Inhibition of endothelial cell growth by macrophage-like U-937 cell-derived oncostatin M, leukemia inhibitory factor, and transforming growth factor beta1.

Conditioned media were collected from phorbol ester-treated human macrophage-like U-937 cells and analyzed for the presence of inhibitors of endothelial cell (EC) proliferation. By a combination of ion exchange and reverse-phase liquid chromatography, three inhibitors were purified to homogeneity as ascertained by microsequencing of 14-17 N-terminal amino acids. These inhibitors were identified as oncostatin M (OSM), leukemia inhibitory factor (LIF), and transforming growth factor β1 (TGF-β1). The identities of the three EC growth inhibitors were confirmed by demonstrating that recombinant human OSM, LIF, and TGF-β1 were inhibitory in the same concentration range. Inhibition of EC proliferation by OSM was a newly described property of this cytokine. OSM was the most potent inhibitor with a half-maximal inhibition by recombinant material of 0.15-.2 ng/ml compared with 0.6-0.9 and 0.9-1.0 ng/ml for LIF and TGF-β1, respectively. The three factors inhibited basal, vascular endothelial cell growth factor-stimulated, and fibroblast growth factor 2-stimulated EC proliferation. Interleukin-6 and ciliary neurotrophic factor, two cytokines related structurally to OSM and LIF, were not active as EC growth inhibitors. It was concluded that macrophage-like cells secrete a variety of potent EC growth inhibitors and that one of these, OSM, is among the most potent EC growth inhibitors yet reported.

An understanding of the mechanisms that control the proliferation and differentiation of vascular endothelial cells (EC) 1 should provide important insights into a wide variety of physiological and pathological events such as embryonic development, wound healing, tumor growth, diabetes, and atherosclerosis (1,2). ECs in adult tissue are normally quiescent despite the wide distribution of angiogenic factors such as acidic fibroblast growth factor (FGF-1), basic FGF (FGF-2), and vascular endothelial growth factor (VEGF) (2). One possible mechanism that might prevent abnormal EC proliferation might be the expression of EC growth inhibitory factors. A number of polypeptides have been shown to inhibit EC growth in culture. These include transforming growth factor ␤1 (TGF-␤1) (3), tumor necrosis factor-␣ (TNF-␣) (4), platelet factor 4 (5), interferon-␥ (6), cartilage-derived inhibitor (7), leukemia inhibitory factor (LIF) (8), thrombospondin (9), and angiostatin (10). Antibodies directed against the integrin ␣ v ␤ 3 also inhibit EC proliferation (11). EC inhibitors in vitro are usually also angiogenesis inhibitors in vivo, but this is not always the case. For example, TGF-␤1 and TNF-␣, which are EC inhibitors in vitro, are actually stimulators of angiogenesis in vivo. It has been postulated that TGF-␤1 and TNF-␣ act indirectly to stimulate angiogenesis by induction of angiogenesis factors from inflammatory cells, for example, macrophages and T lymphocytes (12). Another possibility is that these EC inhibitors block proliferation but stimulate tube formation, an important component of angiogenesis (13).
Since macrophages commonly reside at sites of physiological angiogenesis such as wounds and of pathological angiogenesis, such as tumors and atherosclerotic plaques, it is possible that this cell type plays an important role in the angiogenic response (14 -16). Macrophages have been shown to produce and release a number of angiogenic stimulators such as VEGF (17), FGF-2 (18), and platelet-derived growth factor (19). Based on the idea that macrophages might produce a balance of EC stimulators and inhibitors, we examined the conditioned media (CM) of U-937 cells, a human histiocytic lymphoma cell line that differentiates into macrophage-like cells upon phorbol ester treatment and that has been demonstrated to produce VEGF (17). In a previous report, we demonstrated that macrophages released an EC growth inhibitor that was neither TGF-␤1 nor TNF-␣ (20). This EC inhibitory activity was characterized as being basic and mildly heparin-binding but was not purified to any degree. We subsequently analyzed CM of U-937 cells and found a similar activity. We now report the complete purification, N-terminal microsequencing, and identification of three EC growth inhibitors released by U-937 cells. These are oncostatin M (OSM), LIF, and TGF-␤1. LIF and OSM belong to a family of IL-6-related cytokines that exhibit a similar helical structure and share receptor components, while TGF-␤1 is a member of the TGF-␤ superfamily. Unlike TGF-␤1 and LIF, OSM has not been reported previously to be an EC growth inhibitor and is particularly potent with half-maximal inhibition at 150 -200 pg/ml. We have also found that U-937 cells release VEGF, lending support to the idea that macrophage-like cells produce both EC stimulators and inhibitors.

