On the binding of tumor necrosis factor (TNF) to
heparin and the release in vivo of the
TNF-binding protein I by heparin.
M Lantz, … , E Nilsson, I Olsson
J Clin Invest.
1991;
88(6)
:2026-2031.
https://doi.org/10.1172/JCI115530
.
Tumor necrosis factor (TNF), a protein released by activated macrophages, is a central
mediator of the host response to infection and inflammation. The TNF-binding protein I
(TNF-BP-I) is a soluble fragment of the p60 transmembrane TNF receptor and an antagonist
to TNF. The level of serum TNF-BP-I was found to be increased in patients with renal
insufficiency as a result of a decrease in the glomerular filtration rate. During hemodialysis
of patients with renal failure there was a rapid but transient increase in serum TNF-BP-I.
This increase was found to be caused by heparin given before dialysis and a similar
dose-dependent response to heparin was observed also in healthy individuals. The finding of a
repeated release of TNF-BP-I into the circulation with intermittent injections of heparin
indicates that TNF-BP-I is present both in a storage pool and in a circulating pool. The
mechanism for the heparin-mediated release of TNF-BP-I was not explained; TNF-BP did
not show affinity for heparin. On the other hand, TNF was found to have affinity for heparin
and it could also be dissociated from heparin by TNF-BP-I. It is suggested that heparin-like
molecules of the extracellular matrix can retain TNF in physical proximity with target cells
and restrict the actions of TNF and protect against systemic harmful manifestations.
Research Article
On
the Binding of Tumor
Necrosis Factor (TNF)
to
Heparin
and the Release In Vivo
of the TNF-binding Protein I by
Heparin
Mikael Lantz,* Hans Thysell,t Eva Nilsson,* and Inge Olsson*
*Division ofHematology, DepartmentofMedicine, andtDepartmentofNephrology, University Hospital, S-22185 Lund,Sweden
Abstract
Tumor necrosis factor (TNF), a protein released by activated macrophages, is a central mediator of the host response to
in-fection and inflammation. TheTNF-binding protein I (TNF-BP-I) is a soluble fragment of the p60 transmembrane TNF receptor and an antagonist to TNF. The level of serum TNF-BP-I wasfound to be increased in patients with renal insuffi-ciency as a result of a decrease in the glomerular filtration rate. During hemodialysis of patients with renal failure there was a rapid but transient increase in serum TNF-BP-I. This increase wasfound to be caused by heparin given before dialysis and a
similardose-dependent response to heparin was observed also in healthy individuals. The finding of a repeated release of
TNF-BP-I into the circulation with intermittent injections of heparin indicates thatTNF-BP-I is present both in a storage
pooland inacirculating pool.Themechanism forthe
heparin-mediatedrelease of TNF-BP-I was notexplained;TNF-BP did not show affinity for heparin. On the other hand, TNF was
found to haveaffinity for heparin and it could also be
disso-ciatedfromheparinby TNF-BP-I. It is suggested that heparin-like molecules of the extracellular matrix can retain TNF in physicalproximitywith target cells and restrict the actions of TNF and protect againstsystemic harmful manifestations.(J.
Clin.Invest. 1991.88:2026-2031.)Key words: inflammation-endotoxin*glomerular filtration*lymphotoxin*growthfactors
Introduction
Tumornecrosis factor
(TNF)'
wasdiscovered inanattempt toexplain hemorrhagic necrosis oftumors in relation toacute
bacterial infections (1, 2).TNFisproduced byactivated
macro-phages asaresponsetolipopolysaccharides of Gram-negative bacteria (1, 2)andplaysamajorroleinendotoxic shock(3, 4),
which can lead tocardiovascularcollapse, organ failure,and death. It is also knownas apleiotropicagent which
participates
in a numberofways in the defense ofthe organism
against
infectious agents (5). Obviously some ofthe effects can beharmfulif they are notproperly controlled, leadingtocascade
reactions, suchasinendotoxic shock. Mechanisms of regula-tion are therefore important and mechanisms by which the
AddresscorrespondencetoMikael Lantz,M.D.,ResearchDepartment
2, E-block, LundHospital,S-22185Lund,Sweden.
Receivedfor publication19 June1990 andinrevisedform17July
1991.
1. Abbreviations usedinthis paper: TNF,tumornecrosisfactor;
TNF-BP-I, TNF-binding proteinI.
