Antihemophilic Factor Antigen

LOCALIZATION IN ENDOTHELIALCELLS BY

IMMUNOFLUORESCENTMICROSCOPY

LEONW. Hoyw, RENEP. DELOS SANTOS, and JoHN R. Hoym

From the Department of Medicine, University of Connecticut School ofMedicine, Farmington, Connecticut 06032, and the Department of Pediatrics,University of Minnesota School of Medicine, Minneapolis, Minnesota 55455

A B S T R A C T The tissue localization of antihemophilicfactor (AHF, Factor VIII) has been determined byimmunofluorescent studies using monospecific rabbit anti-body to human AHF. Specific staining demonstratingAHFantigens has been identified in endothelial cells of awide range of human tissues. The staining pattern wasobserved in endothelial cells of arteries, capillaries, andveins as well as the cells lining hepatic and splenicsinusoids. Specific fluorescence was limited to these endo-thelial cells in sections of kidney, liver, spleen, lymphnode, cardiac and smooth muscle, thyroid, umbilical cord,and skin. Absorption studies established that the stain-ing was specific for cells in which there were proteinsthat had AHF antigens. The demonstration of fluores-cence within the cytoplasm of endothelial cells suggeststhat these cells synthesize proteins that have AHFantigens.

INTRODUCTIONAntihemophilic factor (AHF, Factor VIII)1 procoagu-lant activity is reduced in patients with classic hemo-philia (hemophilia A) and von Willebrand's disease.Despite considerable effort by many investigators in re-cent years, the cell type and/or organ responsible forthe production of this coagulation factor has not beenclearly established. Studies suggesting synthesis by liver

This work was presented in part to the 56th AnnualMeeting of the Federation of American Societies for Ex-perimental Biology, 10 April 1972.

Dr. J. R. Hoyer is a recipient of a National Institutes ofHealth Special Postdoctoral Fellowship (AM-50447).

Received for publication 17 April 1973 and in revised form2 July 1973.

1Abbreviations used in this paper: AHF, antihemophilicfactor (Factor VIII); PBS, phosphate-buffered saline; FITC,fluorescein isothiocyanate; VWF, von Willebrand's factor.

(1-5), spleen (2, 3, 6, 7), kidneys (8), lymphocytes(9), and macrophages (10) have been reported. Theseprevious investigations have reported AHF synthesisand/or release from perfused organs (2, 3, 7), specificcell types in tissue culture (9, 10), and in AHF-deficientanimals given organ transplants (4, 5). Such studies arelimited by the lability of AHF procoagulant activity,however, and by the difficulty in distinguishing changesin AHF activity from nonspecific shortening of clotting-time assays by other procoagulant materials. As a re-sult, numerous conflicting claims have not vet beenresolved.

The use of monospecific antibodies to antihemophilicfactor provides an alternative way to study AHF. It hasbeen recently demonstrated that antisera from rabbitsimmunized with highly purified AHF inactivate AHFprocoagulant activity (11-13), form stable complexeswith AHF (13), and react with AHFin inunumoprecipi-tation assays (11, 13). These properties suggest thatanti-AHF might also be an effective reagent for morpho-logic study of AHF distribution. Our studies (14), andthe concurrent independent investigations of Bloom, Gid-dings and Wilks (15), have verified this suggestion andhave identified AHF within endothelial cells in humantissues using indirect immunofluorescent microscopy. Thepresent report extends these findings to additional tissues,provides more detailed evidence for the specificity of thisstaining, and identifies with greater precision the locali-zation of the immunoreactive material.

METHODSPreparation of antisera. Rabbit anti-AHF was obtained

from New Zealand albino rabbits immunized with highlypurified AHF (80-100 U/mg protein) prepared from freshnormal human plasma by precipitation at - 3C with 3%ethanol, precipitation with 10% polyethylene glycol, and gelfiltration through Sepharose 4B (Pharmacia Fine Chemi-

The Journal of Clinical Investigation Volume 52 November 1973 2737-2744 2737

cals, I1nc., Piscataway, N. J.) following the method ofZimmerman, Ratnoff, and Powell (11 ). The antiserumuse(l in the studies reported here was obtained after a totalof four injections over a 7 month period, and it was ab-sorbed with an AHF-poor fraction of normal human plasma(13). A single line was identified when this absorbed serumwas then tested by Ouchterlony gel diffusion against AHF-rich concentrates (3% ethanol at -30C with suspensionof the precipitate in 1loth of the original volume) of normaland hemophilic plasmas, but no lines were observed whenthis absorbed antiserum was tested with plasma concen-trates from patients with severe von Willebrand's disease.Immunoelectrophoresis of concentrates from normal andhemophilic plasmas demonstrated a single line with p-mobility; these findings are identical to those reported byZimmerman and co-workers (11). This monospecific ab-sorbed antiserum had the same capacity to inactivate AHFactivity in normal human plasma as did the rabbit serumfrom which it was prepared. Further details of antigenpreparation, immunization schedule, and absorption proce-dure have been reported (13).

