synovial membranefrom patients withrheumatoid arthritisjoints (patients 1-7) by means of an arthroscopic technique described previously.22 Biopsy specimens weretaken fromthe part of

Annals of the Rheumatic Diseases 1995; 54: 654-661

Detection of cytokine producing cells in thesynovial membrane from patients with rheumatoidarthritis

Ann-Kristin Ulfgren, Staffan Lindblad, Lars Klareskog, Jan Andersson, Ulf Andersson

AbstractObjectives-To develop and evaluate anew immunohistochemical method tostudy the localisation and phenotype ofindividual cytokine producing cells insynovial biopsy specimens in rheumatoidarthritis.Methods,-Cryopreserved sections ofsynovial tissue from nine patients withrheumatoid arthritis were incubated withcarefully selected cytokine specific anti-bodies detecting 19 different cytokines,after fixation of the specimens with para-formaldehyde and using saponin topermeabilise the cell membranes.Results-The immunohistochemical me-thod yielded reproducible and distinctstaining patterns, in which the cytokinesaccumulated mainly in the Golgiapparatus of producer cells, indicatingthat the method preferentially detectedlocal synthesis rather than cytokineuptake. The cytokine production patternsvaried considerably between biopsyspecimens from different patients.Conclusion-The present modified im-munohistochemical method may providea simple and rapid way to determine thelocal production of a wide array ofcytokines in the synovium. The dataobtained with this method also indicatedthat more T cell derived cytokines thanpreviously recognised were present inactive synovitis, as located and sampled byarthroscopy.

(Ann Rheum Dis 1995; 54: 654-661)

The role of different cytokines in thedevelopment of synovitis in rheumatoidarthritis (RA) has been a matter of con-siderable debate during recent years. Earlystudies showed that T lymphocytes infiltratingthe rheumatoid synovial tissue expressedvarious cell surface activation markers, forexample HLA-DR,1'2 suggesting that Tlymphocytes have an active role in theinflammatory process in the synovial mem-brane.3 Later, results were presented thatindicated an abundance of macrophagederived cytokines such as interleukin (IL)-1,IL-6, IL-8, tumour necrosis factor cx (TNFot),and granulocyte macrophage colony stimula-ting factor (GM-CSF),4` but a relativepaucity ofT cell cytokines such as IL-2, IL-3,

IL-4, and interferon gamma IFN-y' 124 in therheumatoid joint. These data provoked somedoubts concerning the importance of T cellactivation in the pathogenesis of RA,5 but alsocalled for more extensive studies on cytokineproduction at various stages of the develop-ment of synovitis.A considerable number of studies have

subsequently addressed the question of localcytokine production in the rheumatoid joint.Most of these studies, however, are associatedwith two distinct problems. First, all studies sofar performed have analysed only a fewcytokines at a time, which has limitedcomparative studies of the production of thewide array of cytokines that are now known tointeract in a synergistic or antagonistic way inan inflammatory lesion. Second, the synovialtissues analysed have in most cases beenobtained at orthopaedic surgery-that is, atlate stages of disease.To address both these problems, we aimed

to develop methodology to detect a largenumber of simultaneously produced cytokinesin arthroscopically obtained biopsy specimensfrom RA patients. We have developed andrecently described a procedure used in humantonsils'6 and rectal biopsy specimens'7 basedon immunohistochemical intracellular stainingof cytokines by means of a combined techniqueof fixation and permeabilisation,'8 whichpermits the quantitation of cells producing atleast 19 different cytokines. Our handling andstaining of the detergent exposed sectionspreserves the morphology of the intracellularlyaccumulated cytokines in a unique way.'120The aldehyde based fixation provides asignificant advantage over acetone fixation interms of morphological maintenance of theintracellular cytokine producing organelles,and saponins (plant glycosides with highaffinity for cholesterol) have been used topermeabilise cells by intercalating cellmembranes in a reversible fashion to replacecholesterol. As saponin interacts predomi-nantly with cholesterol alone, much of themorphology of the membrane structure of cellsfixed with paraformaldehyde is found toremain intact when examined under themicroscope.'8 19The present pilot study of synovial biopsy

specimens from patients with RA aimed todemonstrate the feasibility and reproducibil-ity of the method and to reveal the patternsof cytokine production in the patientsstudied.

Department ofRheumatology,Karolinska Hospital,Stockholm, SwedenA-K UlfgrenS LindbladL KlareskogDepartment ofImmunology,Arrhenius-Laboratoriesfor Natural Sciences,Stockholm University,Stockholm, SwedenJ AnderssonDepartment ofRheumatology,St Goran's Children'sHospital,Stockholm, SwedenU AnderssonCorrespondence to:Ann-Kristin Ulfgren,Department ofRheumatology,Karolinska Hospital,S-17 1 76 Stockholm,Sweden.

