L Totani, A Piccoli, G Pellegrini, A Di Santo and R Lorenzetendothelial cells

Polymorphonuclear leukocytes enhance release of growth factors by cultured

ISSN: 1524-4636 Copyright © 1994 American Heart Association. All rights reserved. Print ISSN: 1079-5642. Online

7272 Greenville Avenue, Dallas, TX 72514Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association.

1994, 14:125-132Arterioscler Thromb Vasc Biol 

http://atvb.ahajournals.org/content/14/1/125located on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://www.lww.com/reprintsReprints: Information about reprints can be found online at   journalpermissions@lww.com410-528-8550. E-mail:

Fax:Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050. Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters  http://atvb.ahajournals.org//subscriptions/Biology is online at Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

125

Polymorphonuclear Leukocytes Enhance Releaseof Growth Factors by Cultured Endothelial Cells

L. Totani, A. Piccoli, G. Pellegrini, A. Di Santo, R. Lorenzet

Abstract Porcine aortic endothelial cells (PAECs) in cul-ture constitutivery secrete polypeptide (endothelium-derived)growth factors (EDGFs) into the surrounding medium. Incu-bation of PAECs with human peripheral blood polymorpho-nuclear leukocytes (PMNs) caused a significant increase inEDGF release as assessed by pH]thymidine incorporation intoBALB/c 3T3 mouse fibroblasts and cell proliferation assay.The effect was time dependent and correlated with the num-ber of PMNs, reaching a maximum with a 1:1 PAEC to PMNratio. Generation of mitogenic activity was prevented bycycloheximide, indicating a requirement for de novo proteinsynthesis. Antibody-mediated inhibition assays suggested thatmitogenic activity was due to platelet-derived growth factorand basic fibroblast growth factor. When supernatant fromN-formyl-methionyl-leucyl-phenylalanine-stimulated PMNswas substituted for PMNs during incubation with PAECs,

powerful mitogenic activity was generated, indicating theinvolvement of soluble mediators. A role for free oxygenradicals was ruled out by experiments in which superoxidedismutase and catalase did not prevent the increase in mito-genic activity. By contrast, serine protease inhibitors such assoybean trypsin inhibitor, a^antitrypsin, and eglin C reducedthe PMN-stimulating activity by 70%, 80%, and 100%, respec-tively. The possible involvement of cathepsin G and elastasewas investigated. Cathepsin G and elastase, when substitutedfor PMNs, increased the release of EDGFs in a dose-depen-dent fashion, mimicking the effect of PMNs. These findingssuggest a new role for leukocyte-vessel wall interactions in theproliferative feature of atherosclerosis. (Arterioscler Thromb.1994;14:125-132.)

Key Wools • growth factors • endothelial cells • poty-morphonuclear leukocytes • serine proteases

Smooth muscle cell (SMC) proliferation accountsfor the intimal hyperplasia of the arterial wall,which is the main feature of fibrofatty athero-

sclerotic lesions. An accelerated form of this prolifera-tive process appears to play a central role in therestenosis that occurs after angioplasty1 and in prema-ture artery disease in transplanted organs.2 Release ofgrowth-promoting factors by endothelial cells, macro-phages, and activated platelets at the site of vascularinjury leads to migration and proliferation of SMCs thatmay ultimately contribute to the intimal thickeningobserved in vascular stenosis.3 The importance of mito-gens released by endothelial cells during intimal forma-tion has recently been emphasized in an aortic organmodel system by Koo and Gotlieb.4 These authors haveshown that the mean number of SMCs present in theintima of the porcine aorta is correlated with thepresence of a nondenuded endothelium or its condi-tioned medium.

Endothelial cells in vitro synthesize polypeptidegrowth factors, collectively named endothelium-derivedgrowth factors (EDGFs), including basic fibroblastgrowth factor (bFGF)5-7 and a platelet-derived growthfactor (PDGF)-like molecule.5-8 Enhanced secretion ofEDGFs has been found after stimulation of culturedendothelial cells with agonists such as thrombin,9 factorXa,10 low-molecular-weight fibrinogen degradationproducts,11 bacterial endotoxin,12 and tumor necrosisfactor.13

Received January 19, 1993; revision accepted October 17, 1993.From the "Antonio Taticchi" Unit for Atherosclerosis and

Thrombosis Research, Istituto di Ricerche Farmacologiche MarioNegri, Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy.

Reprint requests to Lida Totani, MD, Consorzio Mario NegriSud, Via Nazionale, 66030 S Maria Imbaro, Italy.

Electron microscopy and immunohistochemical tech-niques have shown that in addition to SMCs, infiltratingcells such as monocytes/macrophages, T lymphocytes,and granulocytes are constituents of atheroscleroticlesions.1416 Epidemiological studies have shown a directcorrelation between the number of polymorphonuclearleukocytes (PMNs) and the incidence of ischemic dis-ease.17'18 PMNs may contribute to vascular occlusion,damaging endothelial cells by secretion of several prod-ucts, such as arachidonic acid metabolites, toxic oxygenspecies, and proteolytic enzymes such as cathepsin Gand elastase.19 PMNs may "switch" a healthy endothe-lium to a thrombogenic one by increasing endothelialcell plasminogen activator inhibitor activity.20 More-over, cathepsin G has been shown to stimulate plasmin-ogen activator inhibitor release from human umbilicalvein endothelial cells21 and to increase albumin fluxacross the endothelial monolayer.22 In addition, cathep-sin G as well as elastase suppresses thrombin-inducedprostacyclin production in human endothelial cells.23

Among the various mechanisms by which PMNsmight contribute to vascular occlusion, in the presentstudy we investigated whether PMNs could modulatemitogen release by endothelial cells. The results indi-cate that PMNs enhance the release of EDGFs fromendothelial cells. Specific antibody-mediated inhibitionassays reveal that both PDGF and bFGF are producedand secreted after PMN challenge. Our data also sug-gest that the serine proteases cathepsin G and elastasemay be responsible for the increased mitogenic activity.