Preparation of U-937 Cell Conditioned Media
The human histiocytic lymphoma cell line U-937 was purchased from the American Type Culture Collection and was maintained in RPMI 1640 supplemented with 10% fetal calf serum (FCS) and GPS. Cells were seeded in roller bottles at a density of 5 ϫ 10 5 cells/ml in fresh RPMI 1640 culture medium containing 5% (v/v) FCS and TPA. Two days later, cells were washed with serum-free media and were incubated in serum-free RPMI 1640 for 48 h, at which time the supernatants were collected and replaced with fresh serum-free RPMI 1640. CM were collected again at 4 and 6 days after 48 h of serum-free culture incubation. The samples of CM were pooled and used as the starting material for the purification of EC inhibitors.

Purification of Endothelial Cell Growth Inhibitors
Three liters of serum-free U-937 CM were concentrated to 125 ml by ultrafiltration (S1Y10 spiral-wound membrane, molecular weight cut off 10,000; Amicon, Inc., Beverly, MA). The retentate was dialyzed against 0.02 M acetic acid, pH 3.5, and after centrifugation (50,000 ϫ g, 30 min) the supernatant was applied to a TSK gel SP-5PW cation exchange column (7.5 ϫ 75 mm, Toso Haas, Montgomeryville, PA). The column was equilibrated with 0.02 M acetic acid, pH 4.0, and bound material was eluted with a linear gradient of NaCl (0 -1.5 M) at a flow rate of 1 ml/min. Fractions of 1 ml were collected and tested for bioactivity. Three active peaks of EC growth inhibitory activity (ECI) were pooled and designated as endothelial cell inhibitors 1, 2, and 3 (ECI 1, ECI 2, ECI 3).
ECI 1-Fractions collected from the first peak of inhibitory activity eluted from the TSK gel SP-5PW column (with 0.1-0.3 M NaCl) were pooled and injected onto a Vydac C 4 RPLC column (4.6 ϫ 250 mm) equilibrated with 10% acetonitrile, 0.1% trifluoroacetic acid. Fractions (1 ml) were eluted with a linear gradient of 20 -60% acetonitrile at a flow rate of 1 ml/min. Next, the inhibitory fractions from the C 4 RPLC column were pooled, diluted 2-fold with 0.1% trifluoroacetic acid, and injected onto a TSK gel TMS-250 RPLC column (4.6 ϫ 75 mm, Toso Haas), which had been equilibrated with 10% acetonitrile, 0.1% trifluoacetic acid. Fractions (1 ml) were eluted with a linear gradient of 32-42% acetonitrile at a flow rate of 1 ml/min. Finally, the inhibitory fractions pooled from the TMS-250 RPLC column were diluted 2-fold and injected onto a Vydac C 4 RPLC column (4.6 ϫ 50 mm) equilibrated with 10% acetonitrile, 0.1% trifluoroacetic acid. Inhibitory fractions (0.5 ml) were eluted with a linear gradient of 35-45% acetonitrile at a flow rate of 0.5 ml/min. ECI 2-The second peak of EC growth inhibitory activity eluted from the TSK gel SP-5PW cation exchange column (with 0.5-0.7 M NaCl) was injected onto a Vydac C 4 RPLC column (4.6 ϫ 250 mm) equilibrated with 10% acetonitrile, 0.1% trifluoroacetic acid. Fractions (1 ml) were eluted with a linear gradient of 20 -60% acetonitrile at a flow rate of 1 ml/min. Next, EC growth inhibitory fractions obtained from the C 4 RPLC column were injected onto a Vydac C 18 RPLC column (4.6 ϫ 250 mm). EC growth inhibitory fractions were eluted with a gradient of 35-45% acetonitrile. ECI 3-The third peak of EC growth inhibitory activity eluted from the TSK gel SP-5PW column (with 0.8 -0.9 M NaCl) was injected onto a Vydac C 4 RPLC column (4.6 ϫ 250 mm) equilibrated with 10% acetonitrile, 0.1% trifluoroacetic acid. Fractions (1 ml) were eluted with a linear gradient of 20 -60% acetonitrile at a flow rate of 1 ml/min. Next, inhibitory fractions were diluted 2-fold with 0.1% trifluoroacetic acid and injected onto a Vydac C 4 RPLC column (4.6 ϫ 50 mm) equilibrated with 2-propanol. Fractions (1 ml) were eluted with a linear gradient of 25-35% 2-propanol at a flow rate 0.5 ml/min. All the fractions collected during the various purification steps were tested for EC growth inhibitory activity at a final dilution of 1:200 (1 l of fraction/200 l of medium/well).