This monospecific antiserum or a globulin fraction pre-pared from the serum by precipitation with 50% saturatedammonium sulfate followed by dialysis against 0.01 Mphosphate, 0.14 M NaCI, pH 7.5 (PBS) was diluted inPBS before immunofluorescent staining using the indirecttechnique. Entirely consistent results were obtained whetherserum (diluted 1: 16) or globulin (0.2-0.8 mg/ml) wasused.

Iommunofluorescent methods. The studies were carriedout in two separate laboratories using immunofluorescentmethods (16, 17) that are identical except as noted below.5-mm blocks of tissue obtained by surgical biopsy or atautopsy within 6 h of death were rapidly f rozen on asmall cellulose sponge in isopentane prechilled in liquidnitrogen or in a dry ice-ethanol bath and stored at - 70'Cuntil used (less than 3 mo). Sections were cut in a cryo-stat, air dried, and stored at -20'C for periods as longas 3 mo. Smears prepared from peripheral blood, aspiratesof bone marrow, and platelet-rich plasma were air dried,fixed in acetone for 10 min, and stored at - 20'C for 1 wkor less.

Fluorescent staining of tissue and smears was performedaccording to the method of Ortega and Mellors (18) usingthe indirect technique. Sections were air dried, washedthree times with PBS (5 min/wash), and reacted withunlabeled antisera for 30 min. They were then washedthree times with PBS (2 min/wash) and treated withfluorescein isothiocyanate (FITC) -conjugated goat anti-rabbit gamma globulin for 45 min. The sections were againwashed three times with PBS (2 min/wash) and thencovered with a drop of 10% glycerin, 90% PBS, and aglass cover slip. Longer periods of washing with PBSbefore and after exposure to the same unlabeled antiserumin one laboratory (16) did not affect the pattern of fluo-rescence. All incubations and washes were carried out atroom temperature.

The goat antirabbit globulin was prepared in goats im-munized with rabbit IgG that had been isolated by DEAE-cellulose chromatography. An IgG fraction of this goatantiserum, also obtained by DEAE-cellulose chromatogra-phy, was labeled with FITC (19) and rechromatographedusing DEAE-cellulose (20). A single line was identifiedwhen the antiserum was tested with rabbit serum by im-munoelectrophoresis, and it did not react with humanserum in Ouchterlony gel diffusion. The F/P ratio (milli-

grams FITC/milligram protein) of the FITC-goat anti-rabbit globulin used in these studies was 7.3 X 10-'. TheFITC-goat antirabbit globulin was absorbed with humanIgG (3 mg/ml antiserum) before use at a final concentra-tion of 0.5-0.9 mg/ml. Similar immunofluorescence observa-tions were made using FITC-goat antirabbit globulin ob-tained from Hyland Div. (Travenol Laboratories, Inc.,Costa Mesa, Calif.) that was absorbed with lyophilizedmouse liver powder before use at a 1: 16 dilution.

Stained sections were examined in a Leitz (E. Leitz, Inc.,Rockleigh, N. J.) Ortholux photomicroscope using an HBO200 XV light source, a UG1 (2 mm) excitation filter andLeitz UV absorbing barrier filter or with a Zeiss (CarlZeiss, Inc., New York) microscope using a similar lightsource and a BG 12 excitation filter. In some instance, anFITC interference filter was used. Photographs were takenusing Tri-X Pan film (Eastman Kodak Co., Rochester,N. Y.). Selected tissue sections were examined by phasecontrast immunofluorescent microscopy as well using aZeiss Ultraphot II microscope with indirect immunofluores-cent illumination. This permitted more precise definition ofthe distribution of positive immunofluorescence. For thefurther identification of tissue, marrow, or blood cell ele-ments, cover slips were removed after photographs of im-munofluorescent staining were obtained. Sections or smearswere then restained using hematoxylin-eosin for tissue and\Wright's stain for bone marrow and blood smears.

