media and grown for 48 h in serum-depleted medium comprised of RPMI 1640 supplemented with 0.1% heat-inactivated FBS, insulin (5 jLg/ml), transferrin (5 jLg/ml) , selenium (5 ng/ml), 25 mM Hepes, and 2 mM glutamine. Subse-

quently, cultures were incubated in the absence of serum for 24 h in the above medium and concentrated as previously reported (14). HLE demonstrated very high levels of accumulated extracellular immunoreactive uPA (80 ± 20 ng/106 cells) compared with HUVE (0.3 ± 0.2 ng/106 cells). Both cell types displayed appreciable extracellular tPAantigen, although HLE levels (9 ± 1 ng/106 cells) were significantly higher than HUVE (2 ± 0.2 ng/106 cells). Extracellular PAI-I was greater in HUVE (7 ± 0.5 ng/106 cells) than HLE (2 ± 1.2 ng/1Q6 cells). These findings were corroborated by the SDS-PAGE PA zymography (Figure 8). HLE-conditioned medium contained a strong plasminogen-dependent caseinolytic band at M, 55,000 comigrating with human uPA standard. In contrast, HUVE conditioned medium dis-

MW(KD)

·92 ·69

·46

Figure 8. 5DS-PAGE zymography comparing HUVE and HLE conditioned medium. In lanes 1 and 2, the migration of the urokinase (UK) and tPA standards are shown. The HLE conditioned medium contained a strong plasminogen-dependent caseinolytic band at M, 55,000 as well as a higher molecular weight caseinolytic activity (M, 92,000).

628

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL.

played little plasminogen-dependent caseinolytic activity at this molecular weight. HLE also demonstrated a high molecular weight (M, 92,000) plasminogen-dependent caseinolytic activity. This species probably represents a PAI-I/uPA complex with residual plasminogen activity as has been previously described (28). HUVE used in this experiment were passages 3 through 5 from primary and HLE used were passages 4 through 6 from sorting.

Discussion To date, cultures of large vessel cells have been the principal in vitro endothelial cell model from which data have been generalized. Although microvessel endothelial cells share some common features with their large vessel counterparts, the recent literature has clearly shown that it is naive to consider all endothelial cells identical. The relative ease by which long-term cultures of endothelial cells from large vessels can be produced has enabled an ever-expanding knowledge of the specialized functions and properties of endothelium. A number of laboratories have isolated and grown lung microvascular endothelial cells from small animals (3, 4), although long-term culture of endothelial cells from microvessels has often proved difficult, requiring elaborate 'matrix and media preparations and extensive cloning and subcloning procedures. Because specimens from human sources are often difficult to obtain and limited in size, improved methods are required to enable the establishment of microvascular endothelial cells from these tissues. In the present study, we have described a method for the isolation and serial cultivation of microvessel endothelial cells from human lung. The success of this method is based on the three major components: (1) initial dissection, (2) selection of cell populations, and (3) propagation in "growth medium" enriched with FGF. The initial dissection involved selection of peripheral lobe sections with no visible large blood vessels. Removal of the outermost 0.5 em was performed to reduce contamination with mesothelial cells and possible surface microbial contaminants. The selection of cell populations was based on the avid incorporation of DiIAc-LDL by endothelial cells. Acetylated low-density lipoprotein is incorporated by macrophages and endothelial cells via the "scavenger cell pathway" of low-density lipoprotein metabolism (7, 16). The use of the fluorescent derivative, DiI-Ac-LDL, allows the use of flow cytometric instrumentation to sort cell mixtures based on the differential labeling of various cell types in a complex mixture. In the initial isolates, two apparent populations of cells were observed which avidly stained with the DiI-Ac-LDL. A highly stained population was obtained which did not proliferate in culture and which exhibited a "fried-egg" appearance upon attachment to the gelatin matrix. The morphology, avid uptake, and lack of growth of this population is consistent with the properties of macrophages, which were most likely present in the starting material. A second population, also sorted by FACS, was obtained and shown to be endothelial in origin, based on the presence of numerous cell markers for endothelial cells, including the presence of Factor VIII R-Ag (10, 29), ACE (3, 8, 19), uptake of DiI-Ac-LDL by the entire population (7), and binding of the lectin UEA (9). Using the growth medium described in MATERIALS AND METHODS, the HLE could be

