Thrombomodulin is present in human plasma
and urine.
H Ishii, P W Majerus
J Clin Invest.
1985;
76(6)
:2178-2181.
https://doi.org/10.1172/JCI112225
.
Thrombomodulin is an endothelial cell membrane protein that is a cofactor required for the
rapid activation of plasma protein C. We now report that plasma and urine of normal
subjects contains a modified form of thrombomodulin that is soluble. The levels measured
by radioimmunoassay were 292 +/- 60 ng thrombomodulin/ml plasma and 102 +/- 38 ng
thrombomodulin/ml urine. Thrombomodulin was isolated from both plasma and urine by
immunoaffinity chromatography using a polyclonal anti-human thrombomodulin IgG column.
The apparent molecular weight of soluble thrombomodulin was estimated by immunoblot
analysis using 125I-monoclonal anti-thrombomodulin IgG. When run without
2-mercaptoethanol, soluble thrombomodulin appeared as two polypeptides, Mr = 63,000 and
54,000, while samples run with 2-mercaptoethanol migrated mainly at Mr = 85,000. These
results imply that the soluble form of thrombomodulin is smaller than the cellular form,
presumably because of a lack of the membrane-binding domain. Soluble thrombomodulin is
similar to cellular thrombomodulin in its intrinsic protein C-activating cofactor activity as
measured by antibody neutralization. The apparent Km for protein C was the same for
cellular and soluble thrombomodulin, while the soluble form requires a higher concentration
of thrombin (three- to fivefold) for one-half maximal activity than the cellular form.
Thrombomodulin functional activity cannot be directly measured in plasma because of
some inhibitory substance(s). The physiological significance of circulating and urinary
thrombomodulin is presently […]
Research Article
Find the latest version:
Thrombomodulin
Is Present in Human Plasma and Urine
Hidemi Ishii and Philip W.MajerusDivisionofHematology-Oncology, Departments ofInternal Medicine and BiologicalChemistry,
Washington UniversitySchool ofMedicine, St. Louis, Missouri 63110
Abstract
Thrombomodulin
isan endothelial cell membraneprotein
that isacofactorrequired
fortherapid
activation ofplasma
proteinC.Wenowreportthat
plasma
andurine of normalsubjects
con-tainsa modified form ofthrombomodulin that issoluble. Thelevels measured
by
radioimmunoassay
were292±60ngthrom-bomodulin/ml
plasma
and102±38ngthrombomodulin/ml
urine.Thrombomodulin was isolated from both
plasma
and urine byimmunoaffinity chromatography
using
apolyclonal
anti-humanthrombomodulin
IgG
column. The apparent molecularweight
ofsoluble thrombomodulinwasestimated
by
immunoblot analysisusing
1251-monoclonal
anti-thrombomodulin IgG. When runwithout
2-mercaptoethanol,
soluble thrombomodulin appearedas two
polypeptides,
Mr =63,000
and54,000,
while samplesrunwith
2-mercaptoethanol migrated mainly
atMr
= 85,000.These results
imply
that the soluble form ofthrombomodulin
is smaller than thecellularform, presumably
because ofalack ofthe
membrane-binding
domain. Solublethrombomoduiinissim-ilar to cellularthrombomodulin in its intrinsic
protein
C-acti-vating
cofactoractivity
asmeasuredby
antibody
neutralization. The apparent Kmforprotein
Cwas the same for cellular andsoluble
thrombomodulin,
while the soluble formrequires
ahigherconcentration of thrombin
(three-
tofivefold)
for one-halfmax-imal
activity
than the cellular form. Thrombomodulinfunctionalactivity
cannot bedirectly
measuredinplasma
because ofsomeinhibitory
substance(s).
Thephysiological significance
ofcir-culating
andurinary
thrombomodulinispresently
obscure.Introduction
Protein C isavitamin
K-dependent glycoprotein
of62,000molwt,whichcirculatesin humanplasmaataconcentration of -4 ,ug/ml(1).Protein C is converted to a protease withanticoagulant
function
by
thrombin. Theconversion requiresan endothelialcell membrane
protein cofactor, thrombomodulin,
which wasfirstisolated from rabbit
lung
(2).Wehaveisolatedthe human form ofthrombomodulin from placenta (3) and lung(4) and haveprepared
bothpolyclonal
(3)andmonoclonal (5)antibodiesdirected
against
theprotein.
