Cellular Localization of Thrombomodulin in Human Epithelium ... - NCBI

American Journal of Pathology, Vol. 146, No. 4, April 1995 Copyright X) American Society for Investigative Pathology

Cellular Localization of Thrombomodulin in Human Epithelium and Squamous Malignancies

Donna J. Lager,* Edward J. Callaghan,* Stephen F. Worth,* Thomas J. Raife,* and Steven R. Lentzt From the Departments of Pathology* and Internal Medicine,t The University of Iowa College of Medicine,

Iowa City, Iowa

Thrombomodulin is a ceUl surface glycoprotein that functions as an anticoagulant. Although initialy identified on endothelial ceUs, thrombomodulin is also expressed by other vascular ceUs, by mesothelial ceUs, and by epidermal keratinocytes. To determine whether thrombomodulin is expressed by epithelial ceUs in locations other than skin, we conducted a survey of thrombomodulin protein and mRNA in human epithelium. Thrombomodulin protein was detected by immunohistochemistry in aUl samples containing stratified squamous epithelium, including oral mucosa, larynx, esophagus, uterine ectocervix, and vagina. In these tissues, thrombomodulin staining localized to the suprabasal layer, with minimal staining observed in the basal or superficial layers of epithelium. Thrombomodulin was not detected in cuboidal, simple columnar, or pseudostrattifed columnar epithelium and was detected variably in transitional epithelium. Thrombomodulin staining was also observed in 21 of 26 cases of invasive squamous ceU carcinoma and in several examples of squamous carcinoma-in-situ and squamous metaplasia. Expression of thrombomodulin mRNA was confirmed by in situ hybridization in both normal and malignant squamous epithelium. FuUl-length, functionally active thrombomodulin was demonstrated in cultured squamous epithelial ceUs. These data demonstrate that thrombomodulin expression correlates with the squamous phenotype and suggest that hemostasis is regulated by compartmentalization ofprocoagulant and anti-

coagulant epithelial proteins. 1995, 146:933-943)

(Am J Pathol

Thrombomodulin is a membrane-bound glycoprotein that functions within blood vessels as an endogenous anticoagulant. Thrombomodulin inhibits the procoagulant properties of thrombin and accelerates activation of protein C, a potent antithrombotic protease.1-3 Thus, when constitutively expressed on the luminal surface of endothelial cells, thrombomodulin functions to localize coagulation to sites of vascular injury. Abnormalities of the protein C anticoagulant pathway are present in up to 50% of patients with venous thromboembolism4,5 and have been implicated in the pathogenesis of arterial thrombosis, disseminated intravascular coagulation, and cutaneous necrosis.67 Although initially identified as an endothelial cell protein,8 thrombomodulin is now known to be expressed by several other types of cells. Within the vascular system, thrombomodulin has been detected in platelets,9 megakaryocytes,10 monocytes,11 neutrophils,12 and placental syncytiotrophoblast.13 Thrombomodulin antigen has been demonstrated on mesothelial surfaces lining body cavities, where it may function to prevent fibrin deposition.14 Thrombomodulin is also expressed by suprabasal keratinocytes within human epidermis, an extravascular tissue. 15,16 Keratinocyte thrombomodulin is functional as an anticoagulant, and its expression correlates with keratinocyte differentiation.16 These observations suggest that thrombomodulin regulates hemostasis in both vascular and extravascular locations. To determine whether thrombomodulin is also synthesized by other types of epithelial cells, we examined tissue sections of human epithelium by immunohistochemistry and in situ hybridization. Our results demonstrate that thrombomodulin is expressed in both keratinized and nonkeratinized stratified squamous epithelium but not in simple or pseudostratified Supported in part by National Institutes of Health Grants T32 HL07344 and P30 DK-25295 and by American Cancer Society Grant IN-122N, administered through the University of Iowa Cancer Center. Accepted for publication January 12, 1995. Address reprint requests to Dr. Donna J. Lager, University of Iowa Hospitals and Clinics, Department of Pathology, 200 Hawkins Drive, 5239D RCP, Iowa City, IA 52242.

933

934

Lager et al

AJP April 1995, Vol. 146, No. 4

epithelium. Thrombomodulin was also expressed in many examples of invasive squamous cell carcinoma, squamous carcinoma-in-situ, and squamous metaplasia and in cultured squamous epithelial cells. These data demonstrate that thrombomodulin expression is a consistent feature of the squamous phenotype and suggest that extravascular hemostatic pathways are regulated by compartmentalization of thrombomodulin within the epithelium.

Materials and Methods Tissue Sources Formalin-fixed, paraffin-embedded tissue blocks containing normal or malignant epithelium were retrieved from the files of the Division of Surgical Pathology, University of Iowa. The tissues examined included keratinized stratified squamous epithelium (skin), nonkeratinized stratified squamous epithelium (oral mucosa, larynx, esophagus, uterine ectocervix, vagina), transitional epithelium (ureter, urinary bladder), cuboidal epithelium (salivary acini and ducts, mammary lobules and ducts, thyroid follicles, pancreatic acini, renal tubules, fallopian tube, seminiferous tubules), simple columnar epithelium (stomach, intestine, endometrium, endocervix, prostate), and pseudostratified columnar epithelium (tracheobronchial tree, vas deferens). Neoplasms examined included squamous cell carcinoma of the larynx, lung, esophagus, and uterine cervix and squamous cell carcinoma and transitional cell carcinoma of the urinary bladder. Supporting vasculature was present in each case examined. For in situ hybridization, fresh tissue sections of skin, uterine ectocervix, and squamous cell carcinoma were frozen and stored at -70 C.

