Amnerica JournalX of Pctholog p, tiol. 150, No. 6, June 1997Coppight C) American SocietyJor Investigative Patbology

Sublytic Concentrations of the Membrane AttackComplex of Complement Induce EndothelialInterleukin-8 and Monocyte ChemoattractantProtein-1 through Nuclear Factor-KB Activation

Kenneth S. Kilgore,* Elisabeth Schmid,tThomas P. Shanley,* Craig M. Flory,tVipul Maheswari,* Nicole L. Tramontini,*Hillary Cohen,* Peter A. Ward,*Hans P. Friedl,t and Jeffrey S. Warren*From the Departnment ofPathology,* University ofMichiganMedical School, and Pharmaceuitical Research Division,tWarner Lambert-Parke Davis, Ann Arbor, Michigan, andthe Department of Surgery,t University Hospital,Zurich, Switzerland

Activation ofthe complement cascade and subse-quent assembly ofthe membrane attack complex(MAC) occur in a number ofpathophysiologicalsettings. Whenformed on the surface ofendothe-lial ceUls in sublytic concentrations, the MAC caninduce a number ofproinflammatory activities,including the secretion ofsoluble mediators (eg,interleukin (IL)-8 and monocyte chemoattractantprotein (MCP)-1) and the up-regulation of ceUlsurface adhesion molecules. Available data indi-cate that MAC-induced cell activation may occurthrough several complex signal transductionpathways, but little is known about the intranu-clear mechanisms by which complement-derivedproducts promote the up-regulation of inflam-matory mediators. Using purified distal comple-mentproteins (C5-9) to assemblefunctionalMACon early-passage human umbilical vein endothe-lial ceUs (HUVECs), we examined mechanisms ofMCP-1 and IL-8 induction. Formation of sublyticconcentrations ofMAC promoted an increase innuclear factor (NF)-edB DNA binding activitywithin 60 minutes as determined by serial elec-trophoretic mobility shift assay. Cytosolic to nu-clear translocation of NF-#cB was confirmed byWestern immunoblot and immunocytochemicalanalyses. Formation of the C5b-8 complex alsopromoted NF- #cB translocation but to a lesser de-

gree than observed in HUVECs containing com-plete MAC. No cytosolic to nuclear translocationof the p65 NF-#cB subunit was observed in un-stimulated HUVECs or in ceUls incubated with theMAC components devoid of C7. Preincubation ofHUVECs with pyrrolidine dithiocarbamate pre-vented MAC-induced increases in IL-8 and MCP-1mRNA concentrations and protein secretion. Adirect cause and effect linkage between MAC as-sembly and NF-KB activation was establishedthrough examination ofthe pharmacological ef-fect of the peptide SN50 on IL-8 and MCP-1 ex-pression. SN50 is a recently engineered 26-ami-no-acid peptide that contains a lipophilic cell-membrane-permeable motif and a nuclearlocalization sequence that specfi-cally competeswith the nuclear localization sequence of theNF-KBp5O subunit. This study provides direct invitro evidence that the distal complement system(MAC) can promote proinflammatory endothe-lial ceUl activation, specifically, increases in IL-8and MCP-1 mRNA concentrations andprotein se-cretion, and that cytosolic to nuclear transloca-tion of NF-idB is necessary for this response.(AmJ Pathol 1997, 150.2019-2031)

A large body of evidence suggests that both solubleand cell surface mediators coordinately regulate theevents that characterize the acute inflammatory re-sponse.1 Activation of the complement system canmodulate leukocyte and endothelial cell functionsthrough the actions of the anaphylatoxins, C3b and

Supported in part by the National Institutes of Health (HL 48287 and5T 32-HL 07517). K. S. Kilgore was supported, in part, by theAmerican Heart Association of Michigan (21F956 fellowship).Accepted for publication February 13, 1997.

Address reprint requests to Dr. Jeffrey S. Warren, Department ofPathology, Box 0602, University of Michigan Medical School, 1301Catherine Road, Ann Arbor, MI 48109-0602.

2019

2020 Kilgore et alA/P June 1997, Vol. 150, No. 6

its derivatives, and the heteropolymeric membraneattack complex (MAC) and related complexes (eg,C5b-7 and C5b-8).2'3 Assembly of the MAC on somecell surfaces can result in a number of proinflamma-tory actions. Nucleated cells exhibit membrane-as-sociated mechanisms that protect them from lysis bythe C5b-9 complex.34 Injury or death of nucleatedcells occur when there is a high density of poresformed by MAC, whereas lower densities of cell sur-face MAC may induce changes in cellular activity.3Assembly of MAC in sublytic concentrations canmodulate cell function in a variety of ways, includingthrough alteration of the expression of cellular adhe-sion molecules and through up-regulation of the se-cretion of proinflammatory mediators such as tumornecrosis factor (TNF)-a, interleukin (IL)-1, basic fi-broblast growth factor, and platelet-derived growthfactor.5-7 Using purified C5-C9, we recently reportedthat assembly of sublytic densities of functional MACon endothelial cells results in the up-regulation andsecretion of the neutrophil and monocyte chemotac-tic cytokines IL-8 and monocyte chemoattractantprotein (MCP)-1, respectively.8 Furthermore, Torzew-ski et al9 have reported that MCP-1 is also releasedby smooth muscle cells, suggesting that the proin-flammatory actions of the MAC are not limited toendothelial cells.The mechanisms by which assembly of the MAC

promotes increased expression of these proinflam-matory mediators have yet to be fully ascertained.Assembly of the MAC (and C5b-7 and C5b-8) hasbeen associated with a number of signal transduc-tion events. For example, MAC formation on endo-thelial cell membranes results in a rapid increase inintracellular cyclic AMP concentration, suggestingthat the MAC may interact directly with pertussis-toxin-sensitive G proteins.10 Intracellular calcium,derived from both intracellular and extracellularstores, increases after MAC deposition.11 A rise inintracellular calcium concentration allows cell activa-tion via Ca2"-sensitive proteins or through directactivation of various protein kinases.12 Finally, MACdeposition is associated with the appearance ofphosphatidylcholine breakdown products and theactivation of protein kinase C.12

Although it is apparent that the MAC may modu-late cell function through several signal transductionpathways, the mechanisms through which it medi-ates IL-8 and MCP-1 production by endothelial cellsremain to be established. A large body of evidencesuggests that nuclear factor (NF)-KB (and othermembers of the Rel family) plays a key role in theinflammatory response through its capacity to mod-ulate the production of proinflammatory cytokines

and the up-regulation of adhesion molecule expres-sion.13'14 Among genes that encode proinflamma-tory mediators and contain promoter elements thatare regulated by members of the NF-KB family oftranscription factors are E-selectin, intercellular ad-hesion molecule (ICAM)-1, vascular cell adhesionmolecule (VCAM)-i, IL-6, IL-8, and MCP-1.15-17NF-K B, a transcription factor associated with rapidactivation responses, is composed of two subunits,p50 and p65.18 Within inactive cells, the p50 andp65 subunits are bound within the cytoplasm to theinhibitor protein, IKB-a.1920 When the cell is appro-priately stimulated, IKB-a is phosphorylated, ubiquiti-nated, and proteolytically degraded, thus allowingthe liberation of NF-KB and its translocation from thecytosol to the nucleus, where it binds to specificpromoter elements and activates target genes.21We examined the role of NF-KB activation in MAC-