Accepted for publication9 March 1995

654

on February 26, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.54.8.654 on 1 August 1995. D

ownloaded from

http://ard.bmj.com/

Detection of cytokine producing cells in the RA synovial membrane

Patients and methodsPATIENT SELECTION

Synovial biopsy specimens were obtained fromnine patients (six women, three men; ages

27-67 years) who fulfilled the AmericanCollege of Rheumatology criteria for thediagnosis of RA.2" Seven of these patients weresubjected to arthroscopic biopsy of the kneejoints (patients 1-7) by means of an

arthroscopic technique described previously.22Biopsy specimens were taken from the part ofsynovium that was macroscopically mostinflamed; synovial specimens were obtainedfrom the wrists of two patients (patients 8and 9) undergoing surgery in active synovitis.All the patients were receiving NSAIDs andone patient (No 3) was taking a slow actingdrug (auranofin). The history of exacerbationof local synovitis ranged from six weeks to fiveyears.

All patients gave their informed consentand approval was granted by the EthicsCommittees at the Uppsala University and theKarolinska Institute, Stockholm.

HANDLING OF BIOPSY SPECIMENS

Specimens obtained from arthroscopic surgerywere immediately snap frozen in liquidnitrogen and embedded in OCT compound(Tissue-Tek, Mites Elkhart, IN). Thoseobtained at hand surgery were snap frozen inliquid isopentane. All tissue were kept at -70°Cuntil sectioned.

IMMUNOHISTOCHEMISTRY

Cryostat sections 8 ,um thick were mountedon gelatin coated glass slides (Novakemi,Stockholm, Sweden) and fixed for 30 minutesin paraformaldehyde (PFA) and phosphatebuffered saline (PBS) (4% PFA-PBS SigmaChemicals, St Louis, MO, pH 7 4) and were

subsequently stored at -20°C until required forstaining. The initial step of the tissue stainingwas performed with 2% fetal calf serum inbalanced salt solution (BSS) (Gibco Ltd,Paisley, UK) supplemented with 01-/% saponinas a detergent (Riedel de Haen AG, Seelze,Germany) for seven to eight minutes at 37°Cin order to reduce background signals caused

by hydrophobic interactions. Endogenousperoxidase activity was blocked by 1%hydrogen peroxide and 2% sodium nitridedissolved in BSS-saponin for one hour at roomtemperature in the dark. After three additionalwashes in BSS-saponin, the slides were

incubated overnight at room temperature in a

humid chamber with 100 RI of a panel ofcarefully selected cytokine specific mono-

clonal antibodies (MAb) or affinity purifiedpolyclonal antibodies at a concentration of2-5 ,ug/ml in BSS-saponin (table 1). Controlstainings were performed in parallel withspecies and isotype matched myelomaproteins. The slides were then washed threetimes in BSS-saponin and incubated with1% normal goat serum in BSS-saponin for30 minutes at room temperature in order toreduce background signals caused by IgG Fcinteractions by the biotinylated second stepgoat reagents. Biotin labelled secondaryantibodies, absorbed against human Ig (biotingoat antimouse IgGl and IgG2b, Caltag Lab,South San Francisco, CA; biotin goat antiratIgG, Vector Lab, Burlingame, CA) were useddiluted 1:300 during a 30 minute incubation atroom temperature. After subsequent washes inBSS-saponin, the slides were incubated eitherwith avidin-biotin-horseradish peroxidase(Vectastain, ABC-HP-kit, Vector Lab), or withavidin-biotin-alkaline phosphatase (Vectastain,ABC-AP-kit, Vector Lab). Colour reactionwas developed by diaminobenzidine (DAB)0.5 mg/ml or an alkaline phosphatase sub-strate kit including levamisol-blocking endo-genous alkaline phosphatase (Vector Lab).The reaction field was blocked after five to30 minutes by three washes in BSS,counterstained with Mayer's haematoxylin,dehydrated in ethanol, and mounted in a

glycerin buffer.A sequential staining technique was used for

two colour immunohistochemical analysis.The single colour stain was initially performedas described above for visualisation of cytokineproducing cells using the alkaline phosphataseand monoclonal antialkaline phosphatasetechnique.30 The complex of alkaline phospha-tase and MAb antialkaline-phosphatase was

detected by alkaline phosphatase substrate(red). The sections were then incubated with

Table 1 Cytokine specific antibodies usedfor tissue staining

Cytokine Antibody Isotype Producer Reference

IL-la 1277-89-7, 1277-82-29, 1279-143-4 mouse IgGI H Towbin, Ciba-Geigy, Basel 23IL-1iB 2-D-8 mouse IgGl H Towbin, Ciba-Geigy, Basel 23IL-lra 1384-92-17-19 mouse IgGI H Towbin, Ciba-Geigy, Basel 23IL-2 MQl-17H12 rat IgG2a J Abrams, DNAX, Palo Alto, CA 24IL-3 BVD3-IF9, BVD8-6G8 rat IgGi J Abrams, DNAX, Palo Alto, CA 24IL-4 MP4-25D2 rat IgGl J Abrams, DNAX, Palo Alto, CA 24IL-5 JES-39D10 rat IgG2a J Abrams, DNAX, Palo Alto, CA 24IL-6 MQ2-6A3 rat IgG J Abrams, DNAX, Palo Alto, CA 24IL-8 NAP-1 mouse IgGl M Ceska, Sandoz, Vienna 25IL-10 JES3-19FI, JES3-12G8 rat IgG2a J Abrams, DNAX, Palo Alto, CA 24IL-13 JES8-5A2, JES8-30Fl 1 rat IgG2a J Abrams, DNAX, Palo Alto, CA 26GM-CSF BVD2-21Cl 1, BVD2-5A2 rat IgG2a J Abrams, DNAX, Palo Alto, CA 24G-CSF BVD I3-3A5, BVDI 1-37GI 0 rat IgG J Abrams, DNAX, Palo Alto, CA 24TNFa MP9-20A4 rat IgG J Abrams, DNAX, Palo Alto, CA 24TNFf3 LTX22 mouse IgGI G Adolf, Bender Medsystems, Vienna 27IFNy DIKI mouse IgGI G Andersson, Astra-Drako, Lund 28TGF,I K96 polyclonal rabbit IgG K Miyazono, Ludwig Institute, Uppsala 29TGF,2 K94 polyclonal rabbit IgG K Miyazono, Ludwig Institute, Uppsala 29TGFP3 K95 polyclonal rabbit IgG K Miyazono, Ludwig Institute, Uppsala 29LTBP AB39 polyclonal rabbit IgG K Miyazono, Ludwig Institute, Uppsala 29

