R heumatology 1999;38:202- 213
`
`Cytokine production by synovial T cells in
`rheumatoid arthritis
`
`G. Steiner1· \ M. Tohidast-Akrad2
`, G. Witzmann3
`, M. Vesely\
`, P . Zenz7 and
`A. Studnicka-Benke3
`, A. Gal5
`, M. Kunaver6
`J. S. Smolen1
`2
`3
`•
`•
`1 Division of Rheumatology, University of Vienna, 2 Ludwig Boltzmann-Institute
`of Rheumatology, 32nd Department of Medicine and 4Department of Pathology,
`Lainz Hospital, Vienna, Austria, 5 Department of Molecular Biology, National
`Institute of R heumatology, Budapest, Hungary, 6 Department of Medicine, Baden
`HospitaL, Baden and 7 Department of Orthopaedic Surgery, Hospital
`Baumgartner Hohe, Vienna, Austria
`
`Summary
`Objecti11e. To investigate the production of cytokines by T cells in patients with rheumatoid
`arthritis (RA), reactive arthritis (REA) and osteoarthritis (OA).
`Mell10ds. The lymphokines interleukin (IL)-2, IL-4. interferon gamma (IFN-y) and tumour
`necrosis factor beta (TNF-{J) , as well as the monokines lL- 1, TL-6 and TNF-oc, were measured
`by immunoassays in sera and synovial fluid (SF) from patients with RA, REA and OA. ln
`addition, cytokine expression was studied by immunohistochemistry in synovial membrane
`tissue sections from patients with RA and OA.
`Results. Almost 60% of RA sera contained at least one of the cytokines investigated,
`though in low concentrations, whereas cytokines were generally not detectable in sera from
`REA and OA patients. In contrast, cytokines were found in virtually all SF: thus. the
`majority of SF from RA patients contained JFN -y (median level 17 pg/ ml) in addition to the
`monokines IL-6 (4700 pg/ ml ) and TNF-oc ( 157 pg/ ml) . IFN-)• and IL-6 (but not TNF-oc) were
`also frequently measured in SF from REA patients, whereas OA samples typically contained
`only IL-6. Immunohistochemical analysis of tissue sections from RA patients revealed
`lymphokine expression in 0. 1- 0.3% ofT cells, particularly IL-2 and lFN-y, and to a lesser
`extent also IL-4. Interestingly, the expression of TNF-oc and IL-6 by synovia l T cells was also
`observed. The majority of cytokine-expressing T cells were CD4-positivc T-helper cells
`typically found in perivascular areas. whereas cytokine-producing CD8-positive T cells were
`found distributed throughout the synovium. As expected, in specimens from OA patients,
`T cells were much less abundant and expression of cytokines could not be detected.
`Conclusion. These data clearly demonstrate production of cytokines by T cells in RA
`synovia l tissue, indicating that activated T cells p lay a role in the pathophysiological events of
`RA.
`
`KEv WORDS: T cell, Cytokines, fFN-y, IL-4, TNF-o:, Rheumatoid artluitis, Reactive arthritis,
`Osteoarthritis, Pathogenesis.
`
`The aetiology and pathogenesis of rheumatoid arthritis
`( RA) are still unresolved. In recent years. analysis of
`cyto kinc production in RA has attracted particular
`interest since many cytokines are involved in the regula(cid:173)
`tion of the immune and the inflammatory response [ 1 ].
`Although cells of the immune system are considered to
`be their primary so urces. several cytokines can also be
`
`Submitted 14 Aprill998; revised versio n accepted 2 September 1998.
`Correspondence
`to: G. Steiner, Division of Rheumatology,
`Department of Medicine liT, University of Vienna, Waebringer Guertel
`18- 20. A-1096 Vienna, Austria.
`
`produced by other cell types such as fib roblasts. endo(cid:173)
`thelial or even epithelial cells. Thus. in the inflamed
`synovium. monocytes as well as type A and type B
`syuoviocy1es have been shown to produce large amounts
`or cytokincs. particularly pro-inflammatory cytokines
`such as tumour necrosis factor alpha (TNF-a). interleu(cid:173)
`kin ( IL)-1 or IL-8 [2- 41. In contrast to the general
`agreement on the role of highly activated macrophages
`and synoviocytes in the pathophysiology of RA, the
`involvement ofT cells is unclear. RA is considered an
`autoimmune disease. and therefore a decisive role for
`T cells both in initiating and maintaining the disease
`
`202
`
`© 1999 British Society for Rheumatology
`
`

`

`T-cell-derived cytok.ines in RA
`
`203
`
`process should be assumed. However. a lthough the
`majority of synovial T cells bear an activated phenotype.