We have shown that thrombo-modulin iswidely
distributed on the endothelium of humanarteries,
veins, capillaries,
and lymphatics in allorgansexceptAddresscorrespondencetoDr.Majerus, Division of
Hematology-On-cology, Washington UniversitySchoolofMedicine,660South Euclid, St.Louis,MO63110.
Receivedfor publication22July1985.
brain(6,
7).
Thegreatestcontentof thrombomodulin is inlungandplacentaasmeasuredby radioimmunoassay of extracts of these tissues(7).Wenowreportthat humanplasmaand urine contain a soluble form of thrombomodulin that has been de-tectedbyboth
radioimmunoassay
andbyafunctionalassay.Methods
Exceptwhereindicated,allreagentswerefromSigmaChemicalCo.,St. Louis,MO.ProteinC(8),thrombin(9),andantithrombinIII(10)were
isolatedfrom humanplasma.Humanthrombomodulinwasisolated from
placentaaspreviously reported (3). Polyclonal anti-human thrombo-modulinantibody (3)andmousemonoclonalanti-thrombomodulin an-tibody (5)wereprepared using placentalthrombomodulinastheantigen (3). The iodination ofthrombomodulin wasperformedaspreviously
described (7).TheiodinationofproteinAwasperformed accordingto Mellmanand Unkeless(11).Free1251wasremovedbychromatography
on a I-mlcolumn ofSephadexG-25. Protein concentrationwasmeasured usingthe Bio-Rad assay (Bio-Rad Laboratories, Richmond, CA)with humanIgGas astandard.
Immunoaffinity chromatography.Anti-humanthrombomodulin
IgG
agarosewaspreparedbycoupling 10mgof rabbitanti-human throm-bomodulin IgG (10 mg/ml)in0.1Mmorpholino-ethane-sulfonicacid,
pH 6.5,to 1 ml ofAffi-gel 10(Bio-Rad Laboratories) overnightat4VC.
Residualactive esterswereblocked bythe addition of0.1 ml of 1 M ethanolamine, pH 8.0, for 6 h. Thegelwaswashed with 20 vol of 20 mMTris, pH 7.4, containing0.15 M NaCl and 0.2% Nonidet-40,and 10 vol of 20 mMTris,pH 7.4, containing2 MNaSCN,0.15 MNaCl,
and0.2%Nonidet-40.Thegelwasfinallyequilibratedwith 20mMTris,
pH 7.4, containing0.15 MNaCl,0.2%Nonidet-40,and 0.02% sodium azide.60-ml samplesof urine andplasmaweredialyzed against20 mM Tris, pH 7.4, containing 10 mMbenzamidine,0.15 MNaCl,and0.2% Nonidet-40.The dialyzed urine andplasmawereappliedtoseparate 0.5-ml columns of theanti-thrombomodulin IgGat3ml/hat4°C.After washingwith 15 ml of theequilibrating buffer, thrombomodulinwas
eluted with 20 mMTris, pH 7.4, containing2.0M NaSCN and 0.2% Nonidet-40. Fractions (0.5 ml)werecollected, dialyzed against 100vol of 0. 15 MNaCl,0.2%Nonidet-40,and20 mMTris, pH7.4,andassayed
forproteinC-activatingcofactoractivityand forthrombomodulinby
radioimmunoassay.
Radioimmunoassay.Theradioimmunoassayofthrombomodulinwas
performedaspreviously reported(7).Reaction mixturescontainedthe following:testsample,0.25,gofanti-thrombomodulin IgG,and0.5 rd of nonimmune rabbit serum in 20 mMTris, pH 7.4, containing0.15 M NaCIand0.2% Nonidet-40,in0.2 ml.After incubationat4°Cin 1.5 mlEppendorftubes(Brinkman Instruments Co., Westbury, NY) for2 h,2ngof'25I-thrombomoduliii (16,000 cpm,8000cpm/ng)was added and theincubationwascontinued for anadditional 16 h. A 10%
sus-pension of Staphylococcus aureus (Pansorbin,Calbiochemical-Behring
Corp.,SanDiego, CA)in0.1 ml,wasthen added to each reactionmixture. Incubationwascontinuedfor anadditional 1 h at roomtemperature. TheS. aureusorganismswerecollected by centrifugationat 14,000g for 10min,the supernatant wasremoved,and the1251I-thrombomodulin
bound was measured in agamma scintillation spectrometer.
