JOURNAL OF CELLULAR PHYSIOLOGY 142488-495 (1990)

Effect of Inflammatory Cytokines on Human Endothelial Cell Proliferation YASUHIRO SAEGUSA, MORRIS ZIFF, L I N D A WELKOVICH, A N D DRUIE CAVENDER* Department of Internal Medicine, Inflammation Research Unit, and the Harold C. Simmons Arthritis Research Center, The University of Texas Health Science Center, Southwestern Medical School, Dallas, Texas 75235-9030 (Y.S., M.Z., D.C.); and the Department of Microbiology and Immunology, University of Miami School of Medicine, Miami, Florida 33101 (D.C., L.W.). Neovascularization, a common occurrence in chronic inflammatory lesions, requires endothelial cell (EC) proliferation. Because this form of inflammation i s often mediated by immunologically generated cytokines, the effects of such cytokines on human umbilical vein EC proliferation in vitro were investigated. Low concentrations of recombinant interferon gamma (rlFN-y) (1 0-1 00 U/ml), but not a higher concentration (1,000 Uiml), enhanced both basal and endothelial cell growth factor (ECGF)-stimulated EC proliferation. Recombinant interleukin 1 (rlL1) and recombinant tumor necrosis factor-a (rTNF) had minor effects on basal EC proliferation, but significant inhibition was observed in the presence of ECGF. A combination of riFN-y and rTNF induced marked suppression of EC proliferation, which appeared to be due to a cytotoxic effect on the EC, as demonstrated by 5’Cr release. In contrast, the combination of rlFN-y and rlL-1 had only an additive effect on EC proliferation, with no evidence of cytotoxicity. These results suggest that cytokines have important regulatory roles in local vascular proliferation. These effects varied not only with the individual cytokine, but also with the combination of cytokines used. The most striking effects were 1) the stimulation of proliferation by IFN-y at a low concentration and 2) the inhibition by both rlL-1 and rTNF of ECGF-stimulated proliferation.

Angiogenesis, the process of generation of new blood and tumor necrosis factor-a (TNF),produced mainly by vessels, is a n important feature, not only of normal macrophages (Nathan, 1987) and EC (Stern et al., phenomena such a s embryonic development and wound 1985; Miossec et al., 1986), have been demonstrated to healing, but also of pathological states such as tumor modulate the phenotypic expression and functions of growth and chronic inflammation (Folkman, 1985). EC. These include expression of cell surface human The growth of new blood vessels requires the prolifer- lymphocyte antigen (HLA) class I1 (Pober et al., 1983) ation of endothelial cells (EC), and a number of growth and other antigens (Pober et al., 19861, ability to bind factors capable of stimulating EC proliferation have both polymorphonuclear leukocytes (Bevilacqua et al., been isolated and characterized (Maciag et al., 1979; 1985; Gamble et al., 1985) and lymphocytes (Yu et al., Gospodarowicz et al., 1978a; D’Amore et al., 1981; 1985; Cavender et al., 1986, 1987b), expression of proKlagsbrun and Smith, 1980; Miyazono et al., 1987; coagulant activity (Bevilacqua et al., 1984, 19861, and Gospodarowicz et al., 1978b; Thomas et al., 1985). synthesis of prostaglandins (Albrightson et al., 1985). Some of these factors have also been shown to stimu- To investigate the possibility that these cytokines may late angiogenesis, as measured by the chick embryo also regulate angiogenesis, we have examined the efand rabbit corneal assays (D’Amore et al., 1981; Klags- fect of recombinant IFN-y, IL-1, and TNF on human umbilical vein EC (HUVEC) proliferation. brun and Smith, 1980). In immunologically mediated chronic inflammation MATERIALS AND METHODS such a s rheumatoid synovitis, a significant increase in the number of capillaries and postcapillary venules Materials is observed, and a variety of mononuclear cell populations, including lymphocytes, macrophages, and One batch of recombinant IFN-y was purchased from plasma cells, infiltrate the perivascular space. Recent AmGen Biologicals (Thousand Oaks, CA) and had a investigations have demonstrated that inflammatory specific activity of 1.6 x lo7 U/mg, as determined by cells and the factors they produce are potent inducers of the manufacturer in comparison with a n NIH IFN-y EC proliferation (Martin et al., 1981; Ooi et al., 1983; Watt and Auerbach, 1986) and neovascularization (Polverini et al., 1977). Inflammatory cytokines such as gamma interferon (IFN-y), produced by activated Received January 26, 1989; accepted November 8, 1989. T-lymphocytes (Epstein, 1981) and interleukin 1(IL-1) *To whom reprint requestskorrespondence should be addressed. 0 1990 WILEY-LISS, INC

CYTOKINES AND ENDOTHELIAL CELL PROLIFERATION

standard. Unless noted otherwise, this preparation of IFN-y was used in the experiments described below. A second batch of recombinant IFN-y was a gift from Dr. Alan Shaw (Biogen, Geneva Switzerland) and had a specific activity of 3.0 x 10.i U/mg as determined by the manufacturer using a n antiviral cytopathic assay. These two batches of IFN-y had qualitatively similar effects on EC proliferation. Recombinant TNF was TNF-a, also a gift from Biogen, and had a specific activity of lo7 Ulmg, a s determined by the manufacturer using a n internal standard and a 48-hr cytotoxicity assay on L929 cells. Recombinant IL-la and IL-1p were also obtained from Biogen; both had a specific activity of 5 x 1 0 7 U/mg, a s determined by the manufacturer using the lymphocyte-activity factor (LAF) assay with murine thymocytes. Recombinant human endothelial cell growth factor (rECGF) was kindly provided by Dr. Yee-Hon Chin (University of Miami School of Medicine).

489

harvester (Mini-Mash 11) (MA Bioproducts, Walkersville, MD) and counted in a liquid scintillation counter.