Immunohistochemistry All sections were subjected to immunohistochemical staining with two mouse monoclonal antibodies specific for human thrombomodulin. Antibody EC13.3 was raised against full-length human thrombomodulin,17 and antibody TM1009 was raised against the epidermal growth factor-like domains of recombinant human thrombomodulin (Dako Corp., Carpinteria, CA). Both of these antibodies bind to human thrombomodulin with high affinity and inhibit protein C activation.18,19 Formalin-fixed, paraffin-embedded tissue sections were deparaffinized in xylene, rehydrated in graded alcohols to water, and rinsed in phosphate-buffered saline, pH 7.4 (PBS). Endogenous peroxidase activity was blocked with 0.3%

H202 in water for 30 minutes at room temperature. Nonspecific background staining was prevented by application of normal horse serum (Vector Laboratories, Burlingame, CA). Sections were covered with anti-thrombomodulin IgG (diluted 1:100 in PBS) for 2 hours at room temperature and rinsed in PBS. Sections were covered with biotinylated horse anti-mouse IgG for 30 minutes at room temperature, rinsed in PBS, and then covered with avidin-biotin-peroxidase complex (Vector Laboratories). After 30 minutes at room temperature, sections were rinsed in PBS and then incubated with 0.05% 3, 3'-diaminobenzidine tetrahydrochloride dihydrate to develop the peroxidase signal. A counterstain of 10% Harris hematoxylin was applied before coverslipping. Negative control slides were prepared by substituting normal mouse serum or isotype-matched mouse IgG1 (Dako Corp.) for the primary antibody. Specificity of staining was confirmed by preincubating the anti-thrombomodulin antibodies with a 10-fold molar excess of purified soluble recombinant thrombomodulin deletion mutant TM456.20 Staining intensity was scored as absent (0), weak (+), moderate (+ +), or strong (+ ++).

Probe Generation A plasmid for transcribing sense and antisense thrombomodulin riboprobes was generated by subcloning the 1671-bp Sacl-Not fragment of human thrombomodulin cDNA from pTMNC18 into pBluescriptll (Stratagene, La Jolla, CA). After linearization with Nod or Bgl II, sense and antisense thrombomodulin riboprobes were transcribed with RNA polymerases T3 and T7, respectively. Uridine 5'-a-[35S]-

thiotriphosphate (Amersham Corp., Arlington Heights, IL) was incorporated into the riboprobes during transcription. The radiolabeled riboprobes were then subjected to alkaline hydrolysis to generate fragments of approximately 200 bp.

In Situ Hybridization In situ hybridization was performed by using a modification of procedures described previously.21,22 Sections (5 to 10 p) of frozen tissue were cut at -20 C and mounted on glass slides that had been pretreated with 2% 3-aminopropyltriethoxysilane (Aldrich Chemical Co., Milwaukee, WI). The slides were allowed to come to room temperature and then fixed in 4% paraformaldehyde for 10 to 15 minutes. After several rinses in PBS, the sections were digested with 0.125 mg/ml Pronase E in 50 mmol/L Tris-HCI, pH 7.5,

Thrombomodulin in Human Epithelium 935 AJP April 1995, Vol. 146, No. 4

and 5 mmol/L EDTA for 10 minutes at room temperature and rinsed in PBS. After dipping in 0.1 mol/L triethanolamine-HCI, pH 8.0 (TEA buffer), the sections were acetylated for 10 minutes with 0.25% acetic anhydride in TEA buffer, dehydrated in graded alcohols, and air dried. Sections were covered with 35S-labeled antisense or sense riboprobes (250,000 cpm/slide) in 35 p1 of hybridization buffer (20 mmol/L sodium acetate, pH 5.0, 0.3 mol/L NaCI, 1 mmol/L EDTA, 10 mmol/L dithiothreitol, 50% formamide, 1X Denhardt's solution, 10% dextran sulfate, and 2.75 mg/ml yeast tRNA), sealed with a sialinized coverslip, and incubated overnight at 55 C. After hybridization, the slides were washed in 4X standard saline citrate buffer (SSC), incubated for 30 minutes at 37 C with 1.4 U/ml ribonuclease Ti and 20 pg/mI ribonuclease A in 10 mmol/L Tris-HCI, pH 7.5, 0.5 mol/L NaCI, and 1.0 mmol/L EDTA. The slides were rinsed in 10 mmol/L Tris-HCI, pH 7.5, 0.5 mmol/L NaCI, and 1.0 mmol/L EDTA and washed in 2X SSC containing 0.1% 2-mercaptoethanol for 30 minutes at room temperature, 0.1X SSC containing 0.1% 2-mercaptoethanol for 30 minutes at 56 C, and 0.1 X SSC containing 0.1 % 2-mercaptoethanol for 30 minutes at room temperature. Sections were then dehydrated in graded alcohols and air dried. The slides were dipped in a 50% solution of NTB-2 photographic emulsion (Eastman Kodak Co., Kingsport, TN) in water, developed after 3 weeks of exposure at 4 C, and counterstained with hematoxylin. The sections were examined under both bright and dark field illumination.