destructivepower of TNF is suppressed have potential clinical
applications. The action ofTNFis likely to be regulated at
differentlevels, e.g., at the levelof production ofTNF, at the levelof interactionwith TNF anditsreceptor, and at the post-receptor level. In addition, the newly discoveredTNF-binding protein I (TNF-BP-I) may play a protective role against the
adverseactivities ofTNF(6-10).TNF-BP-Ihas beenpurified from urine and consists ofa 30-kDcysteine rich glycoprotein
withouthomologiestoknownproteinsequences (8-10). It in-hibits thebiological activitiesof TNF bybindingto TNF with
high affinityand thus prevents TNF frombindingtoits trans-membrane cellular receptor (6-10). Immunological
cross-reac-tivity of antibodiesto TNF-BP-IwithcellsurfaceTNF
recep-torssuggests that TNF-BP-Iconstitutesasolublefragment of
the p60 transmembrane TNF-receptor (11, 12). This finding
wasconfirmedby molecularcloning ofthep60TNF receptor
(13). Recentlyasecond TNF-binding protein,TNF-BP-II,has
been
purified
fromhumanurineandmolecularly cloned (12,14).TNF-BP-IIexhibits identity with the extracellular ligand-binding domain ofthe p80 TNF receptor(14).Thep60 and p80 TNF receptors show homology, but not identity, in the extracellular
ligand-binding
domain butpossessdifferent intra-cellular domains, suggesting alternative modes in signallinganddifferent function (13-16).
In our initial work on TNF-BP-I (7, 8) excessively high
levels ofTNF-BP-I were observed in serumofpatients with
renalfailure.This findingled us toinvestigatetherelationship
between serum TNF-BP-I and renal function as well as the
effectonTNF-BP-I of
hemodialysis
treatment.The resultsofthese
investigations
arepresentedinthisreport. Wedetectedatransient effect of
heparin
ontheserumlevelofTNF-BP-Iandourresultssuggest thatapool ofTNF-BP-Iispresent invivo from whichTNF-BP-Icanberapidlyreleased. Theaffinity for
heparin
wasinvestigated forboth TNF and TNF-BP-I andwefoundthat TNF representsa
heparin-binding cytokine
while TNF-BP-I showed no affinity forheparin.
Our observationsmay suggest the existence of mechanisms by which TNF is
restricted
to alocalfocusleading
toprotection against
itssys-temic effects. Methods
Materials. Heparin,
Fragmin®,
and protamine chloride were from KabiVitrum,Sweden.Heparin-SepharosewasfromPharmacia,Upp-sala, Sweden. Phorbol12-myristate 13-acetate(PMA)wasfrom
Sigma
Chemical Co.,St.Louis,MO. Endothelial cellgrowthfactorwasfrom
BoehringerMannheim, Bromma, Sweden. RecombinantTNF
(pro-ducedbyGenentechInc.,SanFrancisco, CA)andamonoclonal anti-body to TNF were kindly provided by Dr. Gunter
Adolf,
ErnstBoehringer Institute, Vienna,Austria. Native TNF-BP-Iwas
purified
asdescribed before and containedone30-kD chain under both reduc-ing andnonreducingconditions(8).Amonoclonalantibody
(TBP-1)
wasobtained by immunizing mice with purifiedTNF-BP-I
(Adolf,
G.R., andI. Apfler.Submitted for
publication.) Polyclonal
antibodiestoTNF and TNF-BP-Iwereobtainedbyimmunizingrabbits with
re-2026 M.Lantz, H. Thysell,E.Nilsson, andI.Olsson J.Clin. Invest.
© The American Society for Clinical Investigation, Inc. 0021-9738/91/12/2026/06 $2.00
combinant TNF and native TNF-BP-I, respectively.The antibodies againstTNF werespecific and did not cross-reactwith TNF-BP-Ior
lymphotoxin; neither did the anti-TNF-BP-I antibodies cross-react withTNF orlymphotoxin (1 1).
Collectionofserumsamples.Informedconsent wasobtained from allindividualsparticipatinginthisstudybefore collection of the sam-ples. Serumwaspreparedfromvenousblood drawn frompatientson regularhemodialysis,continuousambulatoryperitoneal dialysis,with different renal disordersnot ondialysis,andhealthyindividuals. The glomerular filtration ratewasdeterminedby useof 51Cr-EDTA (17) andserumcreatinineasdescribed(18). Sampleswerecollected from 19
patientsbeforehemodialysis,at30 min andatthe end ofdialysis.The
following membranes were used for dialysis: cuprophane, polysul-phone, polycarbonate,cellulosa acetate, polyacrylonitrile, and
poly-amide. Forcorrection of changes inserumTNF-BP-I duetoreduction of extracellular volumeduringdialysis,determination ofcystatinC of serum wasperformedasdescribed(19).Allpatientsonhemodialysis
received5,000-15,000Uof heparinatthebeginningofdialysis; sam-plesweredrawnbefore and after theinjectionofheparin.Inaddition, sampleswerecollectedatfrequentintervals fromhealthyindividuals
receivingintravenousinjectionsof500, 2,000,5,000,and10,000Uof heparin, 5,000 U ofFragmin",or20 mgprotaminechloride. Serum sampleswerealso collectedduringcirculation in vitro ofheparinized
blood. Fresh heparinized donor blood (500 ml) was recirculated
throughacuprophanedialyzerfor 2 h. The blood volumewaskept
constantbyinfusion ofaRingersolution.
Statistical parameterswerecalculatedbyuseofStatViewSE, Mac-intosh.