Specificity of altisera. The specificity of staining forAHF was evaluated by absorption studies. Anti-AHFglobulin (0.05 ml of a globulin preparation that had aprotein concentration of 13.2 mg/ml) was incubated with0.25-0.5 ml of the absorbing material at 370C for 1 h andat 40C overnight. After the precipitate was removed bycentrifugation (2,400 g for 30 min), the absorbed anti-AHFwas used for immunofluorescent studies at the same dilutionas anti-AHF to which PBS had been added instead of theabsorbing material.

Ethanol concentrates were prepared from 5 ml of plasmaby slowly adding 80% ethanol to a final concentration of3',Z and holding the material at -3°C for 30 min beforecentrifuging (17,000 g for 20 min at -3° C). The materialused in absorption studies was dissolved in 0.15 M NaClto a final volume of 0.25 ml. A concentrate of normalhuman serum was obtained by precipitating proteins withone-third saturated ammonium sulfate and dialyzing theredissolved material against 0.15 M NaCl. The globulinobtained f rom 5 ml of serum was concentrated to 0.3 mlin this manner. Absorptions of the anti-AHF were, there-fore, carried out with concentrates prepared from 100 timesthe volume of the antibody globulin preparation. Purifiedhuman fibrinogen prepared by the method of Takeda (21)contained greater than 95% coagulable protein and wasused at a ratio of 29 mg/ml antibody globulin. Highly puri-fied human AHF obtained by the method of van Mourikand Mochtar (22) was used at a ratio of 1 mg/ml antibodyglobulin. A comparison of the AHF procoagulant activity(1,400 U/100 ml) with protein content (0.10 mg/ml) indi-cates that the AHF preparation was approximately 10,000-fold purified when compared with normal plasma.

Control studies also included tissues incubated with rabbitantifibrinogen. Monospecific rabbit antihuman fibrinogen(Hyland Div., Travenol Laboratories) used for indirectfluorescent staining gave results consistent with those ob-tained when monospecific FITC-conjugated rabbit anti-human fibrinogen, prepared as previously described (17),was used for direct staining. Both antifibrinogen sera

2738 L. W. Hoyer, R. P. de los Santos, and J. R. Hoyer

q

FIGURE 1 Immunofluorescent micrographs of tissue sections stained with rabbit anti-AHF.(A) Low power micrograph of cardiac muscle with positive staining of venules and capillariescut longitudinally and in cross section (X 150). (B) Higher power micrograph of the samecardiac muscle section. The positive staining is limited to endothelial cells of the smallvenule (X 350). (C) Renal cortex with positive staining of peritubular capillaries (X 500).(D) Lymph node venule showing positive staining of endothelial cell cytoplasm. The sur-rounding lymphoid cells are negative (X 1,200).

formed single immunoprecipitin lines when tested withhuman plasma by Ouchterlony immunodiffusion analysisand by immunoelectrophoresis.

RESULTSSections of liver, spleen, lymph node, kidney, skeletaland cardiac muscle, thyroid, skin, umbilical cord, aorta,and portal vein were examined using indirect immuno-fluorescent microscopy. Specimens that were obtained

by surgical biopsy from individuals with no history ofbleeding or thrombotic disorders included skin, kidney,muscle, and spleen. These included 20 kidney specimensobtained from living, related kidney-transplant donorsimmediately after removal of the kidney for transplan-tation. These individuals gave specific informed consentfor research studies utilizing the biopsy material. Addi-tional material was obtained at autopsy of patients with

Antihemophilic Factor Antigen in Endothelial Cells

va

2739D

no history of bleeding or thrombotic disorders. Sectionsfrom five livers, four spleens, four lymph nodes, twoumbilical cords, and two skin biopsies were examined.One sample was examined for each of the other tissueslisted above.

In all tissues examined, immunofluorescent stainingdemonstrating AHF antigenic material was restricted toendothelial cells (Figs. 1 and 2). This endothelial locali-zation was apparent in arteries, capillaries, and venulesand was observed in sections of aorta and portal vein.The distribution of positive immunofluorescence withinendothelial cells was cytoplasmic and in most cases ap-peared to be granular. The nuclei of these cells did notstain and were seen as negative images (Fig. iD). Inthose tissues in which the endothelial cytoplasm is verythin (e.g., arteries and some glomerular capillaries), theimmunofluorescent staining appeared more linear (Fig.2A). The density of endothelial cells within the red pulpof the spleen made it difficult to be absolutely certain thateach positively staining cell was endothelial. Phase-contrast immunofluorescence improved the definition ofthe distribution, however, and at least 95% of positivelystaining cells in the spleen were identified as endothelialcells. Staining was not identified within vascular lumens.