MethodsChemicals

Dulbecco's modified Eagle medium (DMEM), calf serum(CS), penicillin, streptomycin, glutamine, and 0.05% trypsin/

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

126 Arteriosclerosis and Thrombosis Vol 14, No 1 January 1994

0.02% EDTA were purchased from GIBCO BRL, GrandIsland, NY. Fetal calf serum (FCS) was from Biochrom,Berlin, FRG. JV-2-hydroxyethylpiperazine-W-2-ethanesulfo-nic acid (HEPES) was obtained from Merck, Darmstadt,FRG. Dextran T-500 and Ficoll-Hypaque were from Pharma-cia Fine Chemicals, Uppsala, Sweden. Cytochalasin B,Af-formyl-methionyl-leucyl-phenylalanine (fMLP), soybeantrypsin inhibitor (SBTI; 1 mg protein inhibits 3 to 5 mg trypsinwith an activity of approximately 10 000 BAEE U/mg protein),a,-antitrypsin (a,AT; 2 to 4 mg inhibit 1.0 mg trypsin with anactivity of 10 000 BAEE U/mg protein), superoxide dismutase(SOD; activity, 2500 to 5000 U/mg protein), catalase (activity,2000 to 5000 U/mg protein), serum bovine albumin (BSA), andcycloheximide were purchased from Sigma Chemical Co, StLouis, Mo. Eglin C, cathepsin G, and elastase from humanneutrophils were from Calbiochem Biochemicals, San Diego,Calif). Goat anti-human PDGF and monoclonal anti-bovinebFGF antibodies were purchased from British Bio-technologyProducts Ltd, Oxdon, UK, and UBI, Lake Placid, NY, respec-tively. [Me</ry/-3H]thymidine (['HJTdR), 2 Ci/mmol, was pur-chased from Du Pont de Nemours, Biotechnology SystemsDivision, Bad Homburg, FRG. Tissue-culture dishes werefrom Falcon Labware Division, Becton Dickinson Co, Oxnard,Calif.

Cell CultureEndothelial cells were obtained from porcine aortas as

previously described.24 For experiments, porcine aortic endo-thelial cells (PAECs) were plated in six-well plates at aconcentration of 8x10*/well and grown to confluency in ahumidified atmosphere of 93% air/7% CO2 at 37°C in DMEMsupplemented with 10% FCS, 100 U/mL penicillin, 100 Mg/mLstreptomycin, and 1% glutamine. The number of PAECs atconfluency was 5 x 105/well.

PMNs and PMN-Derived Supernatant PreparationsHuman blood was collected from healthy donors who had

not taken any drugs during the 2 weeks preceding the study,with citrate as the anticoagulant. PMNs were isolated bydextran sedimentation and density-gradient centrifugationwith Ficoll-Hypaque according to B0yum.25 The cells wereresuspended in 15 mmol/L HEPES/DMEM at the desiredconcentration. In some instances PMNs were stimulated byfMLP in the presence of cytochalasin B at 37°C. After 15minutes the cells were sedimented by centrifugation (5 min-utes at 12 000 rpm), and the supernatant was collected andextensively dialyzed against 15 mmol/L HEPES/DMEM toremove cytochalasin B before use in the release studies.

Release StudiesConfluent PAECs were washed with DMEM containing 2

mg/mL BSA followed by three washings with 15 mmol/LHEPES/DMEM at different intervals over a period of 24hours. PAECs, PMNs, or PMN supernatants were incubatedalone or together in 15 mmol/L HEPES/DMEM in a finalvolume of 1 mL at 37°C in 7% CO2 over a period of 15 hoursunless otherwise specified. When SBTI (100 Mg/mL), a, AT (1mg/mL), eglin C (100 Mg/mL), SOD (50 Mg/mL) plus catalase(50 Mg/mL), or cycloheximide (1 Mg/mL) was used, it wasadded at the beginning of incubation. At the desired time theconditioned medium was collected, centrifuged for 5 minutesat 12 000 rpm to remove cellular debris, and stored at -20°Cuntil the mitogenic assay. When cycloheximide was used theconditioned medium was extensively dyalized against 15mmol/L HEPES/DMEM before the mitogenic assay.

Evaluation of Cell DamageCell number was estimated by hematocytometer counting,

and viability was assessed by the trypan blue exclusion test. Toassess cell damage, lactate dehydrogenase released into the

endothelial cell-conditioned medium was measured with acolorimetric method (Sigma).

0-Glucuronidase AssayTo quantitate the release of azurophilic granules by PMNs,

we measured the activity of ^-glucuronidase in the condi-tioned medium of PMNs incubated alone or in the presence ofPAECs for various periods from 15 minutes to 15 hours./J-Glucuronidase activity was measured according to Talalay etal.26 This method is based on the generation of phenolphtha-lein from a phenolphthalein-/S-glucuronic acid complex by£-glucuronidase. The optical density of free phenolphthaleinis measured at 540 nm and is proportional to the enzymeactivity. Total 0-glucuronidase activity of 1O6 PMNs wasdetermined by lysing the cells with Triton X-100. Ten percentof the total enzyme activity generated by this number of PMNswas the detection limit of the assay. When the same number ofPMNs was stimulated by fMLP (10"* mol/L) in the presence ofcytochalasin B (2.5 Mg/mL), 40% of the total /3-glucuronidasewas detected in the supernatant. In contrast no release couldbe detected from unstimulated PMNs. 0-GIucuronidase activ-ity was also undetectable in the conditioned medium ofPAECs coincubated with PMNs for 15 hours.