SDS-Polyacrylamide Gel Electrophoresis (PAGE)
The molecular masses of EC inhibitors were determined by 15% SDS-PAGE under non-reducing conditions and silver stained as described previously (23). In order to extract EC growth inhibitors, nonreduced samples were electrophoresed on a 1-mm thick, 10-cm long 15% SDS-polyacrylamide gel. The polyacrylamide gels were sliced into 1.5-2-mm pieces, which were each crushed and extracted for 6 h at 4°C with 50 l of phosphate-buffered saline. The polyacrylamide gel extracts were assayed for EC growth inhibitory activity.

Amino Acid Analysis and Microsequencing
Non-reduced samples were electrophoresed on 15% SDS-PAGE gels, as described above, and transferred to PVDF membranes (ProBlott, Applied Biosystems, Foster City, CA). After staining with 0.1% Coomassie Brilliant Blue, protein bands corresponding to EC inhibitory activity were excised and dried. One-tenth of the sample was used for amino acid analysis to estimate the total protein amount, and the other nine-tenths was analyzed for N-terminal amino acid sequences by microsequencing. Amino acid analysis and microsequencing were performed by Dr. Bill Lane of the Harvard Microchemistry Facility using Pico-Tag amino acid analysis HPLC and a gas-phase sequenator (Applied Biosystems).

Initial Purification of U-937 Cell-derived EC Inhibitors-
Human U-937 cells treated with 10 nM TPA for 24 h became adherent and acquired many of the characteristics of macrophages (24). The serum-free CM collected from the TPA-treated U-937 cells, but not from untreated U-937 cells, inhibited EC growth. About 50% inhibition of [ 3 H]thymidine incorporation into both capillary and aortic ECs occurred at a concentration of 1-2% (v/v). In order to purify the EC inhibitory activity initially, 3 liters of U-937 cell CM (150 mg of protein) were concentrated, dialyzed, and applied to a strong cation exchange column, TSK gel SP-5PW. At least three peaks of EC growth inhibitory activities were detected that were eluted with about 0.2, 0.6, and 0.85 M NaCl, respectively (Fig. 1). These three peaks were termed ECI 1, ECI 2, and ECI 3 in the order of their increasing concentrations of NaCl needed for elution. Both BCE and BAE cells were inhibited, but only the BAE cell results are shown. Besides inhibiting [ 3 H]thymidine incorporation into ECs, these three inhibitory activities also inhibited increases in cell number by 40 -65% (Table I).
Purification of ECI 1-ECI 1 was purified by a combination of C 4 RPLC ( Fig. 2A), TSK gel TMS-250 RPLC (Fig. 2B), and C 4 RPLC (Fig. 2C) as described under "Experimental Procedures." Active fractions from the last column were pooled, lyophilized, and analyzed by SDS-PAGE (Fig. 3A). A major protein band of 40 kDa and a minor doublet band of about 70 kDa were detected. The SDS-polyacrylamide gels were extracted, and EC inhibitory activity was found to be associated with the 40-kDa band (Fig. 3B). The remainder of the sample was analyzed on a replicate SDS gel, and the bands were transferred to a PVDF membrane. The 40-kDa band was excised and subjected to microsequencing (Table II), and the N-terminal amino acid sequence was found to be SPLPITPVXATXAIRHP, corresponding to the N-terminal sequence of human LIF (25). The final yield of pure LIF from 3 liters of U-937 CM was estimated to be 20 pmol (0.8 g).
Purification of ECI 2-ECI 2 was purified by a combination of C 4 RPLC (Fig. 4A) and C 18 RPLC (Fig. 4B) as described under "Experimental Procedures." The EC inhibitory fractions from the C 18 column were pooled and analyzed by SDS-PAGE, and a major protein band of 30 kDa and minor bands of 25 and 70 kDa were detected (Fig. 5A). After gel extraction, it was found that the inhibitory activity of ECI 2 was associated with the 30-kDa band (Fig. 5B). The remainder of the sample was analyzed on a replicate SDS-polyacrylamide gel, and the bands were transferred to a PVDF membrane. The 30-kDa band was excised and subjected to microsequencing (Table II), and the N-terminal amino acid sequence was found to be AAIGSXS-KEYRVLL corresponding to the N-terminal sequence of human OSM (26). The final yield in pure protein from 3 liters of U-937 cell CM, as estimated by amino acid analysis, was 8 pmol (0.24 g).
An EC mitogenic peak was eluted from the C 4 column with 30% acetonitrile (Fig. 4A). This mitogenic activity was neutralized about 50% by incubation with anti-VEGF neutralizing antibodies (data not shown), consistent with previous reports of VEGF production by U-937 cells (17).
Purification of ECI 3-ECI 3 was purified by a combination of C 4 RPLC using an acetonitrile gradient column (Fig. 6A) and C 4 RPLC using a 2-propanol gradient (Fig. 6B) as described under "Experimental Procedures." EC inhibitory activity from this second C 4 column was pooled, concentrated, and analyzed by SDS-PAGE that showed the presence of a single band of about 28 kDa (Fig. 6C). Since this was the only protein band detected, it was transferred directly to a PVDF membrane, followed by microsequencing. The N-terminal sequence was found to be ALDTNYXFSSTLKN (Table II), corresponding to the N-terminal sequence of human TGF-␤1 (27). The final yield of pure ECI 3 was estimated to be 12 pmol (0.34 g).
The N-terminal sequences obtained for human U-937 cell-derived ECI 1, 2, and 3 corresponded to human but not bovine LIF, OSM, and TGF-␤1 N-terminal sequences, respectively, making it unlikely that the EC inhibitors were artifactually derived from bovine serum during the purification process (Table II).

