Rickettsia rickettsii Infection of Cultured Human ...

Rickettsia rickettsii Infection of Cultured Human Endothelial Cells Induces Tissue Factor Expression By Lee Ann Sporn, Patricia J. Haidaris, Rui-Jin Shi, Yale Nernsrson, David J. Silverman, and VictorJ. Marder response requiresthe intracellular presence ofR rickettsii Microvascular thrombi underlie many of the clinical maniorganisms, because uninfectedrickettsia were ineffective festations of Rocky Mountain spotted fever (RMSF), a disand the responsewas blocked by inhibiting rickettsial entry ease characterized by Rickettsia rickettsii infectionof vasusing cytochalasin B. TF inductionwas not mediated by encular endothelial cells. Studies were designedto determine whether R rickettsii-infection of cultured human umbilicaldothelial cell release of soluble factor, because no response was induced using culture medium conditioned byR rickvein endothelial cells results in tissue factor (TF) induction, ettsii-infected cells. Furthermore, preadsorption of suspena process that could directly activate coagulation in insions of R rickettsiiwith polymyxin B to remove contamifected vessels. Whereas uninfected endothelial cell culdid not eliminate the TF tures showed essentially undetectable TF mRNA and ac- nating lipopolysaccharide response. Inductionof TF in vital endothelial cells during R tivity, both TF mRNA and activity were present after R rickettsiiinfection could bethe trigger for vascular thromrickettsii infection. TF mRNA levels were transient, peakbus formation of RMSF. ing at 4 hours after the initiationof infection, whereas the peak of TF activityoccurred at 8 hours. Induction of the TF 0 1994by The American Societyof Hematology.

and the cytoplasm of endothelial cellsand exhibits earlycell HE SPOTTED FEVERdiseases,ofwhichRocky to cell spread without detectable host cell injury." Mountain spotted fever (RMSF) is the most virulent Tissue factor (TF) is a membrane-bound glycoprotein (caused by Rickettsia rickettsii), result from infection with that is the essential cofactor involved in activation of the insect vector-transmitted rickettsial organisms. These disextrinsic pathway of TF forms complexes eases are manifested by fever, papular skin rash,and microvascular occlusive disease with multiple organ involvementwith factorsVI1 and VIIa, which leadsto activation of factors IX and X, resulting inthe conversion of prothrombin and tissue ischemiathat can lead to a fatal outcome."6Rao to thrombin and, finally, of fibrinogen to fibrin. Increased et al' showed evidence of platelet activationas well as actiTF expression in cultured endothelial cells has been induced vation of the extrinsic coagulation,contact, and fibrinolytic by a variety of perturbations, including exposureto tumor enzyme systems withinhours after infectionof human volnecrosis factor-a (TNF-a),22,23interleukin- 18 (IL18),24 unteers inoculated withR rickettsii as part of a vaccine-asphorbol 12-myristate 13-acetate(PMA)?' or bacterial liposessment study. The organism infects endothelial cells of polysaccharides (LPS).26 Such expression is transient and small vessels in regions of ischemic skin necrosis early in the i l l n e ~ s ~ and , ~ of , ~ larger vessels patients in with advanced regulated at the mRNAlevel due to both enhancement of transcriptional rate and enhancement of mRNA stadisease.' Because the vascular endothelialcell is a primary b ilit~.'~-'~ target of infection by this organism, developmentof a proIn the present study,we have investigated the possibility thrombotic phenotype beforeloss of cell viability could be that R rickettsii infectionof cultured endothelial cells results responsible for much of the pathologic change in this disin TF expression. Changes in both TF activity and steady ease. In studiesof cultured endothelial cells infected in vitro state TF mRNA levels weremonitored during the course. of with R rickettsii, we have previouslydemonstrated changes infection. In addition, we have explored whetheror not TF resulting from in vitro infection and before cell death, ininduction requires the intracellular presence of the rickettsia cluding increased platelet adherence to the surface of inorganisms. We hypothesize that TF expression may occur fectedcells,"increasedreleaseoflargevonWillebrand in R rickettsii-infected endothelial cells in vivo and may be factor multimers," and enhanced E-selectin cell surface expression and neutrophiladherence," and others have From the Hematology Unit,Department of Medicine, University shown increased secretion of plasminogen activator inhibiof Rochester School of Medicine and Dentistry, Rochester, NY: the tor in R rickettsii and R conorrii-infected endothelial cells.I3 Department of Medicine, Mt. SinaiSchool of Medicine, New York, Rickettsialorganismslikely enter the endothelial cell NY; and the Department ofMicrobiology andlmmunology, Univerthrough a process of induced phagocytosis, after which exit sity of Maryland School ofMedicine, Baltimore, MD. from the phagosome presumably involves rickettsial phosSubmittedAugust 16, 1993: accepted November l , 1993. pholipase A acti~ity.'"'~Entry of the organisms requires Supported in part by Grants No. HL-30616, HL-43711, HLboth rickettsial energy and endothelial cell participation. 29019, and AI-l 7416from the Heart, Lung and Blood and the AlPretreatment of endothelial cells with cytochalasin B, which lergy and Infectious Disease Institutes. National Institutes of disrupts the actincontaining cellular cytoskeleton,inhibits Health. Bethesda, MD. Address reprint requests to Lee Ann Sporn, PhD, Hematology rickettsial entry as does inactivation of rickettsia with Nethylmaleimide or formalin." The presence of calcium ion- Unit, PO Box 610, University of Rochester Medical Center, 601 Elmwood Ave, Rochester, NY 14642. ophores inhibits entry of rickettsia, suggesting that formaThe publication costs of this article were defrayed in part by page tion of a calcium gradient is involved in this process.18The charge payment. This article must therefore be hereby marked cholesterol receptor binding drugs digitonin and amphoter"advertisement" in accordance with18 U.S.C.section 1734 solely to icin B reduced plaque formationby R rickettsii on endotheindicate this fact. lial cells, implicating involvement of this receptor inthe in0 1994 by TheAmerican Society ofHematology. fection process.16 R rickettsii replicatesin both the nucleus 0006-4971/94/8304-0015$3.00/0

