Proc. Nail. Acad. Sci. USA Vol. 87, pp. 7095-7099, September 1990 Medical Sciences Infection of vascular endothelial cells with herpes simplex virus enhances tissue factor activity and reduces thrombomodulin expression NIGEL S. KEY*, GREGORY M. VERCELLOTTI*, JOHN C. WINKELMANN*, CHARLES F. MOLDOWt, JESSE L. GOODMAN*, NAOMI L. ESMONt, CHARLES T. ESMONt§, AND HARRY S. JACOB*¶ *Department of Medicine, University of Minnesota, Minneapolis, MN 55455; tVeterans Administration Hospital, Minneapolis, MN 55417; tCardiovascular Biology Research Program, Oklahoma Medical Research Foundation and Oklahoma University Health Sciences Center, and §Howard Hughes Medical Institute, Oklahoma City, OK 73104 Communicated by Laszlo Lorand, June 8, 1990 (received for review March 29, 1990) ABSTRACT Latent infection of vascular cells with herpes- ulant effects of thrombin generated on the endothelial cell viruses may play a pathogenic role in the development of surface by an amplified prothrombinase complex system human atherosclerosis. In a previous study, we found that might be offset by upregulation of a major anticoagulant cultured human umbilical vein endothelial cells (HUVECs) pathway involving the thrombomodulin (TM)/protein C sys- infected with herpes simplex virus 1 (HSV-1) became proco- tem. Here we demonstrate the converse: namely, that the agulant, exemplified both by their enhanced assembly of the potential ameliorating effects of TM are actually lost in HSV prothrombinase complex and by their inability to reduce infections. Thus, TM surface expression and mRNA are adhesion of platelets. We now report two further procoagulant rapidly reduced in endothelial cells that are infected with consequences of endothelial HSV infection: loss of surface HSV-1. Superimposed upon this potentially procoagulant thrombomodulin (TM) activity and induction of synthesis of event, we found a significant increase in the expression of tissue factor. Within 4 hr of infection ofHUVECs, TM activity tissue factor (TF) by infected endothelial cells. We conclude measured by thrombin-dependent protein C activation de- that these procoagulant events may aggravate the thrombotic clined 21 ± 3% (P < 0.05) and by 18 hr, 48 ± 5% (P < 0.001). diathesis associated with HSV vascular infections. Similar significant TM decrements accompanied infection of bovine aortic endothelial cells. Identical TM loss was induced with HSV-2 infection but not with adenovirus infection. De- MATERIALS AND METHODS creased surface expression of TM antigen (measured by the Reagents. Bovine protein C and bovine antithrombin III specific binding of a polyclonal antibody to bovine TM) closely were purchased from Enzyme Research Laboratories (South paralleled the loss of TM activity. As examined by Northern Bend, IN). Protein C lacking y-carboxyglutamic (Gla) resi- blotting, these losses apparently reflected rapid onset (within 4 dues was prepared as described (6). The chromogenic sub- hr of HSV infection) loss of mRNA for TM. In contrast, HSV strate S2238 (D-Phe-pipecolyl-Arg-p-nitroanilide) was from infection induced a viral-dose-dependent increase in synthesis Helena Laboratories. Tunicamycin, human a-thrombin, of tissue factor protein, adding to the procoagulant state. The polymyxin B, bovine serum albumin, Escherichia coli lipo- results indicate that loss of endothelial protein-synthetic ca- polysaccharide (LPS), and Hepes were from Sigma. Trypsin/ pacity is not a universal effect ofHSV infection. We suggest that EDTA solution and Hanks' balanced salt solution were from the procoagulant state induced by reduction in TM activity and GIBCO. Rabbit brain thromboplastin (thromboplastin C) was amplified tissue factor activity accompanying HSV infection of from Baxter, Dade Division (Miami). The RIA kit for inter- endothelium could contribute to deposition of thrombi on leukin la (IL-la) was from Genzyme. atherosclerotic plaques and to the "coagulant-necrosis" state Endothelial Cell Cultures. Primary HUVEC cultures were that characterizes HSV-infected mucocutaneous lesions. grown to confluence as described (7). Bovine aortic endo- thelial cells (BAECs; passages 3-6) were used in some From an earlier suggestion that herpes simplex virus 1 experiments. These cells demonstrated the typical cobble- (HSV-1) may be involved in human atherosclerosis (1), we stone morphology ofendothelial cells and stained positive for became interested in the effects of HSV-1 infection on factor VIII-related antigen by indirect immunofluorescence. endothelial cells in vitro. The possibility that such infection Virus. HSV-1 strain 17+ was propagated in rabbit skin cells might induce a "procoagulant state" is suggested from the and virus was titrated by standard methods (8). Confluent histology of HSV-1 mucosal