Cell Surface CD28 Levels Define Four CD4+ T Cell Subsets: Abnormal Expression in Rheumatoid Arthritis

uvr(te(iC

C

0pC

Clinical ImmunologyVol. 99, No. 2, May, pp. 253–265, 2001doi:10.1006/clim.2001.5003, available online at http://www.idealibrary.com on

Cell Surface CD28 Levels Define Four CD41 T Cell Subsets:Abnormal Expression in Rheumatoid Arthritis1

Laura-Ines Salazar-Fontana,* Eva Sanz,* Isabel Merida,† Antonio Zea,* Ana Sanchez-Atrio,* Luis Villa,*,‡Carlos Martınez-A,† Antonio de la Hera,* and Melchor Alvarez-Mon*,‡,2

*Laboratory of Immunology and Oncology, School of Medicine, University of Alcala-Centro de Investigaciones Biologicas (CSIC)Associated Unit, Ctra. Madrid–Barcelona Km. 33, 28871 Alcala de Henares, Madrid, Spain; ‡Clinical Immunology and Oncology Branch,Prince of Asturias University Hospital, Ctra. Madrid–Barcelona Km. 33, 28871 Alcala de Henares, Madrid, Spain; §Rheumatology Unit,

Ramon y Cajal University Hospital, Ctra. Madrid–Colmenar Km 9.1, 28034 Madrid, Spain; and †Department of Immunology and
Oncology, Centro Nacional de Biotecnologıa, CSIC, Ctra. Madrid–Colmenar Km. 15, 28049 Madrid, Spain
CD28 is a costimulatory receptor expressed in mostCD41 T cells. Despite the long-standing evidence for

p- and downregulation of surface CD28 expression initro, and the key regulatory role assigned to the up-egulation of CD28 counterreceptor [the CD152CTLA-4) molecule], in vivo CD28 induction has at-racted little attention. We studied CD28 and CD152xpression and function in 33 rheumatoid arthritisRA) patients, 20 clinically active and 13 inactive, andn 24 healthy donors. Four subsets of CD282, CD28low,D28int, and CD28high peripheral blood human CD41 T

cells were defined using three-color flow cytometry.The three CD281 subsets displayed a one-, two-, orthreefold quantitative difference in their relativenumber of CD28 antibody binding sites, respectively(P < 0.01). RA patients, whether active or inactive,showed a distinct phenotype when compared tohealthy donors: (i) the percentage of CD41CD28high

cells was increased twofold and the CD41CD28low sub-set was reduced twofold (P < 0.01) and (ii) the

D41CD28high cells from RA patients showed an in vivoactivated phenotype, CD45RO1CD5highIL-2Ra1 (P <.01). Active RA patients were different from inactiveatients. They showed a twofold increase in meanD28 expression (P < 0.05), whereas each of the CD281

subsets in the inactive RA patients showed reducedexpression when compared to healthy donors. Nota-bly, both active and inactive RA patients showed ab-normal CD28 upregulation when T cells were acti-vated in vitro with CD3 antibodies, but only inactive

1 This paper was supported by Plan Nacional de Salud (CICYT)Joint Grant 99-0099-CO3 to the University of Alcala (E.S. andM.A.M.) and the Centro de Investigaciones Biologicas (A.H.). TheDepartment of Immunology and Oncology was founded and is sup-ported by the CSIC and Pharmacia & Upjohn. Salazar-FontanaPh.D. thesis was supported by the CICYT Grant SAF 96-201 toM.A.M. and A.H.

2 To whom correspondence should be addressed at Departamentode Medicina, Facultad de Medicina, Ctra. Madrid-Barcelona Km. 33,28871 Alcala de Henares, Madrid, Spain. Fax: 34-91-8854526. E-
mail: [email protected].
253

RA patients showed a hypoproliferative response toTCR/CD3 triggering when compared to healthy donors(P < 0.01). This defective proliferation was normalizedby concurrent crosslinking with CD28 antibody. Nodifferences were noted in the expression of CD152 orCD80, a CD28 and CD152 shared ligand. The disregu-lated in vivo expression of CD28 was related to the RApatients’ disease activity and suggests that modula-tion of CD28 surface levels may be an additional mech-anism to finely tune the delicate responsiveness/toler-ance balance. © 2001 Academic Press

Key Words: rheumatoid arthritis; CD28 surface mod-ulation; T-cell anergy; costimulatory molecules;CTLA-4; CD152; CD80; CD86; B7.

INTRODUCTION

Optimal stimulation of T cells requires engagementof the T cell receptor for antigen, the TCR/CD3 com-plex, concurrently with second signals provided by co-stimulatory receptors (1). When costimulatory signalsare provided, T cells can escape to a status of functionalclonal inactivation to recall antigens termed anergy(1–4). Accumulating evidence indicates that a primarysource of costimulatory signals is the interaction be-tween CD28 on the T cell with members of the B7family (CD80, CD86) on the antigen-presenting cell(APC) (2–6). Simultaneous engagement of the CD28receptor, by its natural ligands or with antibody, re-duces 5-fold the threshold number of triggered TCRrequired for T lymphocyte activation (5) and greatlyenhances interleukin-2 (IL-2) production, IL-2 receptorexpression, and, consequently, proliferation by TCR/CD3-stimulated T cells (2–4). CD28 is expressed on thesurface of most CD41 and in half of CD81 adult humanperipheral blood T cells (6). There is a second receptorfor CD80 and CD86, termed CTLA-4 (CD152), which isalso expressed by both CD41 and CD81 T cells (7).
Despite its homology to CD28, CD152 exhibits three
1521-6616/01 $35.00Copyright © 2001 by Academic Press

All rights of reproduction in any form reserved.

aimSispNafwe

cwrw

254 SALAZAR-FONTANA ET AL.

distinguishing features (reviewed in Ref. 8): (i) it bindsto its B7 ligands with a 17-fold higher affinity com-pared to CD28, (ii) unlike CD28, which is constitutivelyexpressed in high abundance by T cells, CD152 is notreadily detectable in resting cells, but is inducible inlow abundance upon activation, reaching maximal lev-els on the cell surface only 24–48 h later, and (iii) it isnot functionally redundant with CD28, but seems toact, in part, in opposition to the costimulatory effects ofCD28, downregulating T cell responses to TCR/CD3-triggered signals.

As originally formulated, the restriction of costimu-latory ligands expression to “professional” APC pro-vides a mechanism for peripheral tolerance to tissue-specific antigens expressed by cells that do not expresscostimulatory ligands (1). Costimulatory ligand CD80confers APC function on parenchymal tissue and leadsto autoimmunity in experimental models (9). Blockadeof CD28 or CD152 interaction with their ligands havebeen shown either to suppress (10–12) or to exacerbatethe severity (13) of experimental autoimmunity, re-spectively. This underlines the relevance of anergy as afail-safe mechanism to maintain effective self-toler-ance by autoreactive clones (14). Rheumatoid arthritis(RA) is a highly prevalent autoimmune disease, affect-ing 1–3% of the adult population, whose main clinicalfeature is the inflamation of diarthrodial joints (15), forwhich elevated expression of CD80 has been reportedin synovial fluid mononuclear cells (16). RA is a chronicdisease, but many patients link periods of “active” dis-ease with spontaneous or drug-induced phases of re-mission in the manifestations of inflammation, heretermed “inactive,” which can in turn end in a newinflammatory relapse (17).

The requirements for activation of memory T cellsare less stringent than those for naive T cells, due inpart to their different patterns of costimulatory mol-ecule expression (18). The upregulation of the sur-face expression of several costimulatory moleculesthat occurs after T cell activation is well docu-mented, and their multimodal expression is widelyused to distinguish resting/naive from activated/memory T cell subsets in vivo (18 –20). We reasonedthat, if CD28 were upregulated in in vivo activated Tcells, as it is induced both at the protein and at theRNA level upon in vitro TCR/CD3 stimulation (21,22), measurement of surface CD28 expression levelsmay be valuable to delineate activated/memory T cellsubsets using a better defined costimulatory receptoras an “activation marker.” We have explored thishypothesis in peripheral blood T cells from eitherclinically active or inactive RA patients and healthydonors and found that CD28 levels define four CD41

T cell subsets. There is marked redistribution of theCD41CD281 subpopulations in RA patients when
compared to healthy donors, with distinct phenotypic
and functional features in clinically active and inac-tive patients, which may contribute to the under-standing of tolerance failure in RA.