Lymphocytes and plasma cells in lymph node andspleen sections did not show positive staining for AHF

(Fig. ID). Similar findings were observed in smearsprepared from peripheral blood and from aspirated bonemarrow. Neither normal marrow lymphocytes nor periph-eral blood lymphoblasts from a patient with acute lympho-blastic leukemia were stained by the rabbit anti-AHF.Megakaryocytes were the only marrow cells showingpositive staining for AHF. Similarly, only platelets werestained when peripheral blood smears were examined.Macrophages identified in marrow smears were not stained.

The immunofluorescent staining of renal tissue sectionswas restricted to endothelial cells and was prominent inarteries (Fig. 2A) and arterioles, venules, and capil-laries (Fig. 1C). Within glomeruli the staining, re-stricted to endothelial cells, was linear or granular withinthe cytoplasm according to the plane on which the sec-tion was cut. Sections of liver had prominent fluorescentstaining of veins in portal triads as well as sinusoidallining cells, but parenchymal cells were not stained byrabbit anti-AHF. Control experiments with rabbit anti-fibrinogen demonstrated fluorescence restricted to liverparenchymal cells and there was no staining of endo-thelial cells.

The pattern of immunofluorescence was generally con-sistent from tissue to tissue and within sections. Al-though there was some variability in the intensity ofstaining for AHF, positive endothelial staining was ob-

FIGURE 2 Cross section of small muscular arteries in the renal cortex. (A) Section stainedwith rabbit anti-AHF demonstrating positive staining of endothelial cells lying on the luminalside of the internal elastic lamina. The autofluorescence of the internal elastic lamina shouldbe distinguished from the granular endothelial cell staining. The autofluorescence is presentin control sections that have not been incubated with FITC-conjugated antisera and is identifiedin stained tissues by its blue color which is easily distinguished from the apple green color offluorescein-stained material (X 800). (B) Section stained with rabbit antihuman fibrinogen.There is no positive staining of endothelial cells. Prominent autofluorescence of the internalelastic lamina is also present in this artery which is from a different field from the sametissue section (X 800).

2740 L. W. Hoyer, R. P. de los Santos, and J. R. Hoyer

served in all tissue sections examined. Specimens ob-tained by surgical biopsy had the same immunofluorescentpatterns as those in tissue obtained at autopsy. Althoughthere were minor differences in immunofluorescent meth-ods, the staining patterns observed in the two labora-tories did not differ.

The specificity of the immunofluorescent staining wasestablished in a series of control studies. After absorptionof the anti-AHF with 0.075 mg of highly purified AHF/mg IgG, the antiserum no longer stained tissues thatstrongly reacted with unabsorbed anti-AHF (Table I).Positive staining of endothelial cells was also completelyblocked by absorption of rabbit anti-AHF with concen-trates prepared from normal human plasma, normal se-runm, and from plasma of a patient with hemophilia A.In contrast, absorption with an ethanol concentrate ofplasma from a patient with severe von Willebrand's dis-ease (AHF activity less than 2 U/100 ml [13]) did notdecrease positive staining of endothelial cells. Absorptionwith purified human fibrinogen (2.2 mg/mg antibodyIgG) also failed to affect endothelial cell staining.

Tissue sections incubated with normal rabbit serum,normal rabbit IgG, rabbit antihuman fibrinogen, or PBSfollowed by FITC-goat antirabbit globulin showed nopositive staining of endothelial cells. Hepatic paren-chymal cells were stained when rabbit antifibrinogen wasused, and absorption of this antibody with highly purifiedAHF failed to decrease the staining of parenchymal cellsfor fibrinogen. The absorption studies are summarizedin Table I.

DISCUSSIONThe immunofluorescent microscopic studies reported heredemonstrate the localization of AHF antigen in endo-thelial cells throughout the body. The specific stainingwas detected in the endothelial lining of arterial andvenous vessels as well as in capillaries and cells liningsplenic and hepatic sinusoids. No immunofluorescentstaining was identified in lymphocytes or macrophages-cells previously considered possible sources of AHF-or in hepatic or renal parenchymal cells. The similardistribution in biopsy and autopsy material is evidencethat the localization is not related to diffusion of AHFinto-or adsorption onto-nonviable endothelial cells.The staining pattern, cytoplasmic rather than surface, isadditional support for this conclusion.