71992

serially cultivated and large numbers of cells prepared for use in biochemical, immunologic, and molecular biologic studies. HLE thus obtained maintained the endothelial markers described above for at least seven passages after cell sorting, and all of the biochemical experiments described herein used HLE from passages 4 through 7 post-sorting. Due to limitations in the availability of normal human lung tissue for cell cultures as well as the limited size of the samples received, it was not possible to obtain sufficient cells from primary cultures to determine whether there were changes in cell phenotype with culture. Therefore, we cannot exclude the possibility that some of the observed differences between HLE and HUVE described herein were manifestations of altered cell properties as a consequence of adaptations to culture conditions. The HLE exhibited several characteristics that differed somewhat from the HUVE. Because the two cell types exhibited different medium requirements for optimal growth, all experiments in which the properties of the cells were compared were preceded by a 48-h incubation in RPMIIlO% FBS after which the cells were treated in an identical manner. Although both cell types expressed Factor VIII R-Ag, there were differences in the distribution of this antigen. Staining of HUVE revealed intense intracellular vesicular staining as well as some extracellular fibrillar deposits. Factor VIII R-Ag is localized to rod-like Weibel-Palade bodies in HUVE (30, 31), but in vivo there structures are rare in mature capillaries (6, 31, 32). The Weibel-Palade body is thought to serve as a storage site for Factor VIII R-Ag and to allow for a mechanism of inducible secretion by thrombin and other vasoactive substances (31). Factor VIII R-Ag in HLE exhibited a diffuse cytoplasmic staining, and the majority of the staining appeared in the extracellular fibrillar assays. Attempts to induce vesicular localization of Factor VIII R-Ag using the HUVE growth medium (EGM-UV) or supplementing the HLE growth medium with dibutryl cyclic AMP (1 to 5 mM) or isobutylmethylxanthine (500 ILM) (strategies used previously by others to "differentiate" endothelial cells [33, 34]) or by serum starvation had no effect on the fibrillar distribution of Factor VIII R-Ag in the HLE (unpublished observations). Subendothelial deposits of Factor VIII R-Ag have been described in the intra-alveolar capillaries of the lung (35) and in the subendothelial matrix (36) in situ. The extracellular nature ofthe Factor VIII R-Ag may indicate an altered equilibrium in antigen distribution or the predominance of a constitutive rather than inducible secretory mechanism for Factor VIII R-Ag secretion in HLE. The deposition of Factor VIII R-Ag in the subendothelial matrix could playa role in providing sufficient Factor VIII R-Ag for platelet adhesion in the case of microvessel injury, but limiting microthrombosis. A property common to many microvascular endothelial cells in vitro is the ability of the cells to organize themselves into three-dimensional tubular networks (20, 21). Additionally, tube formation is a property that distinguishes endothelial cells from fibroblasts, epithelial, and mesothelial and smooth muscle cells that normally do not form tubular arrays in vitro (20). However, tube formation on Matrigel is not a property exclusive to endothelium; Vernon and coworkers (37) recently described the formation of cord-like structures by mouse Leydig cells cultured on Matrigel. In