Cell Culture Neonatal human foreskin keratinocytes (HK) were isolated as described previously16 and cultured in serum-free keratinocyte growth medium (Clonetics Corp., San Diego, CA) containing 0.07 mmol/L calcium chloride. When the cells reached 80% confluence, the medium was changed to keratinocyte growth medium containing 1.4 mmol/L calcium chloride, and the cells were cultured for 48 hours before analysis. Human umbilical vein endothelial cells (HUVEC) were purchased from Clonetics Corp. and cultured in endothelial cell growth medium as described previously.23 C4-1 human cervical squamous cell carcinoma cells were obtained from the American Type Culture Collection (Rockville, MD) and cultured in Waymouth's 752/1 medium with 10% fetal bovine serum. CV-1 African green monkey kidney cells and TMnc cells expressing recombinant human thrombomodulin18 were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum.

Northern Hybridization Northern hybridization was performed by modification of a method described previously.24 Total cellular RNA was isolated from cultured cells by acid guanidinium thiocyanate-phenol-chloroform extraction25 and subjected to electrophoresis on 1% agaroseformaldehyde gels. After transfer to nylon membranes and immobilization by UV cross-linking, thrombomodulin and actin mRNAs were detected with probes randomly labeled with a-[32P]dCTP. The 1390-bp human thrombomodulin probe was generated from plasmid pTMNC18 by digestion with Psd, and the 21 00-bp human y-actin probe was generated from plasmid pHFyA-126 by digestion with BamHl.

Immunoblot Analysis Cells were washed three times with PBS and then incubated for 5 minutes at 23 C in 20 mmol/L Tris-HCI, pH 7.5, 150 mmol/L NaCI, 0.6% Triton X-100, and 10 mmol/L iodoacetamide. Cell lysates were centrifuged at 12,000 x gfor 5 minutes, and supernatant fractions containing 50 pg of total protein were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis under nonreducing conditions with 10% polyacrylamide separating gels.27 Proteins were transferred to Hybond-ECL nitrocellulose membranes (Amersham Corp., Arlington Heights, IL) and blocked with 5% nonfat dried milk in 10 mmol/L TrisHCI, pH 8.0,150 mmol/L NaCI, and 0.05% Tween 20. After sequential incubation with monoclonal antibody TM1009 (1:500) and peroxidase-conjugated sheep anti-mouse IgG (1:10,000), thrombomodulin antigen was revealed by enhanced chemiluminescence (Amersham Corp.).

Thrombomodulin Cofactor Activity Cells were washed three times with PBS and lysates prepared in 20 mmol/L Tris-HCI, pH 8.0, 0.6% Triton X-100, 100 mmol/L NaCI, and 3.0 mmol/L CaCI2. Thrombomodulin cofactor activity was measured by a two-stage protein C activation assay described previously.16 Cell lysates were incubated for 30 minutes at 37 C with 2.6 nmol/L human thrombin (Enzyme Research Laboratories, South Bend, IN), 0.84 pmol/L human protein C (generously supplied by Dr. Hans Peter Schwarz, Immuno AG, Vienna, Austria), and 2.5 mmol/L CaCI2. The reaction was stopped by addition of a mixture of human antithrombin III and heparin, and the amidolytic activity of activated protein C was measured with the chromogenic substrate S-2366

936 Lager et al A/P April 1995, Vol. 146, No. 4

(Kabi Pharmacia Hepar, Franklin, OH). The total protein concentration of cell lysates was determined by a modified Bradford protein assay (Bio-Rad Laboratories, Richmond, CA), and thrombomodulin antigen was measured by an enzyme-linked immunosorbent assay (Diagnostica Stago, Franconville, France).

Results Thrombomodulin Protein in Normal Epithelium Thrombomodulin protein was demonstrated consistently by immunohistochemistry in both keratinized and nonkeratinized stratified squamous epithelium (Table 1). Thrombomodulin was detected variably in transitional epithelium but was not detected in cuboidal, simple columnar, or pseudostratified columnar epithelium. With two exceptions (salivary ductal epithelium and pancreatic islets; see below), all tissues showed identical patterns of staining with both antithrombomodulin antibodies. In all samples examined, strong thrombomodulin staining was observed on the endothelium of subepithelial blood vessels. No staining was seen when either nonimmune mouse serum Table 1. Summary of Thrombomodulin Immunobistochemistry in Normal Epithelium Type of epithelium Stratified squamous

