Serum soluble interleukin 6 (IL-6) receptor and
IL-6/soluble IL-6 receptor complex in systemic
juvenile rheumatoid arthritis.
F De Benedetti, … , D Novick, A Martini
J Clin Invest.
1994;
93(5)
:2114-2119.
https://doi.org/10.1172/JCI117206
.
By using a sandwich ELISA, soluble human IL-6 receptor (sIL-6 R) levels were measured in
the sera of 20 healthy children and of 25 patients with systemic juvenile rheumatoid arthritis
(JRA). In patients with systemic JRA, serum sIL-6 R levels (114.6 +/- 37.7 ng/ml) were
significantly lower (P < 0.01) than those of healthy children (161.2 +/- 45.5 ng/ml). Serum
sIL-6 R levels were negatively correlated (r = -0.610, P < 0.001) with serum IL-6 levels
measured with the B9 cells. The serum IL-6/sIL-6 R complex was detected using an ELISA
based on a monoclonal antibody to IL-6 for capture and on a monoclonal antibody to human
sIL-6 R for detection. Healthy controls had little, if any, detectable serum IL-6/sIL-6 R
complex (OD 0.024 +/- 0.027), while the majority of patients with systemic JRA presented
measurable serum IL-6/sIL-6 R complex (OD 0.492 +/- 0.546). IL-6 levels estimated in the
circulating IL-6/sIL-6 R complexes were in the range of nanograms per milliliter and
approximately 20-fold higher than those measured by the B9 cells. Since serum C-reactive
protein concentrations were much more correlated with serum levels of IL-6/sIL-6 R
complexes (r = 0.713, r2 = 0.51, P < 0.0001) than with the serum IL-6 levels measured with
the B9 cells (r = 0.435, r2 = 0.19, P = 0.05), the large quantities […]
Research Article
Find the latest version:
Serum
Soluble Interleukin 6
(IL-6) Receptor
and IL-6/Soluble IL-6
Receptor Complex in Systemic Juvenile Rheumatoid Arthritis
FabrizioDe Benedetti, Margherita Massa, Patrizia Pignatti, SalvatoreAlbani, DanielaNovick,*and AlbertoMartini
Clinica Pediatrica, Universita degliStudi diPavia,Instituto di RicoveroeCuraaCarattereScientificoPoliclinico San Matteo,
27100Pavia,Italy; and*DepartmentofMolecular Genetics andVirology, WeizmannInstitute, 76100Rehovot, Israel
Abstract
By using a sandwich ELISA, soluble human IL-6 receptor
(sIL-6R) levelsweremeasuredintheseraof20
healthy
chil-dren andof 25patientswithsystemicjuvenilerheumatoidar-thritis
(JRA).
Inpatients withsystemic
JRA,
serumsIL-6R levels(114.6±37.7ng/ml)weresignificantly lower (P<0.01) than those ofhealthy children (161.2±45.5 ng/ml). SerumsIL-6 R levels were negatively correlated (r = -0.610, P
<0.001) withserumIL-6levels measuredwith the B9 cells. The serum IL-6/sIL-6 R complex was detected using an ELISA basedon amonoclonalantibodytoIL-6 for capture and on a monoclonal antibody to human sIL-6 R for detection. Healthycontrols hadlittle, ifany,detectableserum
IL-6/sIL-6Rcomplex(OD0.024±0.027),
while themajorityofpatientswithsystemic JRA presented measurableserumIL-6
/sIL-6
Rcomplex(OD0.492±0.546).IL-6 levels estimated in the
cir-culating
IL-6/sIL-6
Rcomplexes
wereinthe range ofnano-grams per milliliter and 20-fold
higher
than thosemeasuredby the B9 cells. SinceserumC-reactive
protein
concentrations were much morecorrelated withserumlevelsofIL-6/
sIL-6Rcomplexes (r
=0.713,
r2 = 0.51,P<0.0001)
than with the serum IL-6 levels measuredwith the B9 cells(r
=0.435,
r2=0.19,P=
0.05),
thelargequantities
ofserumIL-6 present inIL-6/sIL-6Rcomplexesappear to bebiologically relevantin
vivo, at least as far as the induction by IL-6 ofacute phase protein production.(J. Clin.Invest.
1994.93:2114-2119.)
Key words: interleukin 6*interleukin 6receptor *C-reactive protein*juvenile rheumatoid arthritis
Introduction
IL-6 hasmultiple
biological
effectsonimmune andinflamma-tory responses(for reviewseereference 1). Ithasbeenshown
to induce fever andhepatocyteacute phase response, to en-hanceproliferation and maturation ofBcellsaswell as immu-noglobulin production, andtoactivateTlymphocytefunction.
Elevatedlevelsof IL-6havebeenfoundinperipheralbloodor
otherbiological
fluids
inavariety of diseases, including bacte-rialandviral infections, neoplasia,trauma,andchronicinflam-matorydiseases(2).
The receptormediatingthebiological activities of IL-6 has been identified as twodifferent membrane glycoproteins, an
Address correspondence to Alberto Martini, M.D., Clinica Pediatrica, IRCCS San Matteo, P. le Golgi 2, 27100 Pavia, Italy.
Receivedfor publication16September1993.