655



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/

Ulfgren, Lindblad, KIareskog, Andersson, Andersson

a second biotinylated BSS-saponin dilutedMAb for phenotyping of cells, which wasdeveloped by avidin-biotin-horseradish per-oxidase and DAB for peroxidase staining(brown).

SPECIFICITY TESTS

Highly purified natural or recombinant cyto-kines were used to block specific cytokinestaining. The appropriate cytokine was added(in excess, at a concentration of 20-50 jig/ml)to its corresponding cytokine specific antibody(2-5 pug/ml) at 4°C overnight. Staining with thecomplex was performed as described pre-viously and compared with results obtained bycombining anticytokine MAb preincubatedwith other cytokines as a control.The following cytokines were used in these

specificity tests: natural IL-lcx (Dr C Heusser,Basel) and IL- I (Dr C Dinarello, Boston),recombinant (r) IL-Ira (Dr H Towbin, Basel),rIL-2, rIL-3, rIL-4, rIL-5, rIL-6, rIL-8(Genzyme Corp, Boston), rIL-8 (Dr Ceska,Vienna), natural IL-10 (Dr J Abrams, PaloAlto), rTNFa and rTNF, (Bayer Inc,Hannover), rIFN-y (Boehringer Ingelheim Inc,Vienna), rGM-CSF (Sandoz Inc, Basel),recombinant granulocyte colony stimulatingfactor (rG-CSF) (Hoffnan La-Roche, Basel)and bovine transforming growth factor(TGF)1PI (Dr C Snapper, Bethesda), rTGFP2(Genzyme, Boston), and large latent TGFPcomplexes (Dr K Miyazono, Ludwig Institute,Uppsala).

QUANTIFICATION OF STAINED CELLSFor identification, cytokine producing areaswere visualised in a Reichert-Jung microscope(Polyvar II, Reichert-Jung, Vienna) underX 100 magnification. Cytokine expression wasassessed as the number of positively stainedcells of the whole tissue section. Positivelystained cells were scrutinised in high powerfields (X400 magnification). The area of eachsection was measured by a computerised imageanalyser (Quantimet 570, Leica-CambridgeLtd, Cambridge, UK). All microscopicevaluations were performed by one observer(AU) on three different occasions to excludewithin observer variation; very similar resultswere obtained. All samples were also assessedby another observer (LK) and the resultsagreed very well with those ofthe first observer.The incidence of cytokine producing cells pertotal area of cryocut sections was quantifiedaccording to a scoring system: 0 = negative cellstaining; 1 = 0-5 positively stained cells;2 = 5-10 positively stained cells; 3 = more than10 positively stained cells. The scores representmeans of numbers obtained at three differentassessments.

ResultsMORPHOLOGY OF CYTOKINE STAINING IN

SYNOVITISThe intracellular cytokine staining pattern formost interleukins revealed by immunohisto-

chemical staining of synovial membranesections was cytoplasmic (figs 1, 2), and cyto-kine producing cells tended to be present inclusters in the tissue. The morphology of thisintracellular staining in certain cells waslocalised in a characteristic juxtanuclearposition reflecting the accumulation of syn-thesised cytokines in the Golgi organelle ofproducer cells."9 In contrast, but in agreementwith our previous results obtained with in vitroactivated cells in suspension,23 immunohisto-chemical staining in tissue sections for IL-iot,IL-i1 3, and IL-ira showed a diffuse nuclearand cytosolic distribution, without anylocalised intracellular accumulation (fig 1F,G).

In addition to cytoplasmically stained cells,we observed an immunoreactivity extendingover extracellular areas surrounding producercells, mainly in stainings for GM-CSF (fig 2F),G-CSF, IL-Ira (fig IG), IL-3, IL-8, IFNy (fig2D), TGFP2 (fig 1C), TGFP3 and the latentTGFI binding protein (LTBP) (fig 1D), thetransport protein for TGF. The specificitiesof the extracellular and the intracellularimmunoreactivity were verified by theircomplete inhibition by preincubation of thecytokine specific MAb with the relevantcorresponding human cytokines, but not withother cytokines (fig 1B v 1C).A total of nine different synovial tissues were

stained with antibodies against 19 differentcytokines (table 1). Production of one orseveral of these cytokines was detected in all ofthe synovial biopsy specimens, although with avery heterogenous pattern of productionamong the patients studied (table 2). Positivecontrol stainings were performed using anti-CD3 and anti-HLA-DR antibodies whichshowed the expected increased HLA-DRexpression and extensive T cell infiltrationin all specimens examined (fig 2A)-anappearance characteristic for diagnosis ofRA.3'Mouse, rat or rabbit myeloma proteins wereused as negative control staining antibodies;these control stainings were negative in all buttwo biopsy specimens (from patients 2 and 6),both of which contained rheumatoid factorproducing cells in the synovial membranes. Asthe positively stained cells in these biopsyspecimens appeared morphologically to beplasma cells, we assume that the backgroundsignals were caused by intracellularly producedRF interacting with exogenous Ig used forstaining.32