`as indicated by expression of MHC class II antigens.
`only a few a lso express the IL-2 receptor [5, 61. This.
`and the fa ilure unambiguously to demonstrate signifi(cid:173)
`cant production of lymphokirJe proteins iD RA synovial
`tissue [3. 7- ll ]. has led to debate about the role of
`T cells in RA [1 2- 16].
`In an attempt to further o ur understanding of the
`role ofT lymphocytes in RA. we have analysed T-cell(cid:173)
`derived cytokines in sera and synovial fluid (SF) from
`patients with RA and other rheumatic disorders by
`immunoassays, a nd have also investigated cytokine
`expression in synovial membranes by immunohisto(cid:173)
`chemical methods using single- and double-labelling
`teclmiques. The results obtained demonstrate that a
`relatively small subset of activated T cells in the RA
`synovial membrane produce lymphokines. particularly
`IL-2 aud interferon gamma (IFN-y) . suggesting their
`participation in the pathophysiological processes of R A.
`
`Patients alld methods
`
`Patiems
`Synovial fluid was aspirated from knee joiut eiTusious
`and blood samples drawn by venepuncture at the same
`time. Thirty-three paired serum/ SF samples were from
`27 patients with RA (female/male: 22/ 5: median age
`59 yr. range 38- 76 yr: median disease duration 5 yr.
`range 1- 23 yr) : 85% of the patients were rheumatoid
`factor (R F) positive. Patients with reactive arthritis
`( REA) aud osteoarthritis (OA) served as disease cou(cid:173)
`trols: 26 samples were obtained from 17 REA patients
`(fema lejmale: ll f7; median age 34 yr. range 14- 65 yr:
`median disease duration 5 mouths. range 2-42 months)
`and 15 samples were obtained from 15 patients with
`OA (female/male 8/ 7: median age 60 yr, range 47- 81 yr:
`mediarJ disease duration 3 yr, range 0.5- 21 yr). RA was
`classified according to the 1987 revised criteria of the
`American College o f Rheumatology [ 17). REA was
`defined as seronegative o1igoarthritis with cultural evi(cid:173)
`dence of sexually transmitted disease or cultural and/o r
`serological evidence of Sal111onel/a or Yersinia infection.
`OA of the knee joints was diagnosed clinically and
`radiologically. The majority of pa tients were treated
`with non-steroidal anti-inflammatory drugs (diclofcnac.
`ibuprofen or oxicams). In addition. eight RA and
`o ne R EA patient received gl ucocorticoids ( < 10 mg
`prednisone/day). and 15 RA patients were also on slow(cid:173)
`acting anti-rheumatic drugs (methotrexate, gold salts or
`chloroquine) . Iu all patients, joints were examined for
`pain and swelling. and the Ritchie index was determined.
`In RA patients. early morning stiffness was recorded in
`addition. Laboratory parameters included haemoglobin.
`platelet count. white cell counts in peripheral blood and
`SF. serum iron, erythrocyte sedimentation rate (ESR ).
`C-reactive protein (CRP) and RF. All patients were
`seeu at the out-patient clinic of the Centre for Rheumatic
`Diseases at Lainz Hospital and presented with effusions
`
`of the knee joints. Informed consent was obtained from
`all patients. Control sera from 20 healthy subjects (age
`range 27- 58 yr) were also analysed.
`
`Detection of cytokines in serum and SF
`Immunoassays were employed for all cytokine measure(cid:173)
`ments. Assays for IL-l , IL-2, IL-4 and IL-6 were
`obtained from R & D Systems (Abingdon. U K ). the
`assay
`for TNF-a was
`from Medgenix (Fieurus.
`Belgium) . the assay for TNF-,8 was from Bender
`MedSystems ( Vienna. Austria) and the assay for IFN-y
`was from Centocor ( Malvern. PA. USA ). All assays
`were carefully tested with respect to reproducibility (see
`below). and recovery of recombinant cytokines in spiked
`body fluid samples (three concentrations-
`low. medium
`and high. e.g. 30. 100 and 400 pgjml -
`in three sera a nd
`SF each). Possible interference by R F was tested by
`analysis of samples before and after absorption of RF
`on lgG agarose (Sigma. St Louis, MO. USA). as
`described [ 18 ]. The limit of detection was defined as the
`minimum concentration which could be reproducibly
`measured, i.e. with <25% inter-assay variation. It must
`be noted that in all assays used. the detection limit
`defined in this manner was higher than that given by
`the manufacturers. Thus. reproducible detection limits
`were 10 pgjrnJ for IPN-y. 20 pg/ml for TNF-,8. 25 pg/ ml
`for IL- Ia and IL-1 ,8. and 30 pgjml for IL-2. IL-4, IL-6
`and TNF-a. The specificity of the assays was assessed
`by competition assays in which neutralizing anti(cid:173)
`cytokine antibodies (obtained from Endogen, Boston.