Transitional

Cuboidal

Simple columnar

Pseudostratified columnar

Tissue Skin Oral mucosa Larynx

Esophagus Ectocervix Vagina Ureter Urinary bladder* Salivary acini Salivary ductst Mammary lobules Mammary ducts Thyroid follicles Pancreatic acini Pancreatic isletst Renal tubules Fallopian tube Seminiferous tubules Stomach Intestine Endometrium Endocervix Prostate Tracheobronchial tree Vas deferens

Staining intensity +++ ++ +++ +++ ++ ++ ++

0/+ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0, absent; +, weak; ++, moderate; +++, strong staining. *Variably present (weak staining when present). tWeak staining with antibody EC13.3; absent staining with antibody TM1009. tModerate staining with antibody TM1009; absent staining with antibody EC13.3.

or isotype-matched mouse monoclonal IgG1 was applied as the primary antibody (Figure 1 D, F). In agreement with our earlier observations, 16 strong suprabasal staining was observed in keratinized stratified squamous epithelium of skin (Figure 1A). No staining of the basal cell layer was seen, and progressive diminution of staining was observed in the granular and cornified layers of epidermis. An identical pattern of suprabasal staining was observed in nonkeratinized stratified squamous epithelium of esophagus (Figure 1 B) and uterine ectocervix (Figure 1C) as well as larynx and vagina (not shown). In both keratinized and nonkeratinized squamous epithelium, thrombomodulin staining was concentrated along the cell borders. Sections of transitional epithelium from ureter showed a variable pattern, with foci of either suprabasal or superficial staining (Figure 1 E). Four of ten cases of transitional epithelium of the urinary bladder showed thrombomodulin staining that varied from weak and focal to strong and diffuse; no thrombomodulin staining was observed in the remaining six cases. Thrombomodulin protein expression was not detected in nonstratified epithelium such as gastric mucosa (Figure 2). Two tissues demonstrated discrepant patterns of thrombomodulin staining with the two monoclonal antibodies. Salivary ductal epithelium stained weakly with monoclonal antibody EC13.3 but did not stain with monoclonal antibody TM1009 (Table 1). Unlike stratified epithelium, salivary ductal epithelium stained in a cytoplasmic pattern. In agreement with a previous report,28 moderate staining of pancreatic islets was observed with antibody TM1009. However, islet cells did not stain with antibody EC13.3. In both tissues, staining was blocked completely by preincubation of the antibody with purified soluble recombinant human thrombomodulin.

Thrombomodulin Protein in Squamous Malignancies We have demonstrated previously that thrombomodulin is expressed in squamous cell carcinomas of the skin.16 To determine whether thrombomodulin is also expressed in squamous cell carcinomas of other tissues, immunohistochemistry was performed on sections of invasive squamous cell carcinoma of the larynx, lung, esophagus, uterine cervix, and urinary bladder. Thrombomodulin was detected in 21 of 26 cases; some tumors showed strong, diffuse staining, and others showed strong central staining with less intense staining at the periphery of tumor cell nests.

Thrombomodulin in Human Epithelium

937


J.

I

0

Figure 1. Immuinohistocbemical localizationi of thrombomodulin protein in ntormal human epithelium (magnification, X250). In skin (A), thrombomodulin staining is concentrated along the cell borders in the spinous layer of epidermis, with minimal staining of the basal, granular, or cornified layers. Stuprabasal staining is also seen in the nonkeratinized stratified squamous epithelium of the esophagus (B) and uterine ectocervix (C). The transitional epithelium of the ureter (E) shouns a focal pattern of suprabasal staining (arrows). Staining of subepithelial capillary endothelium is observed in all tissues (small arrows). No staining is detected in sections of uterine ectocervix (D) or ureter (F) incuibated with ntegative control mouse monoclonal immunoglobulin GI.

Figure 2. Immunohistochemical staining for thrombomodulin in niormal human gastric mucosa shows staining of capillary endothelium within the lamina propria (arrown) but no staining of the gastric epithelium (magnification, X250).

Areas of keratin pearl formation often stained with diminished intensity (Figure 3A). Like normal squamous epithelium, individual tumor cells demonstrated peripheral accentuation of thrombomodulin staining. The 5 cases in which thrombomodulin was not detected included 3 poorly differentiated carcinomas (1 from larynx and 2 from lung), a laryngeal squamous cell carcinoma in which keratinization was clearly evident, and a lymph node metastasis from a squamous cell carcinoma of the bladder. In this latter case, although the primary bladder tumor showed prominent keratinization and stained strongly for thrombomodu-

lin, the metastasis resembled transitional cell carcinoma and did not stain for thrombomodulin. Thrombomodulin protein was detected in 6 of 10 examples of endocervical, bronchial, or transitional epithelium undergoing squamous metaplasia (Figure 3B) and in 7 of 14 cases of squamous carcinoma-insitu (Figure 3C). Thrombomodulin staining in squamous carcinoma-in-situ was variable in intensity and was focal within the epithelium in 3 cases, localized to the basal layer in 1 case, and suprabasal in 3 cases. Thrombomodulin was also expressed in 7 of 11 cases of invasive transitional cell carcinoma of the

938

Lager et al


4

i 74

-.y.L AF# :5 isl-wkm

urinary bladder. Foci of squamous differentiation were clearly evident in 2 of these tumors, and the intensity of thrombomodulin staining was strongest within these foci.