IDENTIFICATION OF MONOKINE PRODUCTION

There was great variation between patients inproduction of cytokines known to be producedmostly by macrophages (IL-la, IL-i1P, IL-ira,IL-6, IL-8, TNFot, GM-CSF and G-CSF)(table 2). The three different types of IL-1were found to be coexpressed in differentcombinations in individual biopsy specimens,or were not expressed at all. IL-i1 producingcells outnumbered those of IL-la: IL-lao wasfound in four of nine tissues, but IL-i13 inseven of nine patients; IL- 1 ra was evident in sixof eight synovial tissues examined. Staining for

656



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/


1/ ::*t~. ' :; . e ,^

I f f6 , b..ss,etheJ*e% ;|''1'

*_ #

FZ R it4\~~~~~~~ 2> y

M Gt~~~~~~~~~~~~R

~~~~~~~A~~~~~~~~~~~~~~~~~'~~~~~~~~~~~~~~~_

Pr0~

Figure 1 Videoprint photographs illustrating brown immunoperoxidase stainingfor cytokine producing cells in cryopreservedsynovial membrane biopsy specimens arthroscopically obtainedfrom RA patients (A-D = patient 6 in table 2, E, F = patient 3;G, H= patient 5). Counterstain: haematoxylin. A: Stainingfor TGF/31: combined intra- and extracellular deposition ofthecytokine. (Original magnification X80.) B: Elimination of TGF/32 staining shown in C by preincubation of the anti-TGF,B2antibodies with recombinant TGF/32 as a specificity control. (Original magnification X80.) C: TGF/32 staining ofaconsecutive section without any addition ofexogenous TGF,32. The production of this isoform was more pronounced than that ofTGF,BJ. Individual producer cells and prominent extracellular presence of the cytokine are demonstrated. (Originalmagnification X80.) D: Staining in a consecutive section for latent TGF/3 binding protein (LTBP), the transport protein forTGF,B. (Original magnification X800.) E: TNFa production in the lining layer and in scattered cells in the deeper layers.(Original magnification X500.) F: IL-1,producing cells. Note the distinct staining of both the cytoplasm and the nucleus ofproducer cells-a pattern characteristic ofIL-I staining. (Original magnification X 1000.) G: IL-i ra production expressed inthe endothelium ofa blood vessel. Strong nuclear staining, analogous to the appearance with IL-1i3. (Original magnificationx 800.) H: Typical appearance of localjuxtanuclear deposition ofIL-5 in a producer cell indicates the accumulation of thecytokine in the Golgi organelle. No extracellular deposition. (Original magnification X800.)

657



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/

Ulfgren, Lindblad, Kiareskog, Andersson, Andersson

*. to.

ft,

*

*

4

,v*-^ M1

.4D {D '

,s,-> t g ~as

*:r_* ^..o*, ....

a * e

~ a' a+

40' o, ,,".I., ), ^i 0

N~ 7>

GFigure 2 Immunohistochemical staining, mainly oflymphokines, in synovial biopsy specimens from patient 2 (A-F) andpatient 4 (G-H). A: Cluster of infiltrating CD3 positive lymphocytes in the synovial membrane. CD3 staining (brown)performed in the presence of the detergent saponin demonstrates both cytoplasmic and membrane immunoreactivity.(Original magnification X 400.) B: Control staining with irrelevant isotype matched myeloma protein and indirectimmunoperoxidase technique without any background signals. (Original magnification x 320.) C: IL-2 staining ofanactive T cell infiltrate demonstrating multiple producer cells. Note the additional entrapping ofIL-2 in the extracellular areaclose to the synthesising cells. (Original magnification X 200.) D: IFNy producing cells in another lymphoid infiltrate.(Original magnification X320.) E: Two colour staining ofIL-4 producing cells (red alkaline phosphatase) and CD3(brown peroxidase). IL-4 production is demonstrated in both CD3 positive and CD3 negative cells. (Originalmagnification X 1000.) F: GM-CSF production. The cytokine is expressed in the Golgi stack ofproducer cells andextracellularly around synthesising cells. (Original magnification X 1000.) G: IL-1 0 producing cells predominantlylocalised in clusters. (Original magnification x320.) H: IL-13 producing cells. (Original magnification X500.)

658



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/


Table 2 Enumeration of cytokine producing cells per section in synovial biopsiesfrom RApatients

Patient Section area IL-2 IL-3 IL-4 IL-5 IL-1O IL-13 IFNy TNF,8(mm2)

1 5-2 0 1 0 1 0 0 0 02 7-4 2 3 3 0 3 3 3 33 90 1 1 2 0 1 0 1 14 20-7 1 3 0 0 3 3 1 35 7-8 1 0 3 2 2 1 3 16 7-5 2 3 3 3 2 3 2 17 2-0 0 0 0 0 0 0 0 08 3-1 0 3 3 1 0 0 3 09 10-5 0 1 1 1 1 1 0 0

Patient Section area G-CSF GM-CSF TNFa IL-la IL-i,8 IL-Ira IL-6 IL-8(mm2)

1 5-2 0 3 0 0 1 2 2 02 7-4 3 3 1 0 3 0 1 13 90 3 2 2 0 3 1 0 34 20-7 0 1 2 1 0 3 0 15 7-8 3 3 2 1 1 2 2 26 7-5 2 3 0 1 1 2 1 17 2-0 0 0 0 0 0 nd 0 08 3-1 2 1 1 0 3 1 2 19 10-5 2 1 3 1 3 0 1 3

Patient Section area TGFI38 TGF,82 TGF,B3 LTBP Mouse Rat Rat(mm2) IgGI IgGI IgG2a