Carley, Niedbala, and Gerritsen: Human Lung Microvessel Endothelium

this study, we found that both HUVE and HLE exhibited the capacity to form tube-like structures when seeded to a matrix comprised of Matrigel, whereas epithelial (MDCK) and fibroblastic (CCD-37Lu) cells cultivated in the same media as the endothelial cells did not form tubes. The HLE arrays formed rapidly and exhibited an appearance reminiscent of microvascular beds in vivo. The HUVE structures were multicellular and formed at a considerably slower rate. The production of PA and PAI-I have been well documented for cultured large vessel endothelial cells. Human microvessel endothelial cells from foreskin (38) and omentum (39) have been shown to produce tPA and PAl-I, and smaller amounts of uPA were produced by the foreskin microvessel endothelial cells. The HLE produced tPA, uPA, and PAl-I, but uPA was the predominant form of PA. A recent report by Wojta and associates (40) indicated that human renal microvessel endothelial cells produce large amounts of single chain uPA. In our study, the levels of uPA exceeded those reported for renal microvessel endothelium (40) by more than lO-fold. This observation provides further support for the hypothesis of Wojta and associates (40) that the vascular bed of origin may be the major determinant of PA expression. Furthermore, differences in relative production of the two forms ofPA and in the extracellular accumulation of PAI-I may reflect dissimilarities in fibrinolytic potential. A high fibrinolytic capacity of the pulmonary capillary endothelial cells may contribute to "filtration"of systemic venous blood, removing small clots and other particulate materials before they can reach the capillary beds of other vital organs such as the brain. The rapid formation of tube-like structures by HLE and the expression of uPA may be related. Recent studies by Niedbala and associates (41, 42) have shown that treatment of HUVE with TNF-a results in the induction of uPA and a corresponding increase in tube-forming activity on Matrigel. Pretreatment of HUVE with interferon-y blocked the TNF-a-induced uPA expression as well as the increase in tube-forming activity. TNF-a treatment of HLE did not result in a further increase in measurable uPAactivity, perhaps because ofthe very high levels of uPAin otherwise untreated HLE (data not shown). Thus, the elevated expression of PAs by microvascular endothelial cells may be related to their angiogenic potential. Observations from the present study suggest that prost acyclin is not the major eicosanoid synthesized by human lung microvessel endothelial cells, a finding that is in agreement with observations in other human (25) and animal (12, 26) microvessel endothelial cells. The lung microvascular endothelial cells synthesized significant amounts of PGE" and pretreatment of the HLE with cytokines such as TNF-a and IL-I dramatically increased the capacity of the cells to synthesize PGE,. The significance of the preferential synthesis of PGE, by various microvascular endothelia is unknown. According to the work of most investigators, PGE, is pro-inflammatory, evoking pain and participating in increased vascular permeability. However, PGE, also exerts some anti-inflammatory actions, including inhibition of neutrophil secretory responses and of IL-I stimulated lymphocyte activation. Thus, the promotion of PGE, synthesis by cytokines and other pro-inflammatory mediators may form

