ARTHRITIS & RHEUMATISMVol. 56, No. 8, August 2007, pp 2492–2502DOI 10.1002/art.22748© 2007, American College of Rheumatology

Inflammation and Ectopic Lymphoid Structures inRheumatoid Arthritis Synovial Tissues

Dissected by Genomics Technology

Identification of the Interleukin-7 Signaling Pathway inTissues With Lymphoid Neogenesis

Trieneke C. G. Timmer,1 Belinda Baltus,1 Mark Vondenhoff,1 Tom W. J. Huizinga,2

Paul P. Tak,3 Cornelis L. Verweij,1 Reina E. Mebius,1 andTineke C. T. M. van der Pouw Kraan1

Objective. In �25% of synovial tissues from rheu-matoid arthritis (RA) patients, infiltrates of T cells,B cells, and follicular dendritic cells (FDCs) are spa-tially organized into structures resembling lymphnodes with germinal centers. The remainder of thetissues lack FDCs and show either a diffuse or anaggregated T cell and B cell infiltrate. To gain moreinsight into this specific disease process, we sought toidentify the genes expressed in RA tissues with ectopiclymphoid structures.

Methods. Gene expression profiling of RA syno-vial tissues was determined by complementary DNAmicroarray analysis and quantitative real-time polymer-

ase chain reaction. The presence of lymphoid folliclesand localization of interleukin-7 (IL-7) in synovial tis-sue sections was determined by immunofluorescencestaining using specific antibodies.

Results. Findings of gene expression analysisconfirmed previous reports that tissues with lymphoidstructures showed elevated expression of CXCL13,CCL21, CCR7, and lymphotoxin � and � messengerRNA. In addition, the tissues also showed enhancedexpression of the chemokines CXCL12 and CCL19 andthe associated receptors CXCR4 and CXCR5, which areimportant for the attraction of T cells, B cells, anddendritic cells. Pathway analysis revealed increasedexpression of genes involved in JAK/STAT signaling,T cell– and B cell–specific pathways, Fc� receptor typeI signaling in mast cells, and IL-7 signal transduction inthe tissues with ectopic lymphoid follicles, accompaniedby increased expression of IL-7 receptor � (IL-7R�)/IL-2R� chains and IL-7. Protein expression of IL-7 in RAtissues was localized within fibroblast-like synoviocytes,macrophages, and blood vessels and was colocalized withextracellular matrix structures around the B cell follicles.

Conclusion. Activation of the IL-7 pathway mayplay an important role in lymphoid neogenesis, analo-gous to its role in the development of normal lymphoidtissue.

Rheumatoid arthritis (RA) is a chronic systemicautoimmune disease of unknown etiology. Patient vari-ability in clinical presentation and differential responseto treatment suggest that RA is a heterogeneous disease.The heterogeneous character of RA is also evident from

The views expressed herein are those of the authors.Supported in part by the Marie Curie Training Network

(EURO-RA), the European Community Sixth Framework Pro-gramme (project AutoCure), the Innovation Oriented Research Pro-gram on Genomics, The Netherlands Organization for ScientificResearch Genomics Program (grant 050-10-120), and the Centre forMedical Systems Biology (a center of excellence approved by TheNetherlands Genomics Initiative/Netherlands Organization for Scien-tific Research).

1Trieneke C. G. Timmer, MSc, Belinda Baltus, PhD, MarkVondenhoff, MSc, Cornelis L. Verweij, PhD, Reina E. Mebius, PhD,Tineke C. T. M. van der Pouw Kraan, PhD: VU University MedicalCenter, Amsterdam, The Netherlands; 2Tom W. J. Huizinga, MD,PhD: Leiden University Medical Center, Leiden, The Netherlands;3Paul P. Tak, MD, PhD: Academic Medical Center, University ofAmsterdam, Amsterdam, The Netherlands.

Ms. Timmer and Dr. Baltus contributed equally to this work.Address correspondence and reprint requests to Tineke van

der Pouw Kraan, PhD, VU University Medical Center, Departmentof Molecular Cell Biology and Immunology, C264, PO Box 7057, 1007MB Amsterdam, The Netherlands. E-mail: t.vanderpouwkraan@vumc.nl.

Submitted for publication December 15, 2006; accepted inrevised form April 17, 2007.

2492

the type of cellular infiltrates in the synovial tissues. In�25% of patients, cellular infiltrates show a high degreeof cellular organization, resembling structures normallyobserved in lymph nodes and comprising distinct T celland B cell areas and a network of follicular dendriticcells (FDCs) within the B cell area and are thereforereferred to as ectopic or tertiary lymphoid structures. Inthe remainder of the patients, the tissues do not containFDCs and show either a diffuse lymphocytic infiltrate oran aggregated T cell and B cell infiltrate (1,2). Lymphoidstructures in the synovium are likely to contribute to thepathogenesis of RA by local production of (auto)anti-bodies. However, the factors that contribute to theformation of lymphoid follicles in RA are not fullyunderstood.

Ectopic lymphoid structures are not specific forRA, but are commonly associated with chronic inflam-mation, since they have also been identified in otherchronic inflammatory diseases, such as Sjogren’s syn-drome (3), Hashimoto thyroiditis (4,5), Helicobacterpylori–induced gastritis (6), chronic hepatitis C (7),multiple sclerosis (8,9), and Crohn’s disease (10) (forreview, see ref. 11).