Fibroblast proliferation The fibroblast proliferation assay was performed as described previously (Matsubara and Ziff, 1987). Human foreskin fibroblasts, in their fifth to tenth subpassage, were harvested from stock cultures by trypsinization and resuspended in RPMI 1640 supplemented with 10%FCS and antibiotics. Then 1 x 10 cells in 0.2 ml medium were plated in each well of flat-bottomed microtiter plates. Cultures were incubated for 72 hr, and 1 pCi of 3H-TdR was added to the cultures 15-18 h r before harvesting and 3H-TdR incorporation was measured a s described above.

EC cytotoxicity Cytotoxicity for EC was determined by a 51Cr release assay as described previously (Cavender et al., 1987a) with some modifications, and 2 x lo4 EC in 0.2 ml of Preparation of human umbilical vein EC medium were seeded into gelatin-coated wells of 96EC were obtained from human umbilical cords by well plates and allowed to adhere for 2-3 h r in a COs collagenase (Worthington, Freehold, NJ) digestion, us- incubator (95% air, 5% COz, 37" C). They then were ing a modification of the method of Jaffe et al. (19731, incubated with 1 pCiiwell of Na25'Cr04 (ICN Radioas previously described (Johnson, 1980). EC from sin- chemicals) overnight, washed with warm medium, and gle cords were cultured in complete medium, i.e., RPMI incubated with test agents in 0.2 ml of medium for 24 1640 (GIBCO, Grand Island, NY), containing 15% hr. After incubation, the microtiter plates were centriheat-inactivated fetal calf serum (FCS) (GIBCO), 10% fuged, 0.1 ml of su ernatant was collected from each pooled human type 0 serum from normal donors, 25 well, and released Cr was measured usin a gamma ' pg/ml EC growth supplement (ECGS) (Collaborative counter. To determine the total amount of 8 'Cr incorResearch, Lexington, MA), 5 Uiml heparin (Upjohn porated into the EC before the incubation, 0.2 ml of 1% Co., Kalamazoo, MI), and antibiotics, in gelatin-coated Triton XlOO (Sigma Chemical Co., St. Louis, MO) was tissue culture flasks. At confluence, the cells were de- added to one quadruplicate set of EC wells and incutached, using 0.125% trypsin (GIBCO) and 1 mM bated for 24 hr. Percent specific 51Cr release was calEDTA, resuspended in complete medium, and seeded culated as follows: into three flasks for further passage. EC were used a t cpm (experimental) - cpm control the third or fourth subpassage. Cells were confirmed to % specific51Cr release = x 100 cpm (total) - cpm (control) be EC by immunofluorescent staining with a rabbit antifactor VIII antiserum (Cappel Laboratories, West Spontaneous release of 51Cr ranged between 20-30% of Chester, PA). Fewer than 1%of the cells were stained the total counts incorporated, by a monoclonal antimacrophage antibody, 63D3 RESULTS (Nunez et al., 19821, by FACS analysis.

8

EC proliferation The EC proliferation assay was carried out as described previously (Matsubara and Ziff, 1987) with some modifications. EC were detached from confluent monolayers by trypsinization, washed, and resuspended in RPMI 1640 containing 5% FCS and antibiotics. Then 1.5 x lo4 EC in 0.2 ml medium were seeded into each well of gelatin-coated, 96-well, flat-bottomed microtiter plates and allowed to adhere to the gelatin by incubating for 2-3 h r in a CO, incubator (95% air, 5% COz, 37°C). The EC monolayers then were cultured in the presence or absence of varying concentrations of ECGS plus 1U/ml heparin and with or without IFN-y, IL-1, or TNF for 48 hr. Fifteen to 18 h r before harvesting, 1 pCi of tritiated thymidine (3H-TdR) (ICN Radiochemicals Co., Irvine, CA) was added to each well. After completion of the incubation, EC were washed once with warm phosphate-buffered saline and detached by trypsinization. In all experiments, inspection of the EC monolayers by phase-contrast microscopy showed that they remained in a subconfluent state. The EC were harvested onto glass fiber paper, using a n automatic

Enhancement of EC 3H-thymidineincorporation by IFN-.(-and inhibition by IL-1 and TNF Studies of EC proliferation were carried out under basal conditions and in the presence of ECGS and heparin at concentrations previously found to be optimal for such proliferation (Matsubara and Ziff, 1987). As shown in Figure lA, rIFN-y increased both unstimulated and ECGS-heparin stimulated 3H-TdR incorporation by EC. The maximal effect was observed a t concentrations of 10-100 U/ml. At 1,000 U/ml of rIFN-y, EC proliferation was decreased. In fact, ECGS-heparin induced EC 3H-TdR incorporation was suppressed below the control level a t this concentration. Similar results were obtained using IFN-y preparations from two different sources when these were tested in parallel (Table 1). In addition, similar results were obtained if the EC were cultured in medium containing 5% human serum rather than 5% FCS, or if recombinant human ECGF was substituted for the crude ECGS used in Figure 1 (data not shown). In contrast to the results with IFN-y, rIL-1 had almost no effect on basal EC DNA synthesis except for a

490

SAEGUSA ET AL.

A.

B.

TABLE 2. Effects of cytokines on EC proliferation, as determined by direct cell counts

C.

Length of culture (davs)

Cvtokine

IFN-y IL-la IL-lp TNFa -

f

.w "0

1

0L-

404QQ4000

IFN-YIU/ml)

0 0.1 4 40 400 IL-l(U/mil

0LU-L-J0 4 10 100 1000 TNF (U/ml)

Fig, 1. Effect of cytokines on EC proliferation. 1.5 x lo4 EC were incubated in the presence (6.) or absence (0-0) of 12.5 pgiml ECGS plus 1 U/ml heparin with varying concentrations of rIFN-y (A), rIL-lp (B), or rTNF (0.Values represent the mean ? SD of five separate experiments, each done in triplicate.