1 5-2 nd nd 1 nd 0 0 nd2 7-4 nd nd nd nd t t t3 9 0 3 1 3 nd 0 1 04 20-7 2 2 3 2 0 0 05 7-8 nd 0 nd nd 0 0 nd6 7-5 1 3 3 3 t t t7 2-0 0 3 0 3 0 0 08 31 3 3 3 3 0 0 09 10-5 2 3 3 3 0 0 0

Scoring system for incidence of cytokine producing cells per cryocut section: 0 = negative cellstaining; 1 = 0-5 positively stained cells; 2 = 5-1 0 positively stained cells; 3 = >10 positivelystained cells.tSynovial membrane contained rheumatoid factor producing plasma cells.

these three proteins was located primarily inirregularly shaped macrophage like cells, whichwere situated in both the lining layer and thesublining layer (fig 1F). IL-Ira production wasalso demonstrated in endothelial cells of bloodvessels (fig 1 G). Staining for TNFot wasevident in six of nine tissues and was presentmainly within the lining layer (fig 1E).GM-CSF antibody gave a strong positive

reaction in eight of nine tissues, in most casesmore widespread than that for either TNFoe orIL-1. Staining for GM-CSF was particularlyprominent in the lining layer and, notably,there was also a pronounced extracellularstaining for GM-CSF in five of nine tissues(fig 2F). G-CSF antibodies yielded positivestaining in six of nine tissues and this was alsocombined with a substantial extracellularstaining, particularly in the sublining layer.IL-8 staining was present in seven of ninetissues, where it was most conspicuous in thelining layer and in fibrotic areas. Many IL-8positive cells resembled fibroblasts morpho-logically.

IDENTIFICATION OF TGFIThe intensity ofTGF,3 staining was impressiveand the frequency of TGF,2 producing cells,in particular (fig 1C), outnumbered that ofall other studied cytokines. The consistentextracellular localisation of large amounts ofTGF, suggested a localised storage depot inthe tissues. Most regions staining for TGF,coexpressed LTBP, the transport protein forTGF,B, indicating that the extracellular

presence ofTGF, was a consequence of activesecretion, rather than a result of leakage oflatent TGF,3 complexes (fig 1D).

IDENTIFICATION OF LYMPHOKINE PRODUCTION

Staining for IFNy was seen in six of ninetissues (fig 2D) and that for IL-2 in five of ninetissues (fig 2G). Typically, the vast majority ofCD3 positive T cells did not show any signs ofIFN-y or IL-2 synthesis, but positive stainingfor each of these cytokines was preferentiallylocated within certain aggregates of CD3positive cells, other apparently identical T cellclusters being completely negative for IL-2 andIFNy.TNFP was found in five of nine biopsy

specimens, mainly at the sites that stained forTNFot, although the TNF3 staining was morelimited. One biopsy specimen showed TNF,3producing cells, but no evidence of TNFaproduction.With the present methodology, very dis-

tinctly positive IL-4 staining was evident bothin small lymphocyte like cells and within largerirregular cells in six of nine tissues. Two colourstaining for surface CD3 antigens andintracellular IL-4 demonstrated that IL-4production occurred in T cells and in non-Tcells (fig 2E). A further, notable feature of IL-4staining was that the positively stained cellswere often present in lymphoid clusters, whichsometimes also contained IL-2 and IFN-yproducing cells. The synthesis of these variouscytokines occurred in distinctly separate cells.The recently described T cytokine, IL-13 (aT cell lymphokine with many functionalproperties shared by IL-4 and IL-10), wasidentified in five of nine patients; its tissuelocalisation was approximately the same as thatof IL-4 (fig 2H). In addition, IL-10 producingcells were demonstrated in six of nine tissues(fig 2G). Producer cells were either smallround mononuclear cells in perivascularlymphoid aggregates, or larger irregular cellslocated mainly in or close to the lining layer.

Staining patterns for IL-3 and IL-5 (fig 1H),which can be synthesised by T cells and bymast cells, resembled each other: positivelystained cells were in both cases either smallround mononuclear cells or large irregularcells. Most cells expressing these two cytokineswere found in the lining or sublining layer.

DiscussionThree findings of potential interest aredescribed in the present report. The mainfinding is the demonstration of positive intra-cytoplasmic staining for a large number ofsimultaneously synthesised human cytokines insynovial biopsy specimens from RA patients.The second finding concerns the large diversityof patterns of cytokine production seen inbiopsy specimens displaying similar histo-pathological features and staining profiles forconventional lymphocyte and macrophage cellsurface markers. The third observation is thatof a substantial presence of T cell derivedcytokines (IL-2, IL-4, IL-5, IL-1 0, IL- 13,

659



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/

Ulfgren, Lindblad, Klareskog, Andersson, Andersson