The existence of germinal center–like structuresis indicative of a high level of organization within theseectopic lymphoid structures, allowing B cells that are inclose contact with FDCs to become activated andproduce (auto)antibodies. Most of our knowledge oflymphoid tissue development is derived from murinestudies. During embryonic development, a stringentlyregulated developmental program is initiated that even-tually results in the formation of highly organized lym-phoid tissue (for review, see ref. 12). It has become clearthat one of the first events is the initial clustering ofinducer cells (IL-7R�,CD3–,CD4�,CD45�) and orga-nizer cells (VCAM-1�,ICAM-1� stromal cells).Interleukin-7 receptor (IL-7R) signaling is essential forinduction of the initial expression of lymphotoxin (LT)�1�2 on inducer cells, which allows triggering of theLT� receptor (LT�R) expressed on organizer cells.LT�R signaling in concert with interactions between�4�1 and vascular cell adhesion molecule 1 (VCAM-1)leads to the production of chemokines, such as CXCL13,CCL19, CCL21, and CXCL12, by stromal organizercells, which mediate the cellular influx, leading to theformation of cellular aggregates. For the subsequentformation of distinct areas into which B and T lympho-cytes can lodge, an interaction between inducer cells orB cells and stromal cells has been implicated (13,14). Itis likely that similar developmental programs are alsooperative during the formation of tertiary lymphoid

structures at sites of inflammation, although the cellularsubsets that initiate this structure formation may differ.

There is a close relationship between inflamma-tion and lymphoid organogenesis, since chemokines andadhesion molecules are crucially involved in both pro-cesses. Indeed, a higher expression of the cytokines LT�and LT� and the chemokines CXCL13 and CCL21 wasobserved in RA tissues with germinal centers (2,15–17).

In the present study, we examined gene expres-sion profiles in RA synovium with different types oflymphoid infiltrates. We used microarray analysis toexamine which molecular pathways are used and whichgenes determine the different types of lymphoid organi-zation.

MATERIALS AND METHODS

Synovial tissue samples. Twelve patients with RA whowere undergoing total joint replacement surgery were includedin the gene expression profiling study, as described previously(18,19). Patients met the American College of Rheumatology(ACR; formerly, the American Rheumatism Association) 1987criteria for RA (20,21). The characteristics of the patients(duration of disease, erythrocyte sedimentation rate, and whiteblood cell count) were not different between the differenttissue groups (see description below). Tissue samples weresnap-frozen in liquid nitrogen, and the frozen blocks werestored in liquid nitrogen until they were sectioned for analysisby immunofluorescence. All patients gave their informedconsent, and the Medical Ethics Committee approved thestudy protocol.

Gene expression profiling of RA synovial tissue sam-ples. The details of isolating messenger RNA (mRNA), label-ing, and hybridizing have been described elsewhere (18,19).Briefly, synovial tissues (�1 gm) were dissected and quicklyfrozen in liquid nitrogen and stored at –80°C. Total cellularRNA and mRNA were isolated from the tissues with TRIzolreagent (Gibco BRL, Carlsbad, CA) and with the FastTrack2.0 kit (Invitrogen, Carlsbad, CA), respectively, according tothe manufacturers’ instructions. For gene expression profilingby DNA microarray analysis, fluorescent complementary DNA(cDNA) probes were prepared from 1 �g of experimentalmRNA sample by oligo(dT)-primed polymerization usingSuperscript II reverse transcriptase in the presence of Cy5-labeled dCTP (as described at http://cmgm.stanford.edu/pbrown/protocols.html). A common reference mRNA sampleconsisting of a mixture of mRNA isolated from 11 different celllines, synovial tissue, fibroblasts, and activated peripheralblood mononuclear cells was labeled with Cy3.

Microarray procedures, data analysis, and statisticalanalysis. We combined the data from 2 previous tissue profil-ing studies (18,19). All data are publicly available at theStanford Microarray Database (22) (available at http://genome-www.stanford.edu/microarray). Genes with at least80% of the data present were included, and only genes with anabsolute value that was at least 2 times higher than the medianexpression of that gene in at least 1 array were selected. Allgenes were expressed relative to their median expression level

LYMPHOID NEOGENESIS AND IL-7 SIGNALING IN RA SYNOVIAL TISSUES 2493

across the arrays of the same microarray platform, allowing usto combine the data from the 2 platforms.

Statistical analysis of microarray data was performedusing Significance Analysis of Microarrays (SAM) (available athttp://www-stat.stanford.edu/�tibs/SAM) (23). For visualiza-tion of the results, we used Cluster and TreeView software (24)(available at http://rana.lbl.gov/EisenSoftware.htm) to performhierarchical cluster analysis. Pathway level analysis of geneexpression data was performed by gene set enrichment analysis(25), using pathways from Kegg and Biocarta databases.

Quantitative real-time polymerase chain reaction(PCR). For these studies, 20 ng of mRNA was transcribed intocDNA using a RevertAid First-Strand cDNA synthesis kit(MBI Fermentas, St. Leon-Rot, Germany). Briefly, mRNAwas heated at 70°C for 5 minutes with 1 �l of oligo(dT) primers(0.5 �g/�l) and 1 �l of random hexamers (0.2 �g/�l) in a finalvolume of 12 �l and chilled at 4°C for 1 minute. Next, reversetranscription of mRNA was performed in a total volume of20 �l containing first-strand buffer, 2 �l of dNTP (10 mM), 20units of RNase inhibitor, and 200 units of H Minus Moloneymurine leukemia virus reverse transcriptase. The samples wereincubated at 42°C for 60 minutes and then for 10 minutes at70°C.

Quantitative real-time PCR was performed with anABI Prism 7900 Sequence Detection System (PE AppliedBiosystems, Foster City, CA). The reaction mixture was com-posed of SYBR Green Mastermix, 250 or 500 nM of eachprimer, and cDNA in a total volume of 20 �l, according to themanufacturer’s instructions. For each gene, a standard curvewas used to calculate gene expression levels, thus correctingfor different primer efficiencies. Gene expression levels wereexpressed relative to GAPDH. The primer pairs shown inTable 1 were synthesized by Invitrogen (Breda, The Nether-lands). When real-time PCR data were combined with mi-croarray data, real-time PCR data were expressed as log2expression values, analogous to the microarray data.

Immunofluorescence staining of synovial tissues. Cry-ostat sections (5 �m) were mounted on gelatin-coated glassslides. The sections were dried, fixed in dehydrated acetone for10 minutes, and air dried for an additional 10 minutes. Toprevent nonspecific binding of antibody, sections were blockedwith 10% normal goat serum (Invitrogen) for 20 minutes.Sections were then incubated with primary antibody for 1 hourat room temperature, followed by an additional 1 hour ofincubation with fluorescence-labeled conjugates. Sections wereembedded in Vectashield (Vector, Burlingame, CA) and ex-amined with a Nikon Eclipse E800 microscope (Nikon Eu-rope).