TABLE 1. Stimulatory effects of two different preparations of IFN-y on EC uroliferation

IFN-y (Uirnl) 1 10 100 1,000

EC proliferation (mean % increase above control ? SD)' AmGen IFN-y ND 48.7 2 13.2 44.2 t 10.3 8.3 -c 3.6

Biogen IFN-y 12.1 ? 4.1 34.9 ? 10.1 68.5 ? 10.6 19.6 ? 11.6

'EC were cultured with or without the indicated concentrationsof IFN-y In the absence of ECGS and heparin. Values represent mean 2 SD of quadruplicate determinationsand are expressed as the % increase relative to the cpm observed in EC cultures in the absence of IFN-y (18,844 -t 3,560).

small inhibition of 19.4%, observed at a high concentration (100 U/ml) (Fig. 1B). This cytokine, moreover, markedly inhibited ECGS-heparin stimulated EC proliferation in a dose-dependent manner (43.6% a t 100 U/ml) (Fig. 1B). Nearly identical results were obtained with rIL-la (data not shown), which is consistent with the observation that IL-la and IL-lP appear to bind to the same cell-surface receptor (Dower et al., 1986; Matsushima et al., 1986; Kilian et al., 1986). Similarly, rTNF produced a small decrease in basal EC proliferation (20.5% with 100 U/ml rTNF) and markedly inhibited ECGS/heparin-stimulated 3H-TdR incorporation (50.5% with 100 U/ml rTNF) (Fig. 1C). The inhibition of ECGS/heparin-induced EC proliferation produced by rIL-lp or rTNF was not reversed at concentrations of ECGS as high as 100 pg/ml (data not shown), suggesting that the inhibitory actions of rIL-1 and rTNF were not due to competition with ECGF. Trypan blue exclusion staining showed, moreover, that the viability of the EC was not affected by the cytokines at the concentrations used in these experiments.

Effects of cytokines on EC proliferation, as determined by direct cell counts To determine if the above results, obtained by measuring the incorporation of 3H-TdR by the EC, accurately reflected the effects of the cytokines on EC proliferation, a n experiment was carried out in which replicate cultures of EC were incubated with the vari-

IFN-y IL-la IL-1p TNFa

IFN-y IL-la IL-lp TNFa

EC numbers ( x Controls 6.3 4.9 6.5 6.7 6.7 7.2 8.5 6.5 5.2 6.3 3.3 9.2 3.0 3.8 3.8

ECGS

+

heparin 9.2 10.0 5.9 5.4 5.9 11.5 16.0 6.2 9.4 5.7 20.0 25.0 8.5 8.1 8.1

'EC were cultured at a concentration of 5.0 x lo4 cells/well in gelatin-coated, 24-well plates in RPMI/5% FCS 2 ECGS and heparin and in the presence or absence of 30 Uiml IFN-y, 100 U/ml IL-la, 10 Uiml IL-lp, or 100 Uiml TNFa. The cultures were fed every 2 days with fresh cytokines. Triplicatecultures were harvested by trypsinization, pooled, and counted on a hemocytometer on the indicated days.

ous cytokines for up to 6 days, and actual EC numbers were determined every other day. As shown in Table 2, in the absence of ECGS and heparin, incubation of EC in the presence of IFN-y resulted in a n approximately threefold increase in the number of recovered cells on day 6 as compared with controls. A relatively small stimulatory effect of IFN-y was observed in the presence of ECGS/heparin, which probably reflects the fact that EC cultured in the presence of both ECGS/heparin and IFN-y were near the confluent concentration of 30 x lo4 cells/well by day 6 . IL-la, IL-lp, and TNF had little effect on the number of cells recovered in the absence of ECGF, but they greatly inhibited ECGS/ heparin-induced proliferation. Thus, these results confirm the data obtained using the 3H- TdR uptake assay. Effect of IFN-y on EC grown in the absence of heparin or made quiescent by culture in 1% serum In the above experiments, the EC used had been grown for three or four passages in the presence of both ECGS and heparin. Because it is known th a t heparin can bind to IFN--y (Braude, 19841, it was important to rule out that EC-associated heparin might be influencing the effects of the added IFN-y. In addition, it was considered that some of the EC still might have been activated at the time they were treated with IFN-y and that their proliferative state could affect how they responded. To investigate these possibilities, we compared the effects of IFN-y on EC grown in complete medium (rECGF + heparin) or in medium with a tenfold increase in the concentration of rECGF in the absence of heparin. In addition, each of those two types of EC were cultured for 24 h r in medium containing only 1%serum to induce them to become quiescent before the addition of IFN-y. As shown in Table 3, IFN-y, added 1day after the plating of the EC, stimulated the proliferation of EC grown in the absence of heparin as well or better than EC grown in the presence of heparin. In addition, similar results were obtained regardless of whether or not the cells were incubated for 24 hr

491

CYTOKINES AND ENDOTHELIAL CELL PROLIFERATION TABLE 3. Stimulatory effects of IFN-y on EC grown in the presence or absence of heparin'

IFN-)I (Uiml) A. Agents added 1 day after plating:

-

1 10 100 1,000

B. Agents added 2 days after plating:

1 10 100 1,000

ECGFI heparin

+

-

-

-

+ -

-

EC proliferation Cells grown in 10 x ECGF Cells grown in 1 x ECGF/heparin Mean cpm f SD S.I.2 Mean cpm t SD SI 7,760 f 1,052 19,388 i 3,708 13,545 t 1,406 18,686 i 1,557 18,476 -t 2,657 13,752 i 909

(1.00) 2.50 1.75 2.41 2.38 1.77

8,440 i 350 15,496 ? 1,320 11,418 i 1,349 13,535 ? 1,155 13,644 2 1,190 8,964 i 820

(1.00) 1.81 1.34 1.58 1.60 1.05

9,044 f 1,087 14,354 ? 666 10,777 ? 731 12,504 i 1,636 14,176 i 2,365 10,309 -t 478

(1.00) 1.59 1.19 1.38 1.57 1.14

5,853 i 284 12,493 f 1,365 8,384 2 1,519 9.262 t 1,250 10,330 t- 1,578 8,192 f 1,175

(1.00) 2.13 1.43 1.58 1.76 1.40

'EC from a single umbilical cord were grown to confluency in first passage with tenfold (10x 1 rECGF in the absence of heparin. The culture then was divided in half and grown to confluency in either normal complete medium (rECGF + heparin) or with 10 X rECGF in the absence of heparin. EC from the two confluent second passage flasks then were harvested with trypsin and resuspended in RPMU6% human serum. EC from each of the two flasks the.1 were plated out on each of two microtiter plates for the proliferation assays and allowed to incubate overnight. The indicated reagents (IFN-7or ECGFheparin, as a positive control) were added to one of the sets of two plates the next day (partA). The other two microtiter plates were incubated for 24 hr in RPMI/l% human serum before the indicated reagents were added in RPMI/5% serum. The results shown are representative of three such experiments. cpm in experimental cultures (IFN--yor ECGF/heparin) 'SSI = Stimulation index = cpm in control cultures

with medium containing only 1%serum before the addition of IFN-y.