Mitogenic Activity AssaysThe mitogenic activity of the endothelial cell-conditioned

medium was evaluated by using BALB/c 3T3 mouse cell fibro-blasts (clone A31, American Type Culture Collection) as de-scribed.24 The 3T3 cells were plated at a concentration of 104/wellin 96-well microtiter plates, grown to near confluency, andinduced to quiescence by a 48-hour exposure to DMEM supple-mented with 5% human platelet poor plasma-derived serumprepared as described.27 Aliquots of the endothelial cell-condi-tioned medium (5, 10, 20, 40, or 80 ML) were added to the 3T3cell cultures and incubated in 7.5% CO2 for 15 hours. [3H]TdRwas added, and after incubation for 3 hours the cells were washedand fixed to the well bottom with 10% cold trichloroacetic acid.The trichloroacetic acid-precipitable material was dissolved in0.5 mol/L NaOH, collected, and counted in a beta counter. Foreach experiment a standard curve was generated by incubating3T3 cells with increasing concentrations of CS. One unit ofmitogenic activity was arbitrarily assigned as half the maximumactivity obtained by CS.

In selected experiments the mitogenic activity was assessedby a cell proliferation assay. BALB/c 3T3 cells were seeded at2x10*/well in 12-well plates in DMEM supplemented with10% CS. After 24 hours the cells were washed and induced toquiescence by a 48-hour exposure to DMEM supplementedwith 5% human platelet poor plasma-derived serum. Aliquotsof conditioned medium from PAECs and/or PMNs wereadded to the wells and incubated for 24,48, or 72 hours. At theend of the incubation the cells were detached by exposure totrypsin/EDTA, and their number was estimated by hematocy-tometer counting.

Immunoinhibition StudiesThe postculture medium from PAEC/PMN coincubation

was preincubated with goat anti-human PDGF or monoclonalanti-bovine bFGF neutralizing antibodies for 1 hour at 37°Cbefore the mitogenic assay. The antibodies were used atconcentrations sufficient to inhibit 10 ng/mL PDGF and 20ng/mL bFGF, respectively. These antibodies have been shownto neutralize the mitogenic activity of porcine PDGF andbFGF, respectively. Additional samples of postculture me-dium were preincubated with nonimmune goat immunoglob-ulin G or nonimmune mouse immunoglobulin G as describedin the figure legends.

All of the reagents used in this study were dissolved in sterilesolvents and filtered through nonpyrogenic, sterile 0.22-MmMillex filters (Millipore, Molsheim, France). To avoid endo-toxin contamination, sterile, pyrogen-free working conditions

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

Totani et al PMNs and Endothelial Cell Growth Factors 127

X

E3B

6O-1

50-

40-

30-

20-

10-

0 -

Ba•

PAEC + PMNPAECPMN J*

/

/ .s^~ "

10 12 14 160 24 48 72

Incubation time (h) Incubation time (h)FIQ 1. Une plots showing effect of polymorphonuclear leukocytes (PMNs) on the release of mitogenic acttvtty from porcine aorticendotheiiat cells (PAECs). Corrfluerrt PAECs (SxK^/well) and 106 PMNs alone or together were incubated for different time intervals.A, Conditioned medium was collected at the indicated times, sedimented by centrtfugation, and tested for trrtiated thymidlneincorporation into BALB/c 3T3 as described. B, Conditioned medium was collected after 15 hours, sedimented by centrifugation, andtested in a cell proliferation assay on BALB/c 3T3 cells. The cells were detached after 24, 48, or 72 hours and counted as described.

were observed. All results represent the average of at leastthree separate experiments unless otherwise indicated.

ResultsEffects of PMNs on Mitogen ReleaseFrom PAECs

Cultured PAECs release growth factors into the me-dium in a time-dependent fashion that can be detectedafter a 15-hour incubation (Fig 1). By contrast no mito-genic activity is measurable when the PMN-conditionedmedium is tested. When PAECs were incubated withPMNs in a 1:2 ratio, powerful mitogenic activity wasgenerated as assessed by pHJTdR incorporation into 3T3cells. This activity became detectable between 1 and 3hours and increased steadily with time, reaching an eight-fold increase after 15 hours' incubation (Fig 1A). Todetermine whether mitogenesis was actually taking place,a cell proliferation assay was performed as described in"Methods." In a 72-hour incubation, cell counts rose to19±2xl04 (mean±SE, n=3) or 47±10xl04 (n=3) whenconditioned medium from PAECs or PAECs with PMNswas present, respectively (Fig IB). Conditioned mediumfrom PMNs was ineffective.

We next examined the effect of incubation of PAECswith increasing numbers of PMNs. Fig 2 illustrates themitogenic activity of medium from PAECs incubatedfor 15 hours with PMNs in a ratio of PAECs to PMNs of1:0.1, 1:0.2,1:0.5, 1:1, or 1:2. The effect of PMNs wasconcentration dependent and already detectable at aPAEC to PMN ratio of 1:0.1 to a maximum of 1:1(5xlOVmL of each cell type).

To determine whether PMNs were responsible forpostsecretion modification of an inactive form of growthfactor, endothelial cell-conditioned medium collectedfrom PAECs incubated alone for 15 hours was furtherincubated with 106 PMNs/mL for an additional 15-hourperiod. Addition of PMNs to the endothelial cell-conditioned medium did not modify the mitogenicactivity expressed by the endothelial cell-conditionedmedium (6.0±0.8 versus 5.8+1.0 U/mL, mean±SE,n=3), indicating that no postsecretion modification wastaking place.

PMN-induced EDGF production was totally blocked by1 /Ag/mL cycloheximide, indicating a requirement for de

novo protein synthesis (Fig 3). In pilot experiments thisconcentration of cycloheximide blocked protein synthesisby =95% and did not cause cell damage or loss.

Characterization of PDGF and bFGF-like Activityin Conditioned Media

To characterize the nature of the mitogenic activity,the specific neutralizing antibodies anti-PDGF and anti-bFGF were incubated with PAEC/PMN-conditionedmedium before the mitogenic assay. As shown in Fig 4,goat anti-PDGF (10 /xg/mL) reduced the activity by77% while a monoclonal anti-bFGF antibody (10 y.glmT.) caused an inhibition of 52% (no further inhibitionwas obtainable after increasing the concentration of theantibodies), indicating that both mitogens were in-volved. When a mixture of the two antibodies wasincubated with the postculture medium, the mitogenicactivity was completely abolished. Nonimmune goat ormouse immunoglobulin G failed to significantly inhibitmitogenic activity.