T

B I d , Vol83, No 6 (March 15). 1994: pp 1527-1 534

1527

SPORN ET AL

1528

the fundamental component that initiates local thrombotic events that underlie thc pathologic changes of RMSF. MATERIALS AND METHODS

2ol 15

l-

I~ndothdiulWII c11lt11r.c, und infi.cfion. Primary cultures of human umbilical vein endothelial cells were prepared as previously described3”,” after isolation from human Umbikdl cord veins by mildproteolyticdigestion.Secondpassage cells, usedinexperi10 mcnts, were cultured in McCoy’s 5a medium (Flow Laboratories, 50 pg/mL enMcLean, VA) with 20% (vol/vol) fetal bovine serum, dothelial cell mitogen (Biomedical Technologies, Inc, Stoughton, MA), IO0 pg/mL heparin (Sigma ChemicalCO,St Louis, MO), and 25 pg/mL insulin. A plaque-purified sccd of R ric#tcffsiiwith a titer 5 o f 2 X 10’ pfu/mL was prepared in VERO (African green monkey kidney) cells (American Type Culture Collection, Rockvillc, MD) as previously described.3z Endothelialcells were infected by placing suspensions of R rickrttsli contained within lysates of VERO cells n diluted in complete culture medium without antibiotics onto nearCONTROL RR LPS confluent cultures and allowing them to incubate at 37°C in a 5% COz atmosphereforspecifiedtimes.Forcontrolexperiments, Fig 1 . TF activity of endothelial cells after infection with R rickVERO cells were cultured in the absenceof R rick[wt/vol] Na-N-lauroyl sarcosine, 0. I mol/L 0-mercaptific, Orangeburg, NY). toethanol, 0.1% [vol/vol] Antifoam A [30%)cmulsion; Sigma], 5 Inacfivutron undxonicufionof R rickettsiiand ad.~orption oJLPS. mmol/L MgCl2) was added and agitated until viscous. Flasks were R rickcltlsii organisms were inactivated by adding 10 pg/mL tetrathcn rinsed withan additional5 mL RNAlysis buffer, Samples were I hour before placement o n cells. cycline (Sigma) and incubated for Alternatively, R rickett.sii were UV-inactivated by exposure 5 cm frozen at -70°C and thawed before RNA isolation to complete cell a cushionof from a germicidal light in an open container for 30 minutes. R ricklysis. ‘1.0 isolate RNA, samples were layered onto3 mL rtt.sii were disrupted by sonication usinga Sonifier R Cell Disruptor, CsCl(5.7 mol/L CsCl,, 1 mmol/L EDTA) and then centrifuged at 25,000 rpm in an SW41 rotor at 18°C for 18 hours. RNA pellets Model W140 (Heat Systems-Ultrasonics, Inc. Plainview, NY) at full were resuspended four times in 125 power for30 seconds. LPSwas adsorbed from solutions containing pL aliquots of sterile water, after which the RNA was extracted once or until clearanwith equal R ricltrtlsii or disrupted H rickarsii by incubation with polymyxin B-agarose (Sigma; binding capacity,200 to 500 pg LPS fromE coli volume of phcnol:chIoroform:isoamylalcohol (50:48:2),ethanol serotype0128:B12/mL, l0 pL/4.5 X IO6 pfu R rickettsii). Polyprecipitated in 0.2 mol/L potassium acetate, and dissolvedin 20 to myxin B agarose was allowed to settle and supernatant was placed 50 pL water. Twenty micrograms of total RNA per condition was on the endothelialcell cultures. denatured with glyoxal/dimethylsulfoxide (DMSO) and then resolved by electrophoresis in I % agarose gels in 10 mmol/L sodium RESULTS phosphate buffer, pH 7.0.35 Total RNAwas prepared for Northern blot analysis by electrophoretic transfer to Zeta-probe mcmbranes Endothelial cell cultures were infected using6 X IO4 pfu/ (BioRad, Centreville, NY). To prepare TF or y-actin probes, we cm2 R rickettsii organisms or incubated withl Fg/mL LPS used thc IkoRI-IlindIII fragment corresponding to bases I to 1358 subcloned from pKS2h (a kind gift from Dr W.H. Konisberg, Yale for 6 to 8 hours, then lysed for TF activity determination (Fig l). The amount ofR rickettsii used to infect the endoUniversity, New Haven, CT) into pGEM7Zf(-) (Promega, Madithelial cells was based o n previous studies indicating that I , was obtained from son, WI). The human y-actin probe, pHFyAsuch amounts result in uniform infection without loss ofcell Dr L. Kedes (Stanford University, Palo Alto, CA). The 2,200-bp