A.

B

Effect of combination of IFN-y with IL-1 or TNF It is well known that the action of TNF is synergistically enhanced by IFN-y in a number of systems (Pfizenmaier et al., 1987; Sugerman et al., 1985; StoneWolff e t al., 1984; Fransen et al., 1986). We therefore examined the effects of rTNF and rIL-1 on EC proliferation in the presence of rIFN-y. When 100 U/ml of rIFN-y were combined with increasing concentrations of rIL-1, the latter diminished the increased 3H-TdR incorporation observed in the presence of rIFN-y alone in parallel with its effect on the basal level, suggesting 0 0.1 4 40 400 0 4 10 1001000 that rIL-1 and rIFN-y had independent and opposing, TNF (U/ml) I L - 1 (U/ml) additive effects on EC proliferation (Fig. 2A). On the other hand, the growth inhibitory action of rTNF was 2. Effects of combinations of IFN-y with either IL-1 or TNF on markedly enhanced by the presence of 100 U/ml rIFN- Fig. EC proliferation. EC were cultured with increasing concentrations of y, although this concentration of rIFN-y alone had a rIL-lp (A) or rTNF (B) in the presence (-) or absence (0-0) of 100 growth stimulatory effect (Fig. 2B). The latter result Uiml rIFN-y. Values represent the mean i SD of three experiments, indicates that rTNF and rIFN-y act in a synergistic each done in triplicate. manner to decrease EC proliferation.

--

Additive effect of IL-1 and TNF Because IL-1 and TNF have been reported to have a synergistic antiproliferative action on certain tumor cell lines (Ruggiero and Baglioni, 19871, we investigated the combined effects of rIL-1 and rTNF on the growth of EC in the presence of ECGF plus heparin. An approximately additive inhibitory effect was observed (Fig. 3). Cytotoxic action of TNF plus IFN-y, but not IL-1 plus IFN-y To analyze the mechanism of the growth inhibitory activity of rIL-1 and rTNF, we investigated their cytotoxicity for EC. After 24 h r of incubation with up to 1,000 U/ml rIFN-y, 100 U/ml rIL-1, or 1,000 U/ml rTNF alone, no significant cytotoxicity for EC was de-

tected as measured by 51Cr release (data not shown). Moreover, as shown in Table 4,EC were not susceptible to a combination of 10 U/ml rIL-1 and 100 U/ml rIFN-y. However, significant 51Cr release was observed in the combined presence of 100 U/ml rTNF and 100 U/ml rIFN-y. The concentrations of IL1 and TNF used in these experiments were chosen because they produced equivalent inhibition of ECGF-stimulated proliferation (Fig. 1). The results were consistent with the finding of decreased 3H-TdR incorporation in the presence of combined rIFN-y and rTNF described above (Fig. 2). Similar observations were obtained when assays were performed in the presence or absence of ECGS plus heparin (Table 4).Thus, although both rIL-1 and rTNF had the ability to inhibit ECGF-stimulated EC DNA synthesis, they were not, by themselves, cytotoxic for

492

SAEGUSA ET AL h

m

F

r

Yr)

A.

B.

I

I

r

T

-

0

0.01

0.1

1

Fig. 3. Effect of combined IL-1 and TNF on ECGF-induced EC proliferation. EC were cultured with rIL-1p and rTNF in the presence of 12.5 p,giml ECGF plus 1 Uiml heparin. The concentrations of rTNF were 0 Uiml (o-), 0.1 Uiml ( S O ) , 1 Uiml (A-A), and 10 Uiml (n-n). Values represent the mean ? SD of triplicate determinations.

I L - 1 (U/mi)

TNF (U/mll

Fig. 4. Effect of cytokines on fibroblast proliferation. 1 x lo4 human foreskin fibroblasts per well were incubated in the absence (0-0) or presence (0-0) of 100 Uiml rIFN-y with increasing concentrations of rIL-lp (A) or rTNF (B). Values represent the mean i SD of three experiments, each done in triplicate.