Effect of PMNs on PAEC Morphology and ViabilityThe classic "cobblestone" morphology of resting,

confluent PAECs is shown in Fig 5A. PMN incubation

0 0.05 0.1 0.25 0.5 .

PMN ( cell x 109 ml)

FIG 2. Bar graph showing effect of different numbers of poly-morphonudear leukocytes (PMNs) on the release of mitogenicactivity by porcine aortic endothelial cells (PAECs). ConfluentPAECs (5x10*/well) and PMNs were incubated for 15 hours.The endothelial cell-conditioned medium was collected andtested for mitogenic activity on 3T3 cells. Bars represent themean of three experiments ±SE.

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

128 Arteriosclerosis and Thrombosis Vol 14, No 1 January 1994

PAEC + * *

PMN - * •

Cycloheximide +

FIG 3. Bar graph showing effect of cycloheximide on polymor-phonuclear leukocyte (PMN)-induced endothelium-derivedgrowth factor release. Porcine aortic endothelial cells (5x10*/well) and 10° PMNs were incubated for 15 hours in the absenceor presence of cycloheximide (1 jig/mL). The endothelial cell-conditioned medium was collected, dyalized, and tested formitogenic activity on 3T3 cells. Bars represent the mean of threeexperiments±SE.

with PAECs resulted in close apposition of PMNs onthe endothelial cells. Repeated washings were sufficientto remove almost all of the PMNs. In some experimentsno visible damage of the monolayer could be observed.In others, PAECs underwent a shape change character-ized by cell retraction and formation of intercellularspaces between cells that had been adherent. Theseverity of the damage varied among experiments. In-tercellular spaces were sometimes barely visible (Fig5B); alternatively, large areas of the bottom of the wellwere exposed, as shown in Fig 5C. Cell detachmentcould not be detected in any experiment as determinedby cell count. PAEC viability by trypan blue exclusion

FIG 4. Bar graph of immunoinhibition of mitogenic activity byanti-platelet-derlved growth factor (PDGF) and anti-basic fibro-blast growth factor (bFGF). Postculture medium from porcineaortic endothellal cell-polyrnorphonuclear leukocyte cocultureswas preincubated for 1 hour with either 10 jig/mL anti-PDGF, 10jig/mL anti-bFGF (alone or in combination), 100 ng/mL nonim-mune goat immunoglobulin G (IgG), or 100 yuQlmV. nonimmunemouse IgG. The samples were then tested for mitogenic activityby tritiated thymidine incorporation Into BALB/c 3T3 cells. Re-sults are expressed in percent of mitogenic activity. A 100%value was assigned to the activity expressed by the mediumfrom porcine aortic endothelial cell-polymorphonuclear leuko-cytes coctilture (Control)- Bars represent the mean of fourexperiments±SE.

B

i

FIG 5. Phase-contrast photomicrographs of porcine aortic en-dothelial cells maintained In culture in the presence of polymor-phonudear leukocytes. The cells were Incubated for 15 hourswith serum-free Dulbecco's modified Eagle medium (A) or 10*potymorphonudear leukocytes per well (B and C).

after a 15-hour incubation in either the absence orpresence of PMNs was 97±1% and 96±2% (mean±SE,n=5), respectively. Sublytic damage, as assessed bymeasuring the release of the cytoplasmic enzyme lactatedehydrogenase, gave comparable values for PAECsincubated alone or with PMNs for 15 hours. The findingthat no cell detachment or sublytic damage could beobserved indicates that exposure of the subendothelialmatrix could be accounted for solely by cell retraction.

Effect of PMN Supernatants on EDGF ReleaseWe next investigated whether the generation of mi-

togenic activity was attributable to soluble mediators ordirect cell-cell contact. PMNs were incubated alone orwith fMLP in the presence of cytochalasin B. After 15minutes, the supernatants were cleared by centrifuga-tion, diah/zed, and incubated with PAECs for 15 hours.The supernatant from unstimulated PMNs failed tosignificantly increase the mitogenic activity expressed by

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

Totani et al PMNs and Endothelial Cell Growth Factors 129

300

FIG 6. Bar graph showing effect of potymorphonuclear leuko-cyte (PMN) supernatants on endothellum-derived growth factorrelease. Porcine aortic endothelial cells were incubated alone(control), with PMNs, or with PMNs or fMLP-stJmulated PMNsupernatants (sn) prepared as described. The conditioned me-dium was collected after 15 hours and tested for mitogenlcactivity. The activity Is expressed as mean±SE of three experi-ments. fMLP indicates N-formyl-methionyl-leucyl phenylalanine.

PAECs (Fig 6). By contrast the supernatant from fMLP-stimulated PMNs induced powerful activity that waseven greater than that generated by the intact system inwhich PAECs and PMNs were incubated together.These results indicate that soluble mediators are re-sponsible for the induced generation of EDGFs.

Effect of Oxygen-Radical Scavengers andAntiproteases on PMN-Induced EDGF Release

PMN h/sosomal granules contain several proteinasesthat can be released on PMN activation. Moreover,PMNs generate toxic oxygen species that can modulateseveral biological functions of the endothelium andaffect its integrity.28 The possible role of PMN-derivedproducts, such as neutral lysosomal proteases and toxicoxygen radicals, was then investigated. The serine pro-tease inhibitors SBTI (100 Mg/mL), dAT (1 mg/mL),and eglin C (100 jtg/mL) and the oxygen-radical scav-engers SOD (50 /xg/mL) and catalase (50 fig/mL) wereincubated with PAECs and PMNs. SBTI, a,AT, andeglin C but not SOD or catalase prevented the PMN-induced morphological change when this occurred. Formitogenic activity, the effect of these reagents areshown in Fig 7. The mitogenic activity expressed in theconditioned medium from PAECs incubated withPMNs in the absence of inhibitors was assigned a valueof 100%. SOD and catalase did not affect the increaseof EDGF release by PMN-stimulated PAECs. By con-trast, SBTI and a,AT decreased EDGF release by 70%and 80%, respectively. Moreover, addition of eglin C, aspecific inhibitor of elastase and cathepsin G,29 com-pletely prevented the increase of EDGF release in-duced by PMNs. In addition, in those cultures in whichPMNs induced a change in PAEC shape, eglin Cprevented this morphological modification. Addition ofthe inhibitors themselves to the PMN/PAEC-condi-tioned medium did not affect its mitogenic property,indicating that these inhibitors do not neutralize themitogenic activity once released (not shown). These