PATIENTS AND METHODS

Patients. The study protocol was approved by theUniversidad de Alcala Research and Ethics Commit-tees, and informed consent was obtained from all par-ticipating individuals. Thirty-three patients were stud-ied with a diagnosis of RA on the basis of the 1987revised criteria of the American Rheumatism Associa-tion (15). Patients were separated based on their dis-ease activity, according to the criteria for remissiondeveloped by the American College of Rheumatology(17), remission being absence of RA disease activity(here termed inactive RA); 20 patients were active and13 inactive when blood and synovial samples weredrawn. The six criteria for disease activity assessment(duration of morning stiffness, fatigue, joint pain byhistory, joint tenderness or pain in motion, soft tissueswelling in joints or tendon sheaths, and erythrocytesedimentation rate; 17) were monitored for each pa-tient by two rheumatologists. All patients included inthe study were positive for rheumatoid factor, and theydid not suffer clinical manifestations of rheumatoidvasculitis or heart, lung, or other vital organ affecta-tions that are exclusion criteria for remission (17).Also, the patients had neither evidence of current acuteor chronic disease, other than RA, nor a history ofpathological conditions with possible effects on the im-mune system or alterations in their nutritional stage.At the time of the immunological study, the patientswere receiving one of the following treatments: non-steroid anti-inflammatory drugs (NSAID), 5 inactive/3active; low prednisone (,10 mg/each other day), 2 in-ctive/3 active; or methotrexate (10–15 mg/week), 6nactive/8 active. Blood samples were extracted at a

inimum of 18 h after the most recent treatment dose.ix recently diagnosed, untreated active patients were

ncluded in the study. In 4 of these patients a bloodample and a synovial tissue biopsy were obtained inarallel at the initial diagnosis time, before treatment.either these patients nor the 24 healthy sex- andge-matched controls were taking drugs known to af-ect the immune system. A blood and synovial sampleas also obtained concurrently in a patient who wasntering remission after NSAID treatment.

Cell separation. Peripheral blood mononuclearells (PBL) and synovial membrane mononuclear cellsere obtained as described elsewhere (23). Cells were

esuspended in RPMI 1640 medium supplementedith L-glutamine (BioWhittaker, Walkersville, MD),

10% heat-inactivated fetal bovine serum (BiochromKG, Berlin, Germany) and 25 mM Hepes (Biochrom
KG) (complete medium). Cells were counted in a

matbS

tC

255ABNORMAL CD28 EXPRESSION IN RHEUMATOID ARTHRITIS

Neubauer chamber, and cell viability as confirmed bytrypan blue exclusion was always .95%.

Immunofluorescence and quantitative flow cytom-etry. For three-color immunofluorescence staining,105 cells were incubated for 30 min with the indicatedcombinations of fluorescein-(FITC, green), phyco-erythrin- (PE, red), and peridin chlorophyll proteinconjugate- (PerCP, blue) labeled monoclonal antibodies(mAb). The mAb used were anti-CD3 (SK7, mIgG1),CD4 (SK3, mIgG1), CD5 (L17F12, mIgG2a), CD8 (SK1,mIgG1), CD25 (2A3, mIgG1), CD28 (clone L293,mIgG1), CD69 (L78, mIgG1), CD71 (LO1.1, IgG2a),and anti-HLA-DR (L243, IgG2a); all were purchasedfrom Becton–Dickinson (San Jose, CA). CD45RO(UCHL1, mIgG2a) and CD152 (BN13, mIgG2a, biotin-ylated) were obtained from Caltag Laboratories (SanFrancisco, CA) and Pharmingen (San Diego, CA), re-spectively. The biotinylated anti-CTLA-4 mAb was de-veloped with streptavidin–PE (Becton–Dickinson). An-tibodies specific for TCRb families 3.1, 5a, 6.7, 8a, 13,and 17, as well as for CD80, were all from Pharmingen.The immunofluorescence distributions of the lympho-cyte population, electronically gated by size/complexitycriteria and defined as .99% CD451CD142 cells, were

easured in a FACScan cytometer using Cell Questnd Paint-A-Gate software (Becton–Dickinson). In cul-ured cells, viability was monitored by flow cytometryy the addition of 7-aminoactinomycin D (7-AAD,igma, St. Louis, MO) at 1 mg/ml, which allows for the

exclusion of dead cells and quantitation of fluorescencesignals from two surface receptors. A logarithmic scaleof fluorescence intensity was used, and those cells dis-playing fluorescence intensities above the distributionof an isotype- and fluorochrome-matched irrelevantmAbs were considered positive. CD28 surface antigenexpression was quantitated using a modification of theDako QifiKit (Dako A/S, Glostrup, Denmark), as pre-viously described (24, 25). Briefly, standards were sixsets of beads coated with IgG2a (0, 2.4, 8, 36, 140, or360 3 103 molecules of IgG2a mAb) that were stainedwith saturating amounts of FITC- or PE-conjugatedanti-mIgG2a (Southern Biotechnology, Birmingham,AL). Immunofluorescence distribution was analyzedusing FACScan with the fluorescence detectors FL1 orFL2 set at different voltages. The optimal regressioncurves between antigenic binding capacity (ABC) andmean fluorescence intensity (MFI) of all six standardswere found for 473 gain value in the fluorescence de-tectors. This detector voltage was routinely used forthe quantitation of CD28 or CD152 expression in theacquisition of biological samples. The regression for-mulas log ABC 5 2.5231 1 0.9749 log MFI (FITC) andlog ABC 5 2.1045 1 0.99483 log MFI (PE) were used toconvert the MFI data provided by the FACScan soft-
ware into ABC values (24).
Proliferation, CD28 expression induction, cytokineproduction, and Bcl-XL expression assays. In func-ional assays, cells were stimulated either with solubleD3 mAb (OKT3, 1 mg/ml) from Ortho Pharmaceuti-

cals (Raritan, NJ) or with CD28 mAb (clone 15E8,mIgG1, 1 mg/ml) from CLB (Amsterdam, Holland),alone or in the combinations indicated. Optimal dosesof stimuli were selected after titration experiments inPBL cultures from 20 healthy donors. To assess T cellproliferative capacity (24), 2.5 3 105/ml PBL wereplaced in 96-well flat-bottom culture plates (Nunc,Roskilde, Denmark) in triplicate 0.2-ml cultures, incomplete medium alone or supplemented with the in-dicated stimuli. Cells were pulsed with 1 mCi [3H]thy-midine (Amersham, Aylesbury, UK) for the last 15–18h of the indicated culture period. To determine theamount of IL-2 secreted in culture, PBL were activatedas indicated, cell-free supernatants were collected, andIL-2 was quantitated by ELISA using the commerciallyavailable IL-2 kit Predicta (Genzyme, Cambridge, MA)(24). Similar results were obtained after 24 h of culturewhen cells were incubated in the presence of saturat-ing amounts of H-108 anti-IL-2Ra antibody producedin our laboratory and added to the culture to preventIL-2 consumption. Bcl-XL expression in activated PBLwas measured using Western immunoblotting withspecific antibodies (Oncogene Research, Cambridge,MA) and developed using the ECL chemiluminescencereagent (Amersham) as described (26). In the immuno-blotting assays, the optimal amount of protein loaded(25 mg/assay) was determined in preliminary experi-ments using bicinchoninic acid assay (BCA, Pierce),which was also used in the routine monitoring of pro-tein concentration loaded in every experiment. Weused human thymocytes to optimize the Western blotconditions, as they are a cell source expressing Bclxgene products constitutively, in the absence of ex vivostimulation (unpublished data). To assess the in vitroCD28 induction upon anti-CD3 stimulation, cells werecultured as originally reported by Turka et al. for CD28induction in human thymocytes (21), with a slightmodification. Primary cultures were set at the sameconcentration as in the proliferation assays (2.5 3105/ml PBL). After 4 days in culture, the cells weresubcultured at 2.5 3 105/ml every third day in thepresence of 64 U/ml rhIL-2 (R&D Systems, Minneapo-lis, MN). Viability was .95% under these culture con-ditions, as assessed by trypan blue dye exclusion, at alltime points examined from day 4 to day 10, which wasthe latest time point tested.