gesting that it might also increase TNF receptors on the surface of EC. Such a n increase in TNF receptors TABLE 4. Cytotoxic action of combinations of cytokines on EC1 could render EC more susceptible to a cytotoxic effect of TNF. To investigate this possibility, we examined the Concentration % specific W r release' effect on EC cytotoxicity of pretreatment of the EC Cytokine (Uiml) Without ECGS With ECGS with rIFN-y (Table 5). When EC were preincubated IFN-v 100 2.4 i 1.3 3.3 i_ 2.0. with 100 U/ml rIFN-y for up to 24 hr, no significant ~. _ . . _ IL-16 10 4.8 2 2.5 1.5 2 2.6 51Cr release was detected after 24 h r of further incu100 TNFa 1.4 i 3.5 2.2 i 0.4 bation with 100 U/ml rTNF. However, a s noted above, IFN-7 + IL-1 100 + 10 4.9 2 1.7 2.5 2 6.7 100 + 100 21.6 ? 2.3 28.9 * 1.2 IFN-4 + TNF considerable cytotoxicity was observed when the EC ' 5'Cr-labeled EC were cultured with test agents in the presence o r absence of were cocultured with the two cytokines. 12.5 pgiml ECGS plus 1 Uiml heparin for 24 hr. After incubation, the supernaEffect of cytokines on fibroblast proliferation tant of each well was collected and counted. Similar results were obtained in two other experiments. It has been reported that both IL-1 and TNF are 'Values represent the mean 2 SD of quadruplicate determinations. capable of stimulating fibroblast proliferation (Postlethwaite et al., 1984; Vilcek et al., 1986). However, TABLE 5. Effect of pretreatment of EC with IFN-y on cytotoxic conflicting data have been published on the effect of action of TNF' rIFN-y on this proliferation (Brinckerhoff and Guyre, 1985; Duncan and Berman, 1985). To compare our findAgents present ings on EC with published data on fibroblasts, we exPretreatment in TNF % specifiP Cr release amined the effect of these cytokines on fibroblast prowith IFN-7 cytotoxicity (hr) assay Experiment 1 Experiment 2 liferation. This experiment confirmed that both rIL-1 and rTNF alone stimulated proliferation of subconf lu-2 TNF 4.7 f 0.9 0.6 t 4.3 TNF 0.8 ? 1.7 4 1.8t 0.6 ent human foreskin fibroblasts, with maximal effects 8 TNF 6.4 i 1.9 5.9 i 0.8 being observed a t 10 to 100 U/ml of rIL-1 and 100 U/ml TNF 1.0 ? 1.5 1.2 f 1.4 of rTNF (Fig. 4). On the other hand, rIFN- alone 242 TNF + IFN-y3 33.3 t 0.9 23.1 * 3.4 caused a n approximately 20% suppression of 3H-TdR 'EC were cultured with 51Cr for 24 hr. During the final indicated hours of this incorporation a t a concentration of 100 U/ml. However, period, 100 Uiml rIFN-y were added to the cultures. After completion of the labeling and preincubation, the EC were washed and further incubated with 100 when fibroblasts were cultured with 100 U/ml of rIFNUiml rTNF for an additional 24 hr. Released 51Cr was measured as described in y and increasing concentrations of rIL-1 or rTNF, the determinations. Table 2. Values remesent the mean -C SD of ouadruDlicate . . inhibition of proliferation observed with this concen'rIFN- not added: 3Afterx'Cr labeling, EC were cultured with 100 Uiml rTNF plus 100 Uiml tration of rIFN-y became more manifest, being esperIFN-v for 24 hr. cially strong at concentrations of rTNF of 100 U/ml or greater. These changes were associated with no significant cytotoxicity a s measured by 51Cr release over a EC. When cultured in the presence of rIFN-y, however, 24-hr period (data not shown). rTNF, but not rIL-1, exerted a cytotoxic effect. This DISCUSSION cytotoxicity, produced by a combination of rIFN-y and This study demonstrates that a low concentration of rTNF, required a t least a n 8-hr incubation period to rIFN-y can stimulate subconfluent HUVEC to prolifmanifest itself (data not shown). It has been reported that IFN-y increases the num- erate, both in the presence or absence of ECGF. The ber of TNF receptors on the surface of HeLa cells (Tsu- increase in proliferation, confirmed by both 3H-TdR injimoto and Vilcek, 1986; Tsujimoto et al., 19861, sug- corporation and direct cell counts, was unexpected in ~~

~

~~

CYTOKINES AND ENDOTHELIAL CELL PROLIFERATION

view of the fact that interferons, including IFN-y, have in general been considered to be antiproliferative agents (Friedman and Vogel, 1983). With regard to early reports using nonrecombinant IFN-y, in which the antiproliferative effect has been described, it is pertinent to mention that crude preparations of IFN-y were very likely contaminated with other factors, such as lymphotoxins, which are difficult to separate from IFN-y (Stone-Wolff e t al., 1984), and that IFN-y and lymphotoxins are known to exert strong synergistic cytotoxicity and growth inhibition of certain cell types (Stone-Wolff et al., 1984; Lee et al., 1984). Moreover, recent studies have, in fact, demonstrated that IFN-y can stimulate fibroblast and B-lymphocyte proliferation under appropriate conditions (Brinckerhoff and Guyre, 1985; Defrance et al., 1986). Friesel et al. (1987) recently reported that 100-1,000 U/ml of human IFN-y inhibited ECGF-induced human umbilical vein EC proliferation and that murine IFN-y decreased the number of ECGF receptors on the surface of a murine EC line. The reason for the difference between our observations with human EC and those of Friesel et al. are not clear, although differences in the amounts of IFN-y used may be responsible. In the present study, using material from two different commercial sources, maximal growth stimulatory activity of rIFN-y was observed between 10 and 100 U/ml of rIFN-y, and inhibition occurred a t 1,OOOU of IFN-y. Because the specific activity of our IFN-y preparation from AmGen was approximately 1.6 x lo7 U/mg, IFNy was stimulatory a t concentrations of 0.6-6 ng/ml, but inhibitory at 60 ng/ml. Based on the reported specific activity of the IFN-y preparation used by Friesel et al. (1987) (1.5 x lo6 U/mg), we calculate that 1001,000 U/ml of their IFN-y is equivalent to 67-670 ng/ ml. Therefore, if one compares the results of the two studies on the basis of ng/ml of IFN-y, rather than on a Ulml basis, the two studies agree that 60 ng/ml or more of IFN-y is inhibitory. IL-1 and TNF, both produced mainly by cells of the monocytelmacrophage lineage, share a number of biological actions on EC. These include the stimulation of procoagulant activity (Bevilacqua et al., 1984, 19861, expression of certain surface antigens (Pober et al., 1986), induction of morphological changes (Montesano et al., 1985; Stolpen et al., 1986), and stimulation of neutrophil and lymphocyte adherence (Bevilacqua et al., 1985; Gamble et al., 1985; Cavender et al., 1986, 198713). With regard to EC proliferation, however, results have been at variance. Ooi et al. (1983)found that IL-1 stimulated EC proliferation, whereas Libby et al. (1985) did not. In findings similar to ours, Norioka et al. (1987) reported that IL-1 inhibited ECGF-induced proliferation, but it had little or no effect on basal EC proliferation. On the other hand, it has been consistently reported that TNF inhibited EC proliferation (Stolpen et al., 1986; Frater-Schroeder et al., 1987). In the present study, we observed that both rIL-1 (either c1 or p) and rTNF weakly suppressed 3H-TdR incorporation in the absence of ECGF, but markedly suppressed ECGF-stimulated EC proliferation. This suppression occurred in the absence of cytotoxic effects on the EC. It appears unlikely that the inhibition of ECGF-induced EC proliferation we observed was due to simple competitive inhibition of the growth factor,