induced endothelial activation, specifically, the up-regulation of IL-8 and MCP-1 mRNA concentrationsand protein secretion. Assembly of sublytic concen-trations of functional MAC resulted in the transloca-tion of NF-KB from the cytosol to the nucleus. Theseresponses were inhibited in the presence of pyrroli-dine dithiocarbamate (PDTC) and SN50, which havepreviously been shown to inhibit NF-KB-dependentcell activation. Recent reports suggest that PDTCmay interrupt NF-KB-activated translocation by pre-venting inducible phosphorylation of IKB-a.22'23 Inhi-bition of IL-8 and MCP-1 production with SN50, a26-amino-acid peptide that contains a lipophilic cell-membrane-permeable motif and a nuclear localiza-tion sequence that specifically competes with thenuclear localization sequence of the NF-KB p50 sub-unit,24 strongly supports the conclusion that MACmay serve to activate proinflammatory mediatorgenes (IL-8 and MCP-1) through the activation andtranslocation of NF-KB. The data contained in thepresent study suggest that NF-KB translocation/acti-vation is necessary for MAC-induced IL-8 andMCP-1 secretion by human umbilical vein endothe-lial cells (HUVECs). It remains unclear whetherNF-KB activation/translocation is sufficient.

Materials and Methods

ReagentsPurified complement proteins (C5-C9) were pur-chased from Quidel (San Diego, CA). Lipopolysac-charide (LPS) concentrations in purified complementprotein stock solutions were assayed with the E-Toxate (Limulus amoebocyte lysate) assay (SigmaChemical Co., St. Louis, MO; catalog item 210-Al).

MAC-Mediated Nuclear Factor Translocation 2021AJP June 1997, Vol. 150, No. 6

Peptide-specific rabbit antiserum (product SC-372)to NF-KB p65 was purchased from Santa Cruz Bio-technology (Santa Cruz, CA). Additional chemicalsand reagents, unless otherwise indicated, were pur-chased from Sigma.

Endothelial Cell CultureHUVECs were isolated from umbilical veins by treat-ing with 0.1% collagenase in Dulbecco's modifiedEagle's medium (Whittaker Bioproducts, Walkers-ville, MD) as previously described.25 Cells weregrown and maintained in M199 medium (WhittakerBioproducts), supplemented with 20% heat-inacti-vated fetal calf serum, L-glutamine (4 mmol/L), pen-icillin (100 U/ml), streptomycin (100 ug/ml), endothe-lial growth supplement (25 ,ug/ml; CollaborativeResearch, Bedford, MA), and bovine lung heparin(15 U/ml). Cells were grown on gelatin-coated 150 x25 mm plates and allowed to grow to confluence at370 with 5% CO2 before harvesting for extract prep-aration. Cells were characterized as previously de-scribed and used between the first and third pas-sages.8

Non-Enzymatic Formation of C5b-LikeC5C6 Complex (C5C6*)Purified proteins C5 and C6 were suspended inserum-free HUVEC medium. Formation of the C5b-like activation product was accomplished by oxi-dation with chloramine-T as described previous-ly.26-28 Briefly, C5 (10 ,ug) was added to veronal-buffered saline (10 ,ul) (Sigma) and incubated inthe presence of 10 ,ul of 0.32 mmol/L chloramine-T(Sigma) in water for 10 minutes at room tempera-ture. At the end of the incubation period, methio-nine (1 mmol/L; 10 ,ul; Sigma) was added to themixture to inactivate the remaining chloramine-T.To form the modified C5b-like C5C6 complex(C5C6*), purified C6 (20 ,ug) in 300 ,ul of serum-free medium was incubated (24 hours at 37°C)with the chloramine-T-treated C5. The functional(lytic) and structural (reactivity with monoclonalantibody directed against MAC neoantigen) integ-rity of chloramine-T-generated MAC on HUVECswas verified as previously described.8

Assembly of the Membrane Attack ComplexHUVECs were plated in 150 x 25 mm dishes 24hours before use. Cells were then washed with se-rum-free medium (see above), and the MAC was

assembled for 30 minutes. Assembly of the MACwas initiated by a 15-minute preincubation (370C)with the modified C5C6* activation product (5 p,g/ml)and C7 (10 A.g/ml).8 Endothelial monolayers werethen washed with serum-free medium to remove ex-cess C5C6* activation product and C7. Complementcomponents C8 (10 ,ug/ml) and C9 (10 ,ug/ml) werethen added to the monolayers and allowed to incu-bate for an additional 30 minutes. Monolayers werethen washed with serum-free medium to remove ex-cess noncomplexed complement components andincubated with serum-free medium for the times in-dicated. The total volume per dish was 10 ml. Whereindicated, individual complement components wereheat inactivated (100°C for 30 minutes) or polymyxinB (5 to 50 ,ug/ml) was added to the medium.

Extraction of Cell Nuclei

Nuclear extraction was carried out as described pre-viously.29'30 Briefly, HUVECs (5 x 106 to 5 x 107cells) were harvested by scraping and centrifuging(200 x g for 5 minutes) in phosphate-buffered saline(PBS; 40C) containing a protease inhibitor cocktail(10 ,umol/L aprotinin, 1 ,umol/L leupeptin, 1 ,umol/Lbestatin, 0.5 mmol/L phenylmethylsulfonyl fluoride(PMSF); Sigma). Cells were prewashed in PBS (4°C;2 ml) and pelleted (200 x g for 5 minutes at 40C). Thesupernatant was removed and the pellet waswashed twice with 2 ml of ice-cold buffer A (10mmol/L HEPES, pH 7.9, 1.5 mmol/L MgCI2, 10mmol/L KCI, 0.5 mmol/L dithiothreitol (DTT), and 0.5mmol/L PMSF) and centrifuged as described above.The supernatant was aspirated and the pellet resus-pended in 80 p.l of buffer A plus 0.1% Nonidet P-40.The resulting suspension was then transferred to a1.7-ml Eppendorf tube and incubated on ice for 5minutes, followed by centrifugation at 450 x g for 12minutes at 4°C. The crude nuclear pellet was resus-pended in 60 p.l of buffer C (20 mmol/L HEPES, pH7.9, 0.42 mol/L NaCI, 1.5 mmol/L MgCI2, 0.2 mmol/LEDTA, 0.5 mmol/L DTT; 0.5 mmol/L PMSF, 25% glyc-erol (v/v)) and incubated on ice for 15 minutes. Nu-clei were sonicated twice at 15% power output (10seconds) and microcentrifuged at 14,000 rpm for 2minutes. The nuclear fraction was again centrifugedat 14,000 rpm for 12 minutes (40C) to collect thesupernatant containing the nuclear protein extracts.These extracts were subsequently diluted with 60 p.lof modified buffer D (20 mmol/L HEPES, pH 7.9, 0.05mol/L KCI, 0.2 mmol/L EDTA, 0.5 mmol/L DTT, 0.5mmol/L PMSF). Protein concentrations were deter-mined by Folin's assay with bovine serum albumin(BSA) as a standard (Sigma).