ROF R/CK€TTsll INDUCTION

1529

c

2 4 8 24 12

2 3 4 01224

" LPS

RR

Fig 2. Time course of changes in steady-state levelof TF mRNA and TFactivity of endothelial cells after R rickettsii infection. (A) Total RNA was extracted from cultures of endothelial cells harvested at 2,4. 8 , 12, and 24 hours after R rickettsii (RR) infection. RNAwas subjected to Northern blot analysis usinga TF cDNA probe. y-Actin mRNA content was also measured in control cells and cells infected for 2,4, 8, 12, and 2 4 hours after infection as a control for cell viabilityand mRNA integrity. (B) TF activity of cells infected for varying times was determined usinga two-stage clotting assay.

viability during the first 48 hours of infection." Untreated endothelial cellspossessed nearly undetectableamounts of TF activity (averaging lessthan 1 ng/3 X IO6 cells) as measured usinga two-stage clotting assay. TF activity of R rickeftsii-infected endothelial cells increased to 7.0 & 1.7 ng/3 X lo6 cells compared with 16 2.2 ng/3 X IO6 cells after exposure to LPS. Increases in endothelial cell TF activity was paralleled by increases in TF antigen, as determined by ELBA (data not shown). Incubation of endothelial cells with suspensions of lysed VERO cells(the vector in which the rickettsia organisms are propogated) containing no R ricketfsii didnotresultinincreases in TF activity (not shown). Steady state TF mRNA levelsand TF activity of cells infected withR ricketfsiior treated withLPS for varying times between 2 and 24 hours were monitored by Northern blot analysis using a TF cDNA probe (Fig2A). TF mRNA was essentiallyundetectable in untreatedcultures.Infection with R rickeffsiior LPS treatment resulted in a dramatic _+

Iln 12

24

Time (hours) increase in the 2.2-kb translatable TF mRNA. Increases of a 3.4-kb mRNA also occurred (amounts varying among cell cultures), but this untranslatable mRNA species, representing the presence of an unspliced copy of intron 1, always increased in proportion to the 2.2-kb form. The increase in TF mRNA levelsoccumng in response to R rickettsii infection was transient, peaking at 4 hours and declining by 8 hours after the initiation of infection. The increasein mRNA occumng in response to LPS treatment was also transient, peaking at 3 hours after initial stimulation. Amounts of y-actin mRNA remained relatively stable during the time course, indicatingthat there was very little loss of cell viability during the time course of infection. TF activity also increased transiently (Fig 2B), peakingat approximately 8 hours. Therefore,a lag period of approximately 4 hours existed between peakTF mRNA expressionand peak TF activity expression. The increases in TF activity and mRNA levels observed after R rickettsii infection required the presence of viable,

SPORN ET AL

1530

RR

RR

RR UV

Tet

RR

+

Tet

LPS

LPS

Fig 3. TF activity of endothelial cells after exposure t o noninfective R rickettsii. Suspensions of R rickettsii (RR) were treated for1 hour with 5 pg/mL tetracycline orexposed to a germicidalUV lamp (at a distance 5 ofcm for30 minutes) t o render the organisms noninfective. Endothelial cells were then exposed t o these organisms for 8 hours, at which timecultures were lysed andTF activity determined. Treatment with UV (RR,,) or tetracycline (RRTef)eliminated thecells' response to R rickettsii. Tetracycline (5 pg/mL) did not eliminate theresponse of cellsto LPS (RRTet LPS).