TABLE I Inhibition of endothelial cell proliferation by ECI 1, ECI 2, and ECI 3
BAE were plated at 3.0 ϫ 10 3 cells/well/12-well plate. Samples from the three peaks of inhibitory activity shown in Fig. 1 Fig. 1 were applied to a C 4 RPLC column followed by elution with a gradient of 20 -60% acetonitrile. B, fractions 40 -44 from the C 4 column shown in panel A were applied to a TSK gel TMS-250 RPLC column followed by elution with a gradient of 32-42% acetonitrile. C, fractions 26 -29 from the TSK gel TMS-250 column shown in panel B were applied to a C 4 RPLC column followed by elution with a gradient of 35-45% acetonitrile. The fractions from each column were diluted 1:200 and tested for the ability to inhibit [ 3 H]thymidine incorporation into BAE cells. Fig. 2C (fractions 34 -37) were pooled, concentrated, and analyzed by 15% SDS-PAGE under non-reducing conditions and silver stain. B, after SDS-PAGE, the polyacrylamide gel was sliced into 1.5-2-mm pieces and extracted for 6 h at 4°C in phosphate-buffered saline. The extracts were assayed for the ability to inhibit [ 3 H]thymidine incorporation into BAE cells.

EC Growth Inhibition by Recombinant Human LIF, OSM,
and TGF-␤1-To confirm that LIF, OSM, and TGF-␤1 were responsible for the three EC inhibitory activities purified from U-937 cell CM, recombinant human (rh) LIF (rhLIF), OSM (rhOSM), and TGF-␤1 (rhTGF-␤1) were tested. All three recombinant proteins induced a dose-dependent inhibition of BAE (Fig. 7A) and BCE (Fig. 7B) cell proliferation. Half-maximal inhibition of BAE and BCE cells, respectively, was obtained with 0.9 and 0.6 ng/ml, 0.15 and 0.2 ng/ml, and 1.0 and 0.9 ng/ml rhLIF, rhOSM, and rhTGF-␤1, respectively. Thus, OSM appeared to be the most potent of the U-937 cell-derived ECIs, followed by LIF and TGF-␤1. Since LIF and OSM belong to the same Interleukin-6 related cytokine superfamily, two other members of this family, rhIL-6 and rhCNTF, were tested for EC inhibitory activity (Fig. 7). However, these cytokines were inhibitory only in the 100 -200 ng/ml range, or log 3 less than the rhOSM.
The inhibition of BAE cell proliferation by the three EC inhibitors was non-toxic and reversible in that upon removal of the inhibitors, EC proliferation resumed (data not shown). rhLIF, rhOSM, and rhTGF-␤1 inhibited not only basal EC proliferation but VEGF-and FGF-2-stimulated EC proliferation as well and to about the same extent (Table III). DISCUSSION We have demonstrated that in addition to the EC growth stimulator, VEGF, macrophage-like U-937 cells release EC growth inhibitors as well. We have purified three such inhibi-TABLE II N-terminal amino acid analysis of endothelial cell growth inhibitors ECI 1, ECI 2, and ECI 3 were purified from the CM of U-937 cells and analyzed by SDS-PAGE. The ECI 1 and ECI 2 polyacrylamide gels were extracted, and fractions were analyzed for EC inhibitory activity, which were found to correspond to 40-kDa (ECI 1) and a 30-kDa (ECI 2) bands, respectively. SDS-PAGE analysis of ECI 3 detected a single 28-kDa band, and accordingly gel extraction was not performed. Replicate SDS-PAGE was performed, and the protein bands were transferred to PVDF membranes. The 40-kDa (ECI 1), 30-kDa (ECI 2), and 28-kDa (ECI 3) bands were excised and subjected to N-terminal microsequencing analyses. For comparative purposes, the bovine N-terminal sequences for LIF, a OSM (29), and TGF-␤1 (30) are SPLPITPVNATCATRHP, VAT-GKCSGKYHELL, and ALDTNYCFSSTEKN, respectively, each of which differs from the corresponding human N-terminal sequences by at least one amino acid substitution. Thus the purified human U-937 cel CM ECIs do not appear to be artifactually derived from bovine serum.
The bovine LIF gene sequence has been deposited in the GenBank data base with accession number D50377.