The following primary antibodies were used: anti-CD21L (1:20 dilution; kindly provided by Dr. Y.-J. Liu, Centerfor Cancer Immunology Research, University of Texas, M. D.Anderson Cancer Center, Houston, TX), anti-CD3 (1:10 dilu-tion; BD PharMingen, San Diego, CA), biotinylated anti-CD19 (1:25 dilution; Diaclone, Besancon, France), biotinyl-ated anti-CD8 (1:25 dilution; BD PharMingen), anti–type IVcollagen (1:100 dilution; DakoCytomation, Glostrup, Den-mark), and anti–IL-7 (1:50 dilution; Affinity BioReagents,Golden, CO). For detection of the primary antibodies, we usedAlexa 488–labeled goat anti-mouse Ig (1:400 dilution; Molec-ular Probes, Eugene, OR), Alexa 594–labeled goat anti-rat Ig(1:100 dilution; Molecular Probes), and fluoresceinisothiocyanate–labeled streptavidin (1:100 dilution; Vector).Control staining for each antibody was performed by omittingthe primary antibody. The specificity of the IL-7 antibody wassecured, since preincubation of the antibody with an excessamount of recombinant human IL-7 (Strathmann Biotec,Hamburg, Germany) prevented tissue staining.

RESULTS

Characterization of synovial tissues by CD21LmRNA measurements and immunofluorescence. To de-termine the presence of organized lymphoid follicles in12 synovial RA tissues for which microarray data wereavailable (18,19), we performed quantitative real-timePCR for the long CD21 isoform (CD21L). This isoformof CD21 contains an additional exon and represents thefirst characterized human FDC-specific molecule, whichmay confer unique functions of FDCs in germinal centerdevelopment (26). Takemura et al (2) showed thatCD21L mRNA is selectively expressed in germinalcenter–containing synovial tissues from RA patients andis absent in other types of infiltrates. We detectedsignificant levels of CD21L mRNA in 4 tissues, whereasthe other 8 tissues were negative (Figure 1).

To confirm the presence of lymphoid follicles andto establish the type of infiltrate in the CD21L-negativetissues, we performed immunofluorescence staining us-ing antibodies specific for T cells (CD3), B cells (CD19),and the FDC marker CD21L (Figure 1). Tissues withmarked expression of CD21L mRNA indeed showed the

Table 1. Primers used in the present study

Gene* Size, bp Forward primer Reverse primer

CXCR5 81 ACAGCCATGAACTACCCGCTAA GTCCAATCTGTCCAGTTCCCAGCXCL13 135 CTCTGCTTCTCATGCTGCTGG TGAATTCGATCAATGAAGCGTCTCXCR4 115 AGGGCCTGAGTGCTCCAGTAG TCATAGTCCCCTGAGCCCATTCXCL12 110 GATGCCCATGCCGATTCTT GTTCTTCAGCCGGGCTACAATCCR7 85 ATGAAAAGCGTGCTGGTGGT TCTCCGATGTAATCGTCCGTGCCL19 95 TGTCTGTGACCCAGAAACCCA TGAACACTACAGCAGGCACCCCD21L 114 TCCTGAAATACCAGTTTGTGAAAAAG GTGTAGTCTACAGTCATCCCAGAGACAIL-7 133 TTCCTCCCCTGATCCTTGTTCT CCAATTTCTTTCATGCTGTCCAAIL-7R� 102 TCGATCCATCCCTGATCACTATTT TCCATCTCCCCTGAGCTATTATTG

* IL-7 � interleukin-7; IL-7R� � interleukin-7 receptor �.

2494 TIMMER ET AL

typical organization of cellular infiltrates of T cellssurrounding a B cell area containing FDCs, thus resem-bling structures normally observed in lymph nodes.

In the remaining 8 tissues that were CD21L-negative by real-time PCR, no CD21L-expressing FDCscould be detected by immunofluorescence, which sup-ports the lack of CD21L expression as determined byreal-time PCR. These tissues could be further subdi-vided into 4 tissues that contained T cell and B cellaggregates and 4 tissues that showed a diffuse type ofinfiltrate. The typical localization of CD8 cells outsidethe B cell area of the lymphoid follicles has beenreported previously in RA (27).

Distinctive cytokine and chemokine (receptor)expression in tissues with ectopic lymphoid structures.To further draw parallels between developmental pro-cesses and inflammatory reactions, we performed real-time PCR experiments to evaluate mRNA expressionlevels of molecules that have been implicated in thedevelopment of lymph nodes (Figure 2). We confirmedthe previously described increased expression of LT�,LT�, CXCL13, CCL21, and CCR7 in RA tissues con-taining ectopic lymphoid structures, whereas mRNAencoding the LT� receptor was equally expressed in allsynovial tissues (results not shown) (2,15–17).

These cytokines and chemokines are essentialduring normal lymph node development. We had nowestablished that CXCR5, the receptor for CXCL13, isalso increased in lymphoid follicle–positive tissues. An-other key role in lymphocyte trafficking to secondarylymphoid organs is played by CCR7 and its ligandsCCL19 and CCL21, which exhibit chemotactic activityfor T cells, B cells, and mature dendritic cells. Weconfirmed the increased expression of CCR7, and wealso showed that the other CCR7 ligand CCL19 isincreased in these tissues as well.

These results confirm the previously describedchemokine expression levels in germinal center–containing RA synovial tissues. In addition, this is thefirst demonstration of the elevated expression of chemo-kines CXCL12 and CCL19 and of chemokine receptorsCXCR4 (receptor for CXCL12) and CXCR5 (receptorfor CXCL13) in these tissues.