2022 Kilgore et alAJPJuine 1997, Vol. 150, No. 6

Electrophoretic Mobility Shift and SupershiftAssays

Electrophoretic mobility shift assays (EMSA) were

carried out using a gel shift assay system kit (catalogitem E-3050, Promega Corp., Madison, WI).31 Dou-ble-stranded NF-KB consensus oligonucleotideprobe (5'-AGTTGAGGGGACTTTCCCAGGC-3') was

end-labeled with [32P]ATP (3000 Ci/mmol at 10 mCi/ml; Amersham Life Science, Arlington Heights, IL).After T4 kinase end-labeling, oligonucleotide (35fmol; -1 x 105 dpm) probe and nuclear protein (5,g) were incubated for 30 minutes at room temper-ature in binding buffer (50 mmol/L Tris/HCI, pH 7.5,250 mmol/L NaCI, 2.5 mmol/L DTT, 20% glycerol(v/v)), and 0.5 ,ug of poly dl.dC (Pharmacia, Piscat-away, NJ). The reaction volumes were held constantat 10 ,lI. Where indicated, unlabeled competitiveoligonucleotide (NF-KB) or irrelevant oligonucleotide(AP1) was added 10 minutes before the addition ofradiolabeled probe.31 Samples were run on a non-

denaturing 4% polyacrylamide electrophoretic gel in0.25X TBE buffer (10 mmol/L Tris/HCI, pH 8.0, 1mmol/L EDTA) at 100 mV/15 mA for 4 hours. Gelswere vacuum dried and visualized by KodakX-OMAT film exposure to the gel at -70°C for 1 hour.For supershift assays, the reaction mixture was incu-bated with anti-p65 antibody (Santa Cruz Biotech-nology; sc-372) for 20 minutes at room temperature.The oligonucleotide probe was added and the incu-bation carried out as described above.

Western Blot Analysis

Nuclear and cytosolic extracts from stimulatedHUVECs were subjected to electrophoresis on

10% sodium dodecyl sulfate (SDS)-polyacryl-amide gels and transferred to nitrocellulose in 25mmol/L Tris, 192 mmol/L glycine, 5% methanol at100 V for 1.5 hours at 4°C as described previous-ly.32 Blots were analyzed for the p65 NF-KB sub-unit using primary antibody (Santa Cruz Biotech-nology) at a 1:1000 dilution. After incubation withthe primary antibody, blots were washed and in-cubated with a 1:500 dilution of peroxidase-conju-gated goat anti-rabbit secondary antibody (Bio-Rad, Hercules, CA). Western blot detection was

achieved using a colorimetric detection system(Bio-Rad).

Immunocytochemical StainingImmunocytochemical analysis was carried out to as-

sess subcellular NF-KB p65 subunit localization in

HUVECs activated with MAC.31 HUVECs were grownfor 24 hours in eight-well microchamber slides(Nunc, Naperville, IL). After assembly of the MAC,monolayers were fixed with 4% paraformaldehyde inPBS for 20 minutes at room temperature and then for6 minutes with methanol at -200C. Rabbit polyclonalantibody to the p65 subunit of NF-KB (Santa CruzBiotechnology) was layered onto slides for 1 hour atroom temperature and then incubated with a biotiny-lated goat anti-rabbit secondary antibody (1:1000dilution; Vector Laboratories, Burlingame, CA). De-tection of the primary antibody was accomplishedusing a Vectastain ABC kit (Vector Laboratories) with3-amino-9-ethyl-carbazole as the substrate. Controlsincluded chambers in which the primary antibodywas omitted and chambers in which nonspecific rab-bit immunoglobulin was substituted for the anti-p65antibody. To rule out the possibility that sublytic con-centrations of MAC induce artifactual association ofcytosolic NF-KB with nuclear extracts, we examinedthe localization of an irrelevant cytosolic protein, lac-tate dehydrogenase (LDH), in HUVECs exposed toMAC. These studies were conducted as describedabove but employed a monoclonal antibody directedagainst LDH (Accurate Chemical and ScientificCorp., Westbury, NY).

Effects ofPDTC and SN50 on MAC-Mediated MCP- 1 and IL-8 Secretion

Chemokine-specific enzyme-linked immunosorbentassay (ELISA) analyses were used to quantify IL-8and MCP-1 in conditioned media after stimulation bythe MAC in the presence or absence of either PDTCor SN50 at the indicated concentrations. The sand-wich ELISAs used in this study are modifications ofpreviously described methods that used goat poly-clonal and mouse monoclonal antibodies directedagainst each chemokine under investigation.8 Micro-titer plates (Nunc, Roskilde, Denmark) were coatedwith either anti-IL-8 or MCP-1 polyclonal antibody incoating buffer (0.05 mol/L H3B03, 0.120 mol/L NaCI,pH 8.6) overnight at 4°C (100 ,ul/well final volume).The wells were washed three times with washingbuffer (PBS, 0.02% Tween-20) and incubated for 60minutes at 370C with washing buffer containing 2%BSA. Samples of HUVEC culture supernatants (50,lI/well) were added to the wells and incubated for 90minutes at 370C. Standard curves for each chemo-kine were prepared by diluting recombinant humanIL-8 or MCP-1 (Genzyme, Cambridge, MA) in wash-ing buffer at concentrations ranging from 0.01 to1000 ng/ml. After repeated washings, a mouse

MAC-Mediated Nuclear Factor Translocation 2023AJPJune 1997, Vol. 150, No. 6

monoclonal anti-human antibody (Genzyme) wasadded to the wells and allowed to incubate at 370Cfor 60 minutes. The wells were washed and incu-bated with a peroxidase-conjugated rabbit anti-mouse IgG (1:1000 dilution; Dako Corp., Carpinteria,CA) for 1 hour at room temperature followed byaddition of the chromogenic substrate (o-phenylene-diamine hydrochloride) for 30 minutes. The reactionwas quenched with 3 mol/L H2SO4, and the opticaldensity was determined at 490 nm using an EL340automated microplate reader (Bio-Tek Instruments,Winooski, VT). The sensitivity limits for the ELISAswere 0.01 ng/ml for IL-8 and 0.1 ng/ml for MCP-1.

Northern Hybridization Analysis of IL-8 andMCP- 1 mRNA after PDTC Pretreatment ofMAC-Stimulated CellsNorthern hybridization was performed as describedpreviously.8 Briefly, RNA was isolated from HUVECmonolayers stimulated with the MAC or MAC-stimu-lated HUVECs that had been pretreated with PDTC(10 ,mol/L, 2 hours). Monolayers stimulated withLPS (10 ,tg/ml) were used as positive controls. Fivehours after assembly of the MAC, HUVECs wereprepared for Northern hybridization analysis. TotalRNA extraction was carried out using TRIzol reagent(Gibco, Gaithersburg, MD). After removing insolubledebris by centrifugation, the RNA was precipitatedand washed with isopropyl alcohol and ethanol. Afterresuspension of the pellet, the RNA concentration ofa 1:250 dilution was determined by measuring ab-sorbance at 260 nm.