Cofactor activity of thrombomodulin. Reaction mixtures contained thefollowing:0.5gM protein C,40 nMthrombin, andasampleina
total volof30 pAcontaining 20 mM Tris, pH 7.4, 0.15 M NaCI, 1 mM J.Clin. Invest.
© The American
Society
for ClinicalInvestigation,
Inc.CaC12, and 5 mgbovine serum albumin/ml. ProteinCaformed was measured asdescribed previously (3).
Immunoblot ofthrombomodulin. Sodium dodecyl sulfate-polyacryl-amidegelelectrophoresiswasperformedby the method of Laemmli(12) using4%acrylamide in the stacking and 10% acrylamide intherunning gel. Samples of thrombomodulin from urine,andplasmawere electro-phoresed with and without2-mercaptoethanol. Immunoblotting was performed bythemethod ofBurnette(13). The thrombomodulin was detectedbytwomethods. In onemethod,polyclonal anti-thrombom-odulinserum wasusedat a1:250dilutionanddetectionwasmadeusing
'251I-proteinAandautoradiography. Theothermethod used
'251-mono-clonalanti-thrombomodulinIgGfor direct detectionofthrombomodulin.
Results
Properties ofplasma and urine thrombomodulin. In preliminary studies we assayed samples of human urine and plasma for thrombomodulin antigen as described below. In these experi-mentsweconsistently detected thrombomodulin antigen in both fluids eventhough we could not detect thrombomodulin activity byproteinC-activating cofactor assay ofthe fluids. To definitively establish that thrombomodulin wasactually contained in both plasma and urine, we performed immunoaffinity chromatog-raphy as described in methods. The eluates of both plasma and urine were found to have protein C-activating cofactor activity that waslinear with time and completely dependent on both protein C and thrombin. The protein C dependence of urine
thrombomodulin isshown inFig. 1.Similarresults wereobtained
using thrombomodulin isolated from plasma. The apparent Mi-chaelis constant
(Km)
for protein C was 2 uM for both cellular andsoluble thrombomodulin. The effect of varying thrombinconcentrationonthrombomodulin functionalactivity is shown in Fig. 2. Plasma thrombomodulin had an apparent
K.
for thrombin of 1.1nMcompared with 0.33nMfor cellularthrom-E
-E
z
"I-0~ ll
To
0
PROTEIN C ( pM)
Figure1.Activation ofproteinCby thrombin-thrombomodulin
com-plex.Reaction mixtures(30
Ml)
contained 1.25 nMtissue thrombo-modulinstandard(- *-),or10Mlof fraction of urine thrombomodu-linfromanti-thrombomodulin IgGagarose column(-o-),40nMthrombin,and theindicatedconcentrations ofproteinC. Insetshows double-reciprocal plot forthesamedata. TheKm forproteinCwas2 MM.
6
5
4
3
2
B
0.2
0 It
0 2 3 4 5 -3 -2 -I 0 2
THROMBIN(nM) [THROMBINnM]
Figure 2. Thrombin dependence of protein Cactivation.Ashows sat-urationcurve.Reaction mixtures (30ul)contained 1.25 nMtissue thrombomodulin standard (- * -),or10Mlof fraction of plasma thrombomodulin from anti-thrombomodulin IgGagarosecolumn (-0-),0.5 AMprotein C, andtheindicatedconcentrations of throm-bin.Bshowsdouble-reciprocalplotof thesame data. TheKmof
stan-dardandurinethrombomodulinwas0.33and 1.1 nM,respectively.