lesions, which commonly re- monolayers of endothelial cells were generally infected with veals leukocytoclastic vasculitis and intense intravascular 10 plaque-forming units (pfu) per cell (5, 6). Neither the fibrin deposition (2, 3). Additional clinical support for the HSV-infected nor the mock vehicle contained >0.05 ng of concept of hypercoagulability in HSV-1 infections is the endotoxin per ml as assayed by the Limulus amoebocyte severe, often fatal, intravascular coagulation that often oc- lysate assay (Associates of Cape Cod). Where indicated, UV curs in disseminated HSV infection of neonates (4). irradiation was performed by exposing the virus suspension We demonstrated that HSV-infected human umbilical vein in open-topped 35-mm dishes to a germicidal 30-watt UV endothelial cells (HUVECs) became prothrombotic by virtue strip light (General Electric) at a distance of 20 cm for the of enhanced prothrombinase complex assembly on their appropriate length of time. membrane surfaces. In addition, thrombin-stimulated plate- lets exhibited enhanced adhesiveness to HSV-infected cells Abbreviations: HSV, herpes simplex virus; TM, thrombomodulin; (5). However, we acknowledged that the possible procoag- TF, tissue factor; HUVEC, human umbilical vein endothelial cell; BAEC, bovine aortic endothelial cell; LPS, lipopolysaccharide; pfu, plaque-forming unit(s). The publication costs of this article were defrayed in part by page charge ITo whom reprint requests should be addressed at: Division of payment. This article must therefore be hereby marked "advertisement" Hematology, Box 480, University ofMinnesota Hospital and Clinic, in accordance with 18 U.S.C. §1734 solely to indicate this fact. Harvard Street at East River Road, Minneapolis, MN 55455. 7095 Downloaded by guest on September 26, 2021 7096 Medical Sciences: Key et al. Proc. Natl. Acad. Sci. USA 87 (1990) TF and TM Activity Assay. Total cellular TF activity was RESULTS HUVECs by a modified one-stage clotting assay measured in Characteristics of HSV-1 Infection of Endothelial Cells. (9). Both HUVECs and BAECs can be productively infected with For assay of TM activity, human a-thrombin (0.15 NIH HSV-1. Viability (as assessed by trypan blue and 51Cr-release unit/ml) and bovine protein C (27.5 umg/ml) were added to methods) is maintained in both for at least 36 hr (6). The buffer over washed monolayers. The cells were incubated for maximum period of infection reported in these studies is 18 60 min at 370C in 5% CO2. The supernatants were then hr, at which time cell viability is >95% and lactate dehydro- removed and residual thrombin activity was quenched with genase release does not differ significantly in uninfected or excess antithrombin III (600 nM). In the second stage of the HSV-infected HUVECs. Viral replication (measured as virus assay, conditioned medium was added to a cuvette containing recovery in supernatant plus removed adherent cells) is more buffer and the amidolytic substrate S2238 (0.2 mM). The rate efficient in HUVECs than in BAECs (Fig. 1), and after 24 hr, ofincrease in A405 was taken as a measure ofactivated protein HUVECs are 50 times more heavily infected. C enzymatic activity and was expressed per gg ofendothelial Activation of Protein C. Protein C activation was signifi- cell protein (10). The mean activity for triplicate wells was cantly reduced on HUVECs after as little as 4 hr of HSV expressed as a percentage of control wells that had been infection, and progressive reduction in activity occurred with incubated with medium. more prolonged (18-hr) infection (Fig. 2; *); mock infection Measurement ofTM Antigen. Polyclonal rabbit anti-bovine up to 18 hr provoked no loss of thrombin-mediated protein C TM was prepared as described (11). Iodination with Na125I activation when compared to untreated endothelium (data (15 mCi/ptg, Amersham; 1 mCi = 37 MBq) was performed by not shown). Significant, although lesser and more delayed, the iodogen method (12). reductions of protein C activation also accompanied HSV Monolayers of mock-infected or HSV-infected BAECs infection of BAECs (Fig. 2; o) perhaps reflecting the differ- were washed three times with ice-cold Hepes-buffered 0.9o ence in viral replication in the two endothelia (Fig. 1). NaCl containing 0.2% bovine serum albumin. A saturating Since protein C activation requires thrombomodulin, we concentration (0.4 ,ug/ml) of the iodinated anti-TM antibody assessed the surface expression of TM antigen on BAEC by was then added together with normal human immunoglobulin a specific binding assay using a polyclonal anti-bovine TM. (50 pug/ml). Cells were incubated with the labeled antibody After 18 hr of HSV infection the TM activity of BAECs as for 4 hr at 40C, after which the monolayer was washed and the