`MA. USA or from R & D Systems) were added to
`selected samples as described [ 19]. All positive and
`randomly selected negative samples were measured a
`second time for confirmatio u of results. and the meau
`value of the two determinations was used fo r statistical
`analysis. All cytokiues were below the detection limit in
`sera from healthy individuals. Most of the cytokine
`assays were validated by participating in the Cytokine
`Consensus Study Group of the Europeau Workshop for
`Rheumatology Research [201.
`
`Tissue sections
`Synovial tissue was obtained from eight patients with
`RA and from five patients with OA at the time of knee
`joint surgery. a nd snap-frozen in isopentane cooled with
`dry ice immediately after excision. Cryostat sections
`(4-6 Jtm) were air-dried fo r 1 day at room temperature.
`Sections were then fixed in ice-cold acetone for l 0 min
`and used for inmmnohistochemistry.
`
`Detection of cytokines in synovial membranes
`For cytokine detect ion. affinity-purified rabbit poly(cid:173)
`clonal antibodies to IL-2. IL-4. IL-6. IFN-y and TNF-a
`were used (Endogen, Boston, MA. USA) . In addition,
`monoclonal antibodies to IL-2 (Dako). IL-4 and IL-2R
`(Genzyme. Boston, MA. USA) were employed. R abbit
`IgG (Endogen) or isotype-matched monoclonal antibod(cid:173)
`ies (Dako and l3ecton-Dickinson) served as controls.
`Sections were incubated with the primary anti-cytok:ine
`antibody fo r 60 min. After rinsing, endogeneous peroxi-
`
`

`

`204
`
`G. Steiner e1 a/.
`
`dase was blocked with 0.3% hydrogen peroxide in Tris(cid:173)
`buffered saline ( 10 mM Tris- HCI. 140 mM NaCl. pH 7.4)
`for lO min. This was followed by 30 min incubation
`with the second antibody. a biottnylated affinity-purified
`goat anti-rabbit IgG or biotinylated horse anti-mouse
`IgG. Then.
`seetioLJS were
`incubated with
`the
`VECTASTAIN@ABC reageut ( Vector, Burlingame.
`CA. USA) for another 30 min. Colour was developed
`using diaminobenzidine (Sigma).leading to brown stain(cid:173)
`ing. Finally. slides were counterstained with haematoxy(cid:173)
`lin or Mayer's Hamalaun solution (both from Merck.
`Darmstadt, Germany).
`
`Phenotyping of cells
`Monoclonal antibodies to CD3. CD4, CDS. CD20.
`CD45RA and CD45RO were from Becto n Dickinson.
`Mountain View. CA. USA: antibodies to CD68 and to
`HLA-DR were from Dako. Glostrup. Denmark.
`Isotype-rnatched. unrelated monoclonal antibodies
`served as controls. Sections were incubated with the
`primary antibody diluted between 1:20 and 1:100 in
`Tris-b uiTered saline for 60 m.in at room temperature.
`After rinsing, alkaline phosphatase-labelled. affinity(cid:173)
`purified rabbit anti-mouse inm1unoglobulins (Dako)
`were applied as second antibody for 30 min. followed
`by incubation with the APAAP complex (Dako). Colour
`was developed using Fast Blue (Sigma, St Louis. MO.
`USA) as substrate.
`
`Characteri=ation of cytokine-producing cells
`D ouble-labelling experiments were performed by first
`staining for cytokines. followed by immunophenotyping
`as described above. Thus, upon light microscopy. cyto(cid:173)
`kine-expressing cells appeared brown and CD antigens
`stained blue. The percentage o f cytokine-producing cells
`of a given CD subset was calculated by detennining the
`fraction of double-stained
`lymphocytes among all
`lymphocytes counted ( ~20000 cells per cytokine and
`patient) at 400-fold magnification.
`
`Statistical analysis
`Results of cytokine determinations were correlated with
`serological and clinical parameters of disease activity.
`The data obtained in a ll SF investigated were used for
`analysis. even though some joint aspirates (maximally
`two) were from the same patients at dilferent points in
`time. Data were processed using the SAS package.
`Gro up data are indicated as medians and ranges. The
`Wilcoxon rank sum test was used to analyse diflcrences
`between gToups. The non-pa rametric Kendall correla(cid:173)
`tion coefficient was used to assess correla tions between
`cytokines a nd between cytokines and serologica l o r
`clinical parameters. In some analyses. x2 test (with Yates'
`correction when appropriate) was used.