FIG. 4. Purification of ECI 2.
A, fractions 33-50 from the TSK gel SP-5PW cation exchange column shown in Fig. 1 were applied to a C 4 RPLC column followed by elution with a gradient of 20 -60% acetonitrile. B, fractions 37-40 from the C 4 column shown in panel A were applied to a C 18 RPLC column followed by elution with a gradient of 35-45% acetonitrile. The fractions from each column were diluted 1:200 and tested for the ability to inhibit [ 3 H]thymidine incorporation into BAE cells.
FIG. 5. Extraction of ECI 2 from SDS-polyacrylamide gels. A, fractions from the peak of inhibitory activity shown in Fig. 4B (fractions 11-13) were pooled, concentrated, and analyzed by 15% SDS-PAGE as in Fig. 3A. B, after SDS-PAGE, the polyacrylamide gel was sliced into 1.5-2-mm pieces extracted in phosphate-buffered saline and assayed for BAE cell inhibitory activity as in Fig. 3B.   FIG. 6. Purification of ECI 3. A, fractions 69 -74 from the TSK gel SP-5PW column shown in Fig. 1 were applied to a C 4 RPLC column followed by elution with a gradient of 20 -40% acetonitrile. B, fractions 19 and 20 from the C 4 column in panel A were applied to a second C 4 RPLC column followed by elution with a gradient of 25-35% 2-propanol. The fractions from each column were diluted 1:200 and tested for the ability to inhibit [ 3 H]thymidine incorporation into BAE cells. C, fractions from the peak of inhibitory activity shown in Fig. 6B (fractions  11-14) were pooled, concentrated, and analyzed by 15% SDS-PAGE as in Fig. 3A. tors, and by microsequencing analysis, have identified them to be OSM, LIF, and TGF-␤1. These factors inhibit basal EC proliferation in the 0.1-1 ng/ml range. OSM is particularly potent with a half-maximal inhibition of EC growth at 150 -200 pg/ml (5-6.6 pM), making it one of the more potent peptide EC inhibitors reported to date. By comparison, other known EC growth inhibitors such as TNF-␣, angiostatin, cartilage-derived inhibitor, platelet factor 4, and thrombospondin are halfmaximally effective at 3 ng/ml (4), 300 ng/ml (10), 700 ng/ml (7), 1 g/ml (5), and 1.5 g/ml (9), respectively. The three macrophage-derived EC growth inhibitors inhibit growth of small and large vessel ECs, and besides inhibiting basal growth, they antagonize VEGF-and FGF-2-mediated EC growth as well.
Identification of TGF-␤1 and LIF as EC growth inhibitors is not novel. TGF-␤1, a 25-kDa dimer, has been shown to be an inhibitor of both small and large vessel EC proliferation in a number of studies consistent with the general inhibitory effects of TGF-␤1 on numerous cell types (32)(33)(34). The mechanism of TGF-␤1 inhibition is thought to occur at the level of cyclins, possibly by blocking Cyclin-dependent kinase 4 (CDK4) and inhibiting cyclin E expression (35). LIF is a 50 -58-kDa glycoprotein originally identified as an inhibitor of murine myeloid leukemia cells (36) and subsequently found to have diverse and apparently unrelated biological effects, including the ability to induce bone absorption (37), to stimulate the synthesis of acute stage plasma proteins from hepatocytes (38), and to inhibit embryonic stem-cell growth (39). LIF has been previously reported to be an inhibitor of large blood vessel EC proliferation but (unlike in our studies) not an inhibitor of small vessel EC proliferation (8). This difference is not understood since BAE cells and adrenal cortex-derived capillary ECs were used in both studies. On the other hand, the identification of OSM as an EC growth inhibitor appears to be novel. OSM is a 30-kDa protein identified in the CM of U-937 cells that was first identified as an inhibitor of A375 melanoma proliferation (40). In contrast, OSM is a mitogen for smooth muscle cells (41) (as is another EC inhibitor, TGF-␤1 (42)) and a mitogen for AIDS-Kaposi's sarcoma cells (43,44). Interestingly, while OSM and LIF are EC growth inhibitors, IL-6 and CNTF are not, even