IFN-y and TNF3) at a protein level in synovialbiopsy specimens obtained from active in-flammation in rheumatoid arthritis identifiedvisually via the arthroscope.Our technique for both fixation and per-

meabilisation, combined with staining with 19different cytokine antibodies selected verycarefully from a total pool of several thousanddifferent MAbs, should confer certain advan-tages compared with previously describedmethods for detection of local cytokineproduction in synovitis. The technique gives a

considerably clearer staining morphology tointracellular organelles than do other method-ologies, thereby apparently also permitting a

distinction to be made, on frozen sections,between local production of a cytokine and thepresence of a cytokine which had beenproduced elsewhere. 16-18 The intracellularstaining of cytokines confined to the Golgiapparatus (with the exception of IL-1cytokines) has previously mainly been used inin vitro model systems,'9 in which it has beenshown to correlate well with the productionof cytokine mRNA20 and cytokine secretionin supernatants.'9 20 This colocalization was

confirmed by two colour staining with Golgispecific and cytokine specific MAb.'6 19 Nosuch comprehensive methodology has pre-

viously been described for the detection ofdistinct sites of local cytokine production insections taken from tissues in active synovitis,and some cytokines studied here, such as IL-5and IL- 13, have not previously beendemonstrated in the rheumatoid synovialmembrane.None of the selected cytokine specific MAbs

used in the present study has been successfullyused in parallel control experiments withstandard immunohistochemical techniquesbased on acetone fixation. However, usingother cytokine detecting MAbs in acetone fixedsynovial membrane sections, we havedemonstrated positive immunoreactions. Thedisadvantage has been that it has provedimpossible to discriminate between cytokineproducing cells and cytokine binding cells, as thedominating signals in the acetone fixedmaterial were generated from cell surfacemembranes. Acetone fixation seems to permitdetection of antigen on cell surfaces in a muchmore satisfactory way than PFA-saponintreatment, which instead preserves themorphology of intracellular antigens in a

superior state. In order to identify cytokineproducing cells, there is a clear need todistinguish target cells from producer cells.One of the more interesting potentials of the

present metholology is concerned with theneed for a technique for repeated and rapidanalysis of cytokine production in sequentiallyacquired synovial biopsy specimens frompatients during the course of their disease andduring the institution of various treatment

regimens affecting cytokine regulation. Thepresent study demonstrates that distinctstaining patterns can be identified in biopsyspecimens acquired by arthroscopy. However,it will be important to perform additionalstudies of the local variability of cytokine

production within a given joint beforeembarking on larger clinical follow up pro-grammes.The diversity of the patterns of staining for

cytokine production in the synovia studied,and the presence ofT cell cytokines in certainbiopsy specimens deserve special attention.The results from staining studies in some of thesynovial tissues clearly correspond well withpreviously reported data on the predominanceof mainly monocyte/macrophage derivedcytokines (such as IL-1, IL-6, IL-8, TNFac,GM-CSF, and G-CSF) in the rheumatoidjoint.4'-" The strong staining for the threeisoforms ofTGF, in certain tissues also agreeswith findings of earlier reports.33 Studies of Tcell cytokines at a protein level in therheumatoid joint have been more scarce, andit was therefore of particular interest to observedistinct production of many putatively T cellderived cytokines such as IL-2, IL-3, IL-4,IL-5, IL-10, IL-13, IFN-y, and TNFI inclusters of cells having lymphocyte morphol-ogy, in several biopsy specimens obtained fromactive synovitis in rheumatoid arthritis. Webelieve that we have managed to obtain thesenew results because of our meticulous selectionof cytokine detecting antibodies and our tissuehandling technique. The T cell origin of someof the IL-4 production was confirmed by twocolour staining for surface CD3 markers andintracellular IL-4 (fig 2E). Some of theselymphoid aggregates demonstrated simul-taneous production of lymphokines mediatingcytotoxicity and macrophage activation (Thlresponse), such as IL-2 and IFN^y, andlymphokines promoting antibody formation(Th2 response), such as IL-4, IL-10, and IL-13. These clusters were often seen adjacent tolymphoid infiltrates having identical histo-pathological features without any signs ofcurrent cytokine production.As to the meaning of these staining patterns,