`
`ResuJ ts
`
`Cytokines in sera and synovial fluids
`Cytokine patterns. The cytoki.lle patterns obtained for
`individual patients are shown in a qualitative manner in
`
`Table l. Thus. almost 60% oC RA sera contained measur(cid:173)
`able amounts of at least o ne of the cytokines analysed.
`particularly IL-la and IL-6 (found in 24 and 27% of
`the samples. respectively) . but also IFN-y (15% of the
`sera) and TNF-P (18%). whereas ill sera from REA or
`OA patients cytoki nes were
`in general not found
`(Table lA) . In SF. however. cytokines were detected in
`all three pa tient gro ups (Table 1B) . Comparing Tables
`lA and B. it can be seen that in RA patients the presence
`of a given cytokine in serum was usually associated with
`its presence in SF. a phenomenon not observed in REA
`or OA. Of interest. in nine of the 33 RA sera investi(cid:173)
`gated. two or more cytokines were detectable a nd with
`one exception these sera contained at least one T-cell(cid:173)
`derived lymphokine (Table lA). Summarizing the data
`obtained with SF. R A was characterized by the presence
`of TNF-a. IFN-y and IL-6: R EA samples typically
`contained I FN-y and IL-6. and in OA patients only
`IL-6 was detected. Thus. among these three most fre(cid:173)
`quently detected cytokines. TNF-a was the major dis(cid:173)
`criminator between RA and the other two d isorders.
`Quantitative analysis. IFN-y was the most frequently
`detected lymphokine. T hus. the majority of SF from
`RA and REA patients (72 and 85%. respectively) con(cid:173)
`tained IFN-y (> I 0 pgjml) in contrast to SF from OA
`patients (Fig. 1). Although IFN-y levels in REA were
`in general somewhat higher than in RA. this difference
`did not reach
`the level of statistical significance.
`However. concent rations > 100 pgjml were measured in
`eight REA, but in only two RA samples.
`IL-2. IL-4 and TNF-P were detected much less fre(cid:173)
`quently than IFN-y. Interestingly, and in contrast to the
`monokiues IL-6 and TNF-a (see below), SF levels o f
`these cytokiues were si milar to those measured in serum.
`Thus. IL-2 was present in two serum/SF pairs and in
`two additional SF (detection limit 30 pgjml. range
`36- 107 pg/ml) . and IL-4 was detected in three serum/ SF
`pairs and one additional SF (detection limit 30 pg/ml.
`range 30- 149 pg/ml ).
`fn terestingly. IL-2 and IL-4
`appeared to be associated with each other since five of
`the six IL-2-positive samples a lso contained IL-4
`(Table 1). Finally. TNF-P was detected in nine RA
`patients (four SF/ serum pairs, three SF, two sera) and
`in four REA patients (one SF/ serum pair. three SF);
`levels were similar in both diseases (detectio n limit
`20 pg/ml. range 23- 55 pg/ m.l ).
`The monokines IL-l a, IL- l fJ. IL-6 and TNF-cx were
`detennined in order to investigate potential correlations
`with T-cell-derived cytokines (Table I). T he results were
`genera.lly in line with the literature [2- 4. 8. 10. 11].
`Briefly, IL- la and IL-I P (>25 pgjml) were detected in
`"'30% of patients with RA (26- 182 pg/ml, median
`35 pg/ml ), but not in the o ther disorders. In contrast.
`high
`levels of
`IL-6
`( 400- 80 000 pgjml. median
`4500 pg/ml ) were measured tn virtually all SF from RA
`and REA patients. and also in -70% o f OA patients.
`although levels were significantly lower than in the other
`two disorders (P < 0.01 ). Iu serum. however. IL-6
`(> 30 pg/ml) was found in only 27% of RA patients.