80-kD protein, referred to as the ligand-binding protein (IL-6 receptor[IL-6
R]),'
and a 130-kD protein, referred to as the signal transducing protein (3, 4). cDNA encoding both mole-cules has been cloned and expressed and its biochemical proper-ties were analyzed (3-5). In particular, a geneticallyengi-neered truncated form of the 80-kD IL-6
R,
lacking the trans-membrane and the intracytoplasmatic domains, has been expressed successfully and has been shown to be able to asso-ciate with the 130-kD protein in the presence of IL-6 and to mediate IL-6 functions (6). Moreover, a soluble form ofIL-6 R (sIL-6 R) has been purified as a 50-kD protein from human urine(7) and has been shown to be present in human serum and to bind recombinant human IL-6 with a binding affinity similar tothatof theentirerecombinant IL-6Rmolecule (8). Juvenile rheumatoid arthritis (JRA) is a clinicallyheteroge-nousconditioncurrently divided into differentclinkalforms based on symptoms at onset (9). The systemic form is charac-terizedby chronic arthritis associated with highspikingfever,
other systemic features, includingevanescent cutaneous rash,
hepatosplenomegaly, lymphoadenomegaly, and serositis,and
withprominentlaboratory evidence ofinflammation (9).The other JRAonset forms arecharacterized mainly by chronic joint inflammation and are subdivided according to the num-ber ofjoints involved duringthefirst 6 moof diseasein pau-ciarticular(fourorlessjointsinvolved) orpolyarticularJRA
(fiveormorejoints involved)(9). We have shown previously that serum andsynovial fluid IL-6 levels are elevated in pa-tients with JRA during active disease and normalize during remission(10, 11).Inpatients withsystemicJRA,serum IL-6
levelsaremuchhigherthan those presentin patients with
poly-articular or pauciarticular JRA (10, 11) or in patients with adult rheumatoid arthritis (12, 13). In addition, in patients
withsystemicJRA, synovialfluid IL-6 levels are significantly
higher thanthose presentin patientswith polyarticular or pau-ciarticular JRA or inpatients with adult rheumatoidarthritis
(DeBenedetti,F., V.Gerloni,M.Massa, P. Pignatti, R.
Capo-rali, C. M. Montecucco, K. Matsushima, F. Fantini, and A.
Martini,
manuscript in preparation). Moreover, in systemicJRA,serumIL-6 levels are correlated with the extent and
sever-ity ofjoint involvementand withplatelet counts(10).Taken
together, these data suggestthat IL-6playsanimportantrole in the pathogenesis of systemic JRA. To better understand the
biologyof IL-6 anditsrelationto thesIL-6 R in systemic JRA, inthis studywemeasured serum sIL-6 R levels, evaluated their
relation with IL-6levels, and developed an assay system for the
detection of IL-6/sIL-6 Rcomplexesinsera ofpatients with
systemicJRA.
1. Abbreviations used in thispaper:AP,alkalinephosphatase; CRP,
C-reactiveprotein; HGF,hybridoma growth factor;IL-6R,IL-6
recep-tor;JRA, juvenilerheumatoidarthritis;sIL-6R,soluble IL-6 R.
2114 DeBenedetti,Massa,Pignatti, Albani, Novick, andMartini J.Clin.Invest.
©TheAmericanSocietyforClinicalInvestigation,Inc.
Methods
Patients. 60 children(meanage7.2yr, range2-17 yr)whofulfilledthe AmericanCollegeofRheumatology's (formerlytheAmerican Rheu-matismAssociation) proposed criteria for the diagnosis ofJRA(9)
wereincluded in thestudy. 25 had systemic, 15polyarticular, and 20
pauciarticularJRA. HLA B27orrheumatoid factor-positive(by Latex
agglutination test) patientswereexcludedfrom the study. All patients
werestudiedduring active diseaseasdefinedby thepresenceof
synovi-tison examination. Six patients werereceiving no treatment. The othersweretreated with nonsteroidal antiinflammatory drugs,
asso-ciatedinapproximatelyhalfofthemwithmethotrexate (singleweekly
dose) and/or prednisone (onanalternate-day regimen inthe majority
ofcases). Duringthecourseof thisstudy,somepatientsweretested morethanonceduringdifferentphasesofthediseaseorduring
differ-enttreatments.Sera obtained from 20healthychildren(meanage8.1
yr, range3-18yr), hospitalizedfor bonemarrowdonationorfor minor
surgical procedures,wereusedascontrols. Permission fordrawingof extrabloodduringroutinevenipuncturewasobtained fromparentsof all children.
Bloodwasallowedtoclot for2 h at roomtemperature and
centri-fugedat1,500gfor 10min.Serawerestored at-70'Cuntil used. 12
synovialfluidsamples,3 frompatientswithsystemicJRA and 9 from
patientswithpauciarticular JRA,obtainedbefore intraarticular steroid
injectionwerealso studied.Synovialfluidswerecentrifugedatl0,OOOg
for 10minand stored at-70'Cuntiluse.
AssayprocedureforsIL-6R.sIL-6Rlevelsweremeasuredusinga
sandwichELISA,basedontheuseof monoclonal antibodies 34.14 and alkaline phosphatase (AP)-conjugated 22.3. These two monoclonal antibodies were selected for the ELISA because previous studies showed thattheybindtodifferentepitopesonthe sIL-6 R molecule
( 14). Polystyrene plates (Costar Corp., Cambridge, MA)werecoated
overnightat4°Cwith monoclonalantibody34.14(20,g/mlinPBS),
andnonspecific bindingsiteswereblockedbyfurther incubation with
PBS/2%BSA for1h at37°C.Serumandsynovialfluidsampleswere addedata1:100dilution inPBS/1%BSA,and theplateswere incu-bated for 3 h at37°C.Natural sIL-6R,purifiedtohomogeneityfrom human urine asdescribedpreviously ( 15), was usedas astandard.