it is immediately apparent that we do not as yethave enough information from basic modelsystems on similar comprehensive patterns ofcytokine production to provide an adequateinterpretation. Thus we do not know how thepattern of cytokine production develops duringvarious phases of a classically T cell mediatedinflammatory reaction, for example a delayedhypersensitivity reaction. Nor do we knowenough about these events at various stages inanimal models of chronic T cell dependentinflammatory arthritides such as collagenarthritis or adjuvant arthritis. The limitedexperience available has been obtained with invivo model systems of experimental en-cephalitis in rats.34 These studies havedemonstrated substantial differences in cyto-kine patterns at various stages of disease, withmore T cell derived cytokines in the initialphase, and more macrophage derived cyto-kines (and in particular increased amounts ofTGFP) in late and remitting stages of thedisease.34 Present knowledge based on extrap-olations from these experimental encephalitisstudies suggests that patterns of cytokineproduction observed in some biopsy specimensfrom patients with active RA are compatible

660



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/

Detection ofcytokine producing cels in the RA synovial membrane

with that of a T cell dependent chronic inflam-matory disease. Clearly, however, studies todate on cytokine patterns in human and experi-mental disease represent only the beginning ofa more comprehensive descriptive analysiswhich is required to further understanding ofthe role of local cytokine regulation in chronicinflammation and subsequently to direct ourefforts to modify this regulation.Also ofrelevance to the subject ofthe present

study is the need for such descriptive studiesto be undertaken with due consideration of theclinical and local status ofthe disease process-which by implication necessitates the acqui-sition of biopsy specimens via arthroscopy,preferentially repeatedly in longitudinalstudies. We believe that the technique wedescribe for rapid analysis of a large number ofcytokines will be helpful in the exploration ofpathogenetic mechanisms in RA and theevaluation ofnew therapeutic regimens.

This study was supported by grants from the Swedish MedicalResearch Council (grants Nos 09082 and 10850), AxelJohnsson's Foundation, the National Cancer Institute (grantsNos 2490 and 2766), the Swedish Association againstRheumatism, von Kantzow's Foundation, Crafoord'sFoundation, Nanna Svartz's Foundation, af Ugglas'Foundation, King Gustaf V:s Foundation and Thuring'sFoundation. We gratefully acknowledge the gift of cytokinespecific antibodies from John Abrams (Palo Alto), GuntherAdolf (Vienna), Gudrun Andersson (Lund), Miroslav Ceska(Vienna), Kohei Miyazono (Uppsala) and Harry Towbin(Basel).

1 Burmester G R, Yu D T Y, Irani A-M, Kunkel H G,Winchester R J. Ia+ T cells in synovial fluid and tissueof patients with rheumatoid arthritis. Arthritis Rheum1981; 24: 1370-6.

2 Klareskog L, Forsum U, Malmnas-Tjernlund D,Kabelitz D, Wigren A. Appearance of anti-HLA DRreactive cells in normal and rheumatoid synovial tissue.ScandlImmunol 1981; 14: 183-92.

3 Panayi G S, Lanchbury J S, Kingsley G H. The importanceof the T cell in initiating and maintaining the chronicsynovitis of rheumatoid arthritis. Arthritis Rheum 1992;35: 729-35.

4 Fontana A, Hengartner H, Weber E, Fehr K, Grob P J,Cohen G. Interleukin-1 activity in the synovial fluid ofpatients with rheumatoid arthritis. Rheumatol Int 1982; 2:49-53.

5 Buchan G, Barrett K, Turner M, Chantry D, Maini R N,Feldmann M. Interleukin-I and tumor necrosis factormRNA expression in rheumatoid arthritis: prolongedproduction of IL-loa. Clin Exp Immunol 1988; 73:449-53.

6 Hirano T, Matsuda T, Turner M, et al. Excessiveproduction of interleukin 6/B cell stimulatory factor-2 inrheumatoid arthritis. EurJ Immunol 1988; 18: 1797-801.

7 Brennan FM, Zachariae C 0 C, Chantry D, et al. Detectionof interleukin-8 biologic activity in synovial fluids frompatients with rheumatoid arthritis and production ofinterleukin-8 mRNA by isolated synovial cells. Eur JfRheumatol 1990; 20: 2141-4.

8 Chu C Q, Field M, Feldman M, Maini R N. Localisationof tumor necrosis factor a in synovial tissue and at thecartilage-pannus junction in patients with rheumatoidarthritis. Arthritis Rheum 1991; 34: 1125-32.

9 Xu W D, Firestein G S, Taetle R, Kaushansky K,Zvaifler N J. Cytokines in chronic inflammatory arthritisII. Granulocyte-macrophage colony-stimulating factor inrheumatoid synovial effusions. Jf Clin Invest 1989; 83:876-82.

10 Firestein G S, Berger A E, Tracey D E, et al. IL-I receptorantagonist protein production and gene expression inrheumatoid arthritis and osteoarthritis synovium.t Immunol 1992; 149: 1054-62.