LI11 LPS

0

Control

TNF

RR

Fig 4. Effect of polymyxinB adsorption onresponse of endothelial cells toRrickeftsjj. Solutions containing LPS (1 pg/mL), TNF ( l 0 ng/mL), Rrickettsii(RR).or sonicated Rrickettsii(RR,,.) were incubated with endothelial cells for 6 to 8 hours with (m) or without (0) preadsorption using polymyxin B agarose t o remove soluble LPS. Preadsorption of LPS and sonicated R rickettsii with polymyxin B eliminated thecells' response, but preadsorption ofTNF or RR had no effect.

using LPS in parallel with endothelial cells that were previously uninfccted (Fig 5 ) . Preinfected endothelial cells re+ sponded to LPS stimulationby expressing T F levels cquivalenttothose of LPS-stimulatedendothelial cells not prcviously infected. Thescresultsindicatcthatnotonly were cells capable of undergoing a secondary response and infective organisms (Fig 3). X ric,kcmii organisms were eiwere therefore viable, but, furthermore, that the cells likely ther pretreated with tetracycline or exposed toa germicidal had not responded to LPS during the infcction. lamp to render the organisms noninfcctivc. Endothelial cell cultures were then incubated with these noninfcctivc organ- Endothelial cells were incubated with cytochalasin B (10 pmol/L) duringR rickottsii exposure to inhibit intcrnalizaisms for 8 hours and TF activity was determined. Lack of infectivity was confirmed by immunofluorescence staining using anti-K rickclrsii antibody (not shown). No increases in T F activity occurred after exposure to noninfective orl5 ganisms. Tetrdcyche alone did not prevent TF induction, as cells treated with tetracycline-treated R rickel/sii in the presence of LPS exhibited similarTF activity levels as cells treated with LPS alone. Polymyxin B-agarose was used to adsorb LPS from solutions containingR rickcllsii before placing them on cells the (Fig 4). The adsorption process did not affect infectivity of K ric,keltsii as monitored by immunofluorescence staining. Adsorption using polymyxin B did not inhibit the activity response to infective R riclk>l/sii.On the contrary, adsorption of sonicated R rickcttsii with polymyxin B eliminated the response ofthe cells. Thus, rickettsial LPS plays a role in the stimulationof endothelial cells to express T F only when 0 8 24 30 Time (hours) the organisms are disrupted. Polymyxin B completely eliminated the stimulatory effect of LPS ( 1 Fg/mL), but there was Fig 5. Restimulationof R rickettsii-infectedendothelialcells no inhibitory effect o n T N F ( l 0 ng/mL), another known with LPS. Endothelial cells were infected with R rickettsii (RR) for inducer ofendothelial cell TF activity. 24 hours, during which timeTF activity increased and returned to near-baseline levels. At thispoint, LPS (1 pg/mL) wasadded t o one To determine whether R rickettsii-infected endothelial culture and TF activity was assayed 6 hours later (30 hours after cells couldberestimulatedtoinduce T F usinganother the initiation of infection) (a). A parallel Rrickettsii-infected culture agent, endothelial cells were infected for 24 hours, during was incubated for an additional 6 hours, but in theabsence of LPS which time TF levels increased and returned to near-basestimulation (U).A previously uninfected culture was also incubated line levels. At this time, endothelialcells were restimulated with LPS for 6 hours beforeTF activity determination(0).

r

R

1531

R/CKFTE// INDUCTION OF EC TISSUEFACTOR

cy) dl

I

Control

RR

LPS

Fig 6. Effect of cytochalasin B on endothelial cells' response to Rrickertsiiinfection. Endothelial cells were incubatedwith cytochalasin B (l0 rmol/U during infection(RR) to inhibit internalizationof Rfickettsii, CytochalasinB had no effect on LPS induction of TF activity by the endothelid cells. (0)Without cytochalasinB: (m]with cytochasin B.

tion of the organisms by depolymerization of the endothelial cells' actin-containing c y t ~ s k e l e t o n (Fig ' ~ ~ ~6 )~. Lack of uptake of R rickettsii was confirmed by immunofluorescence stainingusing anti-R rickettsii antibody (not shown). Cytochalasin B-treated endothelial cells not didincrease TF activity in response to the presence of R rickettsii organisms. However, cytochalasinB did not disrupt the ability ofthe endothelial cells to express TF in response to LPS exposure. The effect of cytochalasin B was specificto endothelial cell function, because incubation of R rickettsii organisms with cytochalasinB followed by extensive washing did not inhibit their infectivity when placedon untreated endothelial cells (data not shown). To explore the possibility that a soluble factor released by endothelial cells participates in induction of TF during R rickettsii infection, cellswere infected withR rickettsii for 8 or 24 hours, culture supernatant was then collected, 5 @g/ mL of tetracyclin was added, and the solution was filtered through a 0.22-pm filter to remove R rickettsii organisms. This "conditioned medium" was then immediately placed on cultures of endothelial cells for 6 to 8 hours and TF activity was measured (Fig7). A slight but not statisticallysignificant increase in TF activity overcontrol cells ( P = .I8) occurred in cells treated with conditioned medium alone. This increase was smalland significantly lessthan that induced by infective R rickettsii or LPS. DISCUSSION