Afterwashing, plateswereincubated for 2 h atroomtemperaturewith
AP-conjugated monoclonal antibody 22.3 (0.5 ug/ml in PBS/1% BSA).Thesubstratep-nitrophenyl phosphatewasadded,andthe
opti-caldensity read at 405 nM. The detection limit of the ELISA is 40
pg/mlofpurifiedhuman sIL-6 R.
DetectionofIL-6/sIL-6Rcomplex.To detect thepresenceof IL-6/
sIL-6 Rcomplex, weconstructedan ELISA based ontheuse ofa monoclonalantibodytohuman IL-6(34.1 ) (kindly provided by
Inter-pharm Laboratories, Nes-Ziona, Israel) ( 16)forcaptureanda mono-clonalantibodyto the sIL-6 R(AP-conjugated 22.3)for detection. Monoclonalantibody22.3 waschosen because itwasshown not to inhibit thebindingof radiolabeled humanrecombinant IL-6 tohuman
natural sIL-6 R( 14). Polystyrene plateswerecoatedovernightatroom
temperaturewithmonoclonalantibody34.1 to IL-6ataconcentration of 50tg/mlinPBSorwith PBS alone. Afterblockingthenonspecific bindingsites witha2-hincubationat37°CwithPBS/2% BSA,each testserum,diluted 1:1 inPBS,wasadded to wells coatedwith monoclo-nalantibody34.1 to IL-6and to wells coated with PBSalone,and the
plateswereincubated for 2 h at37°C.Plateswerenotwashed,and the
AP-conjugatedmonoclonalantibody22.3tothesIL-6 Rwasaddedto
thewellscontainingtestsamplesatafinalconcentrationof 1.3,g/ml.
Theplateswereincubated for 1hat 37°Cwithgentle swirling.This
approachwaschosen becausepreliminary experimentswithpatients'
seraand withnormalserainthepresence of exogenousIL-6 showed thatwashingtheplatesfollowedbya 1-hincubationwith AP-conju-gated22.3 inPBS/1%BSA resulted inamarkeddecrease in theoptical density, conceivablybecause of dissociation of theIL-6/sIL-6R
com-plex.Afterwashingtheplates,theassaywasdevelopedasdescribed for the sIL-6 R ELISA. Thebackground optical density usingPBS/2%
BSAasatestsamplewasalwaysbelow 0.190. The resultsareexpressed
asspecific optical density obtainedby subtractingthenonspecific opti-cal density found in wells coated with PBS alone to the optiopti-cal density observed in wells coated with the monoclonal antibody 34.1 to human IL-6. To comparethe levelsof IL-6 present in the IL-6/sIL-6 R com-plexwith the amount ofIL-6 measured by thehybridoma growth factor (HGF) assay (see below), the amountof IL-6wasextrapolated from a standardcurveobtained by addingincreasingconcentrations ofhuman recombinant IL-6 derived from Chinese hamster ovary cells (Genzyme Corp., Cambridge, MA)to a referenceserum fromahealthyadult control. Thisserumhad 100 ng/ml of sIL-6 R and contained neither detectable IL-6 in the HGF assaynordetectableIL-6/sIL-6Rcomplex
(specificOD0.000).Toobtainacceptableresolutionover awide range
ofIL-6concentrations, opticaldensityat405 nMwasread after
differ-enttimes ofincubationwith the substratep-nitrophenyl phosphate (see Results andFig. 5).
AssayforIL-6. IL-6 levels weremeasured using the hybridoma cell line B9 (kindly provided by Dr. L. Aarden, Netherland Red Cross, Amsterdam), asdescribed previously (10). Chinese hamster ovary cell-derived recombinant human IL-6 (Genzyme Corp.)wasusedto
culture B9 cells and as a standard in the assay.L-6 levels were ex-pressed, unless otherwise indicated,in HGFunits-(HGFU), where 1 HGF Uis the amount ofIL-6 required to achieve half-maximal prolifer-ation of B9 cells. The conversion from HGF U/ml to pg/ml was per-formed usingthe amount ofhumanrecombinant IL-6 required to achieve half-maximal proliferation of B9 cells in each assay, that is,
- 4.5 pg/mlof human recombinant IL-6.
Statistical analysis.Data wereanalyzedusingtheMann-Whitney
Utestforunpairedsamples and theSpearman correlationcoefficient.
Results
sIL-6Rlevelsinpatients with JRA.Patients with systemicJRA presented serum sIL-6 R levels (114.6±37.7 ng/ml)
signifi-cantlylower(P < 0.01 ) thanthose found in healthy children
( 161.2±45.5ng/ml),whilepatients with polyarticularor pau-ciarticular JRA had serum IL-6 R levels (176.7±66.9 and
165.9±51.5 ng/ml, respectively) comparable with those of healthychildren (Fig. 1). To rule out thepossibility that the
lowserumsIL-6 R levelsinpatients with systemicJRA could be due to thebindingofendogenousIL-6present in patients' sera tothe sIL-6R,thusmaskingthe epitope recognized by the monoclonalantibodiesusedinthe ELISA, we tested theeffect
of preincubation of normal sera
:with
exogenous IL-6. Asli
E
300-CM
250-n -i
LUJ
J200 --J
r
(D 150
--J
100-w 50
0n
St.
CTRL SYST POLY PAUCI
JRA
Figure1.Serum sIL-6Rlevels inhealthycontrols
(CTRL)
and in patientswithsystemic (SYST),polyarticular
(POLY),
orpauciarti-cular(PAUCI)JRA.Thehorizontal continuous linerepresentsthe meanofeach group.