493

because increased concentrations of ECGF did not overcome the inhibitory effects of rIL-1 and rTNF. In combination, rIFN-y and rIL-1 appeared to have independent, additive effects on EC proliferation. However, a marked synergistic inhibition of proliferation was produced by the combination of rIFN-y and rTNF. Synergy between TNF and IFN-y is well known, not only with regard to the induction of cytotoxic/cytostatic effects (Sugerman et al., 1985; Stone-Wolff et al., 1984; Fransen et al., 1986), but also in relation to noncytocidal activities (Pfizenmaier et al., 1987; Stolpen et al., 1986; Trinchieri e t al., 1986). In the present investigation, we found t h a t although rTNF and rIFN-y had opposite effects on EC proliferation when used alone, they strongly inhibited such proliferation in combination. Although either agent alone had no effect on EC viability a s determined by "Cr release, the EC were lysed in the combined presence of rTNF and rIFN-y. This cytotoxicity, induced by the combined action of rIFN-y and rTNF, is probably the major cause of the marked inhibition of EC proliferation produced by the combination of these two agents. Synergistic cytotoxicity of IFN-y and TNF for EC has previously been reported by Stolpen et al. (1986). Thus, two types of inhibition of EC proliferation appeared to exist. In one, both rTNF and rIL-1 inhibited ECGF-stimulated proliferation without exerting a cytotoxic effect. In the second, rTNF plus rIFN-y markedly inhibited basal EC proliferation, and in this inhibition cytotoxicity had a role. The molecular basis for the EC cytotoxicity expressed by rTNF in combination with rIFN-y is not clear. It has been reported that IFN-y increases the number of TNF receptors on the surface of HeLa cells (Tsujimoto and Vilcek, 19861, suggesting t h a t i t might also increase such receptors on EC. However, no correlation between cytotoxicity and the number of TNF receptors on certain cell lines has been observed (Schweigerer et al., 1987). Furthermore, pretreatment of EC with rIFN-y had no effect on the amount of 51Cr released from the EC by rTNF in the present experiments. Therefore, the EC cytotoxicity induced by rTNF and rIFN-y is presumably not directly related to TNF receptor number. It appears, rather, to be related to a phenomenon that requires the simultaneous presence of the two cytokines in the EC culture. The cytotoxic effect of a Combination of IFN-y and TNF is of particular interest, as it is likely that these cytokines would be released together a t chronic inflammatory sites. Their synergistic injurious effect on the endothelium could have a role in the development of a number of abnormal phenomena t h a t occur in small blood vessels in such lesions. These include the emigration of neutrophils and mononuclear cells and the development of the intimal necrosis observed in inflammatory vasculitis. Such endothelial injury could also contribute to the tumor-necrotizing activity of TNF by interfering with the vascular supply to the tumor. Finally, the EC cytotoxicity generated by these cytokines could modulate the proliferation of the microvasculature that occurs in chronic inflammatory states. Several lines of evidence have demonstrated that IL1 and TNF are potent stimulators of fibroblast proliferation (Postlethwaite et al., 1984; Vilcek et al., 1986).

494

SAEGUSA ET AL

On the other hand, in the present experiments, these agents demonstrated inhibitory effects on EC proliferation. It is difficult to reconcile the differing effects of these cytokines on the fibroblast and EC, as fibroblast proliferation and EC proliferation are usually parallel phenomena in the growth of the connective tissue. However, in chronic inflammation, the infiltrating mononuclear cells are potent producers of EC growth factors (Martin et al., 1981; Watt and Auerbach, 1986); and a number of growth factors, including epidermal growth factor (Gospodarowicz, 1978b), platelet-derived EC growth factor (Miyazono et al., 19871, and fibroblast growth factor (Tsujimoto and Vilcek, 1986) are capable of stimulating EC proliferation. Such agents may stimulate EC proliferation at the inflammatory site while IL-1 and TNF exert their negative effects. In this manner, IL-1 and TNF could have a n important role in the modulation of EC proliferation.

ACKNOWLEDGMENTS The authors thank Mr. Dale Edelbaum and Mrs. Carla Suson for their skillful assistance. This work was supported by NIH grants AM09989 and AI27809. LITERATURE CITED Albrightson, C.R., Baenzinger, N.L., and Needleman, P. (1985) Exaggerated human vascular cell prostaglandin biosynthesis mediated by monocytes: Role of monokines and interleukin 1. J . Immunol., 135,1872-1877 Bevilacqua, M.P., Pober, J.S., Majeau, G.R., Cotran, R.S., and Gimbrone, M.A. J r . (1984) Interleukin-1 induced biosynthesis and cell surface expression of procoagulant activity in human vascular endothelial cells. J. Exp. Med., 160t618-623 Bevilacqua, M.P., Pober, J.S., Majeau, G.R., Fiers, W., Cotran, R.S., and Gimbrone, M.A. J r . (1986) Recombinant tumor necrosis factor induces procoagulant activity in cultured human endothelium: Characterization and comparison with the actions of interleukin 1. Proc. Natl. Acad. Sci. U.S.A., 83:4533-4537 Bevilacqua, M.P., Pober, J.S., Wheeler, M.E., Cotran, R.S., and Gimbrone, M.A. J r . (1985) Interleukin 1 acts on cultured human vascular endothelium to increase the adhesion of polymorphonuclear leukocytes, monocytes, and related leukocyte cell lines. J . Clin. Invest., 76t2003-2011 Braude, LA. (1984) Purification of human growth interferon to essential homogeneity and its biochemical characterization. Biochemistry, 23:5603-5609. Brinckerhoff, C.E., and Guyre, P.M. (1985) Increased proliferation of human synovial fibroblasts treated with recombinant immune interferon. J. Immunol., 134:3142-3146. Cavender, D., Haskard, D., Foster, N., and Ziff, M. (1987a) Superinduction of T lymphocyte-endothelial cell (EC) binding by treatment of EC with interleukin 1 and protein synthesis inhibitors. J. Immunol., 138:2149-2154. Cavender, D.E., Haskard, D.O., Joseph, B., and Ziff, M. (1986) Interleukin 1 increases the binding of human B and T lymphocytes to endothelial cell monolayers. J. Immunol., 136:203-207. Cavender, D., Saegusa, Y., and Ziff, M. 11987b) Stimulation of endothelial cell binding of lymphocytes by tumor necrosis factor. J . Immunol., 139t1855-1860. D’Amore, P.A., Glaser, B.M., Brunson, S.K., and Fenselau, A.H. (1981) Angiogenic activity from bovine retina: partial purification and characterization. Proc. Natl. Acad. Sci. U.S.A., 78t3068-3072. Defrance, T., Aubry, J., Vanbervliet, B., and Banchereau, J . (1986) Human interferon-? acts as a B cell growth factor in the anti-IgM antibody co-stimulatory assay but has no direct B cell differentiation activity. J . Immunol., 137:3861-3867. Dower, S.K., Call, S.M., Gillis, S., and Urdal, D.L. (1986) Similarity between the interleukin 1 receptors on a murine T-lymphoma cell line and on a murine fibroblast cell line. Proc. Natl. Acad. Sci. U.S.A., 83:1060-1064. Duncan, M.R., and Berman, B. (1985) y-interferon is the lymphokine and &interferon the monokine responsible for inhibition of fibroblast collagen production and late but not early fibroblast proliferation. J . Exp. Med., 162516.