In response to infection with R rickettsii organisms, cultured human umbilical vein endothelial cells express TF. Although the overall magnitudeof this TF response by the aggregate endothelial cell culture was approximately half that observed with LPS (Fig I), the individual cell responses were likely equivalent, given that the endothelid cells were not uniformly infected withR rickettsii during the 8-hour incubation period." Thus, infectionwith R rickettsii organ-

isms is a potent inducer of in vitro endothelial cell TF expression, thus providingapotentialexplanationfor the striking thrombotic manifestations of RMSF. The increases in TF activity and TF mRNA levels in infected endothelial cells were transient, with the peak ofTF mRNA occurringat 4 hours after the initiation of infection and the peak of TF activity lagging by approximately 4 hours. Although peak TF activity vanes depending on the stimulant the later appearance of peak TF after R rickettsii infection could have been caused by the time required for organisms to gain entry into the endothelial cells. Prior studies of in vitro infection of endothelial cells have shown that the percentage of infected cells increases during the first 1 to 2 hours of incubation with R rickettsii organisms, plateauing with further inc~bati0n.I~ activity TF and TF mRNA levelsreturned to near-baseline levelsby 12 hours after bothLPS stimulation and R rickettsii infection. The TF response of infected cellswas truly a transient phenomenon, because there was no loss of cell viabilityor general decrease in recovery of intact mRNA as reflected by constant levels of actin mRNA during the 24-hour infection. Furthermore, endothelial cells infected withR rickettsii for 24 hours were also capable of undergoing a second, maximal TF response on subsequentstimulation using LPS (Fig 5). The regulation of TF mRNA in response to LPS, PMA, or TNF has been studied by other inve~tigators?~.~~ PMA and TNF exposure were shown, by nuclear run-off assays,

Fig 7. Effectof medium conditioned by Rfic&t?t?iiinfectioofendotheliel cells on TF induction. Mediumwas Edlectedfrom cultures infected for 24 hours with R rickerrsii (RR). filtered. treated with tetracycline, and placed on fresh endathdial cell culturesfor 8 hours. R rickettsii-conditioned medium(cond. med.) did not result in statistically significant increases in TF activ.&y when compared (P = .18). The presenceof infective with untreated contml cultures R rickettsii (P = .03)and LPS treatment (P = .03)resulted inr i g nificant increases in TF activity when compered with untreated controls. A statistically significant difference in TFactivities was noted when cultures treated with conditioned mediumwere m pared with cultures incubatedwith infective R rickettsii (P = .W). Statistical analysiswas conducted using the one-tailed Student's t-test.