Control studies (Table I) established that the stainingwas specific and that the rabbit anti-AHF has the sameproperties in immunofluorescence microscopy as it doesin immunoprecipitation (11), antibody neutralization(11), and radioimmunoassays (13). The antibody reactswith-or is absorbed by-normal and hemophilic plasmaand serum and by highly purified AHF. It does not reactwith plasma from patients with severe von Willebrand's

TABLE I

Effect of Antibody Absorption on EndothelialCell Fluorescence

Endothelialcell immuno-

Antiserum Absorbing antigen* fluorescence:

Anti-AHF +Anti-AHF Purified AHFAnti-AHI I ibrinogen +Aiiti-Afif1 Normal plasmaAnti-AHFI Hemophilic plasmaAnti-AHF von \Willebraiid's

disease plasma +Anti-AHF Normal serumNormal rabbit

seriim -_Antifibriniogei-

* The properties of the absorbing antigens and the ratios ofantigen to anti-AHF are specified in the text.$All sections were incubated with FITC-goat antirabbitglobulin after the absorbed rabbit antiserum.

disease or with plasma proteins other than AHF, e.g.,fibrinogen.

Although the control studies indicate that the rabbitantibody specifically stains endothelial cells, the con-clusion that this represents the identification of AHFantigen in these cells requires evidence that the rabbitantibody does, in fact, specifically react with the samemolecule in normal human plasma as that which isresponsible for AHFprocoagulant activity.

Four kinds of evidence support this conclusion. Themost direct is the demonstration that this antiserumforms a single line in immunodiffusion studies withhighly purified AHF and with AHF-rich concentrates ofnormal human plasma (11, 13). By immunochemicalcriteria, this antiserum has a single specificity. No pre-cipitin line is detected, however, if the rabbit antiserumis tested with concentrates prepared from plasmas ofpatients with severe von Willebrand's disease (11). Theparallel reduction of AHF procoagulant activity andAHF antigenic material in von Willebrand's diseaseplasmas is additional evidence that this antibody reactswith the AHF molecule (11, 13). The similar physico-chemical properties of AHF procoagulant activity andthe antigen identified by rabbit anti-AHF provide athird kind of evidence for the specificity of the rabbitantibody (13).

Unambiguous evidence for the antibody's specificityhas recently been obtained by immunizing rabbits withwashed immunoprecipitates formed by incubating concen-trates of normal human plasma with the rabbit anti-AHF(23). Postimmunization sera from these animals hadantibodies that inactivate AHF procoagulant activity.

Antihemophilic Factor Antigen in Endothelial CeUs 2741

This observation demonstrates the presence of proteinswith AHF determinants in the immunoprecipitates. Thespecificities identified by anti-AHF are, therefore, de-signated "AHF antigens," and the proteins with whichthey react (if nonfunctional) are recognized as being"AHF-like."

It is possible, of course, that the more sensitive im-munofluorescent method might identify immunoreactivematerial that cannot be detected in immunoprecipitationstudies. The absorption studies using concentrates ofnormal and von Willebrand's plasmas strongly suggest,however, that the immunofluorescent pattern reflects tis-sue localization of AHFantigens.

The molecule identified by rabbit anti-AHF appearsto have a second hemostatic role, that of "von Wille-brand's factor" (VWF), in addition to its AHF pro-coagulant function. Defective platelet function-and aprolonged bleeding time-in von Willebrand's disease isattributed to VWFdeficiency and fractions of normalplasma correct these abnormalities in vivo and in vitro(24-28). As AHF procoagulant activity and VWFareusually present in the same fractions obtained from nor-mal plasma, it has been suspected for many years thatthe two activities might be present on a single molecule.Bouma, Wiegerinck, Sixma, van Mourik, and Mochtarhave recently obtained additional immunologic evidencethat closely links AHF and VWF (29). They demon-strated that high molecular weight fractions prepared byagarose gel filtration of normal or hemophilic plasmacorrect the reduced platelet retention (in glass-bead fil-ters) of blood from patients with von Willebrand's dis-ease and that these are the same fractions that have thehighest concentrations of AHFprocoagulant activity andAHFantigen. Moreover, rabbit anti-AHF reduces plate-let retention of normal blood (29).