bomodulin. Urine thrombomodulin had an apparent Km for thrombin of 1.6 nM (data not shown). The protein C-activating cofactor activity isolated from plasma and urine reflected thrombomodulin asevidenced by the finding that theactivity
in both cases was completely inhibited by an excess of rabbit anti-thrombomodulin IgG. The relative catalyticactivityof urine
thrombomodulin vs. cellular thrombomodulin was estimated by theimmunotitrationexperimentshown inFig.3. Weestimate
that urine thrombomodulin has -v3 the activity ofcellular thrombomodulin based onthe findingthat the tangent tothe inactivation curveofeach thrombomodulin extrapolatestothe
same point using -2/3 as much urine as cellular
thrombo-modulin.Asimilar result was obtained usingplasma thrombo-modulin as shown in Fig. 4. In this experiment plasma
throm-bomodulinhad
½1/2
theintrinsicactivityofthatobservedusingcellularthrombomodulin. We next compared plasma
throm-bomodulintocellularthrombomodulinby immunoblotanalysis
using
1251-monoclonal
anti-thrombomodulin IgGfor detectionasshown inFig. 5. This gelwas runin the presenceof
2-mer-captoethanolwherecellularthrombomodulin hasanapparent molecular weight of 105,000. Themajor polypeptidedetected in plasmathrombomodulin has an apparent molecularweight
of85,000. Thebands detectedatlowermolecularweightsmost
2.0 Figure 3.Inhibition of
tis-sueorurine
thrombomodu-E lin by
anti-human-throm-E 6 bomodulinrabbit IgG.
Tis-sue(0.94ng,-* -) and
E 2 urine(aneluatefraction
Gas fromanti-thrombomodulin,
10
Ml,
-o-)thrombomodu-° 08 linwereincubated with
0-U
Ad
150'a
ngpolyclonalanti-hu-z man-thrombomodulin
rab-H0.4 ' bitIgGfor 15 min at
370C.
ai Thecofactoractivityof
thrombomodulin inprotein
0 50 100 150 Cactivationwasassayedas
ANTITHROMBOMODULIN IgG (ng) described in Methods. In
controlexperimentsitwasshown that theanti-thrombomodulin
IgG
04
I
\A
\\A
0 50 100 150
ANTITHROMBOMODULIN IgG(ng)
Figure 4. Inhibition of tissueorplasma thrombomodulin by anti-hu-man-thrombomodulin rabbit IgG. Tissue (0.63ng,-o -) and plasma
thrombomodulin (an eluate fraction from anti-thrombomodulin IgG column, 10gl,-o-)wereincubated with 0-150ngpolyclonal
anti-thrombomodulin rabbit IgG for 15 minat37°C. The cofactor activity ofthrombomodulin in protein C activationwasassayedasdescribed in Methods.
likely reflect further degradation products of thrombomodulin. Since one of the lower molecular weight peptides detected in
lane 2 migrated similarlytothe heavy chain of IgG,we
consid-eredthe possibility thatthesepeptides might reflect,inpart,low
affinitybindingof'25I-anti-thrombomodulintoIgGthateluted fromtheaffinitycolumn. Weran 5,ugof human IgG in lane 3
which is slightly detected by the monoclonal anti-thrombo-modulin. Theamountofthrombomodulin appliedtothe gelin lane2was7ngbasedonfunctional activityandtheproteinin thissamplewas<2,gby Bio-Rad proteinassay.Thus the bands
detectedinlane 2reflectamuchsmalleramountof protein than
the IgGin lane3. We therefore conclude that the material de-tectedin lane 2reflects peptides derived from thrombomodulin. We also performed immunoblots on samples electrophoresed
intheabsence of 2-mercaptoethanol.Underthis conditionboth soluble and cellular thrombomodulin contain high molecular weight peptides thatdonotrunassharpbands.Thereisadiscrete
protein migrating withapparentmolecularweightof75,000 in cellularthrombomodulin.Inthe plasma thrombomodulin there were two discrete bands detected with apparent molecular
weights of 63,000and54,000.
E E c
E,
0
2 3
-92500
-66200 Figure 5.Immunoblotanalysis of
thrombomodulin. Immunoblot for
-45000 plasmathrombomodulin. The immu-noblotwasperformedasdescribedin
_31
000 Methods. '25I-monoclonalanti-throm-bomodulin IgGwasused for detec-tion. Lane1,isolatedtissue
throm-14400 bomodulin(1.0jyg);lane2,
immunoaf-finity-isolated plasma
thrombo-modulin(100,l);lane 3, human
IgG (5ptg).
a
CL.
cI
a.