Total RNA (12 ,ug) was electrophoretically sepa-rated in 1% agarose, 2.2 mol/L formaldehyde dena-turing gel, followed by capillary transfer to nylonmembrane (Zetabind, Cuno, Meridien, CT). Themembrane was dried for 2 hours in vacuo at 80°C,prehybridized for 5 to 6 hours at 650C in 10 ml of 7%SDS, 0.5 mol/L NaH2PO4 (pH 7.0), and 1 mmol/LEDTA, and 1% BSA. Approximately 1 x 107 to 2 x107 of 32P-labeled IL-8 or MCP-1 probe (see below)or ,3-actin cDNA (gift of Paul Killen, University ofMichigan Medical School) was mixed with 300 ,l ofa 10 mg/ml solution of salmon sperm DNA (0.1mg/ml final), boiled for 10 minutes, and placed on icebefore adding to the hybridization bag. The mem-branes were hybridized overnight at 650C, afterwhich they were washed once at room temperature,twice at 650C (10 minutes each wash) in 5% SDS, 10mmol/L NaH2PO4 (pH 7.0), 1 mmol/L EDTA, 0.5%BSA, and then twice in 1% SDS, 10 mmol/L

60 -

o*l 50-

V30-

m20-0

10-

0-

IL-8I I

MCP-lI , T I

T**

TT

Control MAC MAC + Control MAC MAC+PDTC PDTC

Figure 1. Inhibition ofMAC-mediated IL-8 and MCP-1 secretion fromHUVECs after pretreatment with PDTC ( 10 pumotlL for 2 hours). HU-VECs were exposed to MACfor 30 minutes and uashed, with mediumthen collected at 24 hours. 7Tese data represent the mean ± SEMfromone experiment (oftuwo) performed in quadruplicate. *P < 0.05versusresponse ofHUVECs exposed to MAC alone (one-way ANOVA).

NaH2PO4, 1 mmol/L EDTA at 65°C. The membraneswere then exposed to XAR-5 film (Kodak) at -700C.

The sequence of the IL-8 cDNA probe was 5'-GTT-GGC-GCA-GTG-TGG-TCC-ACT-CTC-AAT-CAC-3'. Theprobe sequence for the MCP-1 was 5'-TTG-GGT-TTG-CTT-GTC-CAG-GTG-GTC-CAT-GGA-3'. To confirmequal loading of the RNA, the amount of total RNA perlane was assessed by monitoring the amount of 13-actinmRNA.

Results

PDTC Pretreatment of MAC-StimulatedHUVECs Inhibits IL-8 and MCP- 1 SecretionWe have previously observed that sublytic concen-trations of MAC can induce IL-8 and MCP-1 secre-tion by HUVECs.' Conditioned medium from PDTC-treated HUVECs bearing sublytic concentrations ofMAC were examined by ELISA 24 hours after MACdeposition (Figure 1). Assembly of MAC on HUVECsresulted in IL-8 and MCP-1 secretion substantiallyabove that observed in unstimulated cells. Pretreat-ment of HUVECs with PDTC (10 ,umol/L for 2 hours)inhibited the MAC-induced secretion of both IL-8and MCP-1. Given the inhibitory effect of PDTC inNF-KB-mediated cell activation,33 these data sug-gest that the MAC-induced increase in IL-8 andMCP-1 requires functional NF-KB.

PDTC Pretreatment of HUVECs InhibitsMAC-Mediated Increases in IL-8 and MCP-1 mRNARNA was isolated from unstimulated HUVEC mono-layers, monolayers stimulated with the MAC, and

i

2024 Kilgore et alAJP.June 1997, Vol. 150, No. 6

a - rt

n% on an 1 on C) eS (A

MCP-1

IL-8

Figure 2. Northern blot anzalysis oflL-8 and MCP-1 mRNA after stim-ulation of HUVEGs by MAC. Cells ulere pretreated for 2 hoturs twitheither PDTC ( 10 uiinol L) or vehicle. Total cellular mRNA uwas isolated5 hoturs after stimtulation. Lanes 1 to 4, unstimulated control cells,LPS-stinmulated cells, cells stimulated with MAC, anzd cells pretreatedwith PDTC before MAC deposition, respectively. Eqtal loadinig ofRNAuas confirmed as assessed by densitometric comparisons of (3-actinnmRNA. Thereu'as less than 11%lane-to-lane variation in RNA loadingin the depictecd experimental result (data not shoun). The depicteddata represent a sinigle, representative blot from three experiments.

monolayers incubated with with PDTC (10 ,umol/L/ml,2 hours) before MAC deposition. Monolayers stimu-lated with LPS (10 ,uglml) were used as a positivecontrol. As illustrated in Figure 2, low concentrationsof IL-8 and MCP-1 mRNA were detected in unstimu-lated monolayers. Marked increases in IL-8 andMCP-1 mRNA were observed in monolayers stimu-lated by the MAC as compared with the amountsobserved in HUVEC monolayers pretreated withPDTC. These results indicate that MAC-mediatedincreases in IL-8 and MCP-1 mRNA are inhibited byPDTC and, in conjunction with the data above, sug-

gest a potential role for NF-KB (see below).

Temporal Analysis of MAC-Mediated NF-KBTranslocation in HUVECsTo directly determine whether MAC can induce cy-

tosolic to nuclear translocation of NF-KB and to as-

certain the temporal characteristics of MAC-inducedNF-KB translocation, EMSA analysis was carried outusing nuclear extracts from HUVECs exposed tosublytic concentrations of MAC for varied lengths oftime (Figure 3). No nuclear NF-KB activity was noted20 minutes after MAC deposition (lane 2). However,

Figure 3. Tenmporal characteristics of MlAC-mediated cytosolic to nu-

clear AlF-KB translocation. HUVECs were exposed to fiinctionally in-tact, sublytic conicenztrationts ofMAC(see Materials and Methods) for 0,20, 60, and 120 minutes as incdicated (lanes 1 to 4). Preparationsshoun in lanes 5 to 7 were froni HUVECs exposed to MAC for 120mintultes. NF-KB consenstus oligonuicleotide probe ( 35fmol, -I X 10'dpm) and nuclear protein ( 5 ,g) were incubated for 30 minutes inbinding buffer. Where indicated, znlabeled competitive oligontucleo-tide (NF-KB) or irrelevant oligontuzcleoticle (AP-1) was added 10 min-utes beforce the addition ofradiolabeledprobe (lanes 5 and 6). Sampleswere nun on a nondenaturing 4% polyacrylamide gel. No nuclearNF-KB activity tnas noted at 0 or^ 20 minutes after MAC fonnationi(lanes 1 and 2). Assembly of the inttact MAC uas associated with agraduial increase in nulclear NF-KB activity beginninzg within 60 mimi-uites (lane 3?) after activation of enclothelial cells. Maximal NF-KBactivity uas observed at 2 hours (lane 4) .NF-KB activity vas notfoundin samples incubated with 100-fold exce-s oJ'unlabeledprobe (lane 5)whereas the presence of the irrelevant oligon1ucleotide AP-1 probe didnot affect NF-KB detectioni (lane 6). Siupershift assays tuere coniductedin the presence of anti-p65 antibody, ol.1hich iniduiced a shiJ? in theband for NF-KB (lane 7). The depicted resullts represent one of threeseparate experiments.

deposition of the intact MAC was associated with agradual increase in nuclear NF-KB activity beginningwithin 60 minutes after activation of endothelial cells(lane 3). Maximal nuclear NF-KB activity was ob-served at 2 hours (lane 4). Incubation of sampleswith 100-fold excess of unlabeled probe inhibiteddetection of NF-KB (lane 5) whereas the presence ofthe irrelevant probe (AP-1) did not alter NF-KB de-tection (lane 6). Samples preincubated with anti-p65antibody induced a supershift in the NF-KB band(lane 7). These data indicate that assembly of sub-