Weiss, Rogers, and Brand, using an in vitro assayfor VWFbased on ristocetin-induced platelet aggre-gation, have also demonstrated VWFin high molecularweight fractions obtained by agarose gel filtration ofcryoprecipitates prepared from normal and hemophilicplasmas (30). Rabbit anti-AHF also inhibits VWFac-tivity measured in this way (30). It appears likely,therefore, that the hemostatic defects in von Wille-brand's disease follow from the reduced concentrationof a single molecule (or a relatively stable complex ofmolecules) that has both AHF and VWFactivity. Itis recognized that the close correlation of VWFandthe antigenic determinants detected by rabbit anti-AHFhas important implications for the studies reported here.The tissue localization of both VWFand AHF are

identified by the endothelial cell immunofluorescence.While immunofluorescent studies cannot establish that

the intact and functional AHFor VWFare present in-or synthesized by-the endothelial cells, they do provide

a guide for more probing studies. Human endothelialcells in tissue culture provide a more direct way to studythe relationship of AHFand VWFto this cell type (31).The accompanying paper reports studies that demon-strate positive immunofluorescence in cultured humanendothelial cells (but not other tissue culture cells) whenthey are tested with rabbit anti-AHF (32). In addition,proteins that have AHF antigenic determinants aresynthesized by these endothelial cells (32). In conjunc-tion with these tissue culture studies, the immunofluores-cent histologic observations strongly suggest that theendothelial cell synthesizes proteins that have AHFantigens.

Bloom, Giddings, and Wilks have suggested an al-ternative interpretation, the adsorption of AHF fromiplasma by endothelial cells, for their similar indirectimnmnunofluorescent results (15). A thin but well-demar-cated layer of immunoreactive material lining the endo-thelium of large and small arteries and veins was iden-tified using rabbit anti-AHF (15). Our findings weresimilar when sections were observed at low magnifica-tions, but it is apparent in higher resolution photomicro-graphs that fluorescent staining is detected throughoutthe cytoplasm of endothelial cells (Figs. 1B and ID).This cytoplasmic distribution would appear more con-sistent with AHF antigen synthesis by endothelial cells,an interpretation supported by tissue culture data (32).

Bloom and co-workers also detected, but did not il-lustrate, immunofluorescence in cytoplasm of mononu-clear cells of the spleen and lymph nodes and perceivedfaint and uncertain staining of hepatic parenchymal cells.Lymphoid and hepatic parenchymal cells showed nopositive staining for AHF in our studies. Although therabbit antisera to AHFwere prepared by generally simi-lar methods, differences in antigen preparation and ab-sorption may account for differences in the localizationof immunofluorescence reported.

An earlier immunofluorescent study of AHF withintissues reported localization in bovine splenic lymphocytesand hepatic parenchymal cells (33). That report did notinclude criteria establishing the specificity of the anti-body, however, and it is possible that the findings wererelated to other antigens in the immunizing material.

The localization of AHF antigen in the endothelialcell is consistent with a number of tissue extraction andorgan perfusion studies in which AHF has been identi-fied by assays of procoagulant activity (2, 3, 6-8) andan immunologic study in which antibody neutralizationwas measured (34). Vascular organs-liver, kidney, andspleen-have been reported to be the source of AHF inthese studies, as would be expected if the appearance ofAHF in extracts or perfusates is related to the synthesisand release of AHF by endothelial cells. Although tis-sue culture studies of lymphocytes and fibroblasts have

2742 L. W. Hoyer, R. P. de los Santos, and J. R. Hoyer

been interpreted as indicating AHF synthesis by thesecells, immunologic determinations have established thatthe procoagulant material in cultured cells is not AHFand that the shortened coagulation times are due to tis-sue factor (35-37).

The endothelial cell localization of AHF antigen is ofconsiderable interest in light of the marked lability ofplasma AHF procoagulant activity values. Rapid in-crease is noted after exercise, injection of epinephrine,neurologic procedures, surgery, and a range of otherstressful stimuli (38). Parallel increases in AHF anti-gen values have been identified in many of these condi-tions (39-40). The mechanism by which these stimulicause increased plasma AHF levels is not known.

It may be significant that tissue thromboplastin hasalso been identified in endothelial cells (41). The vas-cular localization of these procoagulant materials, andpossibly others as well, would expedite their rapid in-teraction in the hemostatic sequence and may, therefore,have considerable survival value. It is also possible thattheir endothelial localization is an important factor in thedevelopment of thrombosis. If minimal changes in theendothelial cells effect significant alterations in the bloodcoagulation system, these cells would have a central rolein the pathogenesis of thrombotic diseases. Further stud-ies of the procoagulant functions of endothelial cells areneeded to consider this possibility.

ACKNOWLEDGMENTSWe thank Dr. Naomi F. Rothfield for use of her laboratoryfacilities to carry out some of the immunofluorescent studiesand Mrs. Ruth Chase for assistance in the preparation ofthe manuscript.

These studies were supported in part by a Veterans Ad-ministration Research Grant.

REFERENCES1. Gardikas, C., P. Bakaloudis, J. Hatzioannou, and D.