629

part of a complex feedback loop that acts to balance the inflammatory response. In conclusion, this study documents a relatively simple, reproducible method for the isolation and long-term serial cultivationof microvesselendothelium from human lung. The differences in procoagulant/fibrinolytic (Factor VIII R-Ag, PA), angiogenic (tube formation, growth requirements), and inflammatory (eicosanoid biosynthesis) potentials suggest that these cells may exhibit specific properties, perhaps reflecting some of the specialized attributes of human lung microvessel endothelium. Acknowledgments: The writers express their appreciation to Mr. Keith Kelley for his insights and participation in the FACS component of this project. The technical assistance of Ms. Monica Stein-Picarella, Ms. Chien-Ping Shen, Ms. Kristin Huwiler, and Ms. Carol Perry is gratefully acknowledged. References I. Delvecchio, P. J., and J. R. Smith. 1981. Expression of angiotensin converting enzyme activity in cultured pulmonary artery endothelial cells. J. Cell. Physiol. 108:337-345. 2. Ryan, J. W., and U. S. Ryan. 1977. Pulmonary endothelial cells. Fed. Proc. 36:2683-2691. 3. Carley, W. W., L. Tanoue, M. Merker, and C. N. Gillis. 1990. Isolation of rabbit pulmonary microvascular endothelial cells and characterization of their angiotensin converting enzyme activity. Pulm. Pharmacol. 3:35-40. 4. Habliston, D. L., C. Whitaker, M. A. Hart, U. S. Ryan, and J. W. Ryan. 1979. Isolation and culture of endothelial cells from the lungs of small animals. Am. Rev. Respir. Dis. 119:853-868. 5. D'Amore, P. A., B. M. Glaser, S. K. Brunson, and A. H. Fenselau. 1981. Angiogenic activity from bovine retina: partial purification and characterization. Proc. Natl. Acad. Sci. USA 78:3068-3072. 6. Gerritsen, M. E., W. w. Carley, and A. J. Milici. 1988. Microvascular endothelial cells: isolation, identification and cultivation. In Advances in Cell Culture. Vol. 6. K. Maramorosch and G. H. Sato, editors. Academic Press, San Diego. 35-67. 7. Voyta, J. C., D. P. Via, C. E. Butterfield, and B. R. Zetter. 1984. Identification and isolation of endothelial cells based on their increased uptake of acetylated low density lipoprotein. J. Cell BioI. 99:2034-2040. 8. Caldwell, P. R. B., B. C. Seegal, K. C. Hsu, M. Das, and R. L. Soffer. 1976. Angiotensin-converting enzyme: vascular endothelial localization. Science 191: 1050-1051. 9. Holthofer, H., 1. Virtanen, A. L. Kariniemi, M. Hormia, E. Linder, and A. Miettinen. 1982. Ulex europaeus I lectin as a marker for vascular endothelium in human tissues. Lab. Invest. 47:60-66. 10. Jaffe, E. A., R. L. Nachman, C. G. Becker, and C. R. Minick. 1973. Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and immunologic criteria. J. Clin. Invest. 52:27452756. II. Mosmann, T. 1983. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Methods 65:55-63. 12. Gerritsen, M. E., and C. D. Cheli. 1983. Arachidonic acid prostaglandin endoperoxide metabolism in isolated rabbit and coronary microvessels and isolated and cultivated coronary microvessel endothelial cells. J. Clin. Invest. 72: 1658-1671. 13. Gerritsen, M. E., C. A. Perry, T. Moatter, E. J. Cragoe, Jr., and M. S. Medow. 1989. Agonist-specific role for Na+H+ antiport in prostaglandin release from microvessel endothelium. Am. J. Physiol. 256:C831C839. 14. Niedbala, M. J., and A. C. Sartorelli. 1989. Regulation by epidermal growth factor of human squamous cell carcinoma plasminogen activator mediated proteolysis of extracellular matrix. Cancer Res. 49:3302-3309. 15. Littlefield, B. A., D. S. Johnston, P. C. Manzer, and P. C. Roche. 1985. Glucocorticoid inhibition of urokinase-like plasminogen activators in cultured human Iymphoblasts. Endocrinology 117:1100-1109. 16. Stein, 0., and Y. Stein. 1980. Bovine aortic endothelial cells display macrophage-like properties towards acetylated J25I-labelled low density lipoprotein. Biochim. Biophys. Acta 620:631-635. 17. Sanan, D. A., E. M. Strumpfer, D. R. van der Westhuyzen, and G. A. Coetzee. 1985. Native and acetylated low density lipoprotein metabolism in proliferating and quiescent bovine endothelial cells in culture. J. Cell BioI. 36:81-90. 18. Ryan, U. S., J. W. Ryan, C. Whitaker, and A. Chin. 1976. Localization of angiotensin converting enzyme (kininase II). II. Immunohistochemis-

630

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

try and immunofluorescence. Tissue Cell 48:125-145. 19. Conroy, J. M., H. Hoffman, E. S. Kirk, H. O. Hirzel, E. H. Sonnenblick, and R. L. Soffer. 1976. Pulmonary angiotensin converting enzyme. J. Bioi. Chem. 251:4828-4832. 20. Chung-Weich, N., W. F. Patton, A. Yen-Patton, H. R Hechtman, and D. Shepro. 1989. Phenotypic comparison between mesothelial and microvascular endothelial cell lineages using conventional endothelial cell markers, cytoskeletal protein markers and in vitro assays of angiogenic potential. Differentiation 42:44-53. 21. Folkman, J., and C. C. Haudenschild. 1980. Angiogenesis in vitro. Nature 288:551-556. 22. Montesano, R., L. Orci, and P. Vassalli. 1983. In vitro rapid organization of endothelial cells into capillary-like networks is promoted by collagen matrices. J. Cell Bioi. 97:1648-1652. 23. Gospodarowicz, D., J. Moran, D. Braun, andC. R. Birdwell. 1976. Clonal growth of bovine vascular endothelial cells in tissue culture: fibroblast growth factor as a survival agent. Proc. Natl. Acad. Sci. USA 73 :41204124. 24. Gospodarowicz, D., G. Greenburg, H. Bialecki, and B. Zetter. 1978. Factors involved in the modulation of cell proliferation in vivo and in vitro: the role of fibroblast and epidermal growth factors in the proliferative response of mammalian cells. In Vitro 14:85-118. 25. Charo.T. F., S. Shak, M. E. Karasek, P. M. Davidson, and 1. M. Goldstein. 1984. Prostaglandin I, is not a major metabolite of arachidonic acid in cultured endothelial cells from human foreskin microvessels. J. Clin. Invest. 74:914-919. 26. Chaudhari, A., A. Pedram, and M. A. Kirschenbaum. 1989. PGI, is not a major prostanoid produced by cultured rabbit renal microvascular endothelial cells. Am. J. Physio/. 256:F266-F273. 27. Weksler, B. B., A. J. Marcus, and E. A. Jaffe. 1977. Synthesis of prostaglandin I, (prostacyclin) by cultured human and bovine endothelial cells. Proc. Nat/. Acad. Sci. USA 74:3922-3926. 28. Hekman, C. M., and D. J. Loskutoff. 1985. Endothelial cells produce a • latent inhibitor of plasminogen activator that can be activated by denaturants. J. Bioi. Chern. 260:11581-11587. 29. Piovella, F., G. Nalli, G. D. Malamani et al. 1978. The ultrastructural localization of factor VIII-antigen in human platelets, megakaryocytes and endothelial cells utilizing a ferritin-labeled antibody. Br. J. Haematol. 39:209-213. 30. Verweij, C. L. 1988. Biosynthesis of human von Willebrand factor. 1988.