though these four cytokines are structurally related and work through a similar receptor system. Significant similarities have been identified in the general exon organization, primary amino acid sequences, and predicted secondary structure of OSM, LIF, CNTF, and IL-6, suggesting that they have a common evolutionary origin (45). These structural similarities might explain why they share many properties in common such as induction of acute phase protein synthesis, stimulation of myeloma growth, and differentiation of neuronal and leukemic cells (46). However, there are differences among these cytokines in receptor recognition. They have a common receptor subunit, gp130, that mediates cytokine signal transduction but that binds the cytokines either not at all, or weakly (47)(48)(49). An LIF-binding protein has been identified, which in combination with gp130 acts as a high affinity LIF receptor (47). The high affinity LIF receptor also binds OSM but not IL-6 or CNTF. There are also specific IL-6 and CNTF receptors acting in concert with gp130 (46,50). Thus one explanation for the EC growth inhibitory activities of OSM and LIF, but not IL-6 and CNTF, is that ECs express the high affinity LIF receptor complex but not the IL-6 or CNTF receptor complexes. A specific OSM receptor that does not bind LIF has been reported (51), which might explain why OSM is a more potent inhibitor than LIF. Another possibility that might explain cytokine specificity is that IL-6 family members have different signal transduction pathways. For example, they induce overlapping but distinct patterns of tyrosine phosphorylation (52,53). An investigation of receptor expression patterns might explain not only differences in cytokine specificity but why these cytokines can act as both inhibitors and mitogens depending on the cell type.
Preliminary experiments indicate that in the chorioallantoic membrane assay (54) OSM (up to 5 g) is not an angiogenesis inhibitor, consistent with previous results failing to show that LIF is an angiogenesis inhibitor in the same chorioallantoic membrane system (8). A mouse cornea angiogenesis assay (55) was also attempted, but rhOSM did not inhibit neovascular formation. However, the relationship between EC inhibition and angiogenesis is unclear. Some EC inhibitors in vitro are angiogenesis inhibitors in vivo, e.g. thrombospondin (9). On the other hand, TGF-␤1, which is a potent EC inhibitor in vitro, is an angiogenesis stimulator in vivo probably acting in an indirect fashion (12). Whether OSM is an angiogenesis inhibitor or not might require a more biologically relevant assay such as analyzing its effects on wound healing, a process mediated by macrophages.
In summary, macrophage-like cells release at least three EC growth inhibitors, OSM, LIF, and TGF-␤1, as well as the potent EC mitogen and angiogenesis factor, VEGF. Thus we speculate that the balance of EC mitogen/EC inhibitor levels might be an important modulator of EC proliferation in response to macrophage infiltration. Interestingly, in advanced atherosclerotic lesions, where macrophages heavily infiltrate the vessel wall, arterial endothelium is damaged and fails to regenerate (28). It is tempting to speculate that OSM and LIF derived from macrophages or other cell types may play a role in mediating vessel endothelium stasis after damage. Acknowledgment-We thank Dr. Shigeki Higashiyama for important help on this project.

TABLE III
Inhibition of basal, VEGF-, and FGF-2-stimulated EC proliferation BAE were plated at 6 ϫ 10 2 cells/well (96-well plates) in DMEM, 1% FCS, and after 1 day, they were incubated with no further addition (basal) or were treated with either VEGF 165 (5 ng/ml) or FGF-2 (1 ng/ml), concentrations corresponding to their ED 50 values. The inhibitors were added concurrently, and after 2 days, [ 3 H[thymidine incorporation into DNA was measured as previously described (31