1532

SPORN ET AL

to result in increased transcription of the TF gene:7 whereas for other pathogenic bacteria, including Yersinia?‘ It is also the increase in mRNA content resulting fromLPS stimulapossible that the endothelial cell responds to oxidative intion was shown to depend largely on increased mRNA stajury, which has been implicated in endothelial cell injury bility rather than on increased transcription rate.29 Studies caused by R rickettsii infe~tion,~’ Infection of endothelial of both transcription rate and changes in mRNA stability cells withR rickettsii likely representsa novel stimulus for would be required to show the mechanism(s) responsible for induction of TF, yet the exact nature of the stimulus and the increased steady-stateTF mRNA levels found inendothe intracellular signaling mechanisms involved are not yet thelial cells after R rickettsii infection. known, It has been demonstratedthat the procoagulant properties Severalresultspresentedheresuggest that the TF reof cultured endothelial cells are enhanced in response to sponse observed requires the infection of endothelial cells with R rickettsii organismsas well as intracellular uptakeof other infective agents.For example, infectionof endothelial the organisms. Rendering R rickettsii organisms noninfeccells with herpes simplex virus (HSV) results TF in mRNA tive by treatment with tetracycline or UV exposure before transcription?* The time course of TF mRNA expression after HSV infection was also transient, peaking at 4 hours placement on cells eliminatedthe TF response. Incubation after the initiation of infection and increasing withthe magof endothelial cells with cytochalasin B, which inhibits internalization of R rickettsii without eliminating adherence nitude of infection. Inaddition to increased TF expression, prothrombinase complex assemblyis enhanced on HSV-inof the organisms to the cell surface bydisrupting the endofected endothelial presumably because of expresthelial cells’ actin-containing cyto~keleton,’~.’~ also oblitersion of HSV PC,a viral protein that can actas a binding site ated theTF response (Fig6).These resultssuggest that, even in the presence of infective R rickettsii, the TF response can- for factor X and prothrombinase complex assembly. The procoagulant phenotype of HSV-infected endothelial cells not be elicited unless the organisms are able to enter the is further enhanced by decreased expressionof the protein C cells. The data are also consistent withthe observation that anticoagulant cofactor, thrombomodulin,42 and enhanced tetracycline-treatedor UV-treated R rickettsii fail to induce Furthermore, HSV-infected endothelial the TF response (Fig3), because only viable organisms are platelet adhesion.43 cells exhibit decreased synthesis of plasminogen activator internalized within endothelial cells.14 inhibitor type 1 (PAL 1) and tissue plasminogen activator(tBecause LPS is a potent inducer of endothelial cell TF PA) at the level of tran~cription.‘?~ Procoagulant activityis activity in vitro, and because rickettsial organisms possess enhanced on cytomegalovirus-infected endothelialcells46s47 LPS, which issimilar in chemical composition and biologic properties to that derived fromE coli,38the possibility exists but is independent of increased TF expression and occurs even using heat-inactivated virus47 and is thereforelikely to that TF was induced by rickettsial LPS releasedinto the mebe independent of active infection. von Willebrand factor dium. This is unlikely, because preadsorption of suspenrelease, decreased thrombomodulin expression, and transions of intact, viable R rickettsii using polymyxin B-agasient increases in TF expression have also been demonrose, which binds to the lipid A component ofLPS strated inR conorii-infected endothelial cells.48 responsible forTF induction,” did not eliminate the TF reThis study reportsthat cultured vascular endothelial cells sponse (Fig 4). Furthermore, endothelial cells did not exinfected withR rickettsii rapidly and transiently express TF. hibit a statistically significant increasein TF activity in reBecause RMSF and other spotted fever diseases caused by sponse to the presence of medium conditioned by infected infection of vascular endothelial cells with rickettsial organendothelial cells (Fig7). isms result in microvascular thrombosis, this finding may However, rickettsial LPS isa potent inducer of endotheshed light on the possible function of the perturbed endolial cell TF expression. When suspensions of tetracyclinethelial cell in the development of thrombotic disease. Such treated R rickettsii were sonicated before placement on cells, endothelial cellderived TF could directly activate coagulathus releasing LPS,TF was induced and was inhibitable by tion and could explain localized thrombus formation in inpreincubation withpolymyxinB-agarose.Thisindicated no clear demonstrathat the lipid A component of rickettsial LPS, like LPS fromfected vessels. However, there has been tion of the inducibility of TF by endothelial cells in vivo, E coli, is the portion of the molecule responsible for the celmuch less the role of endothelial cell TF in thrombus forlular response. We hypothesize that rickettsial LPS is not mation. Detailedstudiesof animal modelsystems are involved inTF induction during an actual infection for sevneeded to assess the importance of endothelial cell TF in eral reasons. Cells exposed to LPS remain refractoryto secpathologic thrombosis. ondary stimulation using LPS.4’ The observation that infectedendothelialcellsresponded to subsequent LPS ACKNOWLEDGMENT stimulation suggests that they had not been previously exWe wish to thank Laura Triou and Sarah Lawrencefor excellent posed to LPS during the infection process (Fig 5). R ricketttechnicalassistance,Kristin Leibert for statisticalanalysis,and sii is an obligate intracellular parasite that must replicate Carol Weed and Linda Piedemonte for help in preparation of the within the host cell cytoplasm or nucleus. Therefore, it is manuscript. unlikely that the organisms are lysed to release LPS after REFERENCES entry into the cells. It is possible that a molecule(s) present on the rickettsial surfaceinteracts directly with intracellular I. Hattwick MA, OBrien RJ, Hanson BF Rocky Mountain spotted fever: Epidemiologyof an increasing problem.Ann Intern components of the endothelial cell and results in activation of intracellular signalling pathways, as has been described Med 84:732, 1976