Haemostasis 18:224-245. 31. Zetter, R R. 1981. The endothelial cells ofJarge and small blood vessels. Diabetes 30(Suppl. 2):23-28. 32. Majno, G. 1965. Ultrastructure of the vascular membrane. In The Handbook of Physiology. Section 2. Circulation. Vol. II. W. G. Hamilton and P. Dow, editors. American Physiological Society, Washington, D.C. 2293-2375. 33. Makarski, J. S., Jr. 1982. Evidence for reversible morphologic phenotypes (sprouts) of cultured endothelial cells. Cell. Bioi. Int. Rep. 6:225-233. 34. Bensch, K. G., P. M. Davison, and M. A. Karasek. 1983. Factors controlling the in vitro growth pattern of human microvascular endothelial cells. J. Ultrastruct. Res. 82:76-89. 35. Kadhom, N., C. Wolfrom, M. Gautier, J. P. Allain, and D. Frommel. 1988. Factor VIII procoagulant antigen in human tissues. Thromb. Raemost. 59:289-294. 36. Sakariassen, K. S., P. A. Bolhuis, and J. J. Sixma. 1979. Human blood platelet adhesion to artery subendothelium is mediated by factor VIII-von Willebrand factor bound to the subendothelium. Nature 279:636-638. 37. Vernon, R. R, T. F. Lane, J. C. Angello, and H. Sage. 1991. Adhesion, shape, proliferation and gene expression of mouse Leydig cells are influenced by extracellular matrix in vitro. Bioi. Reprod. 44:157-170. 38. Van Hinsbergh, V. W. M., E. D. Sprengers, and T. Kooistra. 1987. Effect of thrombin on the production of plasminogen activators and PA inhibitor-I by human foreskin microvascular endothelial cells. Thromb. Raemost. 57: 148-153. 39. Speiser, W., E. Anders, K. T. Preissner, O. Wagner, and G. MullerBerghaus. 1987. Differences in coagulant and fibrinolytic activities of cultured human endothelial cells derived from omental tissue microvessel and umbilical veins. Blood 69:964-967. 40. Wojta J., R. L. Hoover, and T. O. Daniel. 1989. Vascular origin determines plasminogen activator expression in human endothelial cells. Renal endothelial cells produce large amounts of single chain urokinase type plasminogen activator. J. BioI. Chem. 264:2846-2852. 41. Niedbala, M. J., and M. Stein. 1991. Tumor necrosis factor induction of urokinase-type plasminogen activator in human endothelial cells. Biomed. Biochim. Acta 50:427-436. 42. Niedbala, M. J., and M. S. Picarella. 1992. Tumor necrosis factor induction of endothelial cell urokinase-type plasminogen activator mediated proteolysis of extracellular matrix and its antagonism by -y-interferon. Blood 79:1-6.