2- I a

MAC-Mediated Nuclear Factor Translocation 2025A/PJune 1997, Vol. 150, No. 6

- 66kD- 46kD

Figure 4. Immnzsioblot analysis of MAC-induced cytosolic (C) to no-clear (N) tranislocation of NF-KB p65 unit. Endothelial cells u'erestimnlulated wvith intact MACfor3O minutes and alloued to incubate for2 hboirs at 37°C. NE KB translocation was tvisualized by addition qfapolyclonal antibody to the p65 subunit flloowed by a biotin-labeledsecondary antibodyf. Deposition of itntact MAC on HUVECs resulted inato increase inti nuclear p65 and a corresponding decrease of theprotein in the cytosolicfraction. Preventioni ofMACformation throuigbooliissiotn of C7 resulted in failure of cytosolic to nuclear tranislocationiof AF-KB p65. LPS-stimulated (10 pg/4nl) cells revealed pronounced(ytosolic to nuclear translocation of NF-KB p65. The depicted resultsrepresent onie offour separate experiments.

lytic concentrations of the MAC on HUVECs results inthe rapid translocation (1 hour) of NF-KB to cell nu-clei. The requirement for intact MAC in NF-KB trans-location is addressed below.

Western Blot Analysis ofp65 Translocation

Immunoblot analysis of nuclear and cytosolic ex-tracts from unstimulated HUVECs revealed the pres-ence of the NF-KB p65 subunit in the cytosolic frac-tion whereas little p65 was detected in the nuclearfraction (Figure 4). Assembly of the intact MAC onHUVECs resulted in a marked increase in nuclearp65 and a corresponding decrease in cytosolic p65.These data are indicative of p65 translocation. Itshould be noted that the translocation of p65 to thenucleus did not entirely deplete the cytosolic p65pool. Identical treatment of HUVECs with MAC com-ponents devoid of C7 resulted in failure of cytosolicto nuclear p65 translocation. As expected, cytosolicand nuclear fractions derived from LPS-stimulated(10 jig/ml) cells, used as a positive control, revealedpronounced cytosolic to nuclear translocation ofp65.

Intact MAC Is Required for NF-KB DNABindingEMSA analysis of nuclear extracts derived fromHUVECs stimulated with either intact (functional)MAC or identical concentrations of MAC compo-nents devoid of either C7 or C9 (ie, nonfunctional)is shown in Figure 5. Assembly of intact MAC (lane3) promoted a marked shift in the binding of the

I

1 2 3 4 5Figure 5. EMSA analysis of nuticlear extracts derived from HUVECsactivated with the intact MAC or MAC components in the absence qfeither C7 or C9. Unstimulated cells are noted in lane 1. The presenceof initact MAC (lane 3) promoted increased band shift of the NF-KBoligonuicleotide fragment as compared uith that notedfbr cells stiniu-lated with MAC components in the absence ofC9 (lane 5). Omission ofC7from the MAC (lane 4) abrogated the band shift. Stimulation ofHUVECsfor 1 houir with LPS ( 10 gl/ml) uwas used as a positive control(lane 2). These resuilts are representative ofthree separate experiments.

NF-KB oligonucleotide fragment band, similar tothat observed in cells incubated with equivalentMAC components but no C9 (lane 5). Exclusion ofC7 from the MAC (lane 4) resulted in a markedlyreduced band shift as compared with that seen inHUVECs exposed to the complete MAC or C5b-8(MAC components without C9). Stimulation of HU-VECs with media alone or 1 hour with LPS (10,tg/ml) were used as negative and positive con-trols, respectively (lanes 1 and 2). These resultssuggest that intact MAC is required for maximalNF-KB DNA binding and that C5b-8 may also in-duce NF-KB DNA binding.

2026 Kilgore et alAJP.June 1997, Vol. 150, No. 6

Figure 6. Immunocytochemical analysis ofMAC-induced translocation ofNF-KB p65from cytosol to nucleus. A: In unstimuilated HUJVECs, the p65suibzunit uas localized primarily in the cytoplasm. B: In contrast, deposition of intact MAC induced a marked increase in nuclear staining and acorresponding loss ofcytoplasmic NF-KBp65staining). C: Nuclearstaining uwas not observed in cells incubated with MAC components in the absenceof C7. D: LPS-stimulated cells ( 10 jig/ml) were uised as a positive controlfor NF-KB translocation.

Immunocytochemical Evidence that theIntact MAC Specifically InducesTranslocation of NF-KB p65 from Cytosol toNucleus

Immunocytochemical analysis of MAC-inducedcytoplasmic to nuclear translocation of NF-KB p65is depicted in Figure 6. Consistent with the resultsderived from Western blot and EMSA analysis, p65is detected primarily in the cytoplasm of unstimu-lated cells (Figure 6A). Little nuclear staining isobserved (Figure 6A). In contrast, assembly ofintact MAC induced a marked increase in theamount of nuclear staining and a corresponding

decrease in cytoplasmic staining (Figure 6B).These data confirm the ability of the MAC to pro-mote NF-KB translocation. Nuclear staining wasnot observed in cells incubated with equivalentconcentrations of MAC components minus C7(Figure 6C). LPS-stimulated cells (10 ,ug/ml) wereused as a positive control for NF-KB translocation(Figure 6D). These results confirm that depositionof intact MAC results in translocation of NF-KB fromthe cytosol to the nucleus of HUVECs. Assembly ofMAC on HUVECs did not induce a cytosolic tonuclear shift in LDH (Figure 7), suggesting thatMAC-induced NF-KB translocation is specific andnot an artifactual phenomenon (eg, osmotic effect).

I.=--

_*._s_

*_..q__

*tF:

_

I

MAC-Mediated Nuclear Factor Translocation 2027AJPJune 1997, Vol. 150, No. 6

Figure 7. Immunocytochemzical analysis of the cytosolic protein LDH after MAC deposition. LDH is localized to the cytoplasm in uinstiznulated cells(A) and MAC-stimulated cells (B). Nuclear localization ofLDH was not obsened in either of the treatment groups.

MAC-Stimulated HUVECs Produce IL-8 andMCP- 1 via NF-KB: SN50 and LPSSpecificity StudiesBecause of recent reports that suggest that PDTCmay interrupt NF-KB activation/translocation throughprevention of inducible phosphorylation of IKB-a (asa kinase inhibitor),20"21 we used SN5022 to morespecifically address the role of NF-KB in MAC-in-duced HUVEC activation. In addition, studies usingheat-inactivated complement components (C5-C9),MAC devoid of either C7 or C8, and intact MAC in thepresence of polymyxin B were carried to address thepotential role of endotoxin contamination in endothe-lial cell activation. As shown in Figure 8, preincuba-tion of HUVECs with SN50 resulted in concentration-dependent reductions in both IL-8 and MCP-1. Heatinactivation of the complement components used toform MAC (C5-C9) reduced IL-8 and MCP-1 pro-duced by HUVECs to concentrations produced byunstimulated cells.