-4:"
4,,0
I
0
t 0
shown in Fig. 2 for two representative sera, the addition of 20 ng/mlofhumanrecombinant IL-6did not have an effect on the measurement of the sIL-6 R in the assay system.
Whenpatients with JRA weredivided accordingtotheir
treatment, nodifferences in serum sIL-6 R levels were observed
(data not shown), thus suggesting that nonsteroidal antiin-flammatory drugs, prednisoneormethotrexate, donot have a
directeffect on the release of sIL-6 R in vivo. No significant
correlationof serum sIL-6 R levelswith erythrocyte
sedimenta-tionrate values or with C-reactive protein (CRP)
concentra-tionswas found in the three JRA onset types (data not shown). Inpatients with systemic JRA,serum sIL-6Rlevelswere
negatively correlated (r=-0.610,P <0.001 ) withserumIL-6 levelsmeasured with the B9 cells (Fig. 3).Tofurther evaluate the relationship between serum IL-6 andsIL-6 Rlevels, we measuredIL-6 and sIL-6Rlevels inserumsamplesobtained during the'febrile peakfrom twopatientswithsystemicJRA. AsshowninFig.4foronerepresentative patient, the increase inserumIL-6 levelswasassociated withadecreasein sIL-6
R,
whose levelswereinverselyrelated to body temperature.sIL-6 Rwasalso found in synovial fluid samplesat nano-gram permilliliter concentrations withnoevident difference
amongpatients with systemic or pauciarticular JRA. As
re-portedpreviously ( 10,11), while IL-6 levelswere muchhigher insynovialfluidsthaninthecorrespondingserum,thiswas not truefor sIL-6 R,
suggesting
that, unlike IL-6, sIL-6Risnotpreferentially producedat,and thus releasedfrom, sites of in-flammation (Table I).
Presence of circulating IL-6/sIL-6Rcomplexinpatients
with
systemicJRA. To identify the circulating IL-6/sIL-6 Rcomplex,weconstructedanELISA basedonthe use ofa
mono-clonal
antibody
tohuman IL-6(34.1 ) forcaptureandamono-clonal
antibody
tothesIL-6R(AP-conjugated
22.3) for detec-tion of theIL-6/sIL-6Rcomplex.Tocontrolforpossiblenon-specific reactivity
secondarytothebinding of theserumsIL-6Rtoplastic,all seratestedwereincubated in wells coated with PBS alone and in wellscoated with the monoclonal
antibody
34.1toIL-6.Alow backgroundwasfound withnosignificant
differences in thebackgroundoptical
density obtained in wells coated with PBS alone between sera from patients withsys-temicJRA(n =25) andserafrom healthy controls (n = 12) (0.030±0.031 and 0.021±0.020,
respectively;
P=0.79). The0
I.--J Li
-J (0
-j
Li (/)
100
50
* .0 0 * 0
* 0
0 * * *
. 0
80 110 140 170 200
SERUM sIL-6 R LEVELS (ng/ml)
Figure 3.Relationship between serum IL-6 and serum
sIL-6
Rlevels in patients with systemic JRA. SerumIL-6levels were measured using anHGF assayasdescribedin Methods. The dottedlinerepresents thedetectionlimit of the HGF assay.r = -0.618;P<0.001.addition
ofincreasing concentrations of human recombinant IL-6to normal seraresulted ina dose-dependent increaseinspecific optical density, demonstratingthat this assay system is
abletodetect theIL-6/sIL-6R
complex
formedby
thebinding
ofexogenous IL-6 to its soluble receptor in human serum.Dose-response curvesobtainedatdifferent times ofincubation with the substrate for the referenceserum areshown inFig. 5.
Asshown in
Fig.
6, healthy children had little ifany detect-able IL-6/sIL-6 R complex in their serum(specific
OD0.024±0.027),
while themajority
ofserafrompatients
withsystemic
JRAhad detectable levels ofIL-6/sIL-6 Rcomplex(specific
OD0.492±0.546)
withahighly significantdifferencewithserafrom healthy controls (P<0.0001).
Comparison
of IL-6 levelsmeasuredby the HGFassayand thoseestimated in theIL-6/sIL-6
Rcomplex,asmeasured by theELISA,
in the same 24 serumsamples,
lead us tounexpected
results. Asshown in TableII,in
patients
withsystemic
JRA,serumIL-6 levelsestimated in the IL-6/sIL-6Rcomplexwerein therangeofnanograms permilliliter andweremuch
higher
thanthose measuredby the HGFassay(mean ratio IL-6 inIL-6/sIL-6
Rcomplex/IL-6 measured by the HGFassay:22.4). Of
interest,
when thesamecomparisonwasmadefor IL-6 levelspresentinsynovial
fluidsamples,
suchadifferencewas notobserved (Ta-bleII).
Moreover,comparison
of the correlation between IL-6 levelsestimated by thetwoassaysystemsinsera orinsynovial
fluids showedthatresultsby thetwoassay systems arestrictly correlatedinsynovialfluid
samples (r2 =0.78) butare muchSERUM 1
SERUM 2
0.0 0.5 t.o
O.D-Figure 2. Effect of preincubation of normal sera with recombinant human IL-6(20 ng/ml) on the detection of sIL-6 R. Normal sera wereincubated at 37°C for 45 min in the absence or in the presence of 20 ng/ml of recombinant human IL-6 and then tested in the ELISA for sIL-6R asdescribed in Methods. Results from two representative
sera areshown. Background, white box; no IL-6, dotted box; IL-6 (20 ng/ml), striped box.