TNF-BP-I- andTNF-freesera wereprepared byapplyingpooled serumfrom healthy individualsto acolumnof
anti-TNF-BP-I-Sepha-rose equilibratedwith column buffer(0.15 mol/literNaCl, 5 mmol/
liter Hepes, pH 7.4)at aflowrateof 10 ml/h. Fractions of 10 mlwere collected and tested forthe presence ofTNF, TNF-BP-I, and anti-TNF-BP-Iwith ELISA technique.
Aninhibition ELISAfordeterminationofTNF-BP-I.Aninhibition ELISA, where TNF-BP-I competes for bindingto aconstantamountof
polyclonalanti-TNF-BP-I,wasdevelopedfor detection ofserum TNF-BP-I.Samples of 50 Mlwereloaded in 96-well round-bottom incuba-tionplates(2-62170; Nunc, Roskilde, Denmark).A 50-Mlaliquot of
polyclonal anti-TNF-BP-I diluted 1:4,000in incubation buffer(0.1 mol/liter NaCl,0.05mol/liter NaH2PO4,0.05% Tween-20,adjustedto pH 7.5 withNaOH)wasaddedtoeach wellfollowed by incubation
overnight at 4°C. In parallel, 96-well flat-bottom immunoplates
(4-39445;Nunc)werecoatedwith 100 Ml per well of TNF-BP-I dilutedto 7.5ng/ml in 0.1 mol/literNaHCO3, and incubatedovernight at room temperature followedby three washes inwashingsolution(0.15 mol/
literNaCl,0.05% Tween-20). Thecontentof the incubation plate was transferredtothe plate coated with TNF-BP-I whichwasincubated for 3 h at4°C and washed three times.Analiquot of 100Ml secondary
antibody,goat anti-rabbitantibody conjugatedtobiotin (Vectastain ABCkit,AK5001; Vector Laboratories Inc., Burlingame, CA), diluted 1:200 inincubation buffer supplemented with 1 mg/ml BSA was added toeach well. The plateswereincubated for 1 h atroom temperature followedbythree washes. To enhance thesignal, 100Mlofasolution
containingavidin-biotincomplexeswithalkaline phosphatase
conju-gated tobiotin(VectastainABCkit, AK5001; Vector Laboratories) wasadded toeach well. The plates were incubated for 1 h at room temperature, washed three timesfollowed by addition of phosphatase substrate(Sigma 104) dissolvedin 1 mol/liter diethanolamine buffer pH 9.8supplemented with 0.5
Mmol/liter
MgCl2, 100Ml/well.
The ab-sorbancewasmeasured at405nminanautomatic Titertek multiscan ELISA plate reader. Valueswere calculated from a standard curve basedonfreshlyprepareddilutions ofTNF-BP-I(0.3-30Mg/liter)
in TNF-BP-I-free poolserum.Thelowest detectable amount was 0.3jg/
liter.Sandwich ELISA technique for determinationofTNFand TNF-BP-I.Asandwich ELISA was developed for detection ofTNF, in incu-bation buffer and serum, and for detection of TNF-BP-I in cellular
supernatants containing 10% FBS. Nunc 96-well immunoplates (4-39445) were coated with monoclonal antibody to TNF or TNF-BP-I dilutedto2.5 ug/ml (40Ml/well) in 0.1 mol/liter NaHCO3 and incu-bated foratleast 3 horovernightatroomtemperature, followedby
three washes inwashingsolution(0.15 mol/liter NaCl,0.05% Tween-20). Samples of 100,lwereloaded in triplicate and incubated over-nightat40C followed by three washes in washing solution. Analiquot
of 100,lofapolyclonal antibodytoTNForTNF-BP-I, diluted1:3,600
in incubation buffer (0.1 mol/liter NaCl, 0.05 mol/liter NaH2PO4, 0.05% Tween-20, adjustedtopH 7.5 withNaOH) supplemented with 1 mg/ml BSA was addedtoeach well. Plates were incubated for 3 hat room temperature followed by three washes in washing solution. A peroxidase-conjugated goat anti-rabbit antibody (170-6513; Bio-Rad Laboratories, Richmond, CA) diluted 1:2,500 in incubation buffer sup-plemented with 1 mg/ml BSAwasprepared andanaliquot of 100M1
was added to each well. Plates were incubated for 1 h at room tempera-ture, washed three times inwashing solution followed by addition of
100,gl
oftetramethyl benzidine peroxidase (172-1066; Bio-Rad Labora-tories)toeach well. The absorbancewasmeasured at 660 nm inan automaticTitertekmultiscan ELISA plate reader. Valueswere calcu-latedfromastandardcurvebased onfreshlyprepared dilutions ofTNF (0.02-2Mg/liter) in incubation buffer or TNF-free serum. The standard for TNF-BP-I (0.03-3 Mg/liter) was diluted in RPMI 10% fetal bovine serum(FBS). The lowest detectableamountofTNFandTNF-BP-Iwas 0.02alg/literand 0.03 Mg/liter, respectively, and the detection of neither TNFnorTNF-BP-Iwasaffected byheparin.Investigation of heparin affinity. Heparin-Sepharose (0.5 g) was swollen and washed in 100 ml of distilledwaterfor 15min. Freshly preparedheparin-Sepharose (1-2 ml)waspacked in acolumn, equili-brated with column buffer (0.15 mol/liter NaCl, 5 mmol/liter Hepes, and 0.1% BSA, pH 7.4). TNF (15-150 ng) or TNF-BP-I (100-1,000 ng) wasapplied,allowed tobind for 30min,andelutedwith column buffer containing 0.2-1.5 mol/literNaClorheparin 1-100 U/ml, withaflow rateof 20 ml/h. Fractions of 1-3ml werecollected anddialyzed against 0.1mol/liter NaCl, 0.05 mol/L NaH2PO4, pH 7.5, followed by determi-nation of TNF or TNF-BP-I with ELISA technique.