Addition of polymyxin B to HUVECs exposed toMAC did not reduce IL-8 or MCP-1 production (com-pared with that observed in media from HUVECsexposed to MAC in the absence of polymyxin B). Theinhibitory effect of SN50 on MAC-induced IL-8 andMCP-1 production, in conjunction with the EMSA,immunocytochemical, and Western blot data pre-sented in the preceding sections, strongly supportthe conclusion that sublytic concentrations of MACinduce HUVECs through NF-KB-dependent mecha-nisms. The failure of heat-inactivated MAC compo-nents and MAC devoid of C7 or C8 to trigger IL-8and MCP-1 production argue against LPS contami-nation of MAC as the explanation for these findings.The latter conclusion is further supported by the lackof effect of polymyxin B on MAC-induced IL-8 andMCP-1 production by HUVECs. It should be noted

that direct measurements of endotoxin in C5-C9 andmedium alone, at the dilutions used in these studies,were all less than or equal to 0.027 EU/mI.

DiscussionAssembly of sublytic concentrations of intact MACcan modulate cell function in a variety of ways. Forexample, Hattori et al34 35 reported that activation ofcomplement in close proximity to the surface of en-dothelial cell monolayers results in the rapid up-regulation of P-selectin. The MAC has been ob-served to interact in a synergistic fashion with TNF-ato promote increased neutrophil adhesion to endo-thelial cells through the up-regulation of E-selectinand ICAM-1.28 Deposition of the MAC on endothelialcells is also associated with the up-regulation andsecretion of soluble proinflammatory mediators.Schonermark et al5 and Lovett et al6 reported thatthe MAC promotes the secretion of TNF-a and IL-1 ,3,respectively. We recently reported that assembly ofintact MAC on endothelial cells is associated with theup-regulation and secretion of the chemotactic cyto-kines IL-8 and MCP-1.8 However, the mechanismsby which assembly of the MAC promotes increasedexpression of selectins, ICAM-1 and various solubleproinflammatory mediators have yet to be fully elu-cidated.A large body of evidence suggests that the ex-

pression of various cytokines and leukocyte adhe-sion molecules is regulated through several signaltransduction pathways including activation of proteinkinase C (PKC), cholera-toxin-sensitive G-proteins,and the generation of oxygen-derived metabolites.12The MAC has been reported to interact directly andindirectly with a number of these second messengerpathways. For example, complete (C5b-9) or partial

2028 Kilgore et alAJPJune 1997, Vol. 150, No. 6

60

50 -

1-1

E 40-

430-

X 20-

10 -

n-1:U -

125 -

I- 100 -

$ 75-

0050-

25 -

n)-

A

..............................

............

.............

BT

.. ...............

T ..

1. ~~~~~~T....,,..,.,,.,,~~~~~...... ...... ..........

...... ...... .. .. .. .....

E~~~~0,0., X tz;XB

Figure 8. IL-8 anld MCP-1 secretion from HUVECs incubated with

eitherMA'AC, MAC devoid ofC7or C8, heat-inactivated MAC(f 1O00Cfor30 minutes), SN5O (added 15 minutes before MAC assembly), or

polymyxin B. HECs were exposed to MAC(nr indicated treatment)for 30 minutes and washed, with medium then collected at 24 hours.

These data represent the means + SEMfrom one experment (of two)

performed in quadruplicate.

(C5b-8) formation of the MAC is associated with

increased intracellular levels of sn-ib 2-diacylglyceroland activation of a variety of protein kinases includ-

ing PKC and protein kinaseA.11m12 Niculescu eta210

have observed thththe MAC may interact directly

with pertussis-toxin-sensitive guanine nucleotide

binding proteins (G-proteins) in a receptor-indepen-

dent manner. Although not a specific signal trans-

duction pathway, oxygen-derived metabolites are

generated after MAC deposition.36 Activation of

these second messenger systems may afford an

explanation as to the capacity of the MAC to promote

nuclear translocation of NF-KB. The signal transduc-tion pathways that interact with the MAC, includingactivation of G-proteins and PKC, share in the ability

to mediate the activation of NF-KB.37 38 Interaction of

the MAC with multiple signal transduction pathwaysthat correlate with those that affect NF-KB suggests a

possible link between MAC formation and the sub-

sequent up-regulation of proinflammatory mediatorsvia NF-KBa . Although the expression of IL-S and

MCP-1 may be mediated via activation of signaltransduction pathways, the possibility remains that

the expression of these mediators is a direct conse-quence of insertion of the C5b-9 pore complex intothe cell membrane. Acosta et al39 have recently re-ported that the MAC pore allows for the release of theproinflammatory mediators IL-1 and basic fibroblastgrowth factor that may, in an autocrine or paracrinefashion, form a positive feedback system to promotefurther endothelial cell activation.

In the present study, using EMSA, Western blotanalysis, and immunocytochemical analysis, we ob-served that assembly of intact MAC on HUVECsresults in the translocation of NF-KB p65 from thecytosol to the nucleus. Initial NF-KB translocationwas noted 60 minutes after MAC deposition with themaximal degree of NF-KB activity noted at 120 min-utes. In addition, we observed that the maximalMAC-induced nuclear translocation requires thepresence of all of the proteins that constitute MAC(C5-C9) or at least C5b through C8. This conclusionis based on the observation that MAC proteins de-void of C7, but not C9, do not promote translocationof the heterodimers to the nucleus to the degreeobserved after assembly of intact MAC. To confirmthat the translocation of NF-KB to the nucleus isspecific and not a random or artifactual event, theintracellular localization of the cytosolic protein LDHwas determined by immunocytochemistry. LDH is acytosolic protein that is not associated with nucleartranslocation. In the present study, LDH was notdetected in association with nuclei after MAC depo-sition, as determined by immunohistochemistry, in-dicating that the MAC does not promote the nonspe-cific movement of cytosolic proteins (eg, LDH) fromthe cytosolic compartment to the nuclear compart-ment. These data suggest that the nuclear translo-cation of p65 is a specific intracellular signalingevent and not simply due to MAC-mediated loss ofnuclear membrane integrity.We previously reported that deposition of the MAC

on HUVECs promotes the expression of the chemo-tactic cytokines IL-8 and MCP-1, both of which areregulated by the NF-KB family of transcriptional reg-ulators.8 15-17 To address a potential direct linkagebetween MAC deposition and NF-KB activation, weexamined the effect of the NF-KB inhibitors PDTCand SN50 on the production of MCP-1 and IL-8.PDTC has been previously reported to selectivelyinhibit NF-KB activity, presumably by functioning inan antioxidant capacity.33 Pretreatment of HUVECswith PDTC (10 ,umol/L for 2 hours) resulted in signif-icant decreases in the concentration of both IL-8 andMCP-1 as demonstrated by ELISA and Northern blotanalysis. More recent studies suggest that PDTCmay interrupt NF-KB activation/translocation by pre-

_A _

v -I cin

......

MAC-Mediated Nuclear Factor Translocation 2029AJPJune 1997, Vol. 150, No. 6

venting inducible phosphorylation of IKB-a.22'23Therefore, to more directly address the cause andeffect linkage between sublytic MAC assembly andNF-KB activation/translocation, we employed SN50,which specifically inhibits entry of NF-KB into cellnuclei24. The failure of heat-inactivated MAC pro-teins, in aggregate, and MAC devoid of C7 or C8, inconjunction with the lack of effect of polymyxin B onMAC-induced IL-8 and MCP-1 production, arguestrongly against LPS contamination of MAC compo-nents as an explanation for our conclusions.