If

FIG 7. Bar graph showing effect of superoxide dismutase (SOD)and catalase (Cat) and different protease inhibitors on potymorpho-nuclear leukocyte (PMN)-medlated release of mitogenic activity byporcine aortic endothelial cells (PAECs). Confluent PAECs andPMNs in a 1:2 ratio were incubated with or without SOD (50/ig/mL)+Cat (50 /ig/mL), soybean trypsin inhibitor (SBTI), (100/ig/mL), a,-antitrypsin (a1-AT), (1 mg/mL), or eglin C (100 /ig/mL).The endothelial cell-conditioned medium was collected after 15hours and tested for mitogenic activity on 3T3 cells. Results areexpressed in percent of mitogenic activity. A 100% value wasassigned to the activity expressed by PAECs incubated with PMNsin the absence of the Inhibitors. Each column represents the meanof four experiments±SE.

results seem to indicate that serine proteases, such aselastase and/or cathepsin G, contribute to an increase inEDGF release.

Effect of Purified Cathepsin G and Elastase onEDGF Release

To further test this hypothesis, purified elastase orcathepsin G was added to confluent PAECs instead ofPMNs. In the presence of each separate enzyme, mito-genic activity was enhanced in a dose-dependent man-ner (Fig 8). The effect was already observed with aslittle as 10 nmol/L of either enzyme. At higher concen-trations (100 nmol/L for elastase and 25 nmol/L forcathepsin G), a plateau was reached. Moreover, at eachconcentration the enhancing effect of elastase was

100o CuhepdnO

A Eluase

• Cuheptta O (boDed)Pi«t—» (boiled)

0 20 40 60 80 100 120 140 160 180 200

Concentration (nM)

FIG 8. Line plot showing effect of purified cathepsin G andelastase on the release of mitogenic activity by porcine aorticendothelial cells (PAECs). Confluent PAECs were Incubated withpurified cathepsin G or elastase at the indicated concentrations.The endothelial cell-conditioned medium was collected after 15hours and tested for mitogenic activity on 3T3 cells.

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

130 Arteriosclerosis and Thrombosis Vol 14, No 1 January 1994

greater than that caused by cathepsin G. Cathepsin Gand elastase are not mitogenic per se. Addition of theenzymes directly to 3T3 cells did not influence [3H]TdRuptake (not shown).

To rule out the possibility that the increase in mito-genic activity was due to endotoxin contamination,cathepsin G and elastase were boiled for 15 minutesbefore incubation with PAECs. No increase in mitogen-ic activity was detected in this condition, indicating thatthe enhancing effect was solely attributable to cathepsinG and elastase (Fig 8).

PAEC morphology was also examined. PAECs incu-bated with cathepsin G were not different from controlcells. By contrast, elastase caused cell retraction thatwas detectable at 20 nmol/L and increased with higherconcentrations (not shown). However, even at highconcentrations (200 nmol/L), cell number and viabilitywere indistinguishable from those of control PAECs.

To exclude the possibility that modulation of EDGFrelease might result from species incompatibility, se-lected experiments were carried out to test humanPMNs on human umbilical vein endothelial cells. Hu-man PMNs consistently stimulated EDGF release fromboth PAECs and human umbilical vein endothelialcells. A complete report of these experiments will bepublished elsewhere.

DiscussionIn this article we report that human PMNs enhance

the release of EDGFs from PAECs. The effect canalready be observed after a 1- to 3-hour incubation andincreases with time. The amount of EDGFs releasedalso correlates with the number of PMNs and reaches amaximum at a PMN to PAEC ratio of 1:1. Although toa different degree, the same results were obtained whenhuman umbilical vein endothelial cells were substitutedfor PAECs, ruling out the possibility that species incom-patibility could be responsible for the increase in growthfactor release.

As opposed to the supernatant of unstimulatedPMNs, the supernatant from fMLP-stimulated PMNsper se was able to induce generation of mitogenicactivity. The activity generated by this supernatant wasgreater than that detected when the same number ofPMNs was incubated with PAECs.

These findings clearly assign a role for soluble medi-ators and suggest that exposure of PMNs to endothelialcells may be sufficient to induce at least a partial PMNlysosomal granule release. Measurement of the specificazurophilic granule marker /3-glucuronidase in the su-pernatant of PMNs, incubated with or without PAECs,gave negative results. However, partial granule releasebelow the detection limit of the assay cannot be ex-cluded. It should also be considered that PMN interac-tion with PAECs may create a stimulatory environmentfor PMNs. In fact, endothelial cells produce substancessuch as platelet activating factor30 or PDGF-like mole-cules, which have been shown to promote PMN activa-tion.31-32 Moreover, the involvement of adhesion pro-teins responsible for PMN interaction with endothelialcells cannot be excluded.

The proposed role for soluble mediators had led us tostudy the nature of the molecule(s) released fromPMNs. A role for toxic oxygen species has been ruledout in our experimental system. By contrast, the serine

protease inhibitors SBTI, a,AT, and eglin C reversedthe PMN-stimulating effect to various degrees. Themost effective was eglin C, a low-molecular-weightinhibitor of cathepsin G and elastase. Cathepsin G andelastase are present in large amounts in PMN azuro-philic granules33 and have been shown to modulateseveral endothelial functions.2123 Cathepsin G andelastase, incubated with PAECs but in the absence ofPMNs, increased the release of EDGFs in a dose-dependent fashion, mimicking the effect of PMNs.Elastase at a concentration of 10 nmol/L, induced anincrease in mitogenic activity that was comparable tothat obtained when 106 PMNs were incubated withPAECs. According to previous observations,34 thiselastase concentration accounts for 20% of the totalcontent of the enzyme in 106 PMNs. Altogether, theseresults indicate that either cathepsin G, elastase, orboth could be responsible for the increase in EDGFrelease.