Preparationofsupernatants from cells treated with heparin. Cells wereincubated at 1 x106/ml for 3, 8, and 24 h at370CinRPMI 1640 with 10% FBS in the presence or absence ofheparin, 50 U/ml, or PMA, 10ng/ml. Themedium of the human umbilical cord endothelial cells was supplemented with endothelial cell growth factor (200Mg/ml).The following cellswereobtained from American Type Culture Collection, Rockville, MD: HL-60, human promyelocytic leukemia; U-937, hu-manhistiocytic lymphoma; Hep G2, human hepatoma; HeLa, human epitheloid cervix carcinoma;A549, humanlung carcinoma and HU-VEC, human umbilical cord endothelial cell. PMN and mononuclear cells (mono)wereisolated from human heparinized whole blood by centrifugation inLymphoprep'according to the manufacturers
de-scription;NyegaardCo., Oslo, Norway. Samples were stored frozen at -20'C until analyzed with a highly sensitive and specific sandwich ELISAforTNF-BP-I.
Results
Serum
levels
ofTNF-BP-I inrenalinsufficiency.
TNF-BP-I
de-termined in serum from 10 healthy individuals was foundto varybetween5and10,ug/liter.Serumfrompatients on regular
hemodialysis or continuous ambulatory peritoneal dialysis contained 117±9.2
jig/liter
(X±SD), n=24, and 114±7.8,g/
liter(X±SD), n=14,TNF-BP-I, respectively. We investigated
ifthe increased levels of serum-TNF-BP-I in patients with
chronic renal failure couldbe explained by a decrease of
glo-merular filtration rate. Clearance of Cr-EDTA correlated nega-tively and S-creatininepositivelywith serum-TNF-BP-I (Fig. 1).However,thelevels ofserum-TNF-BP-I in patients on
A
0
* .0
00
0
VSO@
40 60 80 100 120 140 Cr-EDTA-clearance ml/min-1 (1.73m2)1
B
200 00 00 80100
S-Creatinine (jimol /L)
Figure1.The relationshipbetweenserumTNF-BP-I and the values of
Cr-EDTAclearance,r=0.506,P=0.0003, (A) andserumcreatinine, r=0.774,P=0.000 1,(B)in patientswith varyingglomerular
filtra-tionrate.
levels expected solelyonthe basis of glomerular filtrationrate.
Thus, factors in addition to kidney function may affect the level ofserumTNF-BP-I.
Effects ofhemodialysison serumlevelsofTNF-BP-L Dur-inghemodialysis the level of serum-TNF-BP-I increased
rap-idlyfrom 117±9.2,ug/liter (X±SD)toamaximum of240±8.2
,gg/liter
(X±SD) and decreased slowly to 160±11.3 Ag/liter(X±SD)attermination ofdialysis.Fig.2showsatypical time course for serum TNF-BP-I during hemodialysis. The
in-creasedlevelsofserumTNF-BP-I returnedtothe basal initial
levels after termination of dialysis (datanotshown). The maxi-malincrease ofserumTNF-BP-Iwasreached after5-15minin
allpatients investigated. Thesametimecourseforserum
TNF-BP-Iwasobserved independent onthetypeofdialysis mem-brane used. We furtherinvestigated the role of the dialysis membrane by circulating heparinizedblood in vitro and found
-i
~1-Im
a20
0-LL
z
2 cn,.
1001
120
180
Tmne
(nu*tes)
Figure2. Results from determinationsofserumTNF-BP-Iduring hemodialysiswithacuprophane dialyserinapatientwithchronic
renal failure.
that the levels of serum TNF-BP-I were unaffected (data not shown). Thus the increase in serum TNF-BP during dialysis is not causedby the dialysis membrane itself.
Theeffect ofheparinonreleaseofTNF-BP-Iintothe
circu-lation. Allpatients undergoing hemodialysis receivedabolus
injection of 5,000-15,000 U of heparin before dialysis. Actu-ally, the heparin injection by itself was found to double the level of serum TNF-BP followed by a slow decrease with time exactly as in patientsundergoing hemodialysis (Fig. 2).