Differential activity of multiple processes as re-vealed by large-scale gene expression profiling. Afterhaving established that despite the small sample size, theexpression levels of genes of importance for lymphoidtissue development were consistent with published data,we next analyzed large-scale gene expression data bymicroarrays. SAM analysis was used to identify genes

Figure 1. Characterization of synovial tissue types by immunohistochemical staining and levels of mRNA for CD21L in synovial tissues frompatients with rheumatoid arthritis. A, D, and E, Staining of T cells (CD3; red) and B cells (CD19; green). B, Staining of B cells (CD19; green) andfollicular dendritic cells (FDCs) (CD21L; red). C, Staining of FDCs (CD21L; red) and T cells (CD8; green). (Original magnification � 200.) F,Expression of mRNA for CD21L in each of the 12 RA synovial tissues, as determined by quantitative real-time polymerase chain reaction. Tissuesources were Leydenburg Hospital (LB) and Leiden University Medical Center (LU).

LYMPHOID NEOGENESIS AND IL-7 SIGNALING IN RA SYNOVIAL TISSUES 2495

that are expressed at significantly different levels be-tween the 3 types of tissues (multiclass analysis). In thisanalysis, we included the real-time PCR data. We de-tected 734 cDNA that were expressed at significantlydifferent levels between the 3 tissue types (at a falsediscovery rate of �5%). To visualize the results, weperformed hierarchical clustering of these genes, posi-tioning genes with a correlated expression profile acrossthe arrays in adjacent rows (Figure 3A).

This analysis revealed differential expression ofgenes involved in several processes. Cluster A representsgenes that are expressed at higher levels in tissues withlymphoid follicles, including genes involved in the adap-tive immune response, such as those involved in antigenpresentation (HLA class I and II molecules), and immu-noglobulins. Chemotactic activity is indicated by theelevated expression of CCL4L/macrophage inflamma-tory protein 1� (MIP-1�), CXCL5, CXCL13, IL-16,CXCL12, CXCR4, and CXCR5. The JAK/STAT path-way was also overrepresented in the tissues with lym-phoid follicles, as indicated by the enhanced expressionof STAT-1, STAT-3, STAT-6, and protein tyrosinephosphatase N6/SH2 domain–containing phosphatase 1.Interestingly, a cytokine receptor that is capable of

activating the JAK/STAT pathway and is of crucialimportance in lymph node development, the IL-7 recep-tor, as well as its coreceptor IL-2R�, were also expressedat increased levels in the follicular tissues.

We identified proangiogenic and antiangiogenicfactors that support the differential formation of newblood vessels in the tissue types. The proangiogenicfactor TEK tyrosine kinase was preferentially expressedin tissues with lymphoid follicles. In contrast, diffuselyinfiltrated tissues showed enhanced expression of theantiangiogenic factors thrombospondin 1 and throm-bospondin 2, along with thrombospondin 3, indicatingrepression of angiogenesis (28).

Diffusely infiltrated tissues did not express genesinvolved in inflammation; instead, these tissues showedsigns of extracellular matrix remodeling, as indicated bythe expression of matrix metalloproteinase 9, differenttypes of collagen (types I, IV, V, VI, XI, and XII),secreted protein, acidic and rich in cysteine/osteonectin,and laminin �2 (Figure 3A, cluster B). The tissues withaggregated types of infiltrates shared some gene expres-sion levels with germinal center tissues and other geneswith diffusely infiltrated tissues. However, they showed aselectively up-regulated expression of genes involved incomplement activation, such as clusterin and C1q �, �,and � polypeptides (Figure 3A, cluster C).

Selective expression of genes in tissues withlymphoid follicles. When gene expression levels andcellular composition were compared, tissues with lym-phoid follicles showed greater similarity to aggregatedtypes of infiltrates than to diffuse infiltrates. We there-fore analyzed by SAM the differences between theaggregated and lymphoid follicle–positive tissues (Fig-ure 3B). It turned out that 74 genes were expressed atsignificantly different levels, most of them showing en-hanced expression in tissues with lymphoid follicles,while only 5 genes were expressed at lower levels (matrix�-carboxyglutamic acid protein and 4 genes of unknownfunction). Increased expression was identified for sev-eral chemokines: CXCL13, CCL19, CCL21, and mono-cyte chemoattractant protein 4 (MCP-4)/CCL13, the�-chemokines MIP-1� and RANTES, the chemokinereceptors CXCR4, CCR7, and CXCR5, and a moleculeimportant for signaling of several chemokine receptors,regulator of G protein signaling 1. B cell activation wasindicated by the increased expression of several immu-noglobulin heavy and light chains. The expression ofhemogen may suggest the presence of precursor stemcells (29). Among the cytokines, LT� was listed, whilethe only cytokine receptor pair that came out as signif-icant was the IL-7R/IL-2R� with IL-7, indicating itsimportance in follicle structures in RA.

Figure 2. Distinct chemokine and chemokine receptor expression inthe 3 rheumatoid arthritis tissue types (those with follicular infiltrates[Foll; n � 4], those with aggregated infiltrates [Aggr; n � 4], and thosewith diffuse infiltrates [Diff; n � 4]). Real-time polymerase chainreaction data are expressed relative to GAPDH. Values are the meanand SEM of 4 samples per group. � � P � 0.05; �� � P � 0.01; ��� �P � 0.001 versus tissues containing follicular infiltrates.

2496 TIMMER ET AL

Findings of pathway analysis. To gain moreinsight into the biologic role of the gene expressionsignatures, we performed pathway level analysis, using

gene set enrichment analysis (25). For this analysis,real-time PCR data for cytokines and chemokines wereagain combined with the microarray data, including

Figure 3. Statistical analysis of large-scale gene expression profiles in patients with rheumatoid arthritis (RA).Significant genes were hierarchically clustered (using Cluster and TreeView software) to show the correlatedexpression profiles. A, Graphic overview of the genes that showed significantly different expression in the 3tissue types (those with follicular infiltrates [Foll; n � 4], those with aggregated infiltrates [Aggr; n � 4], andthose with diffuse infiltrates [Diff; n � 4]), as analyzed by multiclass Significance Analysis of Microarrays(SAM). Cluster A is the immune response genes, cluster B the extracellular matrix (ECM) genes, and clusterC the complement/growth factor genes. B, Graphic overview of the genes that showed significantly differentexpression in 2 of the tissue types (those with follicular infiltrates [n � 4] and those with aggregated infiltrates[n � 4]), as analyzed by 2-class SAM. Red indicates relatively high expression, green indicates low expression,black indicates intermediate expression, and gray indicates missing data. Tissue sources were LeydenburgHospital (LB) and Leiden University Medical Center (LU).