`but no t in R EA a.nd OA patients. and levels were much
`
`

`

`T-cell-derived cytok.ines in RA
`
`205
`
`TABLE I. Cytokine patterns observed in (A) sera and ( B) SF from RA patients. Samples are grouped according to tbe number of cytokines detected per
`sample . For compa rison, the pe rcentages' o f positive s.~mples fro m patients with R EA and OA arc s hown at the bo u om
`
`(A ) Cyto kine paltcms in sera from RA patie nts
`
`IL-2
`30
`
`+
`+
`
`Detectio n limit
`( pgfml )
`
`IFN-y
`10
`
`+
`
`+
`+
`
`+
`
`+
`
`Serum
`14
`4
`2
`19
`9
`II
`I
`23
`28
`3
`26
`32
`10
`6
`13
`15
`29
`2-1
`
`No cytokines were detected in the remaining 15 samples
`
`RA %"
`R EA%
`OA %
`
`15
`0
`0
`
`6
`0
`0
`
`t B) Cytokine patterns in synovial fluids from RA patients
`
`IL-2
`30
`
`+
`+
`+
`
`+
`
`Detection limit
`( pg/ rnl )
`
`IFN-1•
`10
`
`SF
`I-I
`4
`2
`19
`9
`I I
`I
`23
`28
`3
`26
`32
`10
`6
`13
`15
`29
`24
`30
`5
`17
`7
`s
`12
`2 1
`16
`IS
`20
`22
`25
`27
`3 1
`33
`RA %"
`REA%
`OA %
`
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`
`+
`+
`+
`+
`+
`+
`+
`
`+
`+
`
`72
`85
`7
`
`IL-4
`30
`
`+
`+
`+
`
`9
`0
`0
`
`lL-4
`30
`
`+
`+
`+
`+
`
`TNF-P
`20
`
`IL- loc
`25
`
`+
`+
`+
`+
`
`+
`+
`
`18
`
`..
`
`0
`
`+
`+
`+
`+
`+
`+
`
`+
`
`+
`
`24
`0
`0
`
`IL- LP
`25
`
`+
`
`+
`
`6
`0
`0
`
`LL-6
`30
`
`T F-oe
`30
`
`+
`+
`+
`+
`+
`
`+
`+
`+
`+
`
`27
`
`..
`
`0
`
`+
`
`+
`
`6
`0
`7
`
`TNF-p
`20
`
`Ur loc
`25
`
`IL- LP
`25
`
`[L-6
`30
`
`TN F-oe
`30
`
`+
`
`+
`+
`+
`+
`
`+
`
`+
`
`2 1
`16
`0
`
`+
`
`+
`+
`+
`+
`
`+
`
`18
`0
`0
`
`+
`+
`
`+
`
`+
`
`+
`
`+
`
`18
`0
`0
`
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`97
`92
`73
`
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`+
`
`+
`+
`+
`+
`+
`
`70
`16
`0
`
`0
`0
`~
`"'
`0
`
`"' Q. "
`Q.
`=l'
`~
`::r
`.g
`~
`:r "' "
`3
`"
`0
`~
`'<
`0
`><
`Q'
`.=.
`~
`" ;;;
`0
`~
`~
`""
`c
`~
`0
`"
`s:
`"'
`'"
`·'"
`'"
`9
`0\
`
`'<
`
`12
`0
`0
`
`12
`0
`0
`
`

`

`G. Steiner e1 a/.
`
`(r = 0.47. P < 0.000 1; data not shown) . These associ(cid:173)
`ations were seen even better in those cases where consec(cid:173)
`utive samples from individual patients (six RA and six
`REA. respectively) were available. Thus. changes or
`IFN-y, TNF-cx and IL-6 generally occurred in parallel.
`and were accompanied by cha nges in parameters related
`to disease activity (such as ESR. CRP. pla telet count
`and number of lymphocytes in the joint): these changes
`were usually more pronounced in patients with R EA
`(Fig. 3).
`Correlation of cytokines with laboratory and clinical
`parameters. Apart from IL-6. which in RA a nd REA
`patients correlated with ESR a11d CRP (data not showo) ,
`none or the cytokines determined was associated with
`any clinical or serological variable analysed nor with
`disease dura tion. although TNF-cx levels tended to be
`higher in RA patients with lo ng-standing disease. There
`was also no obvious correla tion >vith (glucocorticoid or
`methotrexate) treatment. In consideration of the short
`
`A
`
`300
`
`10000
`
`+TNF-<l
`+ IFN-y
`-o-IL6
`
`:g, 200 ~ 7500
`i
`*-----------
`
`r=
`6>
`
`c
`
`2500
`
`.9:
`=:-
`z
`1!:
`<i 100
`u.
`~
`
`0
`
`206
`
`>160
`
`140
`
`120
`
`- 100
`E --Ol s
`?,'--z
`!:!::
`
`80
`
`60
`
`40
`
`0
`
`0
`
`0
`
`0
`
`0
`
`8
`
`0
`0
`
`0
`
`0
`
`0
`0
`
`0
`
`6
`
`20
`
`0
`
`··-···-··-··± -···---I--····-·-·-o······--···---·-
`REA
`RA
`OA
`FtG. I. Levels ofiFN-y in synovial fluids from patients with
`RA, RE A and OA. Tbc medians arc indicated by solid lines.
`The dotted line indicates the limit of detection ( 10 pg/ml) as
`defined in Patients and methods.
`
`lower than in SF ( 31- 86 pg/ ml. median 46 pg/ml).