In more than three independent control experiments the standard for TNF-BP-I was diluted in serum supplemented with increasing amounts ofheparin to rule out that the increase ofTNF-BP-I was caused by heparin itself. Results from one controlexperiment are given as absorbances for the 10 ng/ml value of the standard in the presence of 0, 10, 100, and 1,000
U/mlof heparin and the values were as follows: 0.585±0.080 (X±SD), 0.567±0.071 (X±SD), 0.546±0.039 (X±SD), and 0.597±0.051 (X±SD). No significant difference could be dem-onstrated in the values obtained in the presence of various amounts of heparin.
To rule out that the heparin-induced release of TNF-BP-I was restricted to patientswith renal failure, heparin was
in-jected intravenously into healthy individuals. A twofold in-crease in serum TNF-BP-I was observed within 5 min after
injection of2,000 U or more ofheparin(Fig. 3). Lowmolecular weight heparin, FragminO, also induced release of TNF-BP-I
intothecirculationbut theeffectwas less pronounced as
com-paredtothatofheparin (Fig.3). In oneexperimentheparinwas
injected repeatedly four times with 1-h intervals (Fig. 4). An
identical increasein serumTNF-BP-Iwas observed eachtime.
Theseresults suggest that TNF-BP-I is equilibratedbetweena
circulating and a storage pool. An alternative explanation is
30
Imt 20 30.
z
10°
-I~~~~~~~C
20J
0.
0 60 12018
iniiuaa
fe adinstato oftvaiou doesofhearn:50laiosi bewenhpri n
Smnvausof
eu N-PIi
U-z E 2
CO) 10
0
o
~~
~~do
1~o 18o0
Time(minutes)Figure3.Results from determinationofserumTNF-BP-1 inahealthy
individual after administrationof various dosesofheparin:500U
(.
.),
2,000U(o~~o),
5,000U(.
in), 10,000U(0
o),or5,000UofFragmin* (A~-t)
Adose-dependentre-lationshipbetweenheparinand5-mmnvaluesofserumTNF-BP-I is
2028 M.Lantz,H. Thysell,E.Nilsson,andI Olsson
-i
,30-0C
m
20-z
10.
cn n.
Im60
a)O40
E
22
1.
120
180
Time (minutes)
Figure 4.Results fromdetermination ofserumTNF-BP-I ofahealthy
individualwhoreceivedhourlyintravenousinjectionsof2,000U of heparinfour times(indicatedbyarrows).
thatheparin inducesproteolytic cleavageof cell surfaceTNF
receptors.However, treatmentofvarious cell lines with hepa-rin did not increase the release of TNF-BP-I into thecellular
supernatants (Table I). When protaminechloride (20 mg), a
competitive antagonisttoheparin, wasinjected 5 minbefore
administration ofheparin, the release of TNF-BP-I intothe
circulationwaspartiallyblocked(Fig. 5). However, injectionof anadditional dose ofheparin60 min later gaveafull response
onthe release of TNF-BP-I. Injection ofprotaminechloride
alone did not affect the serum level of TNF-BP-I (data not
shown).
Investigation of
heparin-binding
ofTNFand TNF-BP-L TNF-BP-Idid not show anyaffinityforheparin (Fig.6A).Ontheother hand TNF boundto aheparincolumn in 0. 15mol/ liter NaCland waselutedby0.5mol/literNaCl
(Fig.
6B).In aseparate experiment increasing concentrations ofNaCl were
used for elution and it wasfound that noelution took place
Table I. Production of TNF-BP-I In Vitro in the Presence orAbsenceofHeparin
Cells Control Hepann PMA
HL-60 0.22 0.18 0.60
U-937 0.24 0.25 2.9
HeLa 2.1 2.0 6.0
A549 0.65 0.60 3.0
Hep G2 0.06 0.06 0.20
HUVEC 0.16 0.15 0.29
Mono 0.17 0.15 0.25
PMN 0.25 0.27 0.50
Supernatantswerecollectedfrom HL-60, U-937, HeLa, A 549, Hep
G2,human umbilical cord endothelial cells (HUVEC), mononuclear cells(mono), andPMNafter24 hin the presence or absence of
heparin(50 U/ml)orphorbolester, PMA, (10 ng/ml) as describedin Methods.PMAserved asapositive control. The amount ofTNF-BP-I
wasdetermined with a sensitive sandwich ELISA as described in Methods.Values are given in
gg/liter.
-J
m 20
U-.I-E
10-2
U)
Cl)
II
II__'4ll
120
180
Time
(minutes)
Figure5. Effect of protamine chlorideonheparin-inducedrelease of TNF-BP-I into the circulation. Protamine chloride, 20 mg (arrowI),
wasinjected 5min before administration of 2,000 U ofheparin
(arrow II).Anadditional dose of 2,000 U ofheparinwasgiven 1 h later(arrow III).
untilthe salt concentrationwasincreasedto0.5 mol/liter(data
notshown). TNF could also be eluted from the heparin column by heparin itself (Fig. 6 C). No binding of TNFtoSepharose
alone was observed (data notshown). These results
demon-stratethat TNF binds specificallytoheparin in vitro.