It could be argued that under more physiologicalconditions, PMNs and the endothelium only interact inplasma, which is rich in scavengers and antiproteasesthat would inhibit cathepsin G and elastase that arepossibly released by PMNs. However, it should bementioned that close contact between PMNs and othercell types would create a protected microenvironmentin which cathepsin G and elastase could not be reachedby the enzyme inhibitors present in plasma.35-36

The morphology of PAEC monolayers after chal-lenge with PMNs was different among the experimentsand was related to considerable donor-to-donor vari-ability. In some experiments PAECs retained theirnormal cobblestone morphology; in others the cellsunderwent a change in shape, with consistent enlarge-ment of intercellular spaces. However, in both cases,cell number, viability, and lactate dehydrogenase re-lease were similar to those of PAECs incubated alone.Although injury to cultured endothelial cells causesrelease of growth factors into the surrounding medi-um,37 it has also been shown that mitogenic activity oflysates from endothelial cells accounts for only a smallamount of the activity found in endothelial cell-condi-tioned medium.11-27 Endothelial cells synthesize PDGF,which has been shown to be secreted into the surround-ing medium,5-8'24 and bFGF, which has been reported toremain cell associated5 or deposited in the extracellularmatrix.38 In the present study the enhancement inEDGF release was measured both when cells retractedand when cells retained the normal cobblestone mor-phology. Moreover, although elastase affected the mor-phology of the monolayer, cathepsin G at the concen-trations used caused an increase in mitogenic activitythat was not accompanied by visible cell retraction.Addition of PMNs to the postculture endothelial cell-conditioned medium failed to generate mitogenic activ-ity, ruling out the possibility of the involvement of aninactive form of growth factors. In addition, the activitywas completely suppressed by cycloheximide. Thesefindings support the hypothesis that the mitogenic ac-tivity detected in the endothelial cell-conditioned me-dium of PAECs exposed to PMNs results in de novoprotein synthesis and that the EDGFs are not derivedfrom an internal storage pool or from the extracellularmatrix, as has been reported by Ishai-Michaeli et al.39

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

Totani et al PMNs and Endothelial Cell Growth Factors 131

Modulation of EDGF synthesis and secretion hasbeen previously reported. Several agents increase thePDGF-like mitogen from endotheliaJ cells,7-11 whereasenhanced release of bFGF or a bFGF-like molecule hasbeen documented from irradiated cells27 or PAECs thathave been exposed to low-molecular-weight fibrinogendegradation products.11 Our data show that PMNsinduce production and secretion of at least two peptidegrowth factors, which are either very similar or identicalto PDGF and bFGF. This finding is supported by dataobtained by antibody-mediated inhibition of the mito-genic activity. Although this activity is completely sup-pressed when postculture medium is incubated with amixture of the anti-PDGF and anti-bFGF antibodies,the possibility that other substances such as interleu-kin-1, insulin growth factors, or transforming growthfactors may have contributed to the activity cannot becompletely dismissed. Several studies support the hy-pothesis that mitogens are involved in intimal thicken-ing, a feature characteristic of atherosclerotic plaques.Extracts from human atherosclerotic lesions containhigh levels of transcription of the sis gene (whichencodes for the B chain of PDGF40), and endothelialcells have been reported to be one of the predominantcell types that express PDGF chain mRNA in athero-sclerotic plaques.41

Epidemiological studies have linked PMN counts tothe risk of ischemic vascular disease.17-42 A positivecorrelation between leukocyte count and the severity ofcoronary artery disease as evaluated by coronary angi-ography has been described.43 Evidence for leukocytosisin diet-induced atherosclerosis in rabbits and humanswith familial hypercholesterolemia has also been report-ed.44-45 Moreover, the presence of neutrophils in diet-induced aortic fatty streaks in green monkeys has beendocumented by Trillo.16 SMC hyperplasia is consideredto be an important feature of the atherosclerotic lesion.In experimental models of neointimal formation thatwere induced by perivascular manipulation to avoiddirect injury of endothelial cells, adherence of PMNs tothe endothelium and subsequent invasion into the inti-mal and medial layers of the vessel wall was observed asan early event preceding intimal SMC proliferation.4*"48

PMN adhesion to the endothelium represents an earlyevent in the acute inflammatory response and patho-genesis of vascular disease. Cellular components of theatherosclerotic lesion synthesize cytokines, which areknown to induce adhesion molecules for leukocytes.49 Inaddition, immunohistochemical studies have shown thepresence of the intercellular adhesion molecule-1 onendothelial cells of human atherosclerotic plaques.50

Taken together, these observations suggest that a PMN/endothelial cell interaction may occur during the devel-opment of atherosclerosis. Although few studies reportthe presence of neutrophils in atherosclerotic lesions, itcannot be excluded that under certain circumstances, adynamic, short-lived interaction may occur betweenneutrophils and endothelium-generating conditions inwhich endothelial functions could be drasticallymodified.

In the present study we have shown that PMNsenhance the secretion of growth factors that are immu-nologically related to PDGF and bFGF from endothe-lial cells. This effect seems to be related to proteasessuch as elastase and cathepsin G, which may be released

on incubation of PMNs with PAECs. The results, de-scribed for the first time in this article, suggest a newrole for PMNs in the progression of vascular occlusion,since they offer evidence for a possible pathway of aPMN-induced switch of the endothelium toward an"atherogenic phenotype."