In addition to defined interactions with secondmessenger systems, it has also been reported thatthe MAC can elicit the production of oxygen metab-olites including 2- and H202.36 Numerous studiesindicate that reactive oxygen intermediates influenceintracellular redox status, which in turn modulatescell activation and proinflammatory cytokine geneexpression.'"4 Studies that have addressed lym-phocyte proliferation and activation of latent viralgenomes (eg, HIV-1-infected T lymphocytes) sug-gest that intracellular redox potential is important inthe modulation of cell activation.3132 Changes incytosolic redox potential have been paralleled bychanges in NF-KB binding and correlate inverselywith intracellular thiol content. Increased intracellularthiol content is thought to reflect increased intracel-lular glutathione concentrations. Studies by Israel eta141 and Staal et a142 suggest that cell activatorssuch as TNF-a and phorbol esters can efficientlyexert their effects only when target cells are in ap-propriate redox status. This is converse to the inter-pretation that TNF-a, phorbol ester, and other medi-ators lead to NF-KB activation through changes incellular redox status per se. In the case of someinducers of cell activation (phorbol ester), NF-KBappears to be activated via a PKC pathway, whereasin the cases of other activators (eg, TNF-a and IL-1,3), PKC is not involved.43 It appears from thesestudies that an intracellular redox equilibrium tend-ing toward oxidation is required for full activation oftransduction pathways regulating the activity of NF-KB-dependent genes. Studies by Schreck et al33,44indicate that H202, and other reactive oxygen inter-mediates that are produced in the microenvironmentof an inflammatory process, can induce the expres-sion and replication of HIV-1 in human T-lymphocytecell lines. Again, this effect is mediated by the NF-KBtranscription factor and can be abrogated by eithernoncytotoxic antioxidants such as N-acetyl cysteineor compounds that block NF-KB activation in intactcells (eg, dithiocarbamates and metal chelatorssuch as desferrioxime).33'45

The premise that cellular redox status and NF-KBare potentially important in inflammation is furthersupported by recent in vitro studies that indicate thatthe expression of some inflammatory cytokines isregulated by intracellular redox status and NF-KB.4648 Alterations in ambient oxygen tension affectIL-1 production in mononuclear cells46 and IL-8 ex-pression in endothelial cells.48 TNF-a production inphorbol-ester-stimulated U937 cells can be partiallyblocked by the antioxidant butylated hydroxyani-sole.47 The ability of the MAC to promote the gener-ation of reactive oxygen intermediates supports thepremise that the MAC may be an important mediatorof proinflammatory gene expression via NF-KB acti-vation. Regulation of NF-KB by the MAC affords apotential mechanism by which complement-derivedproducts potentiate the inflammatory response. Forexample, we have recently observed that MAC dep-osition promotes the up-regulation and secretion ofIL-8 and MCP-1.8 Alterations in ambient oxygen ten-sion affect IL-8 expression in endothelial cells, fibro-blasts, pulmonary epithelial cells, and hepatocytesafter direct exposure of the cells to H202.48 Further-more, in vitro studies employing isolated human mes-angial cells have shown that either exogenous orendogenous (intracellular) oxidants can induceMCP-1 expression.47 The present study offers a po-tential mechanism by which the up-regulation ofthese chemokines may be influenced by the MAC.

The ability of the MAC to promote NF-KB translo-cation represents a potential mechanism by whichthe MAC participates in the induction and mainte-nance of the inflammatory response. In that NF-KB isinvolved in the regulation and expression of adhe-sion molecules and cytokines, it is likely that thisregulatory mechanism is intimately involved in medi-ating the secretion of these proinflammatory media-tors after MAC deposition. Consequently, interac-tions between the MAC and multiple signaltransduction pathways and regulators of gene ex-pression may play an important role in amplificationand maintenance of the inflammatory response.

References

1. Carlos TM, Harlan JM: Leukocyte-endothelial adhesionmolecules. Blood 1994, 84:2068-2085

2. Hugli RE: The chemistry and biology of C3a, C4a andC5a and their effects on cell. Biological Response Me-diators and Modulators. Edited by JT August. NewYork, Academic Press, 1983, pp 99-116

3. Nicholson-Weller A, Halperin JA: Membrane signalingby complement C5b-9, the membrane attack complex.Immunol Res 1993, 12:244-257

2030 Kilgore et alA/PJune 1997, Vol. 150, No. 6

4. Asghar SS: Biology of disease: membrane regulators ofcomplement activation and their aberrant expression indisease. Lab Invest 1995, 72:254-271

5. Schonermark M, Deppisch R, Riedasch G, Rother GK,Hansch GM: Induction of mediator release from humanglomerular mesangial cells by the terminal complementcomponents C5b-9. Int Arch Allergy Appl Immunol1991, 96:331-337

6. Lovett D, Hansch GM, Resch K, Gemsa D: Activation ofglomerular mesangial cells by the terminal membraneattack complex of complement. J Immunol 1987, 138:2473-2479

7. Benzaquen LR, Nicholson-Weller A, Halperin JA: Ter-minal complement proteins C5b-9 release basic fibro-blast growth factor and platelet-derived growth factorfrom endothelial cells. J Exp Med 1994, 179:985-992

8. Kilgore KS, Flory CM, Miller BF, Evans VM, Warren JS:The membrane attack complex of complement inducesinterleukin-8 and monocyte chemoattractant protein-1secretion from human umbilical vein endothelial cells.Am J Pathol 1996, 149:953-961

9. Torzewski J, Oldroyd R, Lachmann P, Fitzsimmons C,Proudfoot D, Bowyer D: Complement-induced releaseof monocyte chemotactic protein-1 from human smoothmuscle cells. Arterioscler Thromb Vasc Biol 1996, 16:673-677

10. Niculescu F, Rus H, Shin ML: Receptor-independentactivation of guanine nucleotide-binding regulatoryproteins by terminal complement complexes. J BiolChem 1994, 269:4417-4423

11. Carney DF, Lang TJ, Shin ML: Multiple signal messen-gers generated by terminal complement complexesand their role in terminal complement complex elimina-tion. J Immunol 1990, 45:623-629

12. Wiedmer T, Ando B, Sims PJ. Complement C5b-9-stimulated platelet secretion is associated with a Ca2+_initiated activation of cellular protein kinases. J BiolChem 1987, 262:13674-13681

13. Collins T, Read MA, Neish AS, Whitley MZ, Thanos D,Maniatis T: Transcriptional regulation of endothelial celladhesion molecules: NF-KB and cytokine-inducible en-hancers. FASEB J 1995, 9:899-909

14. Collins T: Biology of disease: endothelial nuclear fac-tor-KB and the initiation of the atherosclerotic lesion.Lab Invest 1993, 68:499-508

15. Hou J, Baichwal V, Cao Z: Regulatory elements andtranscription factors controlling basal and cytokine-in-duced expression of the gene encoding ICAM-1. ProcNatl Acad Sci USA 1994, 91:1641-11645

16. Jahnke A, Johnson JP: Synergistic activation of inter-cellular adhesion molecule-1 (ICAM-1) by TNF-a andIFN-y is mediatied by p65/p5O and p65/c-Rel and in-terferon-responsive factor Stat 1-a (p91) that can beactivated by both IFN-y and INF-a. FEBS Lett 1994,354:220-226

17. Parry GC, Mackman N: A set of inducible genesexpressed by activated human monocytic and endo-thelial cells contain KB-like sites that specifically bind

c-Rel-p65 heterodimers. J Biol Chem 1994, 269:20823-20825

18. Pober JS, Cotran RC: Cytokines and endothelial cellbiology. Physiol Rev 1990, 70:427-451

19. Baeuerle PA: The inducible transcription activator NF-KB: regulation by distinct protein subunits. BiochimBiophys Acta 1991, 107:63-80