Complexes of TNF and TNF-BP-I prepared by mixing
TNF and TNF-BP-I inaratio of1: 10wereappliedto a
heparin-Sepharose column. No TNF-BP-Iandonlyasmall fractionof the applied TNF was found to bind under these conditions (Fig. 7 A). Obviously complexes betweenTNF andTNF-BP-I lack affinity for heparin sincenoTNF-BP-I wasfound inthe TNF containing fractions eluted from the column. Actually it
waspossibletoelute TNF from the heparin-Sepharose column withasolution of TNF-BP-I (Fig. 7 B).
Discussion
The major findings of this work are that TNF representsa
heparin-binding cytokine and that TNF-BP-I, although lacking heparin-binding capacity by itself, is released into the circula-tion by injeccircula-tion of heparin. These observacircula-tions may be of importance for the understanding of how the biological effects ofTNFareexpressed and regulated.
Like other low molecular weight serum proteins, serum
TNF-BP-I was found to increase with decreasing glomerular filtrationrate.Patients with terminal renal failure had 10 to 20 times higher levels ofserum TNF-BP-I than healthy individ-uals. In addition,weobserved that serum TNF-BP-I increased rapidly but transiently during hemodialysis. The transient
in-crease of serumTNF-BP-I during hemodialysis turned outto
be caused by heparin which is routinely injected before dialysis. Intravenous administration of heparin to healthy individuals
alsoresultedin anincrease of serum-TNF-BP-I.
7E
6I--0)
>400
z
200
20
1 1 0
0.5 M NaCI
5 10 15
0.5 M NaCI
I'
A
0.2M OAM 0.6M ANaCI NaCI NaCI
E 3- A -40
0~~~~~~~~~4
U-o10 20
04
0 10 20
20
B
E
z
5 1b 15 20
b___~~T ii
0 5 10
Fraction no.
C
15 20
Figure 6. Investigation of the binding of TNF and TNF-BP-I to hep-arin.TNF(B and C) or TNF-BP-I (A) was applied to a 2-ml column of heparin-Sepharose which was eluted with 0.5 mol/liter NaCI (indi-catedbyarrowsinAandB) or 10(I),100(11),and 1,000 (III)U/ml
ofheparin (indicated by broken arrows inC).
heparin suggests ananalogy with other responses to heparin.
Lipoprotein lipase anddiamine oxidase are both rapidly re-leased into the circulation by heparin (20, 21). These sub-stancescanbindtoendothelial cells,abinding which is sensi-tivetoglycosaminoglycan degradingenzymes(22). However,
unlike lipoprotein lipase and diamine oxidase, TNF-BP-I lackedaffinity for heparin.Itis thereforeunlikelythat the re-lease ofTNF-BP-I is due to dissociation ofTNF-BP-I from heparan sulphateoftheendothelialcells as has been suggested
for
lipoprotein
lipase and diamine oxidase. The heparin-me-diated release ofTNF-BP-I ismorelikely explainedby an indi-recteffect ofheparinand not bydissociation ofa hypotheticalcomplex between TNF-BP-I and the extracellular matrix.
Any-how, ourdata suggestthat TNF-BP-Iispresentin vivo in a storage pool from which it canbe released by heparin. This
may alsoexplainthedifference in peaklevelsofserum TNF-BP-Ifoundforhealthy humansandpatients with chronicrenal
failure. Thus the basal levelsofserum TNF-BP-I mayreflect
thesize ofthesuggestedpool from whichTNF-BP-Iis released
andtherebyexplain theabovedescribeddifferences in the basal
levels ofserumTNF-BP-I. Thefinding ofrepeated releaseof
TNF-BP-I intothecirculation withintermittent injections of heparinalsostrongly supports thetheory ofastoragepool for
TNF-BP-I.Thispoolisinequilibrium with circulating
TNF-BP-Iand theequilibriumbetweenbound andcirculating TNF-BP-Iis affectedbyheparin.Theroleof heparin inthe releaseof TNF-BP-Iisstrengthenedby thefinding ofapartial inhibition
z1
I-B
Figure7. Prevention of binding of TNF to heparin byTNF-BP-I. TNF-BP-I(o o) and TNF ( )were mixed in a ratio of
10:1,applied to a 2-ml column of heparin-Sepharose, and eluted with
NaCiasindicated in A. In another experiment TNF alone was applied to acolumnofheparin-Sepharose, eluted with 200(I),400(II), and 600ng/ml(III)of TNF-BP-I (as indicated by broken arrows in B) followed by washing with 0.6 MNaCI(as indicated by a broken arrow,IV, in B).
of releaseofTNF-BP-Ibyacompetitive antagonisttoheparin, protamine chloride.