0
U-0C
H-THROMBOMODULIN (ng)
2 5 10 20 50 100
PLASMA, URINE OR FRACTION
FROM IMMUNOAFFINITY COLUMN(jil)
Figure6.Radioimmunoassayofthrombomodulin.
Radioimmunoas-sayof thrombomodulinwascarriedoutasdescribed in Methods.(-) Standardplacenta thrombomodulin, (A)urineconcentrated 10times, (n) plasma, (A) affinity purified urine thrombomodulin, (o) affinity
pu-rified plasma thrombomodulin.
Radioimmunoassayofplasma and urinethrombomodulin.
Astandardcurveofthe radioimmunoassay of thrombomodulin
using isolated placental thrombomodulin is shown in Fig. 6. Assays ofurine, plasma, and affinity-purified materials from
bothsourcesare alsoshown and allappeartohave similarslopes. Wealso added known amounts of placental thrombomodulin
toboth plasma and urine, and showed that therewasgood re-coveryof the added material, further indicatingthat thrombo-modulin is accurately measured. We determined thecontentof thrombomodulin in the urine and plasma ofnormalsubjectsas
shown in TableI. Wehave alsomeasuredthrombomodulin
an-tigen in serum andfindlevels identicalto thoseinplasma. In
order toexclude the possibility that soluble thrombomodulin wasnottrulysoluble butwascontained in fragments of mem-brane,we subjected serumto ultracentrifugation at 150,000 g
for I hand carriedoutaradioimmunoassayofthesupernatant
fraction. All ofthe thrombomodulin antigen remained in the
supernatant,indicatingthat the material isactually soluble.
Table I. Thrombomodulin Content in Urine andPlasma of Normal Human Subjects
Thrombomodulin
ng/ml
Urine (n=9) 102±39
Plasma(n=6) 292±60
Thecontentof thrombomodulinwasmeasuredbyradioimmunoassay
asdescribed in Methods. Sampleswereobtained from normalsubjects aged25-48. Sixmenand threewomenprovidedurinesamples.All of theplasma sampleswerefrommen.
E
E
CL.
0 w
0 U-0 z w
0
Discussion
We havefound that human plasma and urine contain soluble forms of thrombomodulin that appear to be smaller than cellular thrombomodulin. The origin ofthis material is presently unclear. We haveconsidered two major possibilities to explain the find-ings: (a) The soluble circulating forms of thrombomodulin are produced insome cell specifically for secretion into plasma and the molecule functions to promote protein C activation in the circulation. According to this hypothesis soluble thrombo-modulin might be synthesized without the appropriate mem-brane binding domain to allow its insertion into membranes. (b) The second possibility is that the circulating forms of thrombomodulin reflect proteolysis of endothelial membrane thrombomodulin that is produced either as a result of the normal turnoverof thrombomodulin or as a result ofendothelial damage in areasof endothelial injury or inflammation. We have noticed that elastase treatment of isolated placental thrombomodulin generatespeptides thatmigrate on sodium dodecyl sulfate gels similarly to circulating thrombomodulin (unpublished obser-vations). It is estimated that adult man has approximately
6 X 1013endothelial cells (14). If the average content of throm-bomodulin is 40,000molecules/cell as wefindoncultured
um-bilical vein endothelial cells (5), then the totalcontentof throm-bomodulin in endothelium is -300 mg. Circulating throm-bomodulin would total -0.8 mg (0.3
,g/ml
X2600 ml plasma). Thus theproteolysis ofasmall fraction of cellularthrombom-odulin couldconceivablyaccountfor thecirculatingform. We
cannotestimate thisfraction since the plasma circulation time of solublethrombomodulinis unknown. We also donotknow the source ofurinary thrombomodulin. It appears essentially identicaltoplasma thrombomodulin byimmunoblotting, sug-gesting that it is probably derived from plasma. We also donot
knowwhyweareunable to measuredirectlythrombomodulin
activity byprotein Cactivation assay in plasma. The material isolated from plasma by immunoaffinity chromatography is nearlyfullyactive; however, theyieldof materialpurified from
plasma is only 2-5%. The lowyieldcould resultfrom the very small amount ofprotein involved, which may be lost during chromatography orthe subsequent dialysis after elution from
theaffinitycolumnto removeNaSCN. Alternatively,it may be thatmostof theimmunoassayedthrombomodulin inplasmais
degraded toinactiveforms that donotbind
sufficiently
tightlytotheaffinity columntoberecovered upon purification.