Thrombomodulin mRNA in Normal Epithelium and Squamous Cell Carcinoma To confirm that the immunohistochemical staining observed with the anti-thrombomodulin antibodies represented thrombomodulin synthesized by epithelial cells, in situ hybridization was performed with antisense and sense riboprobes generated from the human thrombomodulin cDNA. When human skin was hybridized with the antisense riboprobe, thrombomodulin mRNA was detected in both epidermis and dermal capillary endothelial cells (Figure 4A, B). Within the epidermis, hybridization concentrated in the suprabasal layer, with relative sparing of the basal, granular, and cornified layers. Thrombomodulin mRNA was also detected in normal ectocervical epithelium (Figure 4D, E) and in invasive squamous cell carcinoma of the larynx (Figure 4G, H). No localized hybridization was detected when the sense ri-

Figure 3. Immunohistochemical localization of thrombomodulin protein in malignant, metaplastic, and dysplastic buman squamous epithelium (magnification, X250). Invasive squamous cell carcinoma of the larynx (A) stains strongly, with diminished staining in areas of keratin pearlformation (*). A varable staining pattern is observed in squamous metaplasia of the bronchus (B), witb some areas of strong suprabasal staining (arrow) and other areas devoid ofstaining (). Moderate suprabasal staining is seen in carcinoma-in-situ of the uterine endocenrix (C). Staining of subepithelial capillary endothelium is observed in all tissues (small arrows).

boprobe was used (Figure 2C, F, 1). These results indicate that thrombomodulin mRNA is synthesized by both normal and malignant squamous epithelial cells.

Thrombomodulin in Cultured Squamous Epithelial Cells To determine whether thrombomodulin is synthesized by squamous epithelial cells in culture, we performed Northern hybridization analysis on total mRNA isolated from HUVEC, HK, and the human cervical squamous cell carcinoma cell line C4-1 (Figure 5). Hybridization with a human thrombomodulin cDNA probe demonstrated that each cell type contained a 3.6-kb thrombomodulin mRNA. The intensity of thrombomodulin hybridization was greater in C4-1 than in either HUVEC or HK. Hybridization with a human actin cDNA probe detected similar amounts of actin mRNA in all three cell types. Thrombomodulin expression in cultured cells was examined further by immunoblotting with antibody TM1009 (Figure 6). C4-1 and HK each contained a major 75-kd thrombomodulin band that was identical in apparent molecular mass to full-length recombi-

Thrombomodulin in Human Epithelium 939 AJP Aprl 1995, Vol. 146, No. 4

*'

r

-' 49:..

__

V.444

saj'

_

_s_ _Y

-

4.'.C

,,,i

,4

'U

',

RZw *

4: ii..

Figure 4. In situ hybridization of thrombomodulin mRNA in normal and malignant human squamous epithelium (magnification, x200). A and B: Paired bright and dark field photomicrographs of human skin hybridized with an antisense thrombomodulin riboprobe. Localized hybridization is observed in the epidermis and in dermal capillaries (arrows). D and E: Paired bright and darkfieldphotomicrographs of uterine ectocervix hybridized with an antisense thrombomodulin riboprobe. Localized hybridization is observed in the epithelium and in subepithelial capillary endothelium (arrow). G and H: Paired bright and darkfieldphotomicrographs of invasive squamous cell carcinoma of the larynx hybridized with an antisense thrombomodulin riboprobe. Localized hybridization is observed in tuimor cells (arrows), with diminished hybridization in surrounding connective tissue (*). No localized hybridization is detected in dark field photomicrographs of skin (C), uterine ectocervix (F), or invasive squamous cell carcinoma of the larynx (I) hybridized with a sense thrombomodulin riboprobe.

nant human thrombomodulin expressed on TMnc

cells.18 No thrombomodulin was observed in lysates of untransfected CV-1 cells. Again, the intensity of the thrombomodulin band was greater in C4-1 than in HK. The concentration of thrombomodulin antigen in cell lysates was determined by enzyme-linked immu-

nosorbent assay, and thrombomodulin cofactor activity was measured in a protein C activation assay (Table 2). HUVEC and HK contained comparable amounts of thrombomodulin antigen and activity. The thrombomodulin activity per milligram of total protein in C4-1 lysates was fourfold greater than in HK lysates,

940

Lager et al

AJP Apil 1995, Vol.

146, No. 4

0

''U I z Q~~~~

c

_ I

200TM-)

-- 28S

97

-18S

69-

Actin Figure 5. Northern hybridization of thrombomodulin and actin mRNA. A total of 25 ug of total cellular RNA isolated from HUVEC, HK, or C44I cells was subjected to agarose gel electrophoresis, transferred to nylon membranes, and hybridized with radiolabeled human thrombomodulin or human y-actin cDNA probes.

and sixfold greater than in HUVEC lysates. The specific activity of thrombomodulin was similar in each cell type.