AcknowledgmentsThis study was supported in part by the Italian National

Research Council (CNR, Rome, Italy), Progetto FinalizzatoFATMA contract No. 92.00227.41, and Convenzione CNR-Consorzio Mario Negri Sud. Dr Totani was the recipient of afellowship from the Centro di Fonnazione e Studi per ilMezzogiorao -FORMEZ- (Progetto Speciale "Ricerca Scien-tifica e Applicata nel Mezzogjorno"). The authors wish tothank Dr Virgilio Evangelista for fruitful advice and discus-sions, and Silvia Falcone, Raffaella Bcrtazzi, Maria Pia DeSimone, and the G A Pfeiffer Memorial Library staff for theirhelp in the preparation of the manuscript.

References1. Liu MW, Berk BC. Restenosis following coronary balloon angio-

plasty: role of smooth muscle cell proliferation. Trends CardiavascMed 1991;l:107-lll.

2. Salomon RN, Hughes CCW, Schoen FJ, Payne DD, Pober JS,Libby P. Human coronary transplantation-associated arterio-sclerosis: evidence for a chronic immune reaction to activated graftendothelial cells. Am J PathoL 1991;138:791-798.

3. Ross R. The pathogenesis of atherosclerosis: a perspective for the1990s. Nature. 1993^62:801-808.

4. Koo EWY, Gotlieb AI. Endothelial stimulation of intimal cellproliferation in a porcine aortic organ culture. Am J PathoL 1989;134:497-503.

5. VTodavsky I, Fridman R, Sullivan R, Sasse J, Klagsbrun M. Aorticendothelial cells synthesize basic fibroblast growth factor whichremains cell associated and platelet-derived growth factor-likeprotein which is secreted. / Cell PhysioL 1987;131:402-408.

6. Schweigerer L, Neufeld G, Friedman J, Abraham JA, Fiddes JC,Gospodarowicz D. Capillary endothelial cells express basicfibroblast growth factor, a mitogen that promotes their owngrowth. Nature. 1987^325:257-259.

7. Sato Y, Rifkin DB. Autocrine activities of basic fibroblast growthfactor, regulation of endothelial cell movement, plasminogenactivator synthesis, and DNA synthesis. / Cell BioL 1988;107:1199-1205.

8. Di Corleto PE, Bowen-Pope DF. Cultured endothelial cellsproduce a platelet-derived growth factor-like protein. Proc NatlAcadSciU S A. 1983;80:1919-1923.

9. Harlan JM, Thompson PJ, Ross RR, Bowen-Pope DF. a-Thrombininduces release of platelet-derived growth factor-like molecule(s) bycultured human endothelial cells. J Cell BioL 1986;103:1129-1133.

10. Gajdusek C, Carbon S, Ross R, Nawroth P, Stern D. Activation ofcoagulation releases endothelial cell mitogens. J Cell BioL 1986;103:419-428.

11. Lorenzet R, Sobel JH, Bini A, Witte LD. Low molecular weightfibrinogen degradation products stimulate the release of growthfactors from endothelial cells. Thromb HaanosL 1992:68:357-363.

12. Fox PL, DiCorleto PE. Regulation of production of a platelet-derived growth factor-like protein by cultured bovine aortic endo-thelial cells. / Cell PhysioL 1984;121:298-308.

13. Hajjar KA, Hajjar DP, SuVerstein RL, Nachman RL. Tumornecrosis factor-mediated release of platelet-derived growth factorfrom cultured endothelial cells. J Exp Med. 1987;166:235-245.

14. Libby P, Hansson GK. Biology of disease: involvement of theimmune system in human atherogenesis: current knowledge andunanswered questions. Lab Invest 1991;64:5-15.

15. Masuda J, Ross R. Atherogenesis during low level hypercholester-olemia in the nonhuman primate: I. fatty streak formation. Arte-riosclerosis. 1990-,10:164-177.

16. Trillo AA. The cell population of aortic fatty streaks in Africangreen monkeys with special reference to granulocytic cells: anultrastructural study. Atherosclerosis. 1982;43:259-275.

17. Friedman GD, Klatsky AL, Siegelaub AB. The leukocyte count asa predictor of myocardial infarction. N Engl J Med. 1974^90:1275-1278.

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from

132 Arteriosclerosis and Thrombosis Vol 14, No 1 January 1994

18. Ernst E, Hammerschmidt DE, Bagge U, Matrai A, Dormandy JA.Leukocytes and the risk of ischemk diseases. JAMA. 1987;257:2318-2324.

19. Harlan JM. Neutrophil-mediated vascular injury. Acta Med ScandSuppL 1987;715:123-129.

20. Gilboa N, Neumann PH, Del Vecchio PJ, Gudewicz PW. Porymor-phonuclear leukocytes regulate the plasminogen-activating systemof cultured endothelial cells. Biochem Int 1988;17:585-592.

21. Pintucci G, Iacoviello L, Castelli MP, Amore C, Evangelista V,Cerletti C, Donati MB. Cathepsin G-induced release of PAI-1 inthe culture medium of endothelial cells: a new thrombogenic rolefor polymorphonuclear leukocytes? J Lab Clin Med. 1993;122:69-79.

22. Peterson MW, Gruenhaupt D, Shasby DM. Neutrophil cathepsinG increases calcium flux and inositol poh/phosphate production incultured endothelial cells. / Immunol 1989;143:609-616.

23. Weksler BB, Jaffe EA, Brower MS, Cole OF. Human leukocytecathepsin G and elastase specifically suppress thrombin-inducedprostacyclin production in human endothelial cells. Blood. 1989;74:1627-1634.

24. Witte LD, Coraicelli JA, Miller RW, Goodman DS. Effects ofplatelet-derived and endothelial cell-derived growth factors on thetow density lipoprotein receptor pathway in cultured human fibro-blasts. J Biol Chan. 1982057:5392-5401.

25. B0yum A. Isolation of mononuclear cells and granulocytes fromhuman blood: isolation of monuclear cells by one centrifugation,and of granulocytes by combining centrifugation and sedimen-tation at 1 g. Scand J Clin Lab Invest 1968;21(suppl 97):77-89.