Quantitation of IL-2 mRNA expression. Theamount of IL-2 message was measured by competitivereverse-transcriptase polymerase chain reaction (RT-PCR), using as competitor a riboprobe containing the
hIL-2 RNA with an internal deletion, developed at the

sHtmttatIAsGetl(tbDSqA(

iIptpRtgpa

r

efltoimvtp

wab

C


Department of Genetics, Stanford University (L. Her-zenberg and co-workers, unpublished results). TotalRNA was extracted from 3 3 105 cells using the Ultra-pec RNA isolation system (Biotecx Laboratories,ouston, TX) in the presence of the riboprobe and

ranscribed to cDNA using oligo-dT12–18 primers (Phar-acia Biotech, Uppsala, Sweden) and AMV reverse

ranscriptase (Promega, Madison, WI) (26). Amplifica-ion (27) was performed for 33 cycles: 30-s denatur-tion at 94°C, 30-s annealing at 54°C, and 1-min ex-ension at 72°C using Taq Polymerase (Ecogen). TheL-2 gene-specific primer sequences, sense primer 59-TGTACAGGATGCAACTCCTGTCTT-39 and anti-ense primer 59-GTCAGTGTTGAGATGATGCTTT-AC-39, flanked intervening introns. Preliminaryxperiments using titrated numbers of plasmids con-aining IL-2 and IL-13 showed that the conditions al-owed for a linear and sensitive PCR amplificationapprox. 3000 copies, data not shown). The PCR reac-ion products were subjected to electrophoresis and gelands were measured densitometrically using a Gelocumentation System 1000 (Bio-Rad, Hercules, CA).pecificity of the amplification was confirmed by se-uencing the gene products in the labeled bands in anBI PRISM 337 automatic sequencer (Perkin–Elmer)

26).

Statistical analysis. All analysis was performed us-ng SPSS for Windows 6.0 software (SPSS, Chicago,L). In order to account for the significant skew inroliferation and surface receptor expression levels,he nonparametric Mann–Whitney U test was em-loyed for analysis among healthy, active, and inactiveA groups. Differences in the immunological parame-

ers within the healthy donor and the RA patientroups were assessed with the Wilcoxon test for re-eated measurements. Statistical significance was sett P , 0.05 for all tests.

RESULTS

CD28 Expression in CD41 T Cells Derived fromHealthy Donors and Rheumatoid Arthritis Patientswith Clinically Active or Inactive Disease

To study putative differences in the peripheral bloodT cells from RA patients with active or inactive disease,classified according to criteria that define remission(17), three-color immunofluorescence and flow cytom-etry studies were conducted to quantitate activated/memory T cell subsets in RA and healthy donors. An-tibodies to the TCR/CD3 complex, CD4 or CD8coreceptors, and several costimulatory molecules, fre-quently quoted as activation markers (18–20), wereused for this purpose. To expand on previous studiesshowing impaired in vitro activation of RA T cells in
esponse to polyclonal mitogens (28, 29), quantitation
of the CD28 costimulatory molecule was also includedin our study.

CD41 T cells from RA patients suffering active dis-ease display significantly higher CD28 surface levelsthan those of either inactive RA patients or healthydonors (Fig. 1); the average increase in mean CD28expression (MFI) was twofold (P , 0.05). Two findingsxplained this augmented CD28 expression. First, theuorescence distribution curves from active RA pa-ients demonstrated an overall shift to the right. Sec-nd, further analysis of the enlarged profiles of themmunofluorescence distribution curves revealed a

ultimodal distribution of CD28 expression and pro-ided evidence for increases in the relative frequency ofhe T cell subset bearing higher CD28 levels in RAatients. Four CD41 T cell subsets were evident, in-

cluding a small subpopulation of helper cells that didnot express surface CD28 (CD282 subset, ,1%),

hereas three other subsets expressed low, intermedi-te, and high CD28 levels (represented in the figuresy green, red, and cyan, respectively). The three CD281

subsets can be ascertained from the indentations in thedistribution profiles, which indicate the gaps in theoverlap of the components in the multimodal distribu-tion. The CD281 subsets were readily evident in RApatients which led to their identification and quantita-tion.

Table 1 shows the results of the quantitation of an-tibody binding capacity equivalent sites (24) for thethree CD281 subsets in CD41 T cells from healthydonors and RA patients with active or inactive disease.In healthy donors, the CD41CD28low cells displayed 2 3104 ABC equivalents, whereas the CD41CD28int and

D41CD28high populations displayed roughly two- andthreefold higher CD28 site numbers. A similar profileof one- to threefold distribution was observed in thethree CD281 subpopulations from RA, regardless ofwhether the disease was active or inactive. The differ-ences in the CD28 levels between the CD28high and theCD28int, and between the CD28int and the CD28low sub-populations, were statistically significant within thehealthy, inactive, and active RA donor groups (P ,0.01). Two major differences were noted, however, be-tween healthy donors and RA patients (Table 1). First,both active and inactive RA patients showed a markedredistribution of the relative proportions of the threeCD41CD281 subsets compared to the healthy controls;a twofold decrease in the percentage of CD41CD28low

cells (P , 0.01 for inactive and P , 0.05 for active RAgroups) was combined with an almost twofold in-creases in the percentage of CD41CD28high cells (P ,0.01 for both RA groups). Second, the average CD28ABC site distribution for the CD41CD28low,CD41CD28int, and CD41CD28high populations displayedan evident (;30%) increase in the active RA patients
compared to healthy donors (P , 0.01), which contrib-

tido

d.blspm

e

c

t


utes to the higher MFI found in active RA patients. Incontrast, only a slight decrease (;16%) was noted inhe ABC CD28 sites for each of the three subsets in thenactive RA patients analyzed compared to healthyonors. Such a decrease compensates for the expansionf the proportion of CD41CD28high T cells and results in

a global average CD28 level similar to that of healthydonors. Finally, and consistently with previous results(6, 30), CD31CD41CD282 cells are rare in normal in-

ividuals. The occurrence of marked expansions (i.e.,10%) of this compartment which occurs in peripherallood from some RA patients is well documented in theiterature (31, 32). The latter subset of RA patientsuffers oligoclonal expansions of T cell populations andreferentially bears defined HLA haplotypes, which

FIG. 1. CD41 T cells from active RA patients express signifi-cantly increased amounts of surface CD28. PBL from a representa-tive healthy donor (A) and a clinically inactive (B) or active (C) RApatient were stained with anti-CD3, -CD28, and -CD4 antibodiesconjugated with FITC, PE, or PerCP, respectively. Samples wereanalyzed by flow cytometry. T lymphocytes were identified by char-acteristic forward angle and light scatter profiles, as well as anelectronic gate set on CD31 cells. Dot plots on the left show thecorrelated CD28 ( x axis) vs CD4 ( y axis) expression in CD31 T cells.Histograms on the right depict the distribution of CD28 expressionin CD31CD41 T cells. Paint-a-Gate software was used to display thehree CD31CD41 T cell subpopulations expressing low (green), inter-

mediate (red), or high (cyan) levels of surface CD28 staining; theremaining cell subsets are depicted in gray.

ay impose a genetic control (31). Remarkably, the

xpansion of CD31CD41CD282 T cells has been shownto correlate with the development of extraarticular le-sions rather than RA disease activity at the articularlesions (32). We found that the expansion ofCD31CD41CD282 cells above the standard levels inthe healthy population (,2%) did occur in two, but notin the majority of RA patients, in our series. Expansionof CD31CD41CD282 cells in those two cases did notcorrelate (33) with the RA disease clinical activity clas-sification used, which dedicates four of six criteria tothe direct survey of disease activity in articular lesions(17). Notably, the patients in this study were not se-lected by HLA haplotype and did not suffer extraar-

FIG. 2. The expanded CD31CD41CD28high T cell subpopulationfrom RA patients bears a distinctive CD45RO1CD5highIL2Ra1 in vivoactivated phenotype. PBL from a representative healthy donor (A)and a clinically inactive (B) or active (C) RA patient were stainedwith PerCP-conjugated anti-CD4, PE-conjugated anti-CD28, and an-ti-CD5, anti-CD25 (IL-2Ra), or anti-CD45RO FITC-labeled antibod-ies. Samples were analyzed by flow cytometry. CD41 T lymphocyteswere identified by characteristic forward angle and light scatterprofiles, as well as an electronic gate set on bright CD41 cells. Dotplots show the correlated expression of CD28 ( x axis) vs CD5 ( y axis)at the top, CD25 ( y axis) in the middle, or CD45RO ( y axis) at thebottom. Paint-a-Gate software was used to display the three CD41 Tell subpopulations expressing low (green), intermediate (red), or

high (cyan) levels of surface CD28 staining; the remaining cell sub-sets are depicted in gray (see Fig. 1). These results are representative
of 10 simultaneous determinations.

lC

H

A

F


ticular manifestations of RA disease other than rheu-matoid nodules, which may contribute to the lowfrequency of patients with an expanded CD282 Thelper subset in our cohort of RA patients (32, 33).