E
Co
c
a:
ni
100-
90- 80- 70- 60- 50-40
400-
t-\
300-C, I
*s
200-J
D
100-0
- 18 20 22 24
HOURS
-39 C.)_ -38 O
0.-0
0
ui
-37
I-a o
136
2 _
Figure 4. Serum IL-6 andsIL-6Rlevels during the febrile peak in
representative patientswith systemic JRA. Body temperature, closed squares;serumIL-6, closedtriangles; serum IL-6 R, closed circles.
2116 DeBenedetti,Massa,Pignatti,Albani,Novick, andMartini
-TableI.IL-6andsIL-6 RLevelsinSynovial Fluid and
intheCorresponding Serum Samples ofRepresentativePatients
withSystemic orPauciarticularJRA
sIL-6 R IL-6
Synovial Synovial
fluid Serum fluid Serum
ng/ml HGFU/ml Systemic
1 64.8 99.4 18500 100
2 142.6 148.9 13000 39
3 138.3 168.9 5600 19
Pauciarticular
1 175.6 185.1 5800 18
2 80.2 148.1 6200 <4
3 126.7 167.1 7600 <4
4 60.0 134.9 600 <4
lesscorrelatedin serum (r2= 0.44)(Fig. 7). Taken together,
the datainTableIIand Fig.7 suggestthatquantities of IL-6
presentinserum aremuch
higher
than thoseestimated by theHGF assay andthatthefactor(s) interfering with the detection of IL-6 by the HGFassayispresentinserum,butnotin syno-vial fluid.
To evaluate the in vivo
biological
relevance ofthe large quantities of IL-6 estimated incirculating
IL-6/sIL-6 R com-plexes, we compared thecorrelation betweenserumCRPcon-centrations and the levels ofthe IL-6/sIL-6Rcomplex with the correlation betweenserumCRP concentrations andserum
IL-6 levels measuredwith theHGFassay. Asshown in
Fig.
8, CRP concentrationsweremuch morecorrelated withthelevelsofthecirculating IL-6/sIL-6 Rcomplex (n = 24,r = 0.713, r2
=0.51, P<0.0001 ) than with theserumIL-6levelsmeasured
with the HGFassay (n= 24,r= 0.435, r2 =0.19, P=0.05).
Discussion
InthisstudywehavemeasuredserumsIL-6 R levels in
patients
with JRA anddeveloped
an ELISA for the detection of theUt) 1.0
z
0.011
0~
UJ 0A. 00
0.002
0.04 0.1 1 10 100
EXOGENOUS HUMAN REC. IL-6 (ng/ml)
Figure 5. Dose-responsecurvesof the ELISA for theIL-6/sIL-6 R
complex. Results fromonerepresentativeserumfromahealthyadult controlareshown. Thetestsamplewasincubatedat37°Cfor 45 min in the absenceorin the presence ofincreasingconcentrations of humanrecombinant IL-6 (range 0.062-128 ng/ml) and then tested in theELISAfor the IL-6/sIL-6 Rcomplexasdescribed in Methods. Theoptical densityat405nMwasreadaftera30-min(closed square),a1-h(closed triangles),anda4-h(closed circles)incubation withthesubstratep-nitrophenylphosphate.
U)
z
wL
0 -J
0
C-)
CLA
0-C/)
2.0
1.5
1.0
0.5-0.0
Figure 6.Serum IL-6/sIL-6 R complex in healthy controls
(CTRL)and inpatientswith sys-* temic JRA (SYSTJRA). The re-: sultsareexpressedasspecific
opti-caldensity calculatedasdescribed . s...----.k--- in Methods. The horizontal dotted
CTRL SYST line represents the upper limit of
JRA controls (mean +2 SD).
IL-6/sIL-6 R complex in biological fluids, demonstrating that ahighamountof IL-6 circulatesas anIL-6/sIL-6Rcomplex. The serumsIL-6 R levelsfound in our age-matched con-trols areapproximately twofold higher than thosereported
re-cently byHonda et al.(8) and by Gaillardetal.(17)in healthy adults. Thesedifferences could be because ofdifferencesin the
antibodies usedorpossibly differences inthe ageofthe popula-tiontested.Indeed,wefoundthat serumsIL-6Rwashigherin ourage-matchedcontrolpopulation(mean level 161.2ng/ml) thanin a group of healthyadults (n = 10; mean level 104.5 ng/ml).
Inour study, among patients with JRA, only those with
systemicJRA werefoundto have serum sIL-6 R levels
signifi-cantly lowerthanthose ofhealthy controls comparable with age. As far as weknow,only twostudiesareavailable on serum
sIL-6Rlevels in humandiseases. Patientswith HIVinfection and patients with multiple myeloma have been reported to
have serum sIL-6 R levels significantly higher than healthy
controls(8, 17). These datasuggest thatdifferentmodulation ofthe invivo levels of sIL-6Roccursin differentdisease condi-tions. This maypossibly be because of quantitative differences inIL-6production in vivo.Indeed, serum IL-6 levels foundin patients withHIVinfectionorwith multiplemyeloma (18-20) are muchlowerthanthose presentin patientswith systemic JRA(10).Apossibleroleofthe increased production ofIL-6 indownregulating the levels of sIL-6 R in patients with
sys-temicJRAis suggested by the finding ofanegative correlation
TableII.Comparisonofthe Levels ofIL-6Measured by theHGF Assayand Those EstimatedintheCirculating IL-6/sIL-6R ComplexinSerafromPatients with SystemicJRA andin
SynovialFluidsfromPatientswith Systemic orPauciarticularJRA
HGF assay IL-6/sIL-6RELISA
ng/ml ng/ml
Serum
(n=24) 0.335±0.334 6.850±8.971
(0.022-1.44) (0.26-32.0)
Synovial fluid
(n= 12) 45.2±43.2 39.8±35.4
(1.4-150.0) (1.6-118.0)
24 serafrompatientswithsystemicJRA and 12synovialfluid
sam-ples, 3 frompatientswithsystemicand 9frompatientswith pauciar-ticular JRA,weretested in both assay systems. Resultsareshownas
A
r=0.665
R2=0.44
0.1
***
.0
0 * * *
3 10 100
SERUM IL-6 (HGF U/ml)
B
r=0.881
R2=0.78
0.1
0.0 10
0
0
300 1000
0
10000
SYNOVIAL FLUID IL-6 (HGF U/mi)
Figure7.Comparisonof the correlation between IL-6 levels measured
bythe HGFassaywiththe levels ofIL-6/sIL-6Rcomplexinserum
(A)and insynovialfluid(B).