26. Talalay P, Fishman WH, Huggins C. An improved assay for0-glucuronidase. / Biol Chan. 1946;166:757-773.

27. Witte L, Fuks Z, Haimovitz-Friedman A, Vlodavsky I, GoodmanDS, Eldor A. Effects of irradiation on the release of growth factorsfrom cultured bovine, porcine, and human endothelial cells. CancerRes. 1989;49-.5066-5072.

28. Sacks T, Moldow CF, Craddock PR, Bowers TK, Jacobs HS.Oxygen radicals mediate endothelial cell damage by complement-stimulated granulocytes: an in vitro model of immune vasculardamage. / Clin Invest 1978;61(suppl 5):1161-1167.

29. Seemuller U, Fritz H, Eulitz M. Eglin: elastase-cathepsin Ginhibitor from leeches. Methods EnzymoL 1981;80:804-817.

30. Whatley RE, Zimmermann GA, Mclntyre TM, Prescott SM.Endothelium from diverse vascular sources synthesizes platelet-activating factor. Arteriosclerosis. 1988;8:321-331.

31. O'Flaherty JT, Miller CH, Lewis JC, Wykle RL, Bass DA, McCallCE, Waite M, DeChatelet LR. Neutrophil responses to platelet-activating factor. Inflammation. 1981^:193-201.

32. Tzeng DY, Deuel TF, Huang JS, Senior RM, Boxer LA, BaehnerRL. Platelet-derived growth factor promotes polymorphonuclearleukocyte activation. Blood 1984;64:1123-1128.

33. Plow EF. The major fibrinorytic proteases of human leucocytes.Biochim Biophys Acta. 1980;630:47-56.

34. Travis J, Baugh R, Giles PJ, Johnson D, Bowen J, Reilry CF.Human leukocyte elastase and cathepsin G: isolation, character-ization and interaction with plasma proteinase inhibitors. In:Havemann K, Janoff A, eds. Neutral Proteases of Human Polymor-phonuclear Lekocytes. Baltimore, Md; Urban & Schwarzenberg:1978:118.

35. Weitz JI, Huang AJ, Landman SL, Nicholson SC, SOverstein SCElastase-mediated fibrinogenolysis by chemoattractant-stimulatedneutrophils occurs in the presence of physiologic concentrations ofantiproteinases. J Exp Med 1987;166:1836-1850.

36. Evangelista V, Rajtar G, de Gaetano G, White JG, Cerletti CPlatelet activation by fMLP-stimulated polymorphonuclear leu-kocytes: the activity of cathepsin G is not prevented by antipro-teinases. Blood 1991;77:2379-2388.

37. McNeil PL, Muthukrishnan L, Warder E, D'Amore PA. Growthfactors are released by mechanically wounded endothelial cells.JCeUBioL 1989;109:811-822.

38. Vtodavsky I, Folkman J, Sullivan R, Fridman R, Ishai-Michaeli R,Sasse J, Klagsbrun M. Endothelial cell-derived basic fibroblastgrowth factor synthesis and deposition into subendothelial extra-cellular matrix. Proc Natl Acad Sd U S A. 1987;84:2292-2296.

39. Ishai-Michaeli R, Eldor A, Vlodavsky I. Heparanase activityexpressed by platelets, neutrophils, and h/mphoma cells releasesactive fibroblast growth factor from extracellular matrix. Cell Reg-ulation. 1990;l:833-842.

40. Barrett TB, Benditt EP. sis (platelet-derived growth factor Bchain) gene transcript levels are elevated in human atheroscleroticlesions compared to normal artery. Proc Nad Acad Sd U S A.1987;84:1099-1103.

41. Wilcox JN, Smith KM, Williams LT, Schwartz SM, Gordon D.Platelet-derived growth factor mRNA detection in human athero-sclerotic plaques by in situ hybridization. J Clin Invest 1988;82:1134-1143.

42. Prentice RL, Szatrowski TP, Fujikura T, Kato H, Mason MW,Hamilton HH. Leukocyte counts and coronary heart disease in aJapanese cohort. Am J Epidemiol 1982;116:496-509.

43. Kostis JB, Turkevich D, Sharp J. Association between leukocytecount and the presence and extent of coronary atherosclerosis asdetermined by coronary arteriography. Am J CardioL 1984;53:997-999.

44. Feldman DL, Mogelesky TC, Liptak BF, Gerrity RG. Leuko-cytosis in rabbits with diet-induced atherosclerosis. ArteriosclerThromb. 1991;ll:985-994.

45. Dresel HA, Via DP, Stohr M, Elchner U, Gnasso A, Postiglione A,Blin N, Augustin J, Schertler G. Observations on leukocytes frompatients with severe familial hvpercholesterolemia.yHrterJoucferojij.1986;6:259-264.

46. Kling D, Holzschuh T, Strohschneider T, Betz E. Enhanced endo-thelial permeability and invasion of leukocytes into the artery wallas initial events in experimental arteriosclerosis. Int Angiol 1987;6:21-28.

47. Kockx MM, De Meyer GRY, Jacob WA, Bult H, Herman AG.Triphasic sequence of neointimal formation in the cuffed carotidartery of the rabbit. Arterioscler Thromb. 1992;12:1447-1457.

48. Prescott MF, McBride CK, Court M. Development of intimallesions after leukocyte migration into the vascular wall. Am JPathol 1989;135:835-846.

49. Mover CF, Sajuthi D, Tulli H, Williams JK. Synthesis of IL-1 alphaand IL-1 beta by arterial cells in atherosclerosis. Am J Pathol1991;138:951-960.

50. Poston RN, Haskard DO, Coucher JR, Gall NP, Johnson-TideyRR. Expression of intercellular adhesion molecule-1 in athero-sclerotic plaques. Am J Pathol 1992;14O:665-673.

by guest on July 13, 2011http://atvb.ahajournals.org/Downloaded from