Despite the fact that this report focuses on CD41

helper T cells, it is worth mentioning that CD81 cyto-toxic T cells show a similar multimodal distribution ofsurface CD28 expression with CD282, CD28low,CD28int, and CD28high cell subpopulations (33–35). Theamount of CD28 expressed on CD281 cells is, however,ower in CD81 than in CD41 T cells, and theD81CD282 cells are frequent and express higher CD8

levels than the CD81CD281 subsets (33, 35). Althoughthe three CD281 subsets show a similar one-, two-, andthreefold distribution, the CD81281 cells express ap-proximately half the number of equivalentCD41CD281 subsets (i.e., the CD81CD28low subsetsdisplayed 104 CD28, data not shown, Ref. 33).

The CD41CD28high T Cells Are Largely CD45RO1

Cells That Coexpress IL2Ra Chain and CD5high

Costimulatory Molecules in RA Patients butNot in Healthy Donors

Since CD28 can be upregulated or downmodulatedfollowing in vitro activation by engagement of TCR/CD3 complex (21) and CD28 receptors (36), respec-tively, we studied whether each of the threeCD41CD281 T cell subsets expresses a distinct in vivodistribution of molecules to define activated/memory Tcells. Three-color immunofluorescence and flow cytom-etry analyses were done in PBL from the two RA pa-

TABDifferences between the CD281 Subpopulatio

and Clinically Active

Study groupSubpopulations ofCD31CD41 cellsa

Antibody bin(ABC) CD28 eq

ealthy donors (n 5 24) CD28low 20,006 6CD28int 36,245 6CD28high 58,552 6

Inactive RA (n 5 13) CD28low 16,813 6CD28int 30,711 6CD28high 55,378 6

ctive RA patients (n 5 20) CD28low 25,998 6CD28int 47,115 6CD28high 83,019 6

a The three CD281 T helper cell subsets from healthy donors andig. 1.

b The number of CD28 sites per cell was estimated from the mean fl2.1045 1 0.99483 log MFI, r 2 5 1. The regression curve was obtaine(as analyzed in Methods).

c Results represent means 6 SD.

tient groups and the healthy donors, stained with anti-

CD4, CD28 plus antibodies specific for one of severalmolecules including CD25, CD45RO, CD69, CD71, orHLA-DR. To study the expression of the activationmolecules in the CD41CD28low, CD41CD28int, andCD41CD28high populations, these were gated in green,red, and cyan in the dot plots using Paint-a-Gate soft-ware, as indicated in Fig. 1. Differences can be ob-served in the pattern of activation molecules in thethree CD41CD281 cell subsets and between their dis-tribution in RA patients and healthy controls (Fig. 2).The CD41CD28high population had features of acti-vated/memory cells, as they were largely CD45ROcells, whereas CD41CD28int cells contained bothCD45RO1 and CD45RO2 subsets and the CD41-CD28low populations were largely CD45RO2 cells. Thedistribution patterns of CD45RO were similar withinthe three CD41CD281 subsets from RA patients andhealthy donors, although the expansion of theCD41CD28high population and decrease in theCD41CD28low population in RA account for the higherproportion of CD41CD45RO1 cells reported in the pa-tients (P , 0.01) (37). In contrast, the proportion anddistribution of activated cells expressing the IL-2 re-ceptor alpha chain (IL-2Ra or CD25) was markedlydifferent between RA patients and healthy controls(P , 0.01). Most cells in the three CD41CD281 sub-sets from healthy donors were CD252, regardless ofCD28 expression levels. In the active RA patients, how-ever, the CD41CD28high population was largely CD251

and showed a unimodal distribution, whereas theCD41CD28int or low cells showed a clear bimodal distribu-tion and only a proportion were CD251. CD41-

int or low

1of CD31CD41 T Cells from Healthy Donors

Inactive RA Patients

g capacityalent sitesb

Number of CD28 sites relativeto the CD28low subset from

healthy donorsb (%)

Percentage of theCD31CD41 CD281

cell subset

5801 100 14 6 5,968 181 58 6 7,768 293 28 6 6

7347 84 6 6 4,496 154 51 6 10,582 277 43 6 14,108 130 8 6 7,783 236 48 6 10,056c 415 44 6 11

ically inactive or active RA patients were quantitated as shown in

rescence intensity (MFI) of each subset using the formula log ABC 5sing a set of six calibration beads with titrated amount of antibodies

LEnsor

dinuiv

1016

1223152750

clin

uod u

CD28 cells expressed higher CD25 levels than the

ti

wpwtdhRo

cghRu

(hoec


CD41CD28high CD251 subset, which expressed lowamounts of the IL2Ra that were still clearly evidentwhen compared to those in the CD252 subsets ofCD41CD28int or low cells. Interestingly, a transitional pat-ern was observed in RA patients with inactive disease,n which the CD41CD28high subset showed a clear bi-

modal IL2Ra distribution with CD252 and CD25dull

cells, and a minor proportion of CD41CD28int CD251

cells occurred. The RA patients also showed a CD5distribution pattern distinct from that of healthy do-nors (P , 0.01). CD5 is a costimulatory molecule, for

hich a biochemical association to the TCR/CD3 com-lex during T cell activation has been reported (38),hich displays a homogeneous or unimodal distribu-

ion in healthy donors (25). We also found uniform CD5istribution in the three CD41CD281 subsets fromealthy donors. CD5 distribution was not uniform inA patients, however, whether their disease was activer inactive; the CD41CD28high subset showed clearly

higher levels of CD5 expression than did theCD41CD28int or low subsets. Finally, we observed no dif-ference in distribution of activation markers CD69,CD71 (transferrin-R), and HLA-DR or monitorizedTCR-Vb chain usage in the CD41CD281 subpopula-tions from RA patients and healthy donors (data notshown, 33).

CD41 T Cells from Active and Inactive RA PatientsShow Increased Upregulation of CD28 ExpressionFollowing in Vitro Activation with CD3 AntibodiesCompared to Healthy Donors

Although active and inactive RA patients showedsignificant differences in their mean CD28 expres-sion level in T cells and in their clinical classificationfor disease activity (remission criteria Yes/No), it canbe argued that CD281 subset differences reflect aasual heterogeneity of the human genetic back-round, as there is CD28 level variation betweenuman subjects, whether healthy or suffering fromA (Table 1, 33, 35). To address this, the ability topregulate CD28 expression in CD41 T cells was

tested in a CD3 mAb-driven in vitro activation assay(21). When CD41 T cells were compared from activeFig. 3C) and inactive RA patients (Fig. 3B) andealthy donors (Fig. 3A), marked differences werebserved in the upregulation of surface ABC CD28quivalent sites at 96 h after TCR/CD3 antigen re-eptor triggering. The CD41CD281 subpopulation

strongly augmented its CD28 expression in both RApatients and healthy donors; however, ABC CD28equivalent sites were 2.5 higher in RA patients whencompared to healthy donors (67,045 6 18,934 vs166,728 6 19,073, P , 0.01). Moreover, active andinactive RA patients, who expressed significantly dif-
ferent CD28 levels in vivo (P , 0.05) (Fig. 1, Table
1), equalized CD28 expression levels after in vitroactivation (169,636 6 21,082 and 160,728 6 18,523ABC sites, respectively); both were now significantlyhigher for CD28 staining than the healthy donors(P , 0.01). The differences between RA patients andhealthy donors do not obey distinct kinetics of CD28induction upon activation (Fig. 4). RA patients,

FIG. 3. CD41 T cells from active and inactive RA patients showa selectively disregulated upmodulation of CD28 expression follow-ing TCR/CD3 engagement. PBL from a representative healthy donor(A) and a clinically inactive (B) or active (C) RA patient were ana-lyzed by flow cytometry either ex vivo or after a 96-h in vitro incu-bation with unlabeled anti-CD3 antibodies (OKT3, 1 mg/ml). Histo-grams represent the distribution profiles of CD28 (left) or CTLA-4(right) within the CD41 T cell subset and superimpose the expressionpattern observed either before (gray filled histograms) or after invitro activation (empty histograms). Cells were stained with PE-conjugated anti-CD28 and FITC-conjugated anti-CD4 or with FITC-conjugated anti-CTLA-4 and PE-conjugated anti-CD4, respectively;dead cells were excluded by 7-AAD staining (see Patients and Meth-ods). CD41 T cells were identified by characteristic forward angleand light scatter profiles, as well as electronic gates set on brightCD41 cells and 7-AAD negative cells. Results are representative ofdeterminations done in PBL from six independent healthy donors,six independent inactive RA patients, and six independent RA active
patients.


whether with active or inactive disease, reached asignificantly higher plateau of upregulated CD28 af-ter 48 h than did healthy donors. We found that thelevels of the B7-family CD80 ligand were similar inAPCs from healthy donors and RA patients (data notshown, 33).