Kokkinos. 1965. The factor VIII concentration of he-patic venous blood. Br. J. Haematol. 11: 380.

2. Norman, J. C., J. P. Lambilliotte. Y. Kojima, and H.S. Sise. 1967. Antihemophilic factor release by perfusedliver and spleen: relationship to hemophilia. Science(Wash. D. C.). 158: 1060.

3. Dodds, W. J. 1969. Storage, release and synthesis ofcoagulation factors in isolated perfused organs. Am. J.Physiol. 217: 879.

4. Marchioro, T. L., C. Hougie, H. Ragde, R. B. Epstein,and E. D. Thomas. 1969. Hemophilia: the role of organhomografts. Science (Wash. D. C.). 163: 188.

5. Webster, W. P., C. F. Zukoski, P. Hutchin, R. L.Reddick, S. R. Mandel, and G. D. Penick. 1971. Plasmafactor VIII synthesis and control as revealed by canineorgan transplantation. Am. J. Physiol. 220: 1147.

6. Pool, J. G. 1966. Antihemophilic globulin (AHG, factorVIII) activity in spleen. Fed. Proc. 25: 317.

7. Webster, W. P., R. L. Reddick, H. R. Roberts, andG. D. Penick. 1967. Release of factor VIII (antihemo-

philic factor) from perfused organs and tissues. Nature(Lond.). 213: 1146.

8. Barrow, E. M., and J. B. Graham. 1968. Kidney anti-hemophilic factor: partial purification and some prop-erties. Biochemistry. 7: 3917.

9. Bouhasin, J. D., P. Monteleone, and C. Altay. 1971.Role of the lymphocyte in antihemophilic globulin pro-duction: a rise in antihemophilic globulin levels in ahemophilic subject with acute lymphoblastic leukemia.J. Lab. Clin. Med. 78: 122.

10. Ponn, R. B., F. A. Kellogg, J. M. Korft, C. A. S.Pegg, H. S. Sise, and J. C. Norman. 1971. The role ofthe splenic macrophage in antihemophilic factor (factorVIII) synthesis. Arch. Surg. 103: 398.

11. Zimmerman, T. S., 0. D. Ratnoff, and A. E. Powell.1971. Immunologic differentiation of classic hemophilia(factor VIII deficiency) and von Willebrand's disease.J. Clin. Invest. 50: 244.

12. Stites, D. P., E. J. Hcrshgold, J. D. Perlman, andH. H. Fudenberg. 1971. Factor VIII detection byhemagglutination inhibition: hemophilia A and vonWillebrand's disease. Science (Wash. D. C.). 171: 196.

13. Hoyer, L. W. 1972. Immunologic studies of antihemo-philic factor (AHF, factor VIII). IV. Radioimmuno-assay of AHF antigen. J. Lab. Clin. Med. 80: 822.

14. de los Santos, R. P., and L. W. Hoyer. 1972. Anti-hemophilic factor in tissue: Localization by immuno-fluorescence. Fed. Proc. 31: 262. (Abstr.)

15. Bloom, A. L., J. C. Giddings, and C. J. Wilks. 1973.Factor VIII on the vascular intima: possible impor-tance in hemostasis and thrombosis. Nat. Newo Biol.241: 217.

16. Gonzalez, E. N., and A. F. Rothfield. 1966. Immuno-globulin class and pattern of nuclear fluorescence insystemic lupus erythematosus. N. Engi. J. Mled. 274:1333.

17. Michael, A. S., K. N. Drummond, R. A. Good, andR. L. Vernier. 1966. Acute poststreptococcal glomerulo-nephritis: immune deposit disease. J. Clini. Invest. 45:237.

18. Ortega, L. G., and R. C. Mellors. 1956. Analyticalpathology. IV. The role of localized antibodies in thepathogenesis of nethrotoxic nephritis in the rat. J. Exp.Mled. 104: 151.

19. Wood, B. T., S. H. Thompson, and G. Goldstein. 1965.Fluorescent antibody staining. III. Preparation of fluo-rescein-isothiocyanate-labeled antibody. J. Immunol. 95:225.

20. Gajl-Peczalska, K. J., A. J. Fish, H. J. Meuwissen, D.Frommel, and R. A. Good. 1969. Localization of im-munological complexes fixing p,1C (C3) in germinalcenters of lymph nodes. J. Exp. Med. 130: 1367.

21. Takeda, Y. 1966. Studies of the metabolism and dis-tribution of fibrinogen in healthy men with autologous'"I-labeled fibrinogen. J. Clin. Invest. 45: 103.