Epstein, L.B. (1981) Interferon-gamma: Is it really different from the other interferons? In: Interferon 3. I. Gresser, ed. Academic Press, New York, pp. 13-44. Folkman, J. (1985)Tumor angiogenesis. Adv. Cancer Res., 43:1753n2

Fransen, L., Heyder, J.V.D., Ruyssehaert, R., and Fiers, W. (1986) Recombinant tumor necrosis factor: Its effect and its synergism with interferon-? on a variety of normal and transformed human cell lines. Eur. J. Cancer Clin. Oncol., 22:419-426. Frater-Schroeder, M., Risau, W., Hallman, R., Gautschi, P., and Bohlen, P. (1987) Tumor necrosis factor type a , a potent inhibitor of endothelial cell growth in vitro, is angiogenic in vivo. Proc. Natl. Acad. Sci. U.S.A., 845277-5281. Friedman, R.M., and Vogel, S.N. (1983) Interferons with special emphasis on the immune system. Adv. Immunol., 34t97-140. Friesel, R., Komoriya, A,, and Maciag, T. (1987) Inhibition of endothelial cell proliferation by gamma-interferon. J . Cell. Biol., 104: 689-696. Gamble, J.R., Harlan, J.M., KlebanoE, S.J., and Vadas, M.A. (1985) Stimulation of adhesion of neutrophils to umbilical vein endothelium by human recombinant tumor necrosis factor. Proc. Natl. Acad. Sci. U.S.A., 8253667-8671. Gospodarowicz, D., Bialecki, H., and Greenburg, G. (1978a) Purification of the fibroblast growth factor activity from bovine brain. J. Biol. Chem., 253:3736-3743. Gospodarowicz, D., Brown, K.D., Birdwell, C.R., and Zetter, B.R. (1978b)Control of proliferation of human vascular endothelial cells. Characterization of the response of human umbilical vein endothelial cells to fibroblast growth factor, epidermal growth factor and thrombin. J . Cell. Biol., 77:774-788. Jaffe, E.A., Nachman, R.L., Becker, C.G., and Minick, C.R. (1973) Culture of human endothelial cells derived from umbilical veins. J . Clin. Invest., 52:2745-2756. Johnson, A.R. (1980) Human pulmonary endothelial cells in culture: activities of cells from arteries and veins. J . Clin. Invest., 65:841850. Kilian, P.L., Kaffka, K.L., Stern, AS., Woehle, D., Benjamin, W.M., Dechiara, T.M., Gubler, U., Farrar, J.J., Mizel, S.B., and Lomedico, P.T. (1986) Interleukin l a and interleukin l p bind to the same receptor on T cells. J . Immunol., 136t4509-4514. Klagsbrun, M., and Smith, S. (1980) Purification of a cartilage-derived growth actor. J . Biol. Chem., 255r10859-10866. Lee, S.H., Aggarwal, B.B., Rinderknecht, E., Assisi, F., and Chiu, H. (1984) The synergistic anti-proliferative effect of y-interferon and human lymphotoxin. J. Immunol., 133t1083-1086. Libby, P., Wyler, D.J., Janicka, M.W., and Dinarello, C.A. (1985) Differential effects of human interleukin-1 on growth of human fibroblasts and vascular smooth muscle cells. Arteriosclerosis, 5t186191. Maciag, T., Cerundolo, J . , Ilsley, S., Kelly, P.R., and Forand, R. (1979) An endothelial cell growth factor from bovine hypothalamus: identification and partial characterization. Proc. Natl. Acad. Sci. U.S.A., 76:5674-5678. Martin, B.M., Gimbrone, M.A., Jr., Unanue, E.R., and Cotran, R.S. (1981) Stimulation of nonlymphoid mesenchymal cell proliferation by a macrophage-derived growth factor. J. Immunol. 126t15101515 Matsubara, T., and Ziff, M. (1987) Inhibition of human endothelial cell proliferation by gold compounds. J. Clin. Invest., 79t14401446. Matsushima, K., Akahoshi, T., Yamada, M., Furutani, Y., and Oppenheim, J.J. (1986) Properties of a specific interleukin (IL 1) receptor on human Epstein Barr virus-transformed B lymphocytes: Identity of the receptor for IL la and IL I@. J. Immunol., 146: 4496-4502. Miossec, P., Cavender, D., and Ziff, M. (1986) Production of interleukin 1 by human endothelial cells. J . Immunol., 136t2486-2491. Miyazono, K., Okabe, T., Urabe, A., Takaku, F., and Heldin, C.H. (1987) Purification and properties of a n endothelial cell growth factor from human platelets. J . Biol. Chem., 262t4098-4103. Montesano, R., Orci, L., and Vassalli, P. (1985) Human endothelial cell cultures: Phenotypic modulation by leucocyte interleukins. J. Cell. Physiol., 122t424-434. Nathan, C.F. (1987) Secretory products of macrophages. J . Clin. Invest., 79t319-326. Norioka, K., Hara, M., Kitani, A,, Hirose, T., Hirose, W., Harigai, M., Suzuki, K., Kawakami, M., Tabata, H., Kawagoo, M., and Nakamura, H. (1987) Inhibitory effect of human recombinant interleukin-la and p on growth of human vascular endothelial cells. Biomed. Biophys. Res. Comm., 14Sr969-975. Nunez, G.V., Ugolini, V., Capra, J.D., and Stastny, P. (1982) Mono-