Whatmechanismsareresponsiblefortheheparin-induced
releaseofTNF-BP-I? Recentobservationsbased on immuno-logical cross-reactivity ofTNF-BP-Iwithp60transmembrane TNFreceptorsindicatethatTNF-BP-I isasoluble fragment of
the
p60
TNF receptor,which canbegenerated by cleavage oftheextracellular
ligand-binding
domain (1 1, 12). Inadditionresultsfrom sequencing ofcomplementaryDNAclones encod-ingthep60TNF receptorchain have shownthatthe
extracellu-larligand binding domain encodesthe soluble TNF-BP-I (13-16).There was noindicationthatalternative
splicing
isrespon-sible forthe
generation
ofTNF-BP-I.Ourresults have shown that theproduction
ofTNF-BP-Iisrapidly
increased whenthe TNF receptors aredownregulated incellsby activation ofpro-tein kinase C with
phorbol
esters(1 1). Heparin
maythereforeactby activation of
protein
kinaseCorby direct activation ofaproteaseleadingtocleavage ofthe
p60
TNF receptorresulting
in release ofTNF-BP-I.Thisexplanation
isunlikely
becausewecouldnotdemonstrateanydirect effectonthe releaseof TNF-BP-Ifrom cells treated with
heparin
invitro(Table
I).
Thefinding of
heparin affinity
forTNFhasimplications
forthe
understanding
ofregulation
oftheeffects ofTNF.Interac-tion of cellswith
heparin-like
componentsoftheextracellular matrix seems to beimportant
forregulation
ofgrowth
anddifferentiation.Thusseveral
growth
factorssuchasacidicandbasic fibroblast
growth factor, granulocyte macrophage
colony
stimulating factor, interleukin-3,
andinsulin-likegrowth
factorhave
heparin-binding
abilities(23-25). Obviously
aboundfac-torcanbeastimulus for
growth.
Growth factorsnecessary forhematopoiesis
arebeing synthesized by
thestromal cells ofthe bonemarrowand immobilizedoncomponents of the surfaceof these cells, such asheparan sulphate, before being presented tohematopoieticstemcells(24). Bindingtoheparan sulphate may protectagainstenzymatic degradation.It is also possible thatsequestration by molecules in the extracellular matrix can be a way of protection against systemic effects of cytokines.
Therefore,the presentfinding ofaheparin affinityfor TNF is ofbiological importance.It suggests that TNF may be retained by extracellular matrix,e.g., by heparan sulphatein physical
proximity withtarget cells such asfibroblasts,endothelial cells, andhematopoieticcells.However,injection of heparin did not result inincreasedserumTNF,which may beexplained bya local releasewithoutresultingin increasedserum TNF(data notshown).Atthe same time the extracellularmatrixcould act as abuffer and retain TNF to protect the host against its
sys-temiceffects, whichmay be harmfulasinendotoxic shock (3, 4). Itwillbeinterestingtoinvestigateifaffinityforheparinand extracellular matrix is a feature of othercytokinestoo.
The finding that heparin-bound TNF can be dissociated
withTNF-BP-Iisalsoofinterestbecause TNF-BP-I is a soluble
fragment ofthep60transmembrane TNF receptor(1 1). Thusa
highaffinity binding betweenTNF and TNF-BP-Iis expected.
Alow affinity bindingbetween TNF and aheparin-like
sub-stance wouldallowpresentationof bound TNF to a TNF re-ceptor on the same(autocrineregulation) or anadjacent cell (paracrine regulation) with concurrent dissociation and
spe-cificbindingtothe receptorwithtriggering ofabiological re-sponse.
Weconclude that TNF, butnotTNF-BP-I, has
affinity
for heparinand suggest thatbinding ofTNF tothe extracellularmatrix of cellsmayrestrict its actions. Thefinding that TNF-BP-Icoulddissociate TNFfrom heparin, suggeststhat TNF-BP-I may alsofunctionas acarrierforTNFinvivo. The
mecha-nismbywhich heparin inducesthe observedreleaseof TNF-BP-Iinto the circulationremainsto beelucidated.
Acknowledgments
This work was supported by the Swedish Cancer Society, John and Augusta PerssonFoundation,AlfredOsterlundFoundation, the Swed-ish Associationfor Physicians,and theMedical Faculty of Lund.
References
1.Carswell, E. A., L. J. Old, R. L. Kassel, S. Green, N. Fiore, and B.
William-son.1975.Anendotoxin-induced serum factor that causes necrosis of tumors. Proc.Nat!.Acad. Sci. USA 72:3666-3670.
2. Old,L.J. 1987. Tumornecrosis factor. Polypeptide mediator network. Nature(Lond.).326:330-33 1.
3.Beutler, B.,I.W. Milsark,and A. Cerami. 1985. Passive immunization against cachectin/tumor necrosis factor protects mice from lethal effect of endo-toxin. Science (Wash. DC). 229:869-871.
4.Tracey,K.J., Y. Fong, D. G. Hesse, K. R. Manogue, A. T. Lee, G. C. Kuo,
S. F. Lowry, and A. Cerami. 1987.Anti-cachectin/TNF monoclonal antibodies preventseptic shock during lethal bacteraemia. Nature(Lond.).330:662-664.
5.Sherry, B., and A. Cerami. 1988. Cachectin/tumor necrosis factor exerts endocrine, paracrine, and autocrine control of inflammatory responses. J. Cell. Biol. 107:1269-1277.