LYMPHOID NEOGENESIS AND IL-7 SIGNALING IN RA SYNOVIAL TISSUES 2497

IL-7R, IL-7, and the IL-2R� chain. We observed anumber of significant pathways, represented by theexpression profiles of tissues containing lymphoid folli-cles, as compared with tissues lacking these structures(Table 2). The JAK/STAT signaling pathway was themost significant pathway, followed by the T cell receptorsignaling pathway, the Ras-independent pathway in nat-

ural killer cell–mediated cytotoxicity, Fc� receptor type Isignaling in mast cells, IL-7 signal transduction, theB cell receptor signaling pathway, the IL-2R� chain inT cell activation, and the costimulatory signal duringT cell activation. Figures 4A and B present a graphicoverview of the JAK/STAT and IL-7 signal transductionpathways.

Figure 4. Results of pathway analysis in rheumatoid arthritis synovial tissues, as determined by gene setenrichment analysis. Gene set enrichment analysis was performed on all genes expressed by tissues withfollicular infiltrates (Foll) as compared with tissues lacking these structures, but containing aggregated (Aggr)or diffuse (Diff) infiltrates. See Table 2 for the significant pathways in tissues containing lymphoid follicles. Theinterleukin-7 (IL-7) signaling pathway was found to be the most important cytokine-driven process in folliculartissues. Shown is a detailed graphic overview of A, the IL-7 signal transduction pathway and B, the JAK/STATsignaling pathway. Red indicates high expression; blue indicates low expression.

Table 2. Significant pathways in rheumatoid arthritis synovial tissues containing lymphoid follicles*

Size, bp P FDR q value

JAK/STAT signaling pathway 31 0.014 0.07T cell receptor signaling pathway 14 0.028 0.10Ras-independent pathway in NK cell–mediated

cytotoxicity8 0.000 0.10

Fc�RI signaling in mast cells 9 0.000 0.12IL-7 signal transduction 8 0.025 0.17BCR signaling pathway 8 0.035 0.18IL-2R� chain in T cell activation 11 0.000 0.19Costimulatory signal during T cell activation 12 0.048 0.21

* Gene set enrichment analysis was performed on all genes expressed by tissues with follicular infiltratesas compared with tissues lacking these structures, in order to determine significant pathways. FDR � falsediscovery rate; NK � natural killer; Fc�RI � Fc� receptor type I; IL-7 � interleukin-7; BCR � B cellreceptor; IL-2R� � interleukin-2 receptor �.

2498 TIMMER ET AL

Synovial localization of IL-7. Because of thecrucial role of IL-7 in normal lymphoid organ develop-ment, we determined the localization of IL-7 in diffuseand aggregated infiltrates and lymphoid follicle–containing tissues. We confirmed the presence of IL-7 insome of the fibroblast-like synoviocytes and macro-phages and in blood vessel endothelial cells (results notshown) (30). The diffusely infiltrated synovial tissuesshowed a patchy staining pattern, whereas theaggregate-containing and lymphoid follicle–containingtissues showed a more consistent IL-7 staining through-out the whole tissue. In contrast to the strictly cellularstaining of IL-7 in tissues with diffuse and aggregatedinfiltrates, a large proportion of IL-7 in lymphoid

follicle–containing tissues was extracellular, with a retic-ular pattern. Therefore, we determined the extracellularmatrix component type IV collagen to identify thelocation of IL-7. IL-7 was localized in an extracellularmatrix formed by a double-layered circular structurearound the B cell follicle. Similar structures were iden-tified in tonsils (Figures 5C and D). Such stainings werenot found in tissues with diffuse and aggregated infil-trates (Figures 5A and B). Double staining for IL-7 andlaminin showed the same pattern (results not shown).

DISCUSSION

We previously reported that expression profilesof synovial tissues from RA patients show a high degree

Figure 5. Immunofluorescence staining of rheumatoid arthritis synovial tissues and tonsil for interleukin-7 (IL-7) and type IV collagen. Staining forIL-7 (red) and type IV collagen (green) in synovial tissues containing A, diffuse infiltrates, B, aggregated infiltrates, and C, lymphoid follicles is shownin comparison with staining in D, tonsil. The B cell region (B) (enclosed by the broken line) and the T cell region (T) were localized by B cell/T cellstaining of consecutive slides. (Original magnification � 20 in A, B, and C; � 10 in D.)

LYMPHOID NEOGENESIS AND IL-7 SIGNALING IN RA SYNOVIAL TISSUES 2499

of heterogeneity (18,19). In this study, we analyzed theexpression profiles related to the type of tissue infil-trates in synovial tissues and revealed a correlation ofseveral cytokines and chemokines with the existence oflymphoid follicular structures in the synovium of RApatients.

Recently, it was demonstrated that there is in-creased angiogenesis in aggregated and follicular infil-trates within the synovium, compared with diffuselyinfiltrated synovium, which is related to a reducedexpression of the angiogenic inhibitor thrombospondin 2(28). We confirmed and extended these observationsby showing increased expression of the proangiogenicfactor TIE/TEK, the receptor for angiopoietins, in fol-licular synovium and decreased expression of the anti-angiogenic factors thrombospondins 1 and 2, accompa-nied by reduced expression of thrombospondin 3, intissues with lymphoid tissues and aggregated types ofinfiltrates.