CYTOKINES AND ENDOTHELIAL CELL PROLIFERATION clonal antibodies against human monocytes. 11. Recognition of two distinct cell surface molecules. Scand. J. Immunol., 16.515-523. Ooi, B.S., MacCarthy, E.P., Hsu, A., and Ooi, Y.M. (1983) Human mononuclear cell modulation of endothelial cell proliferation. J. Lab. Clin. Med., 102t428-433. Pfizenniaier, K., Scheurich, P., Schluter, C., and Kronke, M. (1987) Tumor necrosis factor enhances HLA-A,B,C and HLA-DR gene expression in human tumor cells. J. Immunol., 138t975-980. Pober, J.S., Bevilacqua, M.P., Mendrick, D.L., Lapierre, L.A., Fiers, W., and Gimbrone, M.A. Jr. (1986) Two distinct monokines, interleukin 1 and tumor necrosis factor, each independently induce biosynthesis and transient expression of the same antigen on the surface of cultured human vascular endothelial cells J . Immunol., 136: 1680-1687. Pober, J.S., Gimbrone, M.A. Jr., Cotran, R.S., Reiss, C.S., Burakoff, S.J., Fiers, W., and Auk, K.A. (1983) Ia expression by vascular endothelial cells is inducible by activated T cells and by human y-interferon. J. Exp. Med., 157t1339-1353. Polverini, P., Cotran, R.S., Gimbrone, M.A. Jr., and Unanue, E.R. (1977) Activated macrophages induce vascular proliferation. Nature, 269t804-806. Postlethwaite. A.E.. Lachman. L.B.. and Kanp, A.H. (1984) Induction of fibroblast proliferation by interleukin-7 derived from human monocytic leukemia cells. Arthritis Rheum., 27t995-1001. Ruggiero, V., and Baglioni, C . (1987) Synergistic anti-proliferative activity of interleukin-1 and tumor necrosis factor. J. Immunol., 138:661-663. Schweigerer, L., Neufeld, G., Friedman, J., Abraham, J.A., Fiddes, J.C.. Gosuodarowicz. D. (1987) Caoillarv endothelial cells exoress basic fibrbblast growth factor, a mitogen that promotes their- own growth. Nature, 325r257-259. Stern, D.M., Bank, I., Nawroth, P.P., Cassimeris, J., Kisiel, W., Fenton, J.W. 11, Dinarello, C., Chess, L., and Jaffe, E.A. (1985) Selfregulation of procoagulant events on the endothelial cell surface. J . Exp. Med., 162t1223-1235. Stolpen, A.H., Guinan, E.G., Fiers, W., and Pober, J.S. (1986) Recom~~

495

binant tumor necrosis factor and immune interferon act singly and in combination to reorganize human vascular endothelial cell monolayers. Am. J. Pathol., 123r16-24. Stone-Wolff, D.S., Yip, Y.K., Kelker, H.D., Le, J., Henriksen-Destefano, D., Rubin, B.Y., Rinderknecht, E., Aggarwal, B.B., and Vilcek, J. (1984) Interrelationship of human interferon-gamma with lymphotoxin and monocyte cytotoxin. J. Exp. Med., 1593428-843. Sugerman, B.J., Aggarwal, B.B., Hass, P.E., Figari, I.S., Palladino, M.A. Jr., and Shepard, H.M. ( I 985) Recombinant human tumor necrosis factor-a: Effects on proliferation of normal and transformed cells in vitro. Science, 230t943-945. Thomas, K.A., Rios-Candelore, M., Gimenez-Gallego, G., Disalvo, J., Bennet, C., Rodkey, J., and Fitzpatrick, S. (1985) Pure brain-derived acidic fibroblast growth factor is a potent angiogenic vascular endothelial cell mitogen with sequence homology to interleukin 1. Proc. Natl. Acad. Sci. U.S.A., 82:6409-6413. Trinchieri, G., Kobayashi, M., Rosen, M., Loudon, R., Murphy, M., and Perussia, B. (1986) Tumor necrosis factor and lymphotoxin induce differentiation of human myeloid cell lines in synergy with immune interferon. J . Exp. Med., 164t1206-1225. Tsujimoto, M., Feinman, R., and Vilcek, J. (1986) Differential effects of type I IFN and IFN-y on the binding of tumor necrosis factors to receptors in two human cell lines. J . Immunol., 137.2272-2276. Tsujimoto, M., and Vilcek, J . (1986) Tumor necrosis factor receptors in HeLa cells and their regulation by interferon-y. J. Biol. Chem., 261 :5384-5388. Vilcek, J., Palombella, V.J., Henriksen-Destefano, D., Swenson, C . , Feinman, R., Hirai, M., and Tsujimoto, M. (1986) Fibroblast growth enhancing activity of tumor necrosis factor and its relationship to other polypeptide growth factors. J. Exp. Med., I63r632-643. Watt, S.L., and Auerbach, R. (1986)A mitogenic factor for endothelial cells obtained from mouse secondary mixed leukocyte cultures. J . Immunol., 136t197-207. Yu, C.-L., Haskard, D.O., Cavender, D., Johnson, A.R., and Ziff, M. (1985) Human gamma interferon increases the binding of T lymphocytes to endothelial cells. Clin. Exp. Immunol., 62t554-560.