46Figure 6. Immunoblot of thrombomodulin protein. Detergent lysates containing 50 .g of total protein from TMnc, CV-1, C4-I, or HK were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis under nonreducing conditions, transferred to nitrocellulose membranes, blotted with monoclonal antibody TM1009, and developed by ECL. The Mr ofprotein standards is indicated on the left.

Table 2.

Discussion Thrombomodulin was identified first as an endothelial cell protein,8 and immunohistochemical studies performed in several species have confirmed that thrombomodulin is expressed uniformly on the luminal surface of vascular endothelium.13,28629 Endothelial cell thrombomodulin functions to prevent intravascular coagulation by 1), limiting activation of coagulation proteases; 2), stimulating the protein C anticoagulant pathway; and 3), enhancing fibrinolysis.1-3 Recent data indicate that thrombomodulin is also expressed in extravascular locations, including mesothelial surfaces14 and epidermis.16 However, the function of thrombomodulin in these locations is poorly understood. This study demonstrates that thrombomodulin expression correlates with differentiation in both keratinized and nonkeratinized squamous epithelium. With two exceptions (salivary ducts and pancreatic islets), all tissues stained with an identical pattern with two separate monoclonal antibodies that have been shown to be specific for human thrombomodulin.171828 Expression of thrombomodulin in normal

Thrombomodulin Antigen and Cofactor Activity in Cell Lysates

Antigen

Cofactor activity

Cell

(pg/mg)*

(nmol/h/mg)t

Specific activity (mollh/mol)t

HUVEC HK

0.4 ± 0.1 0.6 ± 0.1 2.8 + 0.6

0.12 0.02 0.20 0.02 0.77 ± 0.11

22 ± 6 24 ± 4 21 ± 3

C4-I

All values represent the mean + SD of five determinations. *Thrombomodulin antigen was measured by enzyme-linked immunosorbent assay and is expressed as micrograms of thrombomodulin per milligram of total protein. tThrombomodulin cofactor activity was measured in a protein C activation assay and is expressed as nanomoles of activated protein C generated per hour per milligram of total protein. tSpecific activity is expressed as moles of activated protein C generated per hour per mole of thrombomodulin.

and malignant squamous epithelium was confirmed by in situ hybridization of thrombomodulin mRNA, and full-length, functionally active thrombomodulin was demonstrated in cultured squamous epithelial cells. These observations indicate that thrombomodulin is synthesized by both normal and malignant human epithelial cells. The differential staining of salivary ductal epithelium and pancreatic islets indicates that these tissues contain antigenic epitopes that


overlap with thrombomodulin but are recognized differently by the two antibodies. Whether these epitopes represent thrombomodulin or crossreacting antigens cannot be determined from these studies. The distribution of thrombomodulin in different types of epithelium was nonuniform. Thrombomodulin was detected consistently in keratinized and nonkeratinized stratified squamous epithelium and variably in transitional epithelium. However, thrombomodulin was not detected in simple or pseudostratified epithelium. Within stratified squamous epithelium, thrombomodulin protein and mRNA were concentrated in the suprabasal layer, with minimal expression in the basal or superficial layers. This pattern of expression indicates that thrombomodulin is localized in epithelial compartments separate from those containing tissue factor, a major procoagulant.30 Unlike thrombomodulin, tissue factor is concentrated in the superficial granular layer of stratified squamous epithelium31 and is also expressed in nonstratified epithelium.31,32 These observations suggest that hemostasis is regulated in part through compartmentalization of procoagulant and anticoagulant proteins within the epithelium. A reciprocal relationship between the expression of thrombomodulin and tissue factor has also been observed in vascular endothelium. Unstimulated endothelial cells synthesize thrombomodulin but do not express tissue factor. However, when stimulated with inflammatory cytokines such as tumor necrosis factor or interleukin 1, endothelial cells markedly decrease thrombomodulin synthesis, and begin to synthesize tissue factor.33'34 These effects of cytokines have been implicated in the pathogenesis of intravascular coagulation associated with inflammatory diseases.7 Whether cytokines induce similar changes in the expression of thrombomodulin and tissue factor in epithelial cells has not yet been determined. Thrombomodulin expressed in stratified epithelium may also regulate nonhemostatic processes that are stimulated by thrombin. In addition to its role in promoting blood coagulation, thrombin stimulates inflammatory and proliferative responses in a variety of target cells, including endothelial cells, fibroblasts, smooth muscle cells, and keratinocytes.3538 Many of the hormonal and growth regulatory properties of thrombin are mediated through a proteolytically activated thrombin receptor.39 In mouse embryos, thrombin receptor mRNA is localized predominantly in the basal layer of epidermis, suggesting that basal keratinocytes may respond preferentially to thrombin.32 Selective expression of thrombomodulin in the suprabasal layer of stratified squamous epithelium

may facilitate the localization of thrombin's cell stimulatory activities to the basal layer. Previous studies have suggested that thrombomodulin may be a marker for endothelial15 or mesothelial40 tissues. Our data indicate that thrombomodulin expression is also associated with squamous differentiation. We detected thrombomodulin expression in 21 of 26 cases of invasive squamous cell carcinoma, 7 of 14 cases of squamous carcinoma-insitu, and 6 of 10 examples of epithelium undergoing squamous metaplasia. Similar observations have been reported by Tamura et al,41 who detected thrombomodulin in 4 of 11 cases of squamous cell carcinoma of the lung. Thrombomodulin is also expressed transiently in a variety of vascular and nonvascular structures during murine development.4243 These observations indicate a potential role for thrombomodulin in cellular differentiation and suggest that regulation of thrombomodulin expression differs between endothelial and epithelial cells.