6.Seckinger,P., S. Isaaz, and J. M. Dayer. 1988. A human inhibitor for tumor necrosis factor. J. Exp. Med. 167:1511-1516.
7.Peetre, C., H. Thysell, A. Grubb, and I. Olsson. 1988. A tumor necrosis factor binding protein is present in human biological fluids. Eur. J. Haematol. 41:414-419.
8. Olsson, I., M. Lantz, E. Nilsson, C. Peetre, H. Thysell, A. Grubb, and G. Adolf. 1989. Isolation and characterization ofa tumornecrosis factor binding protein from urine. Eur. J. Haematol. 42:270-275.
9. Engelmann, H., D. Aderka, M. Rubinstein, D. Rotman, and D. Wallach. 1989.A tumornecrosisfactor-bindingprotein purified tohomogeneityfrom humanurine protects cells from tumor necrosis factor toxicity. J. Biol. Chem. 264:11974-11980.
10.Seckinger, P., S. Isaaz, and J. M. Dayer. 1989. Purification and biologic characterization of aspecifictumor necrosis factor inhibitor. J. Biol. Chem. 264:11966-11973.
1 1.Lantz,M., U. Gullberg, E.Nilsson,and I.Olsson. 1990. Characterization invitroofahuman tumornecrosis factorbindingprotein.Asolubleformof the tumornecrosisfactor receptor. J. Clin. Invest. 86:1396-1402.
12. Engelmann, H., D.N.Novick,andD.Wallach. 1989. Twotumornecrosis
factor-binding proteins purified fromhuman urine. Evidence forimmunological crossreactivitywith cellsurfacetumornecrosis factor receptors. J. Biol. Chem. 265:1531-1536.
13.Himmler, A., I. Maurer-Fogy,M.Kr6nke, P.Scheurich,K.Pfizenmaier,
M.Lantz, I.Olsson, R. Hauptmann, C. Stratowa, and G.R.Adolf. 1990. Molecu-larcloningandexpressionof human andrat tumornecrosis factor receptor chain
(p60)andits solublederivative,tumornecrosisfactor-binding protein.DNACell Biol. 9:705-715.
14.Nophar, Y., 0. Kemper, C. Brakebusch, H. Engelmann, R. Zwang, D. Aderka,H.Holtmann, and D. Wallach. 1990. Solubleforms of tumor necrosis factor receptors (TNF-Rs). The cDNA for the typeITNF-R, clonedusingamino acid sequence data of its solubleform,encodes both the cellsurface and a soluble
formof the receptor. EMBO(Eur.Mol.Biol.Organ.)J.9:3269-3278. 15.Loetscher, H.,Y.C.E.Pan,H. W.Lahm, R. Gentz, M.Brockhaus, H. Tabuchi, and W. Lesslaer. 1990. Molecular cloning and expression of the 55 kD
tumornecrosisfactor receptor.Cell.61:352-359.
16.Schall,T.J., M.Lewis,K.J.Koller, A. Lee,G. C. Rice, G. H. W. Wong, T.
Gatanaga,G.A.Granger,R. Lentz, H.Raab, et al. 1990. Molecularcloningand
expressionof receptor for human tumor necrosis factor. Cell. 61:361-370.
17.BrOchner-Mortensen, J. 1972. A simple method for determination of glomerular filtration rate.Scand.J.Clin. Lab. Invest. 30:271-274.
18.HeinegArd,D., and G.Tidestrom.1973.Determinationof serum creati-ninebyadirectcolorimetricmethod.Clin.Chim. Acta. 43:305-3 10.
19.Lofberg,H., and A.0.Grubb. 1979. Quantitation of gamma-trace in humanbiological fluids: indications for production in the central nervous system. Scand. J.Clin.Lab. Invest. 39:619-626.
20.Olivecrona,T., G. Bengtsson, S. E. Marklund, U.Lindahl,and M. H66k. 1977.Heparin-lipoprotein lipaseinteractions. Fed. Proc. 36:60-69.
21.Robinson-White,A., S. B.Baylin,T.Olivecrona,and M.A. Beaven. 1985.
Bindingof diamine oxidase activity to rat and guinea pig microvascular endothe-lial cells.J.Clin.Invest. 76:93-100.
22.Shimada,K., P. J.Gill, J. E. Silbert, W. H. J. Douglas, and B. L. Fanburg. 1981. Involvement of cell surface heparin sulfate in the binding of lipoprotein
lipasetoculturedbovineendothelial cells.J.Clin. Invest.68:995-1002. 23. Folkman, J., andM.Klagsbrun. 1987. Angiogenic factors. Science (Wash.
DC).235:442-447.
24. Roberts, R., J. Gallagher, E. Spooncer, T. D. Allen, F. Bloomfield, and
T.M. Dexter. 1988. Heparan sulphate bound growth factors: a mechanism for stromalcellmediated haemopoiesis. Nature (Lond.). 332:376-378.