R R/CK€TTS// INDUCTION OF EC TISSUE FACTOR

2. Hattwick MA, Retailliau H, OBrien RJ, Slutzker M, Fontaine RE, Hanson B: Fatal Rocky Mountain spotted fever. JAMA 240:1499, 1978 3. Helmick CG, Bernard KW, DAngelo LJ: Rocky Mountain spotted fever: Clinical, laboratory, and epidemiological features of 262 cases. J Infect Dis 150480, 1984 4. Moe JB, Mosher DF, Kenyon RH, White JD, Stookey JL, Ragley LR, Fine D P Functional and morphologic changesduring experimental Rocky Mountain spotted fever in guinea pigs. Lab Invest 35:235, 1976 5. Walker DH, Cain BG, Olmsted PM: Laboratory diagnosis of Rocky Mountain spotted fever by immunofluorescent demonstration of Rickettsia rickettsii in cutaneous lesions. Am J Clin Pathol 69:619, 1978 6. Dimmitt SK, Miller DK: Rocky Mountain spotted fever. U1trastructural findings. Am J Clin Pathol78: 13 1, 1982 7. Rao AK, Schapira M, Clements ML, Niewiarowski S, Budzynski AZ, Schmaier AH, Harpel PC, Blackwelder WC, Schemer J-R, Sobel E, Colman R W A prospective study of platelets and plasma proteolytic systems during the early stages of Rocky Mountain spotted fever. N Engl J Med 3 18: 1021, 1988 8. Woodward TE, Pedersen CEJr, Oster CN, Bagley LR, Romberger J, Snyder MJ: Prompt confirmation ofRocky Mountain spotted fever: Identification of rickettsiae in skin tissues. J Infec Dis 134:297, 1976 9. Harrell G T Rocky Mountain spotted fever. Medicine28:333, 1949 10. Silverman DJ:Rickettsia rickettsii-induced cellularinjury of human vascular endothelium in vitro. InfectImmun 44:545, 1984 I 1. Sporn LA, Shi R-J, LawrenceSO, Silverman DJ, Marder VJ: Rickettsia rickettsii infection ofcultured endothelial cells induces release of large von Willebrand factormultimers from Weibel-Palade bodies. Blood78:2595, 199 I 12. Sporn LA, Lawrence SO, Silverman DJ, Marder VJ: E-selectin-dependent neutrophil adhesion to Rickettsia ricketrsii-infected endothelial cells. Blood8 1:2406, 1993 13. Drancourt M, Alessi M-C, Levy P-Y,Juhan-Vague I, Raoult D: Secretion of tissue-type plasminogenactivator and plasminogen activator inhibitor by Rickettsia conorii- and Rickettsia rickettsiiinfected cultured endothelial cells. InfectImmun 58:2459, 1990 14. Walker TS, Winkler HH: Penetration of cultured mouse fibroblasts (L cells) by Rickettsia prowazeki. Infect Immun 22:200, 1978 15. Winkler HH, Miller E T Phospholipase A and the interaction of Rickettsia prowazekii and mouse fibroblasts (L-929 cells). Infect Immun 38:109, 1982 16. Walker DH, Firth WT,Ballard JG, Hegarty BC: Roleof phospholipase-associatedpenetration mechanism in cell injury by Rickettsia rickettsii. InfectImmun 40:840, 1983 17. Silverman DJ, Santucci LA, Meyers N, Sekeyova Z: Penetration of host cells byRickettsia rickettsiiappears to be mediated by a phospholipase of rickettsial origin. Infect Immun 602733, 1992 18. Walker TS: Rickettsial interactions with human endothelial cells in vitro: Adherence and entry. Infect Immun 44:205, 1984 19. Silverman DJ, Bond S B Infection of human vascular endothelial cellsby Rickettsia rickettsii. J Infect Dis149:201, 1984 20. Nemerson Y: Tissue factor and hemostasis. Blood7 1 :1,1988 2 1. Bach R Initiation of coagulation by tissue factor. Crit Rev Biochem 23:339, 1988 22. Bevilacqua MP, Pober JS, Majeau GR, Fiers W, Cotran RS, Gimbrone MA: Recombinant tumor necrosis factor induced procoagulant activityin cultured human vascular endothelium: Characterization and comparison with the actions of interleukin. Proc Natl Acad Sci USA 83:4553, 1986