ever, preincubation of control sera withincreasing concentra-tions of exogenous human recombinant IL-6 resulted in a
dose-dependent
increase inoptical density, demonstrating
that thisassay systemis abletodetect theIL-6/sIL-6
Rcomplex
formed
by
thebinding
ofIL-6toits solublereceptorpresentin human serum. This also confirms thefindingsof Honda et al.(8)
and ofGaillardetal.( 17)thatcirculatingsoluble IL-6 R isabletobind IL-6. Almost allseraof
patients
withsystemic
JRAdisplayed high optical density
inthisassaysystem,demonstrat-ingthepresenceof elevated serumIL-6/sIL-6Rcomplex lev-els. These resultssuggest
strongly
thatIL-6,
oratleastpartofit,
circulates bound toits soluble receptorin patients with
sys-temicJRA. To the best ofourknowledge,this isthe first demon-stration of thepresenceofcirculating cytokine/soluble
cyto-kinereceptorcomplexesin human diseases. Our data are con-sistent with the following hypothesison thebiology ofIL-6: IL-6 produced at sites ofinflammation (e.g., joints) or
pro-duced in theperipheralblood inresponsetostimuli releasedby inflammatorysites(e.g., IL-lorTNF-a)binds the serum sIL-6 R and it is vehiculated in the blood as anIL-6/sIL-6 R
com-plex.Thisbinding may protectIL-6 fromprotease
degradation,
in a manner similar to what has been described for IL-4 and the
soluble IL-4receptor(24), and/or stabilize IL-6 bioactivity,as has been demonstrated for TNF-a and its soluble receptors (25). TheIL-6/sIL-6 R complex will thus associate withthe 130-kDsignal transducing proteinonthe surface oftarget cells and mediate IL-6bioactivities.
Comparison ofserum IL-6 levels measured by the HGF
assay andthose estimated in circulating IL-6/sIL-6 R
com-betweenserumsIL-6 R levels withserumIL-6 levels, measured
with the HGF assay, and by the decrease in serum sIL-6 R
levels thatwasassociated with the increase inserumIL-6 levels
during the febrile peak. The mechanisms of the production of sIL-6Rhavenotyetbeen clarified (21, 22). Nevertheless,our
datasuggestthattwopossible mechanisms,notnecessarily
mu-tually exclusive, may account for the decrease in theserum
sIL-6 R levels in patients with systemic JRA: (a) since IL-6 has
beenreportedtoinhibittheexpressionof the IL-6 Rgeneboth
in vivo and in vitro (23), adirect effect of increased in vivo
IL-6 productionontheexpression ofthe IL-6Rgeneis
conceiv-able; and (b) the decrease inserum sIL-6 R levelsmaybea
consequenceofaconsumption ofcirculating sIL-6Rduetothe
formation of IL-6/sIL-6 R complexes in thepresenceofIL-6
andof the subsequent binding andinternalizationofthe
com-plex bytarget cellsexpressing the 130-kD signaltransducing
protein.
This second hypothesis is based on the assumption that
circulating IL-6/sIL-6Rcomplexesarepresentinpatients with
systemic JRA. Since theserumsolubleformof IL-6Rand the
genetically engineered truncated form ofIL-6 R have been
showntobind IL-6 withabinding affinitysimilartotheentire
80-kD molecule (6, 8)andgiventheserumconcentrationsof
the sIL-6 R found in thisstudy,it is conceivabletohypothesize
that IL-6 circulates boundtoits solublereceptor.Toverify this hypothesis, we constructed an ELISA based on the use ofa
monoclonal antibody toIL-6 forcaptureand amonoclonal
antibodytosIL-6 Rfor revealingthepresenceof the
IL-6/sIL-6 Rcomplex. Using this approach we found noevidence of
circulatingIL-6/sIL-6 R complex in healthy controls.
How-r=0.435 R2=0. 19
p=0.05
E
D
l
en
D,
100
10
A
0
0 * .
0
* 0 0
.0 *
0
0.6 10
SERUM CRP (mg%)
a
C:
co
-J
en
r=0.713 R2=0.51
p<0.0001
1.0
0.1
0.0
0.6 1
B
.
0
-@00
0000
* *. *
*
10
SERUM CRP (mgX)
Figure8.Correlation ofserumCRPconcentrations withserumIL-6
levels measuredby the HGFassay(A)orwith circulatingIL-6/sIL-6 Rcomplexlevels(B)inpatients with systemicJRA.
2118 De Benedetti, Massa, Pignatti, Albani, Novick, and Martini
a
0
Ix
-J
W
(D
-J
D en
Co o
J
(O
N CD
-J
0
-J
La.