22. van Mourik, J. A., and I. A. Mochtar. 1970. Purifica-tion of human antihemophilic factor (factor VIII)by gel chromatography. Biochim. Biophys. Acta. 221:677.

23. Hoyer, L. W. 1973. Specificity of precipitating anti-bodies in immunological identification of antihaemophilicfactor. Nature (Loiid.). In press.

24. Nilsson, I. M., M. Blombdck, E. Jorpes, B. Blombiick,and S. Johansson. 1957. V. von Willebrand's disease andit correction with human plasma fraction 1-0. ActaMAed. Scand. 159:179.

Antihemophilic Factor Antigen in Endothelial Cells 2743

25. Biggs, R., and J. M. Matthews. 1963. The treatmentof haemorrhage in von Willebrand's disease and theblood level of factor VIII (AH.G). Br. J. Haematol.9: 203.

26. Cornu, P., M. J. Larrieu, J. Caen, and J. Bernard.1963. Transfusion studies in von Willebrand's disease:effect on bleeding time and factor VIII. Br. J. Haema-tol. 9: 189.

27. Zucker, M. B. 1963. In vitro abnormality of blood invon Willebrand's disease correctable by normal plasma.Nature (Lond.). 197: 601.

28. Weiss, H. J., and J. Rogers. 1972. Correction of theplatelet abnormality in von Willebrand's disease by cryo-precipitate. Amn. J. Med. 53: 734.

29. Bouma, B. N., Y. Wiegerinck, J. J. Sixma, J. A. vanMourik, and I. A. Mochtar. 1972. Immunological char-acterization of purified anti-haemophilic factor A (fac-tor VIII) which corrects abnormal platelet retention invon Willebrand's disease. Nat. New Biol. 236: 104.

30. Weiss, H. J., J. Rogers, and H. Brand. 1973. Defectiveristocetin-induced platelet aggregation in von Wille-brand's disease and its correction by factor VIII. J.Clin. Invest. 52: 2697.

31. Jaffe, E. A., R. L. Nachman, C. G. Becker, and C. R.Minick. 1973. Culture of human endothelial cells de-rived from umbilical veins. Identification by morpho-logic and immunologic criteria. J. Clin. Invest. 52: 2745.

32. Jaffe, E. A., L. W. Hoyer, and R. L. Nachman. 1973.Synthesis of antihemophilic factor antigen by culturedhuman endothelial cells. J. Clin. Invest. 52: 2757.

33. Barnhart, M. I. 1967. Immunochemistry. Immuno-

chemistry in Blood Clotting Enzymology. W. H. See-gers, editor. Academic Press, Inc., New York 217.

34. Bloom, A. L., and J. C. Giddings. 1972. Factor VIII(antihaemophilic factor) in tissues detected by antibodyneutralization. Br. J. Haematol. 23: 157.

35. Green, D., C. Ryan, N. Malandrucolo, and H. L. Nad-ler. 1971. Characterization of the coagulant activity ofcultured human fibroblasts. Blood. 37: 47.

36. Zacharski, L. R., L. W. Hoyer, and 0. R. McIntyre.1973. Immunologic identification of tissue factor (throm-boplastin) synthesized by cultured fibroblasts. Blood.41: 671.

37. Rickles, F. R., J. A. Hardin, F. A. Pitlick, L. W.Hoyer, and M. E. Conrad. 1973. Tissue factor activityin lymphocyte cultures from normal individuals andpatients with hemophilia A. J. Clin. Invest. 52: 1427.

38. Green, D. 1970. Factor VIII (anti-hemophilic Factor).J. Chronic Dis. 23: 213.

39. Bennett, B., and 0. D. Ratnoff. 1972. Changes in anti-hemophilic factor (AHF, factor VIII) procoagulantactivity and AHF-like antigen in normal pregnancy,and following exercise and pneumoencephalography.J. Lab. Clin. Med. 80: 256.

40. Prentice, C. R. M., C. D. Forbes, and S. M. Smith.1972. Rise of factor VIII after exercise and adrenalineinfusion, measured by immunological and biologicaltechniques. Thromb. Res. 1: 493.

41. Zeldis, S. M., Y. Nemerson, F. A. Pitlick, and T. L.Lentz. 1972. Tissue factor (thromboplastin): localiza-tion to plasma membranes by peroxidase-conjugatedantibodies. Science (Wash. D. C.). 175: 766.

2744 L. W. Hoyer, R. P. de los Santos, and J. R. Hoyer