1533

23. Nawroth P, Stem D: Modulation of endothelial cells hemostatic propertiesby tumor necrosis factor.J Exp Med163:740,1986 24. Bevilacqua MP, Pober JS, Majeau GR, Cotran RS, Gimbrone MA: Interleukin 1 ( I L I ) induces biosynthesis and cell surface expression ofprocoagulant activity in human vascular endothelial cells.J Exp Med 1606 18, 1984 25. Lyberg T, Galdal K!$ Evensen SA, Prydz H: Cellular coop eration in endothelial cell thromboplastin synthesis.Br J Haematol 53:85, 1983 26. Colucci J, Balconi G, Lorenzet R, Pietra A, Locati D, Donati MB, Semararo N: Cultured human endothelial cells generate tissue factor in response to endotoxin. J Clin Invest7 1: 1893, 1983 27. Scarpati EM, SadlerJ E Regulation ofendothelial cell coagulant properties. Modulation oftissue factor, plasminogen activator inhibitors, and thrombomodulin by phorbol 120 myristate 13-acetate and tumor necrosis factor.J Biol Chem 264:20705, 1989 28. Conway EM, Bach R, Rosenberg RD, Konigsberg WH: Tumor necrosis factorenhances expression of tissue factor mRNA in endothelial cells. Thromb Res 53:231, 1989 29. Crossman D C , Carr DP, Tuddenham EGD, Pearson JD, McVey JH: Theregulation of tissue factor mRNAin human endothelial cells in responseto endotoxin or phorbol ester.J Biol Chem 265:9782, 1990 30. Gimbrone MA Jr, Cotran RS, Folkman F: Human vascular endothelial cells in culture. Growth and DNA synthesis.J Cell Biol 60:673, 1974 3 1 . Wagner DD, Olmsted JB, Marder VJ: Immunolocalization of von Willebrand protein in Weibel-Palade bodies human of endothelial cells.J Cell Biol95:355, 1982 32. Silverman DJ, Santucci LA: Potential forfreeradical-induced lipid peroxidation as a cause of endothelial cell injury in Rocky Mountain spotted fever. InfectImmun 56:3110, 1988 33. Bach R, Gentry R, Nemerson Y: Factor VI1 binding to tissue factor in reconstituted phospholipid vesicles: Induction of cooperativity by phosphatidyl serine. Biochemistry 25:4007, 1986 34. Pitlick FA, Nemerson Y:Purification and characterization oftissue factor apoprotein. Methods Enzymol45:37, 1976 35. Simpson PJ:An improved method for mRNA isolationand characterization of in vitro translation products by Western blotting. Gene 56:161, 1987 36. Haidaris PJ, Wright TW, Gigliotti F, Haidaris C G Expression and characterization of a cDNA clone encoding an immunodominant surface glycoprotein of Pneumocystis carinii. J Infect Dis 166:1113,1992 37. Wagner DD, Olmsted JB, Marder VJ: Immunolocalization of vonWillebrand proteinin Weibel-Palade bodies human of endothelial cells. J Cell Bioi 95:355, 1982 38. MiragliottaG, Fumarola D, Colucci M, SemeraroN Platelet aggregation and stimulation of leucocyte procoagulant activity by rickettsial lipopolysaccharides in rabbits and in man. Experientia 37:47, 1981 39. Schorer AE, Rick PD, Swaim WR, Moldow CF: Structuralfeatures of endotoxin required for stimulation of endothelialcell tissue factor production; exposureof preformed tissue factor &er oxidantmediated endothelialcell injury. J Lab Clin Med 106:38,1985 40. Busso N, Huet S, Nicodeme E, Hiernaux J, Hyafil F Refractory period phenomenon in the induction oftissue factor expression on endothelial cells. Blood78:2027, 199 1 41. Bliska JB, Galin JE, Falkow S : Signal transduction in the mammalian cell during bacterial attachment and entry. Cell 73: 903,1993 42. Key NS, Vercellotti GM, Winkelmann JC, Moldow CF, Goodman JL, Esmon NL, EsmonCT, Jacob HS: Infection ofvascular endothelial cells with herpes simplex virus enhances tissue fac-

1534 tor activity and reduces thrombomodulin expression. Proc Natl Acad SciUSA 87:7095,1990 43. Visser MR, Tracy PB, Vercellotti GM, Goodman JL, White JG, Jacob HS: Enhanced thrombin generation and platelet binding on herpes simplex virus-infectedendothelium. Proc Natl Acad Sci USA 85:8227, 1988 44. Etingin OR, Silverstein RL,Friedman HM, Hajjar D P Viral activation of the coagulation cascade: Molecularinteractions at the surface of infected endothelial cells. Cell6 I :657, 1990 45. Bok RA, Jacob HS,Balla J, Juckett M, Stella T, Shatos MA, Vercellotti GM: Herpes simplex virus decreases endothelial cell plasminogen activator inhibitor. Thromb Haemost 69:253, 1993

SPORN ET AL

46. Van Dam-Mieras MCE, Bruggeman CA, Muller AD, Debie WHM, Zwaal RFA: Induction of endothelial cell procoagulant activity by cytomegalovirus infection. Thromb Res 47:69, 1987 47. Van Dam-Mieras MCE, Muller AD, van Hinsbergh VWM, Mullers WJHA, BomansPHH, Bruggeman CA: The procoagulant response to cytomegalovirusinfected endothelial cells. Thromb Haemost 68:364, 1992 48. Teysseire N, Arnoux D, George F, Sampol J, Raoult D: von Willebrand factor releast and thrombomodulin and tissue factor expression in Rickettsia conorii infected endothelial cells. Infect Immun 60:4388,1992