Note Added in Proof In agreement with our findings, another group recently reported expression of thrombomodulin by human keratinocytes (Mizutani H, Hayashi T, Nouchi N, Ohyanagi S, Hashimoto K, Shimuzu M, Suzuki K: Functional and immunoreactive thrombomodulin expressed by keratinocytes. J Invest Dermatol 1994, 103:825-828).

Acknowledgments The authors thank Dr. Jeanne Snyder and Christine Wohlford-Lenane for advice in performing in situ hybridization, Ms. Mary T. Sturm, HT-ASCP, for performing the immunohistochemical studies, and Yan Chen for technical assistance.

References 1. Dittman WA, Majerus PW: Structure and function of thrombomodulin: a natural anticoagulant. Blood 1990, 75:329-336 2. Esmon CT: Molecular events that control the protein C anticoagulant pathway. Thromb Haemost 1993, 70: 29-35 3. Lentz SR, Sadler JE: The molecular basis of thrombomodulin function. Thrombin, Thrombomodulin, and the Control of Hemostasis. Edited by JC Giddings. Austin, TX, RR Landes Co., 1994, pp 91-120 4. Svensson PJ, Dahlback B: Resistance to activated

942

Lager et al


protein C as a basis for venous thrombosis. N EngI J Med 1994, 330:517-522 5. Bertina RM, Koeleman BPC, Koster T, Rosendaal FR, Dirven RJ, de Ronda H, van der Velden PA, Reitsma PH: Mutation in blood coagulation factor V associated with resistance to activated protein C. Nature 1994, 369:64-67 6. Walker FJ, Fay PJ: Regulation of blood coagulation by the protein C system. FASEB J 1992, 6:3561-2567 7. Esmon CT: Cell mediated events that control blood coagulation and vascular injury. Annu Rev Cell Biol 1993, 9:1-26 8. Esmon CT, Owen WG: Identification of an endothelial cell cofactor for thrombin-catalyzed activation of protein C. Proc Natl Acad Sci USA 1981, 78:2249-2252 9. Suzuki K, Nishioka J, Hayashi T, Kosaka Y: Functionally active thrombomodulin is present in human platelets. J Biochem 1988, 104:628-632 10. Ogura M, Ito T, Maruyama I, Takamatsu J, Yamamoto S, Ogawa K, Nagura H, Saito H: Localization and biosynthesis of functional thrombomodulin in human megakaryocytes and a human megakaryoblastic cell line (MEG-01). Thromb Haemost 1990, 64:297-301 11. McCachren SS, Diggs J, Weinberg JB, Dittman WA: Thrombomodulin expression by human blood monocytes and by human synovial tissue lining macrophages. Blood 1991, 78:3128-3132 12. Conway EM, Nowakowski B, Steiner-Mosonyi M: Human neutrophils synthesize thrombomodulin that does not promote thrombin-dependent protein C activation. Blood 1992, 80:1254-1263 13. Maruyama I, Bell CE, Majerus PW: Thrombomodulin is found on endothelium of arteries, veins, capillaries, and lymphatics, and on syncytiotrophoblast of human placenta. J Cell Biol 1985, 101:363-371 14. Boffa M, Burke B, Haudenschild CC: Preservation of thrombomodulin antigen on vascular and extravascular surfaces. J Histochem Cytochem 1987, 35:12671276 15. Yonezawa S, Maruyama I, Sakae K, Igata A, Majerus PW, Sato E: Thrombomodulin as a marker for vascular tumors: comparative study with factor VIII and Ulex europaeus lectin. Am J Clin Pathol 1987, 88:405-411 16. Raife TJ, Lager DJ, Madison KC, Piette WW, Howard EJ, Sturm MT, Chen Y, Lentz SR: Thrombomodulin expression by human keratinocytes: induction of cofactor activity during epidermal differentiation. J Clin Invest 1994, 93:1846-1851 17. Maruyama I, Majerus PW: The turnover of thrombinthrombomodulin complex in cultured human umbilical vein endothelial cells and A549 lung cancer cells. J Biol Chem 1985, 260:15432-15438 18. Tsiang M, Lentz SR, Sadler JE: Functional domains of membrane-bound human thrombomodulin: EGF-like domains four to six and the serine/threonine-rich domain are required for cofactor activity. J Biol Chem 1992, 267:6164-6170 19. Lin J, McLean K, Morser J, Young TA, Wydro RM, An-