Fur-thermore, it is difficultto measureproteinCactivationdirectly
in plasma since antithrombin III inhibits thrombin relatively rapidly.
The finding that thrombomodulin exists ina form that is both soluble and functionally active haspotential therapeutic implications. Thus, if thestructureof soluble thrombomodulin
canbe
determined,
it may bepossibletoproduce largeamountsofthis moietyby molecularcloningtechniques. Such aprotein could be used as a natural anticoagulant that could inhibit thrombus formation in pathological states.
We thank Dr. Ikuro Maruyama, Dr. Thomas M. Connolly, and Dr. DavidB.Wilsonfor helpful discussions.
References
1.Kisiel,W. 1979. Human plasmaprotein C. Isolation, character-ization,andmechanismofactivation bya-thrombin.J.Clin. Invest.64: 761-769.
2. Esmon, N. L., W.G. Owen,andC.T. Esmon. 1982. Isolation of membrane-boundcofactor for thrombin-catalyzed activation of protein C.J.Biol.Chem. 257:859-864.
3.Salem,H. H., I. Maruyama, H.Ishii, andP. W. Majerus. 1984. Isolationandcharacterization of thrombomodulin fromhumanplacenta. J. Biol. Chem. 259:12246-12251.
4. Maruyama, I., H. H.Salem, H.Ishii, andP. W.Majerus. 1985. Humanthrombomodulin isnot anefficient inhibitor of procoagulant activityof thrombin. J. Clin. Invest. 75:987-991.
5.Maruyama, I., and P. W.Majerus. 1985. Theturnover ofthrombin-thrombomodulin complex in cultured human umbilical vein endothelial cellsandA549lungcancercells:endocytosis anddegradationofthrombin. J. Biol. Chem.Inpress.
6. Maruyama, I.,C. E.Bell,and P. W. Majerus. 1985. Thrombo-modulin is foundon endothelium ofarteries, veins, capillaries,
lym-phatics, and onsyncytiotrophoblast of human placenta. J Cell Biol. 101:363-371.
7. Ishii, H., H. H. Salem, C. E. Bell,E. B. Laposata, and P. W. Majerus. 1986. Thrombomodulin,anendothelialanticoagulantprotein,
isabsentfromhumanbrain.Blood. Inpress.
8. Suzuki, K., J. Stenflo, B. Dahlback, and A.Toedorsson. 1983. Inactivationof human coagulation factorVby activatedprotein C. J. Biol.Chem. 258:1914-1920.
9.Miletich,J. P.,C.M.Jackson, andP. W.Majerus. 1978.Properties ofthefactorXabinding siteon human platelets.J. Biol.Chem. 253: 6908-6916.
10.Owen,W.G. 1975.Evidence for theformation ofan esterbetween thrombin andheparin cofactor. Biochim.Biophys.Acta.405:380-387. 11.Mellman, I.S.,and J.C. Unkeless. 1980.Purification ofa func-tionalmouseFc receptorthrough theuseof monoclonalantibody.J. Exp. Med. 152:1048-1069.
12.Laemmli,V. K. 1970.Cleavage of structuralproteins duringthe assembly oftheheadof bacteriophageT4. Nature(Lond.).227:680-685. 13. Burnette,W.N.1981. "Westernblotting": electrophoretic transfer ofproteins from sodium dodecyl sulfate-polyacrylamide gelsto un-modified nitrocellulose and radiographic detection with antibodyand radioiodinatedproteinA.Anal. Biochem. 112:195-203.
14.Wolinsky, H. 1980.Aproposallinkingclearanceofcirculating