The present study confirms the increased expres-sion of CXCL13, CCL21, CCR7, LT�, and LT� in RAtissues that contain lymphoid follicles (2,15–17). Inaddition, we observed increased expression of the che-mokines CXCL12, CCL19, MCP-4/CCL13, the CCR5ligands MIP-1� and RANTES, and the chemokinereceptors CXCR4 (ligand for CXCL12) and CXCR5(the receptor for CXCL13) in these tissues. Theseresults are consistent with the current model of nor-mal lymph node development, in which B cell and Tcell homing to secondary lymphoid tissues is dependenton constitutively expressed CCL19 and CCL21 in theT cell zone, attracting CCR7� cells, and CXCL13 inthe follicular compartment, attracting CXCR5� cells(13,31,32). Thus, the enhanced expression of CXCL13 intissues with ectopic lymphoid follicles may explain theelevated levels of CXCR5 mRNA in these tissues, dueto the influx of CXCR5� B cells and activated T cells.These findings are also consistent with the observationthat both CXCR5 and CXCL13 are necessary for theformation of B cell follicles in mice (33).

LT�1�2 causes growth and maturation of thefollicular stromal cells and up-regulation of FDC mark-ers (34). It has been postulated that prolonged interac-tion of LT�1�2 with the LT�R on stromal cells will, overtime, lead to a gradual shift from most cells using theclassic NF-�B signaling pathway at early time points toan increasing number of cells initiating an alternativeNF-�B pathway downstream of the LT�R (35). As aconsequence, sustained levels of LT�1�2 may tip thebalance from inflammatory to homeostatic chemokines,resulting in the predominant production of chemokinesthat are implicated in lymphoid organogenesis and ho-

meostasis, such as CXCL13, CCL21, and CCL19 (36).Thus, a local microenvironment that permits lymphoidorganization becomes established.

If prolonged ligation of LT�R is the criticalfactor in the development of both lymphoid organs andgerminal center structures during chronic inflammation,then the question arises how the sustained expression ofLT�1�2 is realized in the follicular infiltrates. Block-ade of IL-7R early during development totally blocksthe induction of LT�/� on lymphoid tissue inducercells, and thus, all subsequent events that depend onLT�/� and are necessary for the further formation oflymphoid organs (37). Furthermore, ectopic expressionof IL-7 in mice induces the development of lymphoidinfiltrates at the site of expression, while in the absenceof IL-7, development of lymphoid lineages and lymphnodes is severely affected (38–40). The enhancedtranscript levels of IL-7 and the IL-7R/IL-2R� chain inthe lymphoid follicle–containing synovial tissues aretherefore likely to contribute to the increased LT�1�2expression, thus creating an environment that is benefi-cial for the attraction and organization of leukocyteswithin the follicular structure. In support of this, wefound a strong correlation between IL-7 mRNA expres-sion and LT� expression (r � 0.90, P � 0.0001) (data notshown).

Consistent with our findings, enhanced levels ofIL-7 have been detected in synovial fluid from a sub-group of RA patients (approximately one-third of theRA patients) as compared with synovial fluid fromosteoarthritis (OA) patients (30), while fibroblast-likesynoviocytes from a subpopulation of RA patients pro-duce increased amounts of IL-7 (41) as compared withfibroblast-like synoviocytes from OA patients. In thisstudy, we show that increased levels of IL-7 and theIL-7R are associated with tissues that contain lymphoidstructures.

IL-7 may also contribute to the inflammatoryprocess in RA. It has been demonstrated thatinterferon-� and tumor necrosis factor � are produced insynovial fluid mononuclear cells upon treatment withIL-7 (42).

Binding of IL-7 to the IL-7R�/IL-2R� chainresults in activation of the JAK/STAT pathway (43). Weidentified an up-regulation of this pathway in tissueswith lymphoid follicles. In fact, within our list of genes ofthe JAK/STAT pathway, IL-7R was the most signifi-cantly up-regulated cytokine receptor gene, followed byits coreceptor, IL-2R� chain, suggesting that the inter-action between IL-7 and IL-7R contributed the most tothis pathway and, thus, determines the cytokine specific-ity of this pathway. There was no differential mRNA

2500 TIMMER ET AL

expression for the alternative ligand of the IL-7R,thymic stromal lymphopoietin protein (data not shown),suggesting that IL-7 is the operating ligand for theIL-7R� chain within these infiltrates.

The present study confirms that both fibroblast-like synoviocytes and macrophages contribute to theproduction of IL-7 (30). There was also marked IL-7staining of blood vessels, which may relate to the capac-ity of IL-7 to bind to extracellular matrix molecules, suchas heparan sulfate proteoglycans (44). In tissues withlymphoid follicles, we showed for the first time that IL-7is colocalized with the extracellular matrix componentstype IV collagen and laminin in a circular structuresurrounding the B cell follicle, a phenomenon we alsoobserved in lymphoid organs, such as the tonsil. Thisstructure is reminiscent of the cortical ridge identified atthe T cell–B cell interface in murine lymph nodes (45),providing a reticular network of extracellular matrixfibers that support the cellular architecture. The corticalridge is enriched for dendritic cells that present antigento T cells. Since antigen-specific B cell and T cellinteractions occur at this B cell–T cell interface, IL-7may contribute to the survival of T cells (46), thussustaining B cell activation. In addition, interactionbetween extracellular matrix–bound IL-7 and T cellsmay induce the adhesive properties of these cells (47),which suggests the possibility of involvement of IL-7 inthe arrangement of T cells around the B cell follicle.

The notion that IL-7 plays a crucial role in thepathogenesis of arthritis is supported by observations ina spontaneous arthritis model. In gp130-mutated mice,increased gp130/STAT-3 signaling in nonhematopoieticcells is responsible for the production of IL-7, which, inturn, is crucial for development of the disease (48). Wesuggest that the observed expression of IL-7 infibroblast-like synoviocytes could be the driving force forgenerating an environment in which B cells and T cellscan survive, interact, and organize into tertiary lymphoidstructures.

From the list of genes and pathways distinguish-ing tissues with lymphoid follicles from tissues lackingthese structures, it has become clear that several pro-cesses take place that act in concert to maintain thehighly organized structures. Overall, the results point toa crucial role of the interaction between LT�/� andLT�R, the interaction between the chemokinesCXCL13, CXCL12, CCL19, and CCL21 and their re-spective receptors CXCR5, CXCR4, and CCR7, and inparticular, the IL-7 signaling pathway in organizing andmaintaining lymphoid follicles in the RA synovium.