Since CD152 (CTLA-4) expression is also promotedafter TCR/CD3 engagement, shares ligands withCD28, and may counteract its function (8), we askedin parallel whether CD152 expression was also dis-regulated in RA (Figs. 3 and 4). In contrast to CD28,CD152 expression was slightly lower in RA patients(Fig. 4); no statistical differences were observed inCD152 levels between RA patients and healthy do-nors ex vivo, at the expression peak after 24 h of invitro CD3 mAb stimulation, or in the expressiondecay that followed thereafter. CD152 expressiondecays more rapidly than does CD28 expression (Fig.4), which remains in the plateau after 10 days (data

FIG. 4. CD41 T cells from RA patients display a sustained over-induction of CD28, but not CTLA-4 (CD152), surface expression afterin vitro T cell activation. Time course analyses of the expression ofCD28 (A) and CTLA-4 (B) molecules were carried out in CD41 T cellsfrom healthy donors (circles) and clinically active (triangles) or inac-tive (squares) RA patients as described in Fig. 3. The y axis repre-sents the antibody binding capacity (ABC) equivalent sites for CD28and CTLA estimated as detailed under Patients and Methods, eitherex vivo or after the periods of culture with anti-CD3 mAb indicatedon the x axis. Values represent means 6 SD of six different deter-minations. **Significant differences were found between RA patientsand healthy donors (P , 0.01). For the sake of clarity, inactive (left)and active (right) RA patients showing no significant differencesbetween them are compared to normal volunteers separately.

not shown).

In RA Patients, the CD28 Expression Levels in CD4 TCells from Inflamed Synovium Are IncrementedWhen Compared to Peripheral Blood AutologousT Cells

CD41 T cells infiltrate the RA synovial membranewhereas they are rarely seen in normal or osteoar-thritic joints. Interestingly, previous phenotypic stud-ies revealed that synovial T cells extracted from in-flamed RA joints are activated and most of themexpress CD45RO, indicating their primed state (39),and cause inflammatory arthritis upon transfer intoSCID mice (40). Because our results showed that theamount of surface CD28 receptor correlated with theactivity of the disease, and peripheral blood CD28high

CD41 T cells from active RA patients were primedCD45RO1IL-2Ra1 cells, we next compared side by sidethe CD28 levels in CD41 T cells extracted from bothperipheral blood and synovial tissue from RA patients.As shown in Fig. 5, the mean ABC CD28 equivalentsites were, in all five RA patients examined, higher inthe synovial membrane than in the peripheral blood.Samples from patients numbered 1 to 4 were obtainedfrom newly diagnosed RA patients. Patient number 5had reached clinical remission after NSAID treatment,except for one joint that remained inflamed at the timeof the study. Although the sample size precludes fur-ther statistical analyses, the mean CD28 sites were10–44% higher in the synovial T cells than in thepaired blood T cells, the maximal difference being ob-served in patients 5 (44%) and 3 (41%).

CD28 Levels in CD4 T Cells from RA PatientsCorrelate with Their T Cell Proliferative Capacity

RA patients have been reported in some studies toshow defective signal transduction via the TCR/CD3complex and to display reduced T cell proliferationupon CD3 stimulation when compared to healthy do-nors (28, 29). In contrast, other studies of the PBL Tcell proliferative response of active, untreated RA pa-tients to TCR/CD3 1 CD28 costimulation show a sim-ilar response in RA patients and healthy controls (41).These previous studies did not, however, allow directcomparison of the proliferative capacity of T cells fromactive and inactive RA patients (28, 29, 41). Cell pro-liferation was therefore induced with CD3 mAb in thepresence or absence of CD28 mAb, and thymidine in-corporation was quantitated in kinetic assays in bothRA patient groups and healthy donors (Fig. 6). Only Tcells from inactive RA patients showed significantlydiminished proliferation under optimal conditions ofCD3 engagement when compared after 96 h in cultureto T cells from both active RA patients and healthy
donors (P , 0.01). This difference did not reflect dis-

ieeRw*a

v

e


tinct kinetic profiles in the three groups studied. More-over, the reduced response of inactive RA patients toCD3 stimulation was almost normalized when CD28was additionally crosslinked using CD28 mAb. Thisindicates that, in our series, the TCR/CD3 signalingdefect observed in RA patients is restricted to inactivedisease and can be overcome by costimulatory signals.It may be argued that CD152 engagement, which canoccur at the initiation of the response despite very lowCD152 levels (8), may inhibit IL-2 secretion (42). Tostudy this possibility, IL-2 expression was quantitatedat the mRNA level by competitive RT-PCR. Titratedamounts of a riboprobe with a 50-bp internal deletionin the IL-2 message were added to aliquots of PBLstimulated with CD3 mAb for 6 h, from which RNAwas purified, and RT-PCR was performed to measureIL-2 gene expression (Fig. 7A). Both active and inactiveRA patients showed an approximately 10-fold increasein IL-2 mRNA expression compared to healthy donors.The increase in RNA accumulation was paralleled by asignificant increase in the amount of IL-2 protein in thesupernatants 24 h after CD3 mAb stimulation, as de-termined using a specific ELISA. Indeed, up to 3.5-foldincreases were found when healthy and RA donorswere compared (180 pg/ml vs 325 pg/ml, P , 0.05 ininactive RA, and 620 pg/ml in active RA patients, P ,0.05). Similar differences were observed following co-stimulation with CD3 plus CD28 mAb (data notshown). We conclude that the defective proliferation isnot due to a limited availability of IL-2. Finally, weexamined Bcl-XL expression as another model to assessthe function of the CD28 signaling pathway in RA T

FIG. 5. Augmented CD28 expression in CD41 T cells from syno-ial tissue from RA patients. CD28 levels were quantitated in CD41

T cells from either peripheral blood or synovial tissue from five RApatients by flow cytometry. Cells were stained with PE-conjugatedanti-CD28 and FITC-conjugated anti-CD4; dead cells were excludedby 7-AAD staining. The linked dots depict the average ABC CD28sites (24) estimated in the CD41 T cells obtained, in parallel, fromach of the two anatomical locations.

cells activated with CD3 mAb alone or in combination a

with CD28 mAb. Bcl-XL has been implicated in theprevention of T cell death in response to TCR/CD3 andFas-L or TNF-R crosslinking (43, 44). We questionedwhether the increased levels displayed by active RApatients were paralleled by enhanced Bcl-XL induction.Active RA patients upregulated Bcl-XL expression tohigher levels than did healthy donors in response toCD3 mAb, as monitored by Western blot analyses (Fig.7B). The increased Bcl-XL expression shown by activeRA patients was sustained over the culture period ex-amined. Unlike active RA patients, the healthy donorswho expressed lower CD28 levels required additionalcrosslinking of CD28 with antibodies to reach maximalBcl-XL induction (data not shown, 33).