-J
0 z
(n
.
plexes led us totheunexpected finding that, in patientswith
systemic JRA, IL-6 was present in nanogram per milliliter amountsinthecirculating IL-6/sIL-6 R complexes compared withthefewhundreds of picogramspermilliliter measured by
the HGF assay. Therefore, ourdata suggestthat in systemic
JRA, butalsopossibly in other disease conditions, IL-6
circu-lates in much higher amounts than those measured by the
HGFassay.Thepossibility that the HGFassaydetects onlya
portion of IL-6presentin humanserumis alsosupported by
theconsistent findingthatIL-6 levels measured with the
hepa-tocyte-stimulating
factor assayhave been found tobemuchhigher than those measured by the HGFassayina
variety
of disease states(26-29). More recently, May et al. (30) have demonstratedthat the bulk of IL-6presentinperipheral
blood circulates intwohigh molecular weight complexes (100-150and400-500 kD) and that
this
IL-6isnotdetectedby the HGFassay (30). Ofnote, amongtheproteinspresentinthe 100-150-kD complex, a 50-kD material was present, consistent with the molecular weight of the human natural sIL-6 R.
Taken together, the
discrepancies
betweenserum IL-6 levels measuredby the HGFassayandby thehepatocyte-stimulating factorassay(26-29), ourresults, and thoseof
Mayetal.(30) raisetheissue of the biological role ofthelargequantities of IL-6 present in the peripheral blood. In ourstudy, thestrict correlationof
serumIL-6 levels measured in the IL-6/sIL-6 R complex ELISA withCRP concentrations in patients withsys-temic JRAsuggests thatthe
high
amountsof
IL-6presentin thecirculating IL-6/sIL-6
Rcomplex
arebiologically
relevant in vivo,at least as farasthe induction by IL-6of
acutephase proteinproduction.Inconclusion, in this studywereportedadecrease in sIL-6
Rlevels inpatients with
systemic
JRA,adisease characterized byamarked increase in serumIL-6 levels. Wedeveloped
an assay systemfor detectionand measurementofcirculating
IL-6/sIL-6Rcomplexes and demonstrated that IL-6
circulates
as anIL-6/sIL-6Rcomplex. We have alsofound
thatlargequan-tities of IL-6,notdetectedby the HGFassay, are presentin the
peripheral
blood ofpatients
withsystemic
JRAasIL-6/
sIL-6Rcomplexes, whose in vivo
biological
relevanceis suggested by their strict correlation withserumCRP concentrations.Acknowledgments
This workwassupportedby theIstituto di Ricovero e Cura a Carattere
ScientificoPoliclinico San Matteo, Pavia, Italy and by the Istituto di
RicercaCesare Serono, Ardea, Italy.
References
1. Van Snick,J. 1990. Interleukin 6:anoverview. Annu. Rev. Immunol.
8:253-278.
2. Hirano, T., S. Akira, T. Taga, and T.Kishimoto. 1990. Biological and
clinicalaspectsof interleukin6.Immunol.Today.11:443-449.
3.Yamasaki,T., T. Taga, Y.Hirata,H.Yawata,Y.Kawanishi,B.Seed,T.
Taniguchi,T.Hirano,andT.Kishimoto.1988. Cloning and expression of the humaninterleukin-6(BSF-2/IFN#2)receptor.Science (Wash.DC). 241:825-828.
4.Taga, T., M.Hibi,T.Hirata,K.Yamasaki,K.Yasukawa,T.Matsuda,T.
Hirano,and T.Kishimoto. 1989. Interleukin 6triggerstheassociation of its receptor with apossiblesignal transducer, gp 130.Cell. 58:573-579.
5. Yawata, H.,K.Yasukawa, S.Natsuka, M.Murakami,KYamasaki,M.
Hibi,T.Taga, and T. Kishimoto. 1993.Structure-functionanalysis ofthe human IL-6 receptor dissociation of amino acid residues requiredforIL-6binding and for IL-6signaltransduction throughgpl30.EMBO(Eur. Mol.Biol.Organ)J. 12:1705-1712.
6.Yasukawa, K.,T.Saito,T.Fukunaga,Y.Sekimori,Y. Kishihara,H.Fukui,
Y.Osugi,T.Matsuda,H.Yawata,T.Hirano,etal.1990.Purificationand
charac-terization ofsoluble humanIL-6receptorexpressedin CHO cell. J.Biochem. 108:673-681.
7.Novick, D., H. Engelmann, D. Wallach, and M. Rubinstein. 1989. Soluble
cytokinereceptorsarepresent in normal human urine. J.Exp.Med. 170:1409-1414.
8.Honda, M., S. Yamammoto, M. Cheng, K. Yasukawa, H. Suzuki, T.Saito,
Y.Osugi,T.Tokunaga,and T.Kishimoto. 1992. Human soluble IL-6 receptor itsdetection and enhanced releasebyHIV infection. J. Immunol. 148:2175-2180.
9.Brewer, E. J., Jr., J. Bass, J. Baum, J. T. Cassidy, C. Fink, J. Jacobs,V.
Hanson, J. E. Levinson, J. Shaller, and J. S. Stillman. 1977. Current proposed revision of JRA criteria. Arthritis Rheum.20(Suppl.):195-199.
10. DeBenedetti, F., M. Massa, P. Robbioni, A. Ravelli, G. R. Burgio, and A. Martini. 1991. Correlation of serum interleukin 6 levels with joint involvement andthrombocytosis in systemic juvenile rheumatoid arthritis. Arthritis Rheum. 34:1158-1163.