AUTHOR CONTRIBUTIONS

Dr. van der Pouw Kraan had full access to all of the data inthe study and takes responsibility for the integrity of the data and theaccuracy of the data analysis.Study design. Mebius, van der Pouw Kraan.Acquisition of data. Timmer, Baltus, Huizinga, Verweij, van der PouwKraan.Analysis and interpretation of data. Timmer, Baltus, Vondenhoff,Huizinga, Tak, Mebius, van der Pouw Kraan.Manuscript preparation. Timmer, Baltus, Huizinga, Tak, Verweij,Mebius, van der Pouw Kraan.Statistical analysis. Van der Pouw Kraan.

REFERENCES

1. Young CL, Adamson TC III, Vaughan JH, Fox RI. Immunohis-tologic characterization of synovial membrane lymphocytes inrheumatoid arthritis. Arthritis Rheum 1984;27:32–9.

2. Takemura S, Braun A, Crowson C, Kurtin PJ, Cofield RH,O’Fallon WM, et al. Lymphoid neogenesis in rheumatoid synovi-tis. J Immunol 2001;167:1072–80.

3. Edwards JC, Wilkinson LS, Speight P, Isenberg DA. Vascular celladhesion molecule 1 and �4 and �1 integrins in lymphocyteaggregates in Sjogren’s syndrome and rheumatoid arthritis. AnnRheum Dis 1993;52:806–11.

4. Armengol MP, Cardoso-Schmidt CB, Fernandez M, Ferrer X,Pujol-Borrell R, Juan M. Chemokines determine local lymphoneo-genesis and a reduction of circulating CXCR4� T and CCR7 Band T lymphocytes in thyroid autoimmune diseases. J Immunol2003;170:6320–8.

5. Aust G, Sittig D, Becherer L, Anderegg U, Schutz A, Lamesch P,et al. The role of CXCR5 and its ligand CXCL13 in the compart-mentalization of lymphocytes in thyroids affected by autoimmunethyroid diseases. Eur J Endocrinol 2004;150:225–34.

6. Mazzucchelli L, Blaser A, Kappeler A, Scharli P, Laissue JA,Baggiolini M, et al. BCA-1 is highly expressed in Helicobacterpylori-induced mucosa-associated lymphoid tissue and gastric lym-phoma. J Clin Invest 1999;104:R49–54.

7. Mosnier JF, Degott C, Marcellin P, Henin D, Erlinger S, Ben-hamou JP. The intraportal lymphoid nodule and its environmentin chronic active hepatitis C: an immunohistochemical study.Hepatology 1993;17:366–71.

8. Serafini B, Rosicarelli B, Magliozzi R, Stigliano E, Aloisi F.Detection of ectopic B-cell follicles with germinal centers in themeninges of patients with secondary progressive multiple sclerosis.Brain Pathol 2004;14:164–74.

9. Corcione A, Casazza S, Ferretti E, Giunti D, Zappia E, Pistorio A,et al. Recapitulation of B cell differentiation in the central nervoussystem of patients with multiple sclerosis. Proc Natl Acad SciU S A 2004;101:11064–9.

10. Kaiserling E. Newly-formed lymph nodes in the submucosa inchronic inflammatory bowel disease. Lymphology 2001;34:22–9.

11. Hjelmstrom P. Lymphoid neogenesis: de novo formation of lym-phoid tissue in chronic inflammation through expression of hom-ing chemokines. J Leukoc Biol 2001;69:331–9.

12. Mebius RE. Organogenesis of lymphoid tissues. Nat Rev Immunol2003;3:292–303.

13. Cupedo T, Mebius RE. Cellular interactions in lymph nodedevelopment. J Immunol 2005;174:21–5.

14. Nishikawa S, Honda K, Hashi H, Yoshida H. Peyer’s patchorganogenesis as a programmed inflammation: a hypotheticalmodel. Cytokine Growth Factor Rev 1998;9:213–20.

15. Shi K, Hayashida K, Kaneko M, Hashimoto J, Tomita T, LipskyPE, et al. Lymphoid chemokine B cell-attracting chemokine-1(CXCL13) is expressed in germinal center of ectopic lymphoidfollicles within the synovium of chronic arthritis patients. J Immu-nol 2001;166:650–5.

LYMPHOID NEOGENESIS AND IL-7 SIGNALING IN RA SYNOVIAL TISSUES 2501

16. Manzo A, Paoletti S, Carulli M, Blades MC, Barone F, Yanni G,et al. Systematic microanatomical analysis of CXCL13 and CCL21in situ production and progressive lymphoid organization in rheu-matoid synovitis. Eur J Immunol 2005;35:1347–59.

17. Page G, Miossec P. Paired synovium and lymph nodes fromrheumatoid arthritis patients differ in dendritic cell and chemo-kine expression. J Pathol 2004;204:28–38.

18. Van der Pouw Kraan TC, van Gaalen FA, Huizinga TW, Pieter-man E, Breedveld FC, Verweij CL. Discovery of distinctive geneexpression profiles in rheumatoid synovium using cDNA microar-ray technology: evidence for the existence of multiple pathways oftissue destruction and repair. Genes Immun 2003;4:187–96.

19. Van der Pouw Kraan TC, van Gaalen FA, Kasperkovitz PV,Verbeet NL, Smeets TJ, Kraan MC, et al. Rheumatoid arthritis isa heterogeneous disease: evidence for differences in the activationof the STAT-1 pathway between rheumatoid tissues. ArthritisRheum 2003;48:2132–45.

20. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

21. Arnett FC. Revised criteria for the classification of rheumatoidarthritis. Bull Rheum Dis 1989;38:1–6.

22. Ball CA, Awad IA, Demeter J, Gollub J, Hebert JM, Hernandez-Boussard T, et al. The Stanford Microarray Database accommo-dates additional microarray platforms and data formats. NucleicAcids Res 2005;33:D580–2.

23. Tusher VG, Tibshirani R, Chu G. Significance analysis of microar-rays applied to the ionizing radiation response. Proc Natl Acad SciU S A 2001;98:5116–21.

24. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysisand display of genome-wide expression patterns. Proc Natl AcadSci U S A 1998;95:14863–8.

25. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL,Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles.Proc Natl Acad Sci U S A 2005;102:15545–50.

26. Liu YJ, Xu J, de Bouteiller O, Parham CL, Grouard G, Djossou O,et al. Follicular dendritic cells specifically express the long CR2/CD21 isoform. J Exp Med 1997;185:165–70.

27. Weyand CM, Goronzy JJ. Ectopic germinal center formation inrheumatoid synovitis. Ann N Y Acad Sci 2003;987:140–9.

28. Park YW, Kang YM, Butterfield J, Detmar M, Goronzy JJ,Weyand CM. Thrombospondin 2 functions as an endogenousregulator of angiogenesis and inflammation in rheumatoid arthri-tis. Am J Pathol 2004;165:2087–98.

29. Yang LV, Nicholson RH, Kaplan J, Galy A, Li L. Hemogen is anovel nuclear factor specifically expressed in mouse hematopoieticdevelopment and its human homologue EDAG maps to chromo-some 9q22, a region containing breakpoints of hematologicalneoplasms. Mech Dev 2001;104:105–11.

30. Van Roon JA, Verweij MC, Wijk MW, Jacobs KM, Bijlsma JW,Lafeber FP. Increased intraarticular interleukin-7 in rheumatoidarthritis patients stimulates cell contact–dependent activation ofCD4� T cells and macrophages. Arthritis Rheum 2005;52:1700–10.

31. Ebert LM, Schaerli P, Moser B. Chemokine-mediated control of Tcell traffic in lymphoid and peripheral tissues. Mol Immunol2005;42:799–809.

32. Luther SA, Bidgol A, Hargreaves DC, Schmidt A, Xu Y, PaniyadiJ, et al. Differing activities of homeostatic chemokines CCL19,

CCL21, and CXCL12 in lymphocyte and dendritic cell recruitmentand lymphoid neogenesis. J Immunol 2002;169:424–33.

33. Ansel KM, Cyster JG. Chemokines in lymphopoiesis and lymphoidorgan development. Curr Opin Immunol 2001;13:172–9.

34. Ansel KM, Ngo VN, Hyman PL, Luther SA, Forster R, SedgwickJD, et al. A chemokine-driven positive feedback loop organizeslymphoid follicles. Nature 2000;406:309–14.

35. Muller JR, Siebenlist U. Lymphotoxin � receptor induces sequen-tial activation of distinct NF-�B factors via separate signalingpathways. J Biol Chem 2003;278:12006–12.

36. Dejardin E, Droin NM, Delhase M, Haas E, Cao Y, Makris C,et al. The lymphotoxin-� receptor induces different patterns ofgene expression via two NF-�B pathways. Immunity 2002;17:525–35.

37. Honda K, Nakano H, Yoshida H, Nishikawa S, Rennert P, IkutaK, et al. Molecular basis for hematopoietic/mesenchymal interac-tion during initiation of Peyer’s patch organogenesis. J Exp Med2001;193:621–30.

38. Rich BE, Campos-Torres J, Tepper RI, Moreadith RW, Leder P.Cutaneous lymphoproliferation and lymphomas in interleukin 7transgenic mice. J Exp Med 1993;177:305–16.

39. Watanabe M, Ueno Y, Yajima T, Okamoto S, Hayashi T,Yamazaki M, et al. Interleukin 7 transgenic mice develop chroniccolitis with decreased interleukin 7 protein accumulation in thecolonic mucosa. J Exp Med 1998;187:389–402.

40. Freeden-Jeffry U, Vieira P, Lucian LA, McNeil T, Burdach SE,Murray R. Lymphopenia in interleukin (IL)-7 gene-deleted miceidentifies IL-7 as a nonredundant cytokine. J Exp Med 1995;181:1519–26.

41. Harada S, Yamamura M, Okamoto H, Morita Y, Kawashima M,Aita T, et al. Production of interleukin-7 and interleukin-15 byfibroblast-like synoviocytes from patients with rheumatoid arthri-tis. Arthritis Rheum 1999;42:1508–16.

42. Van Roon JA, Glaudemans KA, Bijlsma JW, Lafeber FP. Inter-leukin 7 stimulates tumour necrosis factor � and Th1 cytokineproduction in joints of patients with rheumatoid arthritis. AnnRheum Dis 2003;62:113–9.

43. Lin JX, Migone TS, Tsang M, Friedmann M, Weatherbee JA,Zhou L, et al. The role of shared receptor motifs and commonstat proteins in the generation of cytokine pleiotropy and redun-dancy by IL-2, IL-4, IL-7, IL-13, and IL-15. Immunity 1995;2:331–9.

44. Borghesi LA, Yamashita Y, Kincade PW. Heparan sulfate proteo-glycans mediate interleukin-7-dependent B lymphopoiesis. Blood1999;93:140–8.

45. Katakai T, Hara T, Lee JH, Gonda H, Sugai M, Shimizu A. Anovel reticular stromal structure in lymph node cortex: an im-muno-platform for interactions among dendritic cells, T cells andB cells. Int Immunol 2004;16:1133–42.

46. Lenz DC, Kurz SK, Lemmens E, Schoenberger SP, Sprent J,Oldstone MB, et al. IL-7 regulates basal homeostatic proliferationof antiviral CD4� T cell memory. Proc Natl Acad Sci U S A2004;101:9357–62.

47. Ariel A, Hershkoviz R, Cahalon L, Williams DE, Akiyama SK,Yamada KM, et al. Induction of T cell adhesion to extracellularmatrix or endothelial cell ligands by soluble or matrix-boundinterleukin-7. Eur J Immunol 1997;27:2562–70.

48. Sawa S, Kamimura D, Jin GH, Morikawa H, Kamon H, NishiharaM, et al. Autoimmune arthritis associated with mutated interleu-kin (IL)-6 receptor gp130 is driven by STAT3/IL-7-dependenthomeostatic proliferation of CD4� T cells. J Exp Med 2006;203:1459–70.

2502 TIMMER ET AL