DISCUSSION

Our data show four subsets of CD282, CD28low,CD28int, and CD28high CD41 human T cells. Becauseheterogeneous levels of the CD28 costimulatory mole-cule also occur in CD81 T cells (30, 33–35), the multi-modal regulation of CD28 expression is a general fea-ture of human T helper and cytotoxic cells. Ouranalyses show a marked increase in the percentage ofCD28high CD41 T cells in RA patients, which is accom-panied by the acquisition, to different extents in clini-cally active and inactive patients, of an in vivo acti-vated/memory CD45RO1CD5highIL-2Ra1 phenotypenot observed in healthy donors. Notably, the disregu-

FIG. 6. RA defective T cell proliferative response to TCR/CD3engagement occurs selectively in clinically inactive patients and canbe restored by crosslinking of CD28. PBL from healthy donors (A),inactive (B), or active (C) RA patients were stimulated with optimalamounts of anti-CD3 mAb alone (circles) or concurrently with anti-CD28 mAb (squares), and the proliferative response was assessedkinetically by [3H]thymidine incorporation after the indicated timesn culture. Ten donors in each category were analyzed. Results arexpressed as means 6 SD. *Statistically significant differences werevident between the proliferative responses of T cells from inactiveA patients and healthy donors or active RA patients stimulatedith CD3 alone at the peak of proliferation (day 4, P , 0.01).

*Significant differences were also noted between the proliferationttained in the inactive RA samples stimulated for 4 and 6 days with
nti-CD3 alone or in the presence of CD28 (P , 0.05).

experiments.


lation in CD28 expression showed further distinctivefeatures in either clinically active or inactive RA pa-tients. Active patients showed a twofold average in-crease in the CD28 levels expressed by CD41 T cellscompared to healthy donors and inactive RA patients,suggestive of upregulation. In contrast, inactive RApatients showed a slight reduction in the MFI valuesfor each of the three CD281 subsets, consistent withdownmodulation. The differences between healthy do-nors and active or inactive RA patients cannot be at-tributed to the treatment, as markedly increased CD28levels were found in six untreated active RA patients atdiagnosis, and were also found among clinically activeor inactive patients receiving similar therapy. Differ-ences in the CD28 levels between active and inactiveRA patients are not an intrinsic feature of the individ-uals in each group of disease activity, as both studygroups upregulate CD28 expression to similarly highlevels after in vitro activation, markedly exceedingthat of healthy donors studied simultaneously. Themodifications in the distribution of CD28 subpopula-tions cannot be attributed to changes in the absolutenumber of lymphocytes, as we did not observe statisti-cally significant differences in the number of bloodlymphocytes between healthy controls and active orinactive RA patients. CD28 disregulation is specific, asno significant differences were found in the upregula-tion of CD152 (CTLA-4) expression or CD80 levelsamong active and inactive RA patients or healthy do-nors, in vivo or after triggering the TCR/CD3 complexin vitro. This suggests that the CD28 level may be auseful marker for disease activity/remission in RA pa-tients. The mechanisms favoring in vivo up- and down-regulation of CD28 in RA patients are unknown. BothTCR/CD3 complex ligands (21, 30, 45–48) and distinctcostimulatory molecules (35, 36, 47, 49) have been pro-posed to up- and downregulate CD28 expression. Withboth classes of ligands, under conditions that promoteeither CD28 up- or downmodulation, T cells upregulateCD25 (IL-2Ra) levels; cells with downmodulated CD28also become anergic to CD3 mAb engagement, whereastheir defective proliferative responses can be restoredby CD28 mAb crosslinking (35, 36, 46), as shown herefor RA patients. Twofold differences in the CD28 cellsurface level appear sufficient to cause functional dif-ferences. Indeed, CD8128int human T cells showmarked hypoproliferative responses to CD3 engage-ment in vitro when compared to autologous CD8128high

cells that display just twofold higher CD28 levels (35);these data are consistent with the functional resultsshown here. The alteration of the CD28 homeostasis isstrikingly profound in blood T cells. It has been sus-pected that the inflammatory lesions in RA result froman autoimmune response to joint-specific antigens andwould be limited to a minority of T cells; however,

FIG. 7. IL-2 production and Bcl-XL expression do not limitCD3-driven proliferation of RA T cells. Quantitative IL-2 RT-PCR(A). Seven serial 3-fold dilutions of a reference IL-2 riboprobe withan internal deletion were mixed with aliquots of RNA extractedfrom PBL from a healthy donor (left) or an inactive (center) and anactive (right) RA patient PBL that had been stimulated for 6 hwith CD3 mAb, as described under Patients and Methods. RT-PCR was performed and the amplified gene products were visual-ized and quantitated after agarose gel electrophoresis withethidium bromide using a Gel-Doc 1000 documentation system.Arrowheads indicate the riboprobe dilution at which the endoge-nous and deleted bands are in equimolar amounts. Quantitationshowed 10-fold higher amounts in RA patients than in healthydonors in side-by-side experiments with healthy donor and activeor inactive RA patients. The riboprobe standards contained 3-folddilution of defined copy numbers (1.3 3 106, 4 3 105, 1.3 3 105, 4 3104, 1.3 3 104, 4 3 103, 1.3 3 103). Western blot assay of Bcl-XL

protein expression (B). PBL from a healthy donor (left) or anactive RA patient (right) that had been stimulated for 24 or 48 hwith CD3 mAb were lysed, the amount of expressed Bcl-XL proteinwas detected by immunoblotting with a specific antibody, and thereactivity was developed by chemiluminiscence, as described un-der Patients and Methods. The density of the 28-kDa Bcl-XL bandswas quantified using the Gel-Doc 1000 documentation system, andthe results are indicated in the accompanying panels. Results arerepresentative of three similar RT-PCR and three Western blot

recent data indicate that several abnormalities involve

po

gocp

sduT(f

dte

cmTra


a majority of circulating T cells (50). To our knowledge,this is the first report to present results compatiblewith in vivo up/downmodulation of surface CD28 byhuman CD41 T cells from autoimmune patients indifferent disease activity groups (RA active/remission).

It is noteworthy that the CD28 upregulation phe-nomenon is restricted neither in vivo to active RA

atients nor in vitro to antigen-independent activationf CD41 human T cells. Marked CD28 upregulation

also occurs upon in vitro challenge of naive/resting Tcells with APC presenting specific antigen/MHC com-plexes (47) and after reactivation of memory T cells byantigen challenge in vivo, using antigen stimulation inenetically controlled mouse models (48). The findingf marked CD28 upregulation in bona fide memory Tells is relevant when considering the increases in theercentage of CD28highCD45RO1 T cells in the context

human chronic inflammatory responses. It lends sup-port to the notion that CD28 levels can be used at leastas an “activation marker” in both naive and memory Tcell responses, even if it has been reported that reacti-vation of memory cells has less stringent requirementsin some assays for CD28-mediated costimulation (51).

It is possible that CD28 modulation may have sev-eral functional consequences other than the prolifera-tive responses examined here. Indeed, CD28 deliverssignals that have been experimentally dissociated fromproliferation like IL-2 production or regulation of sur-vival via Bcl-XL expression, among others (47, 52). Weshow enhanced Bcl-XL protein levels in preliminarykinetic Western blot analyses when active RA patientswere compared to healthy donors. Although CD28 caninduce Bcl-XL protein, this upregulation is not an ab-olute requisite for modulation of TCR-induced celleath (52, 53). Instead, CD28 engagement preventspregulation of FasL expression, which, together withNF, is instrumental in TCR/CD3-induced cell death

43, 53, 54). In this regard, different T-helper-cell ef-ector responses (i.e., levels of TNFa, INF-g, or IL-2

production) show strikingly distinct in vivo CD28 dose-ependence requirements to achieve the maximal cy-okine production (55). Heterozygous CD281/2 T cellsxpress twofold less CD28 receptors than CD281/1 T

cells, and serum TNFa and INF-g levels in superanti-gen-injected CD281/2 mice were half of those found inCD281/1 mice, whereas IL-2 levels did not differ be-tween CD281/2 and CD281/1 mice (55). Similarly, wefind no differences in IL-2 production levels betweenactive and inactive RA patients, who show twofolddifferences in CD28 levels. Further studies are neededto address the question of whether the production ofTNFa and INF-g, two cytokines relevant to the pathog-eny and therapy of RA (56), are critically influenced bythe modulation of CD28 levels in human T cell subsets.A final CD28 function relevant to our results is that
CD28:CD80 interactions promote T cell adhesion (57).
We show here that T cells in the RA synovial tissueshowed higher CD28 levels than the autologous cells inperipheral blood. In contrast, the expansion of CD282

T cell subsets appears to be ligated to extraarticularsites of inflammation (32) and reflects a systemic anti-gen challenge, for example, persistent infection (58).Our findings parallel reports on CD81 T cells whichshow: (i) that CD81CD28high are enriched among thecells that migrate across endothelial monolayers (35)and (ii) that, in contrast to peripheral blood, RA syno-vial fluid T cells are almost exclusively CD281 (41),which lead us to independently suggest that CD28high Tcells may preferentially migrate to certain sites of in-flammation (33, 34, 41).