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

drews WH, Light DR: Modulation of glycosaminoglycan addition in naturally expressed and recombinant human thrombomodulin. J Biol Chem 1994, 269: 25021-25030 Lentz SR, Chen Y, Sadler JE: Sequences required for thrombomodulin cofactor activity within the fourth epidermal growth factor-like domain of human thrombomodulin. J Biol Chem 1993, 268:15312-15317 Angerer LM, Angerer RC: In situ hybridization to cellular RNA with radiolabelled RNA probes. In Situ Hybridization: A Practical Approach. Edited by DG Wilkinson. New York, Oxford University Press, 1992, pp 15-32 Wohlford-Lenane CL, Snyder JM: Localization of surfactant-associated proteins SP-A and SP-B mRNA in rabbit fetal lung tissue by in situ hybridization. Am J Respir Cell Mol Biol 1992, 7:335-343 Lentz SR, Sadler JE: Inhibition of thrombomodulin surface expression and protein C activation by the thrombogenic agent homocysteine. J Clin Invest 1991, 88: 1906-1914 Tsiang M, Lentz SR, Dittman WA, Wen D, Scarpati EM, Sadler JE: Equilibrium binding of thrombin to recombinant human thrombomodulin: effect of hirudin, fibrinogen, factor Va, and peptide analogues. Biochemistry 1990, 29:10602-10612 Chomczynski P, Sacchi N: Single-step method of RNA isolation by acid guanidinium thiocyanate-phenolchloroform extraction. Anal Biochem 1987, 162:156-159 Gunning P, Ponte P, Okayama H, Engle J, Blau H, Kedes L: Isolation and characterization of full-length cDNA clones for human a, 3, and y-actin mRNAs: skeletal but not cytoplasmic actins have an aminoterminal cysteine that is subsequently removed. Mol Cell Biol 1983, 3:787-795 Laemmli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970, 227:680-685 Fink LM, Eidt JF, Johnson K, Cook JM, Cook CD, Morser J, Marlar R, Collins CL, Schaefer R, Xie S, Hsu S, Hsu P: Thrombomodulin activity and localization. Int J Dev Biol 1993, 37:221-226 DeBault LE, Esmon NL, Olson JR, Esmon CT: Distribution of the thrombomodulin antigen in the rabbit vasculature. Lab Invest 1986, 54:172-178 Edgington TS, Mackman N, Brand K, Ruf W: The structural biology of expression and function of tissue factor. Thromb Haemost 1991, 66:67-79 Drake TA, Morrissey JH, Edgington TS: Selective cellular expression of tissue factor in human tissues. Am J Pathol 1989, 134:1087-1097 Soifer SJ, Peters KG, O'Keefe J, Coughlin SR: Disparate temporal expression of the prothrombin and thrombin receptor genes during mouse development. Am J Pathol 1994, 144:60-69 Bevilacqua MP, Pober JS, Majeau GR, Fiers W, Cotran RS, Gimbrone MA Jr.: Recombinant tumor necrosis


34.

35.

36.

37.

38.

factor induces procoagulant activity in cultured endothelium: characterization and comparison with the actions of interleukin 1. Proc Natl Acad Sci USA 1986, 83:4533-4537 Nawroth PP, Stern DM: Modulation of endothelial cell hemostatic properties by tumor necrosis factor. J Exp Med 1986, 163:740-745 Garcia JGN, Aschner JL, Malik AB: Regulation of thrombin-induced endothelial barrier dysfunction and prostaglandin synthesis. Thrombin: Structure and Function. Edited by LJ Berliner. New York, Plenum Press, 1992, pp 397-430 Van Obberghen-Schilling E, Pouyssegur J: a-Thrombin receptors and growth signaling. Semin Thromb Hemost 1993, 19:378-385 McNamara CA, Sarembock IJ, Gimple LW, Fenton JW, II, Coughlin SR, Owen GK: Thrombin stimulates proliferation of cultured rat aortic smooth muscle cells by a proteolytically activated receptor. J Clin Invest 1993, 91:94-98 Talwar HS, Fisher GJ, Harris VA, Voorhees JJ: Agonistinduced hydrolysis of phosphoinositides and forma-

39.

40.

41.

42.

43.

tion of 1,2-diacylglycerol in adult human keratinocytes. J Invest Dermatol 1989, 93:241-245 Vu TH, Hung DT, Wheaton VI, Coughlin SR: Molecular cloning of a functional thrombin receptor reveals a novel proteolytic mechanism of receptor activation. Cell 1991, 64:1057-1068 Collins CL, Ordonez NG, Schaefer R, Cook CD, Xie S, Granger J, Hsu P, Fink L, Hsu S: Thrombomodulin expression in malignant pleural mesothelioma and pulmonary adenocarcinoma. Am J Pathol 1992, 141:827833 Tamura A, Matsubara 0, Hirokawa K, Aoki N: Detection of thrombomodulin in human lung cancer cells. Am J Pathol 1993, 142:79-85 Imada S, Yamaguchi H, Nagumo M, Katayanagi S, Iwasaki H, Imada M: Identification of fetomodulin, a surface marker protein of fetal development, as thrombomodulin by gene cloning and functional analysis. Dev Biol 1990, 140:113-122 Ford VA, Wilkinson JE, Kennel SJ: Thrombomodulin distribution during murine development. Roux Arch Dev Biol 1993, 202:364-370