20. Lenard MJ, Baltimore D. NF-KB: a pleiotropic mediatorof inducible and tissue-specific gene control. Cell1988, 7:318-332

21. Zabel U, Baeuerle PA: Purified human IKB can rapidlydissociate the complex of the NF-KB transcription fac-tor with its cognate DNA. Cell 1990, 61:255-265

22. Finco TS, Beg AA, Baldwin AS: Inducible phosphory-lation of IKBa is not sufficient for its dissociation fromNF-KB and is inhibited by protease inhibitors. Proc NatIAcad Sci USA 1994, 91:11884-11888

23. Miyamoto S, Maki M, Schmitt MJ, Hatanaka M, VermaIM: Tumor necrosis factor a-induced phosphorylationof IKBa is a signal for its degradation but not dissoci-ation from NF-KB. Proc Natl Acad Sci USA 1994, 91:12740-12744

24. Lin Y-Z, Yao SY, Veach RA, Torgerson TR, Hawiger J:Inhibition of nuclear translocation of transcription factorNF-KB by a synthetic peptide containing a cell mem-brane-permeable motif and nuclear localization se-quence. J Biol Chem 1995, 270:14255-14258

25. Jaffe EA: Culture and identification of large vessel en-dothelial cells. Biology of Endothelial Cells. Edited byEA Jaffe. The Hague, Martinus Nijoff, p 13

26. Vogt W, Hesse D: Activation of the fifth component ofhuman complement by oxygen-derived free radicals,and by methionine oxidizing agents: a comparison.Immunobiology 1992, 184:384-391

27. Vogt W, Zimmerman B, Hesse D, Nolte R: Activation ofthe fifth component of human complement, C5, withoutcleavage, by methionine oxidizing agents. Mol Immu-nol 1992, 29:251-256

28. Kilgore KS, Shen JP, Miller BF, Ward PA, Warren JS:Enhancement by the complement membrane attackcomplex of tumor necrosis factor-a induced endothe-lial cell expression of E-selectin and ICAM-1. J Immu-nol 1995, 155:1434-1441

29. Kravchenko VV, Pan, Z, Han J, Herbert JM, UlevitchRJ, Ye RD: Platelet-activating factor induces NF-KBactivation through a G protein-coupled pathway. J BiolChem 1995, 270:14928-14934

30. Menon SD, Qin S, Guy GR, Tan YH: Differential induc-tion of nuclear NF-KB by protein phosphatase inhibitorsin primary and transformed human cells. J Biol Chem1993, 268:26806-26812

31. Molitor JA, Walker WH, Doerre S, Ballard DW, GreeneWC. NF-KB: a family of inducible and differentially ex-pressed enhancer-binding proteins in human T cells.Proc Natl Acad Sci USA 1990, 87:10028-10032

32. Osborn K, Kunkel S, Nabel GJ: Tumor necrosis factorand interleukin 1 stimulate the human immunodefi-

MAC-Mediated Nuclear Factor Translocation 2031AJPJune 1997, Vol. 150, No. 6

ciency virus enhancer by activation of nuclear KB. ProcNatl Acad Sci USA 1989, 86:2336-2340

33. Schreck R, Meier B, Mannel DN, Drogew W, BaeuerlePA: Dithiocarbamates as potent inhibitors of nuclearfactor KB activation in intact cells. J Exp Med 1992175:1181-1194

34. Hattori R, Hamilton KK, Fugate RD, McEver RP, SimsPJ: Stimulated secretion of endothelial von Willebrandfactor is accompanied by rapid redistribution to the cellsurface of the intracellular granule membrane proteinGMP-140. J Biol Chem 1989, 264:7768-7771

35. Hattori R, Hamilton, KK, McEver RP, Sims PJ: Comple-ment proteins C5b-9 induce secretion of high molecu-lar weight multimers of endothelial von Willebrand fac-tor and translocation of granule membrane proteinGMP-140 to the cell surface. J Biol Chem 1989, 264:9053-9060

36. Adler S, Baker PJ, Johnson RJ, Ochi F, Pritzi P, CouserWG: Complement membrane attack complex stimu-lates production of reactive oxygen metabolites by cul-tured rat mesangial cells. J Clin Invest 1986, 77:762-767

37. Baeuerle PA, Baltimore D: Activation of DNA-bindingactivity in an apparently cytoplasmic precursor of theNF-KB transcription factor. Cell 1988, 53:211-217

38. Hirano M, Hirai S, Mizuno K, Osada S, Hosaka M, OhnoS: A protein kinase C isozyme nPKCE is involved in theactivation of NF-KB by 12-O-tetradecanoylphorbol-13-acetate (TPA) in rat 3Y1 fibroblasts. J Biol Chem 1995,206:429-436

39. Acosta JA, Benzaquen LR, Goldstein DJ, Tosteson MT,Halperin JA: The transient pore formed by homologousterminal complement complexes functions as a bidi-rectional route for the transport of autocrine and para-crine signals across human cell membranes. Mol Med1996, 2:755-765

40. Read MA, Whitley MZ, Williams AJ, Collins T: NF-KB

and IKBa: an inducible regulatory system in endothelialactivation. J Exp Med 1994, 179:503-512

41. Isral N, Gougerot-Pocidalo MA, Aillet F, Virelizier JL:Redox status of cells influences constitutive or inducedNF-KB translocation and HIV long terminal repeat ac-tivity in human T and monocytic cell lines. J Immunol1992, 149:3386-3393

42. Staal JT, Roederer M, Herzenberg LA, Herzenberg LA:Intracellular thiols regulate activation of nuclear factorKB and transcription of human immunodeficiency virus.Proc NatI Acad Sci USA 1990, 87:9943-9947

43. Meichle A, Schutze S, Hensel G, Brunsing D, KronkeM: Protein kinase C-independent activation of nucleartranscription factor KB by tumor necrosis factor. J BiolChem 1990, 265:8339-8343

44. Schreck R, Rieber P, Baeuerle PA: Reactive oxygenintermediates as apparently widely used messengersin the activation of the NF-KB and transcription factorand HIV-1. EMBO J 1991, 10:2247-2258

45. Ghezzi P, Dinarello CA, Bianchi M, Rosandich ME,Repine JE, White CW: Hypoxia increases production ofinterleukin-1 and tumor necrosis factor by humanmononuclear cells. Cytokine 1991, 3:189-194

46. Metinko AP, Kunkel SL, Standiford TJ, Strieter RM:Anoxia-hyperoxia induces monocyte-derived interleu-kin-8. J Clin Invest 1992, 90:791-798

47. Satriano JA, Shuldiner, Hora MK, Xing Y, Shan Z,Schlondorff D: Oxygen radicals as second messen-gers for expression of the monocyte chemoattractantprotein JE/MCP-1 and the monocyte colony-stimulatingfactor CSF-1 in response to tumor necrosis factor-aand immunoglobulin G. J Clin Invest 1993, 92:1564-1571

48. DeForge LE, Preston AM, Takeuchi E, Kenney J, BoxerLA, Remick DG: Regulation of interleukin 8 gene ex-pression by oxidant stress. J Biol Chem 1993, 268:25568-25576