`Finally. TNF-cx (> 30 pgjml) was measured in 70%
`of SF from RA patients (39- 382 pgjml. median
`157 pgjml ) , but in only fo ur SF from REA patients
`(P < 0.001 vs RA) and in none of the OA samples.
`Correlations between cytokines. In RA patients. lFN-y
`levels in SF correlated with both those of TNF-a
`(r = 0.31, P < 0.01 : Fig. 2)
`and
`IL-6
`(r = 0.32.
`P < 0.01 ) . TNF-a also correlated strongly with IL-6
`
`500
`
`400
`
`300
`
`200
`
`-E
`OJ s
`I u.. z
`1-
`
`tl
`
`0
`
`e
`
`0
`0
`
`0
`
`0
`
`B
`
`300
`
`~
`~ 200
`.....
`z
`!!:
`:!.
`z
`I-
`
`100
`
`2
`
`Observation
`
`10000
`
`7500
`
`;;=
`0>
`000 B
`<e.
`3
`""'
`
`2500
`
`+lFN-y
`+TNF-<z.
`-o-IL6
`
`100
`
`0
`
`0
`
`0
`
`0
`
`0
`
`r=0.32
`p<0.01
`
`i o
`0
`j
`0
`:
`0
`0~~~--. . ---------.--------.
`50
`100
`150
`lFN-y (pglml)
`
`•-•·•·1•• ••••~•--•-••--·-•••••-·uoo--••• •- •••~------ •-•••·••• ••••
`
`FIG. 2. Correlation between IFN-y a nd TNF-a levels in RA
`synovial fluids. T he detection limits of the two assays (10 and
`30 pg/ ml. respectively) are indicated by dotted lines.
`
`2
`
`Observation
`FIG. 3. Consecutive levels of TFN-y, TN F-a a nd IL-6 in SF of
`a patient with RA (A) and a patient wit h R EA (B) at the first
`visit ( I) and 14 weeks Ia ter (2). The observed decreases in
`cytokine levels were accompanied by clinical improvement (as
`determined by erythrocyte sedimentation rate, C-reactive pro(cid:173)
`tein. platelet count. number of lymphocytes in SF), which was
`more pronounced in the R EA patient; this was o ne of tbe
`four TNF-a-positive REA patients.
`
`

`

`T-cell-derived cytok.ines in RA
`
`207
`
`biological half-life and the primarily auto- or paracrine
`mode of action of lymphokines. clear-cut quantitative
`correlations with systemic markers or general measures
`o f disease activity were not necessarily to be expected.
`However. as mentioned above. associations were seen
`when looking at time courses of individual patients.
`
`Immuno/iiswchemical detection of cytoldnes
`Phenotyping of synovial membrane T cells. Synovial
`membranes from seven of the eight patients with RA
`were heavily
`infiltrated with CD 3-positive T cells
`(100- 700 cel ls per field at x400 magnifica6on) . In
`accordance with the literature. ~60-80% ofT cells were
`CD45R0 + and 40- 60% expressed HLA-DR [5. 6. 13.
`14]. The ratio of CD4:CD8 cells ranged from 2 to 2.5.
`Synovial membranes from four of the five OA patients
`contained o nly a few T cells. In contrast. in sections
`from the fifth OA patient. more pronounced T-cell
`in(iltrates were present: however. clusters were much
`smaller than those seen in RA patients ( 10- 100 cells per
`field) . The CD4:CD8 ratio in OA patients was compar(cid:173)
`able to that seen in patients with RA.
`Cylokine expression in synovia/membranes. In Fig. 4.
`single stainings for the cytokines IFN-y. IL-2 and IL-4
`are shown in tissue sections from a patient with RA.
`The three lymphokines were typically produced by cells
`infiltrating the synovium a nd were not found within the
`lining layer. These lymphokine-expressing cells tended
`to occur in small clusters and were particularly found
`in perivascular locations in close proximity to endothel(cid:173)
`ial cells. In contrast. IL-6 and TNF-a were found
`thro ugl1out the synovium, including the Lining layer. as
`described ( [3. 21. 22J: data not shown). Although some
`sections from OA patients contained T-cell infiltrates.
`neither cytokines nor expressio n of IL-2R could be
`detected (not shown) .
`To determine the relative frequency (i.e. percentage)
`and to characterize phenotypically cytokine-producing
`T cells. double stainiDgs were performed using mono(cid:173)
`clonal antibodies to CD3. CD4 and CD8. Pronounced
`expression of cytokines was seen in o nly ~0.1 -0.3% o f
`all T cells investigated: however. within some T-cell
`aggregates. cytokines could be detected in up to 3% of
`the cells. Although cytokine-expressing cells stained
`relatively weakly fo r CD antigens (presumably because
`the anti-cytokine antibody (applied first) interfered with
`binding of the anti-CD antibody). double-labelled cells
`can be seen in Fig. 5 where representative results
`o btained for IFN-y. I L-2 and IL-4 are shown. In general.