11 Yanni G, Farahat M N M R, Poston R N, Panayi G S.Intramuscular gold decreases cytokine expression andmacrophage numbers in the rheumatoid synovialmembrane. Ann Rheum Dis 1994; 53: 315-22.

12 Husby G, Wiams R. Immunohistochemical studies ofinterleukin-2 and interferon--y in rheumatoid arthritis.Arthritis Rheum 1985; 28: 174-81.

13 Firestein G S, Xu W-D, Townsend K, et al. Cytokines inchronic inflammatory arthritis I. Failure to detect T celllymphokines (interleukin 2 and interleukin 3) andpresence of macrophage colony-stimulating (CSF-1) anda novel mast cell growth factor in rheumatoid synovitis.J Exp Med 1988; 168: 1573-86.

14 Miossec P, Naviliat M, D'Angeac A, Sany J, Banchereau J.Low levels ofinterleukin-4 and high levels oftransforminggrowth factor-13 in rheumatoid synovitis. Arthritis Rheum1990; 33:1180-7.

15 Firestein G S, Zvaifler N J. How important are T cells inchronic rheumatoid synovitis? Arthritis Rheum 1990; 33:768-73.

16 Andersson J, Bj6rk L, Agren K, Abrams J, Litton M,Andersson U. Concomitant in vivo production of 19different cytokines in human tonsils. Immunology 1994;83: 16-24.

17 Raqib R, Lindberg A A, Wretlind B, Bardhan P K,Andersson U, Andersson J. Persistence of local cytokineproduction in Shigellosis in acute and convalescent stages.Infect Immun 1995; 63: 289-96.

18 Andersson U, Andersson J. Immunolabelling of cytokine-producing cells in tissues and in suspension. In:Fradelizie D, Emelie D, eds. Cytokine producing cells.Paris: INSERM, 1994; 48-64.

19 Sander B, Andersson J, Andersson U. Assessment ofcytokines by immunofluorescence and the paraformalde-hyde/saponin procedure. Immunol Rev 1991; 119: 65-93.

20 Dolhain R J E M, Andersson U, ter Haar N T, et al.Detection of intracellular IFNy by light microscopy usingan immunoperoxidase technique: Correlation with thecorresponding mRNA and protein product. J LeukocyteBiol 1993; 54: 545-51.

21 Arnett F C, Edworthy S M, Bloch D A, et al. The AmericanRheumatism Association 1987 revised criteria for theclassification of rheumatoid arthritis. Arthritis Rheum1987; 31: 315-24.

22 Lindblad S, Klareskog L, Hedfors E, Forsum U,Sundstrom C. Phenotypic characterization of synovialtissue in situ in different types of synovitis. Arthritis Rheum1983; 26: 1321-32.

23 Andersson J, Bjork L, Dinarello C A, Towbin H,Andersson U. Lipopolysaccharide induces humaninterleukin-I receptor antagonist and interleukin-Iproduction in the same cell. Eur J Immunol 1992; 22:617-23.

24 Abrams J S, Roncarolo M-G, Yssel H, Andersson U,Gleich G I, Silver J E. Strategies of anti-cytokine mono-clonal antibody development: Immunoassay of IL- I andIL-5 in clinical samples. Immunol Rev 1993; 127: 5-24.

25 Agace W, Hedges S, Andersson J, Andersson U, Svanborg C.Selective cytokine production in epithelial cells afteractivation with E. coli. Infect Immun 1993; 61: 602-9.

26 McKenzie A N J, Culpepper J A, de Waal Malefyt R, et al.Interleukin-13, a novel T cell-derived cytokine thatregulates human monocyte and B cell function. Proc NatlAcad Sci USA 1993; 90: 3735-42.

27 Andersson U, Adolf G, Dohlsten M, Moller G,Sj6gren H-O. Characterization of individual tumornecrosis factor a- and 13-producing cells after polyclonalT cell activation. _lImmunol Methods 1989; 123: 233-8.

28 Andersson G, Ekre H-P, Alm G, Perlmann P. Monoclonalantibody two-site ELISA for human IFN-y. Adaptationfor determinants in human serum or plasma. J ImmunolMethods 1989; 125: 89-95.

29 Olofsson A, Miyazono K, Kanzaki T, Colosetti P,Engstr6m U, Heldin C-H. Transforming growth factor3- 1, -2 and -3 secreted by a human glioblastoma cell line.I Biol Chem 1992; 267: 19482-8.

30 Cordell J L, Falini B, Erber W N, Ghosh A K,Abdulaziz Z, Macdonald S. Immunoenzymatic labeling ofmonoclonal antibodies using immune complexes ofalkaline phosphatase and monoclonal anti-alkaline phos-phatase (APAAP complexes). J Histochem Cytochem 1984;32: 219-29.

31 Lindblad S, Hedfors E. Intraarticular variation in synovitis.Local macroscopic and microscopic signs ofinflammatoryactivity are significantly correlated. Arthritis Rheum 1985;28: 977-86.

32 Klareskog L, Rubin K, Holmdahl R. Binding of collagentype II to rheumatoid synovial cells. Scand J7 Immunol1986; 24: 705-14.

33 Taketazu F, Kato M, Gobl A, et al. Enhanced expressionof TBF-,3s and TGF-,B type II receptor in the synovialtissues of patients with rheumatoid arthritis. Lab Invest1994; 70: 620-30.

34 Issazadeh S, MustafM, Ljungdahl A, et al. IFN-y, ILA4 andTGF-,B in experimental autoimmune encephalomyelitisin Lewis rats; dynamics of cellular mRNA expression inthe central nervous system and lymphoid cells. J Neuro-science 1995. In press.

661



j.com/

Ann R

heum D


ownloaded from

http://ard.bmj.com/

Documents

synovial membranefrom patients withrheumatoid arthritisjoints (patients 1-7) by means of an arthroscopic technique described previously.22 Biopsy specimens weretaken fromthe part of