In summary, our study shows subsets of CD281 Tells and suggests that CD28 up- and downregulationight occur in either active or inactive phases of RA.his may constitute an overlooked mechanism in theegulation of tolerance to self components in healthnd disease.

ACKNOWLEDGMENTS

We thank Prof. Len Herzenberg for the generous gift of cytokineriboprobes and E. C. Alcocer and C. Mark-Tiemann for editorial help.

REFERENCES

1. Jenkins, M. K., and Schwartz, R. H., Antigen-presentation bychemically modified splenocytes induces antigen-specific T cellunresponsiveness in vitro and in vivo. J. Exp. Med. 165, 302–319, 1987.

2. June, C. H., Bluestone, J. A., Nadler, L. M., and Thompson, C. B.,The B7 and CD28 receptor families. Immunol. Today 15, 321–331, 1994.

3. Linsley, P. S., Brady, W., Grossmaire, L., Arufo, A., Damle, N. K.,and Ledbetter, J. A., Binding of the B cell activation antigen toCD28 costimulates T cell proliferation and interleukin 2 mRNAaccumulation. J. Exp. Med. 173, 721–730, 1991.

4. Lindsten, T., June, C. H., Ledbetter, J. A., Stella, G., and Thomp-son, C. B., Regulation of lymphokine messenger RNA stability bya surface-mediated T cell activation pathway. Science 244, 339–343, 1989.

5. Viola, A., and Lanzavecchia, A., T cell activation determined byT cell receptor number and tunable thresholds. Science 273,104–106, 1996.

6. Hara, T., Fu, S. M., and Hansen, J. A., Human T cell activation.II. A new activation pathway used by a major T cell populationvia a disulfide-bonded dimer of a 44 kilodalton polypeptide (9.3antigen). J. Exp. Med. 161, 1513–1524, 1985.

7. Harper, K., Balzano, C., Rouvier, E., Mattei, M. G., Luciani,M. F., and Goldstein, P., CTLA-4 and CD28 activated lympho-cyte molecules are closely related in both mouse and human as tosequence, message expression, gene structure, and chromosomallocation. J. Immunol. 147, 1037–1044, 1991.

8. Chambers, C. A., Krummel, M. F., Boitel, B., Hurwitz, A., Sul-livan, T. J., Fournier, S., Cassel, D., Brunner, M., and Allison,J. P., The role of CTLA-4 in the regulation and initiation of T-cellresponses. Immunol. Rev. 153, 27–46, 1996.

9. Guerder, S., Picarella, D. E., Linsley, P. S., and Flavell, R. A.,
Costimulator B7-1 confers antigen-presenting-cell function to

1

1

1

1

1

1

1

22

2

2

2

2

3

3

3


parenchymal tissue and in conjunction with tumor necrosis fac-tor alpha leads to autoimmunity in transgenic mice. Proc. Natl.Acad. Sci. USA 91, 5138–5142, 1994.

0. Finck, B. K., Linsley, P. S., and Wofsy, D., Treatment of murinelupus with CTLA4-Ig. Science 265, 1225–1227, 1994.

11. Tada, Y., Nagasawa, K., Ho, A., Morito, F., Ushiyama, O., Su-zuki, N., Ohta, H., and Mak, T., CD28-deficient mice are highlyresistant to collagen-induced arthritis. J. Immunol. 162, 203–208, 1999.

12. Miller, S. D., Vanderlugh, C. L., Lenschow, D. J., Pope, J. G.,Karandikar, N. J., del Canto, M. C., and Bluestone, J. A., Block-ade of CD28/B7-1 interaction prevents epitope spreading andclinical relapses of murine EAE. Immunity 3, 739–745, 1995.

13. Karandikar, N. J., Vanderlugh, C. L., Walunas, T. L., Miller,S. D., and Bluestone, J. A., CTLA-4 a negative regulator ofautoimmune disease. J. Exp. Med. 184, 783–788, 1996.

4. Kroemer, G., and Martınez-A, C., The fail-safe paradigm of im-munological self-tolerance. Lancet 338, 1246–1249, 1991.

5. Arnett, F. C., Edworthy, S. M., Bloch, D. A., McShane, D. J.,Fries J. F., Cooper, N. S., Healey, L. A., Kaplan, S. R., Liang,M. H., and Luthra, H. S., The American Rheumatism Associa-tion 1987 revised criteria for the classification of rheumatoidarthritis. Arthritis Rheum. 31, 1637–1646, 1988.

6. Ranheim, E. A., and Kipps, T. J., Elevated expression ofCD80(B7/BB1) and other accessory molecules on synovial fluidmononuclear cell subsets in rheumatoid arthritis. ArthritisRheum. 37, 1637–1646, 1994.

7. Pinals, R. S., Baum, J., Bland, J., Fosdick, W. M., Kaplan, S. B.,Masi, A. T., Mitchell, D. M., Ropes, M. W., Short, C. L., Sigler,J. W., and Weinberger, H. J., Preliminary criteria for clinicalremission in rheumatoid arthritis. Bull. Rheum. Dis. 32, 7–10,1982.

8. Sanders, M. E., Makgoba, M. W., Sharrow, S. O., Stephany, D.,Springer, T. A., Young, H. A., and Shaw, S., Human memory Tlymphocytes express increased levels of three cell adhesion mol-ecules (LFA-3, CD2, LFA-1) and three other molecules (UCHL1,CDw29, Pgp-1) and have enhanced IFN-gamma production.J. Immunol. 140, 1401–1407, 1988.

9. Springer, T. A., Adhesion receptors of the immune system. Na-ture 346, 425–434, 1990.

0. Sprent, J., T and B memory cells. Cell 76, 315–322, 1994.1. Turka, L. A., Ledbetter, J. A., Lee, K., June, C. H., and Thomp-

son, C. B., CD28 is an inducible T cell surface antigen thattransduces a proliferative signal in CD31 mature thymocytes.J. Immunol. 144, 1646–1653, 1990.

2. Lesslauer, W., Koning, F., Ottenhoff, T., Giphart, M., Goulmy,E., and van Rood, J. J., T90/44 (9.3 antigen): A cell surfacemolecule with a function in T cell activation. Eur. J. Immunol.16, 1289–1296, 1986.

3. Andreu, J. L., Trujillo, A., Alonso, J. M., Mulero, J., and Mar-tınez-A. C., Selective expansion of T cells bearing the gamma/delta receptor and expressing an unusual repertoire in the sy-novial membrane of patients with rheumatoid arthritis.Arthritis Rheum. 34, 808–814, 1991.

24. Roman, I. L., Manzano, L., de la Hera, A., Abreu, L., Rossi, I.,and Alvarez de Mon, M., Expanded CD41CD45RO1 phenotypeand defective proliferative response in T lymphocytes from pa-tients with Chron’s disease. Gastroenterology 110, 1008–1019,1996.

25. Bikoue, A., George, F., Poncelet, P., Mutin, M., Janossy, G., andSampol, J., Quantitative analysis of leukocyte membrane anti-gen expression: Normal adult values. Cytometry 26, 137–147,
1996.
26. Sanz, E., and de la Hera, A., A novel anti-Vpre-B antibodyidentifies immunoglobulin-surrogate receptors on the surface ofhuman pro-B cells. J. Exp. Med. 183, 1693–2698, 1996.

27. Dave, V. P., Cao, Z., Browne, C., Alarcon, B., Fernandez-Miguel,G., Lafaille, J., de la Hera, A., Tonegawa, S., and Kappes, D. J.,CD3d deficiency arrests development of the ab but not the gdTcell lineage. EMBO J. 16, 1360–1370, 1997.

8. Mirza, N. M., Relias, V., Yunis, E. J., Pachas, W. N., and Das-gupta, J. D., Defective signal transduction via T-cell receptorstructure in T cells from rheumatoid arthritis patients. Hum.Immunol. 36, 91–98, 1993.