`lymphokine-expressiog cells were CD3 positive (or found
`within clusters of CD 3-positive cells). belonged primarily
`to the CD4 populatioD. and were typically seen in
`perivascular areas and in the more superficial layers of
`the synovial membrane. CD8-positive cells expressing
`cyto kines were also detected. but were more diffusely
`distributed. occurring also in the sublining zone either
`within small clusters or even as single cells. As expected.
`IL-6
`and TNF-a were
`produced mainly
`by
`CD 68-positive cells: nevertheless. both cytokines also
`co-localized with CD3-positive T cells (Fig. 6).
`
`Although due to the heterogeneity of tissue sections
`exact quantification of immunohistochemical data is
`difficult. analysis of double-stained sections revealed
`IL-2 to be the most frequently detected lymphokine (at
`an average of ~ 70 double-labelled cells/ 20 000 cells
`analysed/ patient). fo llowed by IFN-y ( 45 cells/ patient)
`and IL-4 (25 cells/ patient). which was definitely detect(cid:173)
`able in all RA patients investigated even though in one
`patient as few as 11 IL-4-positive cells could be detected.
`The average ratio of IFN-y:IL-4-producing cells was
`~ 2: I . but varied considerably between
`individual
`patients. ranging between 5: I and 1: I. Remarkably.
`T-cell-specific expression of TNF-a was compa rable to
`that of IL-2. whereas IL-6 expression was similar to
`that ofiFN-y. Thus. the ratio ofTNF-cdL-6-producing
`T cells was compa rable to t11e ratio found for IFN-y
`and IL-4.
`
`Discussion
`T he involvement of T cells in the pathophysio logy of
`RA has been heavily disputed in the past decade [ 12- 16).
`Although it is well establ ished that the rheumatoid
`synovium is infiltrated with T cells showing an activated
`phenotype (as indicated by the expression of MHC class
`II molecules and other activa tion markers) . the majority
`of these cells appear to be in a state of unresponsiveness
`since they lack expression of the IL-2 receptor [5. 6 ].
`Moreover. data on the presence of lymphokines in RA
`synovia l tissue and fluid are controversial: even though
`mRNA for T-cell-specific cytokines has been detected
`by some investigators. conclusive evidence for expression
`of cytokine proteins by synovial T cells is still scarce (3.
`7- 11. 22- 25).
`T he data presented here now unambiguously demon(cid:173)
`strate cytokine production by T cells in the synovium o f
`patients with RA. Thus. apart from detecting the Thl
`cytokine IFN-y in the majority of SF, we have been
`able to demonstrate directly expressio n of cytokine
`proteins by synovial membrane T cells using histochem(cid:173)
`ical methods. In contrast, in T -cell infiltrates fro m OA
`patients. cytokine expression was generally not observed
`nor could
`lymphokines be detected
`in
`their SF.
`Interestingly, cytokiue expression was no t o nly seen in
`CD4-positive T-helper cells. but with sim ilar (relative)
`frequency a lso in the CD8-positive population. With
`RA being a MHC IT-associated disease. a pathophysio(cid:173)
`logical involvement of CD8 cells has been largely neg(cid:173)
`lected so fa r. although these cells might play an impo rt(cid:173)
`ant regulatory role [26 ]. F urthermore. one should also
`bear in mind that CD8 cells are activated by endogene(cid:173)
`ous antigens presented by MHC I molecules which may
`include both pathogen-derived and 'true' self structures.
`Compared to RA. the role of T cells in REA is less
`controversial (27]. Therefore. it was somewhat surpris(cid:173)
`ing to find SF levels of IFN-y in REA and RA to be
`sinlilar. This is in line with o bservations reported by
`Simon et a/. [28 j. who investigated cytokine mRNA
`expression in synovial tissue. and suggests that both
`diso rders bear substantial similarities as far as Th 1 cells
`
`

`

`208
`
`G. Steiner e1 a/.
`
`IFNy
`
`IL-2
`
`IL-4
`
`IFNy
`
`A
`
`CD3
`
`CD4
`
`CDS
`
`FIG. 4. Immunohistochemical detection of IFN-y. IL-2 and
`TL-4 by peroxidase staining (brown colour) in a synovial tissue
`section from a patient with RA As shown here, lymphokine(cid:173)
`producing cells were often located in perivascu lar areas in
`close proximity to endothelial cells. Original magnifications of
`all photographs x 400.