11. DeBenedetti, F., P. Robbioni, M. Massa, S. Viola, S. Albani, andA.
Martini. 1992. Serum interleukin 6 and jointinvolvementin polyarticular and
pauciarticular juvenilechronicarthritis. Clin. Exp.Rheumatol.10:493-498. 12.Hirano, T., T. Matsuda, M. Turner, N. Miyasaka, G. Buchan, B. Tang, K.
Sato,M.Shimizu,R.Maini,M.Feldmann,and T.Kishimoto. 1988. Excessive production of interleukin 6/B cell stimulatory factor-2 in rheumatoid arthritis. Eur. J. Immunol. 18:1797-1801.
13.Houssiau,F. A., J. P.Devogelaer,J.Van Damme, C. Nagant de Deux-chaisnes, and J. Van Snick. 1988. Interleukin 6 in synovial fluid and serum of
patientswith rheumatoid arthritis and otherinflammatoryarthritis. Arthritis Rheum. 31:784-788.
14.Novick, D., H. Engelmann, D. Wallach, M. Revel, 0. Leitner, and M. Rubinstein. 1991. Monoclonalantibodiestothesoluble human IL-6 receptor
affinity purification, ELISA, and inhibition of ligand binding. Hybridoma. 10:137-145.
15.Novick, D.,H.Engelmann,D.Wallach, 0. Leitner, M. Revel, and M. Rubinstein. 1990. Purification of solublecytokinereceptors from normal human urine byligand-affinityandimmunoaffinity chromatography.J. Chromatogr. 5 10:33 1-337.
16.Novick, D.,Z.Eshhar,M.Revel,and Y. Mory. 1989. Monoclonal antibod-ies foraffinity purificationofIL-6/IFN-#2andforneutralizationof HGF
activ-ity.Hybridoma.8:561-567.
17.Gaillard,J.P.,R.Bataille,H.Brailly,C. Zuber, K. Yasukawa, M. Attal, N.
Maruo,T.Taga,T.Kishimoto,and B.Klein. 1993.Increased and highly stable levels of functional interleukin 6 receptor in sera of patients with monoclonal
gammopathy.Eur. J. Immunol. 23:820-824.
18.Honda, M.,K.Kitamura,Y.Mizutani,M.Oishi, M. Arai, T. Okura, K.
Igarahi,K.Yasukawa,T.Hirano,T.Kishimoto,etal.1990.Quantitativeanalysis ofserumIL-6 and its correlation with increased levels ofserumIL-2R in HIV-in-duced diseases. J. Immunol. 145:4059-4064.
19.Bataille, R.,M.Jourdan,X.-G.Zhang,and B.Klein. 1989. Serum levels of
interleukin 6,apotentmyelomacell growthfactor,as areflectofdisease severity inplasmacelldyscrasias.J.Clin. Inv.84:2008-2011.
20.Ludwig, H.,D. Nachbaur,E.Fritz,M. Krainer, and H. Huber. 1991. Interleukin 6 isaprognostic factorinmultiplemyeloma.Blood.77:2794-2795.
21.Lust,J.A.,K.A.Donovan,M. P.Kline,R. R.Greipp, R. A. Kyle, and N. J. Maihle. 1992. Isolation ofanmRNAencodingasoluble form ofthe human interleukin-6 receptor.Cytokine.4:96-100.
22.Mullberg, J.,H.Schooltink,T.Stoyan,M.Gunther,L.Graeve, G. Buse, A.Mackiewicz,P. C.Heinrich,and S. Rose-John. 1993. The soluble interleukin 6 receptor isgeneratedby shedding.Eur.J.Immunol.23:473-480.
23.Portier, M.,D.Lees,E.Caron,M.Jourdan,J.M.Boiron, R. Bataille, and B. Klein. 1992.Up-regulationofinterleukin (IL)-6receptor geneexpression in vitro and in vivo in IL-6 deprived myeloma cells. FEBS (Fed. Eur. Biochem.
Soc.)Lett. 302:35-38.
24. Fernandez-Botran,R.,and E.S. Vitetta. 1991. Evidence that natural murine solubleinterleukin4 receptors mayactastransportproteins. J. Exp. Med. 174:673-681.
25.Aderka, D.,H.Engelmann,Y.Maor,C.Brakebusch,and D. Wallach. 1992. Stabilization ofthebioactivityoftumornecrosis factorbyits soluble
recep-tors.J.Exp.Med. 175:323-329.
26.Sehgal,P.B.,G.Grieninger,and G.Tosato,editors. 1989.Regulationof theacutephase andimmune responses: interleukin 6. Ann. NYAcad. Sci. 557:583 pp.
27.Fong, Y.,L.L.Moldawer,M.Marano,H.Wei,S. B.Tatter,R. H.Clarick,
U. Santhanam, D.Sherris, L. T.May,P. B.Sehgal,and S. F. Lowry. 1989. Endotoxemia elicits increasedcirculating ,2-IFN/IL-6 inman. J. Immunol. 142:2321-2324.
28.Helfgott,D.C.,S. B.Tatter,U.Santhanam,R. H.Clarick,N.Bhardwaj,
L.T.May,and P.B.Sehgal.1989.Multipleforms ofIFN-,B2/IL-6inserumand
bodyfluidsduringacutebacterial infection.J.Immunol. 142:948-953. 29.Jablons,D.M.,J. J.Mule,J. K.McIntosh,P. B.Sehgal,L. T.May,C. M.
Huang,S.A.Rosemberg,and M. T. Lotze.1989.IL-6/IFN-j#2as acirculating
hormone. Induction by cytokine administration in humans. J. Immunol. 142:1542-1547.
30.May, L. T., H. Viguet, J. S. Kenney, N. Ida, A. C. Allison, and P. B.Sehgal.