9. Lotz, M., Tsoukas, C. D., Robinson, C. A., Dinarello, C. A.,Carson, D. A., Vaughan, J. H., Basis for defective responses ofrheumatoid arthritis synovial fluid lymphocytes to anti-CD3 (T3)antibodies. J. Clin. Invest. 78, 713–721, 1986.

0. Azuma, M., Philips, J. H., and Lanier, L. L., CD282 T lympho-cytes. Antigenic and functional properties. J. Immunol. 150,1147–1159, 1993.

1. Goronzy, J. J., Bartz-Bazzanella, P., Hu, W., Jendro, M. C.,Walser-Kuntz, D. R., and Weyand, C. M., Dominant clonotypesin the repertoire of peripheral CD41 T cells in rheumatoid ar-thritis. J. Clin. Invest. 94, 2068–2076, 1994.

32. Namekawa, T., Wagner, U. G., Goronzy, J. J., and Weyand,C. M., Functional subsets of CD4 T cells in rheumatoid arthritis.Arthritis Rheum. 40, 1106–1114, 1997.

33. Salazar-Fontana L., Caracterizacion de los niveles de expresionde la molecula CD28 en linfocitos humanos. Relevancia en lapatologıa autoinmune artritis reumatoide. Ph.D. Thesis, Univer-sidad Autonoma de Madrid, 1998.

34. Berman, J. S., Mahoney, K., Saukkonen, J., and Masuyama, J.,Migration of distinct subsets of CD81 blood cells through endo-thelial cell monolayers in vitro. J Leukocyte Biol. 58, 317–324,1995.

5. Lloyd, T. E., Yang, L., Tang, D. N., Bennett, T., Schober, W., andLewis, D. E., Regulation of CD28 costimulation in human CD81

T cells. J. Immunol. 158, 1551–1558, 1997.36. Linsley, P. S., Bradshaw, J., Urnes, M., Grosmaire, L. S., and

Ledbetter, J. A., CD28 engagement by B7/BB1 induces a tran-sient down-regulation of CD28 synthesis and prolonged unre-sponsiveness to CD28 signaling. J. Immunol. 150, 3161–3169,1993.

37. Arinbjarnarson, S., Jonsson, T., Steinsson, K., Sigfusson, A.,Jonsson, H., Geirsson, A., Thorsteinsson, J., and Valdimarsson,H., IgA rheumatoid factor correlates with changes in B and Tlymphocyte subsets and disease manifestations in rheumatoidarthritis. J. Rheumatol. 24, 269–274, 1997.

38. Osman, N., Lazarovits, A. I., and Crupton, M. J., Physical asso-ciation of CD5 and the T cell receptor/CD3 antigen complex onthe surface of human T lymphocytes. Eur. J. Immunol. 23,1173–1176, 1993.

39. Thomas, R., McIlraith, M., Davis, S. L., and Lipsky, P. E., Rheu-matoid synovium is enriched in CD45Rbdim mature memory Tcells that are potent helpers for B cell differentiation. ArthritisRheum. 35, 1455–1465, 1992.

40. Mima, T., Saeki, Y., Ohshima, S., Nishimoto, N., Matsushita, M.,Shimitzu, M., Kobayashi, Y., Nomura, T., and Kishimoto, T.,Transfer of rheumatoid arthritis into severe combined immuno-deficiency mice. J. Clin. Invest. 96, 1746–1758, 1995.

41. Sfikakis, P. P., Zografou, A., Viglis, V., Iniotaki-Theodoraki, A.,Piskontaki, I., Tsokos, G. C., Sfikakis, P., and Choremi-Papado-poulou, H., CD28 expression on T cell subsets in vivo and CD28-mediated T cell response in vitro in patients with rheumatoid
arthritis. Arthritis Rheum. 5, 649–654, 1995.

4

4

4

4

4

4

5


42. Blair, P. J., Riley, J. L., Levine, B. L., Lee, K. P., Craighead, N.,Francomano, T., Perfetto, S. J., Gray, G. S., Carreno, B., and June,C. H., Cutting edge: CTLA-4 ligation delivers a unique signal toresting human CD4 T cells that inhibits interleukin-2 secretion butallows Bcl-XL induction. J. Immunol. 160, 12–15, 1998.

43. Lin, R. H., Hwang, Y. W., and Lin, C. S., TNF-receptor-2 trig-gered apoptosis is associated with down-regulation of Bcl-XL onactivated T cells and can be prevented by CD28 costimulation.J. Immunol. 158, 598–603, 1997.

4. Boise, L. H., Minn, A. J., Noel, P. J., June, C. H., Accavitti, M. A.,Lindsten, T., and Thompson, C. B., CD28 costimulation canpromote T cell survival by enhancing the expression of Bcl-XL.Immunity 3, 87–98, 1995.

5. O’Hehir, R. E., and Lamb, J. R., Induction of specific clonalanergy in human T lymphocytes by Staphylococcus aureus ente-rotoxins. Proc. Natl. Acad. Sci. USA 87, 8884–8888, 1990.

6. Lake, R. A., O’Hehir, R. E., Verhoef, A., and Lamb, J. R., CD28mRNA rapidly decays when activated T cells are functionallyanergized with specific peptide. Int. Immunol. 5, 461–466, 1993.

7. Hwang, I., Huang, J.-F., Kishimoto, H., Brunmark, A., Peterson,P. A., Jackson, M. R., Surh, C. D., Cai, Z., and Sprent, J., T cellscan use either T cell receptor or CD28 receptors to absorb andinternalize cell surface molecules derived from antigen-present-ing cells. J. Exp. Med. 191, 1137–1148, 2000.

8. Kim, S.-K., Schluns, K. S., and Lefrancois, L., Induction andvisualization of mucosal memory CD8 T cells following systemicvirus infection. J. Immunol. 163, 4125–4132, 1999.

9. Yu, X., Fournier, S., Allison, J. P., Sharpe, A. H., and Hodes, R.,The role of B7 costimulation in CD4/CD8 T cell homeostasis.
J. Immunol. 164, 3543–3553, 2000.
0. Koetz, K., Bryl, E., Spickschen, K., O’Fallon, W. M., Goronzy,J. J., and Weyand, C. M., T cell homeostasis in patients withrheumatoid arthritis. Proc. Natl. Acad. Sci. USA 97, 9203–9208,2000.

51. Van Parijs, L., and Abbas, A. K., Homeostasis and self-tolerancein the immune system: Turning lymphocytes off. Science 280,243–248, 1998.

52. Dahl, A. M., Klein, C., Andres, P. G., London, C. A., Lodge, M. P.,Mulligan, R. C., and Abbas, A. K., Expression of bcl-X(L) restorescell survival, but not proliferation of effector differentiation, inCD28-deficient T lymphocytes. J. Exp. Med. 191, 2031–2038,2000.

53. Collette, Y., Razanajaona, D., Ghiotto, M., and Olive, D., CD28can promote T cell survival through a phosphatidylinositol 3-ki-nase-independent mechanism. Eur. J. Immunol. 27, 3238–3289,1997.

54. Zheng, L., Fisher, G., Miller, R. E., Peschon, J., Lynch, D. H., andLenardo, M. J., Induction of apoptosis in mature T cells by tumornecrosis factor. Nature 377, 348–351, 1995.

55. Saha, B., Harlam, D. M., Lee, K. P., June, C. H., and Abe, R.,Protection against lethal toxic shock by targeted disruption ofthe CD28 gene. J. Exp. Med. 183, 2675–2680, 1996.

56. Feldmann, M., Brennan, F. M., Maini, R. N., Role of cytokines inrheumatoid arthritis. Annu. Rev. Immunol. 14, 397–440, 1996.

57. Turkovski-Corrales, S. M., Fenton, R. G., Peltz, G., and Taub,D. D., CD28:B7 interactions promote T cell adhesion. Eur. J. Im-munol. 25, 3087–3093, 1995.

58. Liuzzo, G., Goronzy, J. J., Yang, H., Kopecky, S. L., Holmes,D. R., Frye, R. L., and Weyand, C. M., Monoclonal T-cell prolif-eration and plaque instability in acute coronary syndromes. Cir-
culation 101, 2883–2888, 2000.
Received August 22, 2000; accepted with revision January 10, 2001

Documents

Cell Surface CD28 Levels Define Four CD4+ T Cell Subsets: Abnormal Expression in Rheumatoid Arthritis