`
`are concerned. Although relatively low when compared
`to TNF-a. the levels of IFN-y in SF were comparable
`to those observed in peripheral blood during acute
`transplant rejection episodes or acute viral infections
`[29]. Given that IFN-y is a potent activator of macro(cid:173)
`phages and (together with other stimuli) can induce or
`enhance TNF-a p roduction (30- 32]. one might a lso
`have expected high TN F-a levels in REA: however, this
`was not the case. Hence, IFN-y a lone cannot be held
`responsible for the chronic overproduction of TNF-o: in
`RA. although the observed correlation between IFN-y
`and TNF-C!. levels is suggestive of a mutual relationship
`between these two cytokines. It has been suggested by
`severa l authors that activation of Th2 cells. and thus
`production of anti-inflammatory cytokines such as IL-4.
`may be deficient in RA and other autoimmune diseases
`[28. 33. 34]. This assumption is primarily based on in
`vitro cytokine patterns of synovial T cells f35- 37] and
`
`FIG. 5. Immunohistochemical characterization of synovial
`cells expressing IFN-y (A), IL-2 ( B) and IL-4 (C) in synovial
`tissue sections from a patient with RA . The cytokines were
`detected by peroxidase staining ( brown colour): the surface
`markers CD3, CD4 and CD8 we re visualized by alkaline
`phosphatase staining (blue colour). D ouble-stained cells are
`indicated by arrows. Note that T cells expressing IL-2 typically
`occurred within aggregates of CD3- or CD4-positive cells,
`whereas IFN-y and IL-4 were also expressed by single CD3-
`or CD4-positive cells, a phenomenon generally observed with
`the CD8 population. Original magnifications of all photo(cid:173)
`graphs x 400.
`
`supported by recent observations in animal models [38,
`39]. Our histochemical data may a lso point in this
`direction since fewer cells stained positive for IL-4 than
`for IL-2 or IFN-y. even though IL-4-expressing T cells
`were definitely seen in all R A patients investigated.
`Determination of IL-4 in SF. on the other hand. did
`not provide further information since. in contrast to
`IFN-y. this cytokine (as well as IL-2 and TNF-/3) was
`only rarely detected, most likely due to the lower sensi(cid:173)
`tivity of the immunoassays used. In any case. cytokine
`data must be interpreted with caution since it would
`probably be too simplistic a view to explain a complex
`
`

`

`T-cell-derived cytok.ines in RA
`
`209
`
`IL-4
`
`IL-2
`
`B
`
`CD3
`
`CD4
`
`c
`
`CD3
`
`CD4
`
`CDS
`
`CDS
`
`FIG. 5. (Cominued)
`
`autoimmune disease like RA solely on the basis of a
`Thl - T h2 imbalance.
`In the light of the apparent pivotal impotiance of
`TNF-a and IL-l and their production by mesenchymal
`a nd myelomonocyticcells of the pannus 12- 4] , it has been
`argued that the pannus was an autonomous tissue with
`macrophage-like cells being centTa1 to the pathogenic
`events. and that T cells entered the synovial membrane as
`·innocent bystanders' rather than as cells specifically
`in
`the
`inflammatory process [ 12. 40- 42 ].
`involved
`According to this assumption, T -cell activation would
`merely reflect a secondary phenomenon or just a reaction
`o fT cells to the ongoing destructive process. Mo reover.
`the n umbers of synovial T cells secreting cytokines could
`be regarded as being far too low to exert significant effects.
`However. several lines o f evidence suggest tha t T cells
`play an important ro le in the pathogenesis ofRA. (i) It is
`well established that small numbers of antigen-specific
`T cells are sufficient to drive inflammatory (immune and
`autoimmune) processes [43, 44 ]. For example. the freq uen(cid:173)
`cies of antigen-specific T cells seen in the cerebrospinal
`fluid of patients with multiple sclerosis o r in skin lesions
`
`of patients suffering from leprosy or allergic contact der(cid:173)
`matitis are well below 1%. and thus of the same order o f
`magnitude as the frequency of cytokine-expressing T cells
`reported here [ 45-47]. (ii) RA is a disease strongly associ(cid:173)
`ated with certain MIIC class II alleles. suggesting involve(cid:173)
`ment of certain T-cell receptors [ 13. 15. 48]. (iii) T cells
`a re already seen i11 the synovium at very early stages o r
`RA and with progressing disease there appears to be little
`change of the synovium's immunohistological features
`[ 49]: furthennore, cytokine-expressing (naive) T cells have
`been observed in the peripheral blood ofRA patients [50].
`(iv) Oligoclooality of synovial T-cell populations has bee