TL1A - The Journal of Immunology

TL1A - The Journal of Immunology

This information is current as of January 12, 2018. Soluble TNF-Like Cytokine (TL1A) Production by Immune Complexes Stimulated Monocytes in Rheumatoi...

2MB Sizes 0 Downloads 9 Views

This information is current as of January 12, 2018.

Soluble TNF-Like Cytokine (TL1A) Production by Immune Complexes Stimulated Monocytes in Rheumatoid Arthritis Marco A. Cassatella, Gabriela Pereira da Silva, Ilaria Tinazzi, Fabio Facchetti, Patrizia Scapini, Federica Calzetti, Nicola Tamassia, Ping Wei, Bernardetta Nardelli, Viktor Roschke, Annunciata Vecchi, Alberto Mantovani, Lisa M. Bambara, Steven W. Edwards and Antonio Carletto

Why The JI? • Rapid Reviews! 30 days* from submission to initial decision • No Triage! Every submission reviewed by practicing scientists • Speedy Publication! 4 weeks from acceptance to publication *average

References Subscription Permissions Email Alerts Errata

This article cites 34 articles, 13 of which you can access for free at: Information about subscribing to The Journal of Immunology is online at: Submit copyright permission requests at: Receive free email-alerts when new articles cite this article. Sign up at: An erratum has been published regarding this article. Please see next page or: /content/179/2/1390.2.full.pdf

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606.

Downloaded from by guest on January 12, 2018

J Immunol 2007; 178:7325-7333; ; doi: 10.4049/jimmunol.178.11.7325

The Journal of Immunology

Soluble TNF-Like Cytokine (TL1A) Production by Immune Complexes Stimulated Monocytes in Rheumatoid Arthritis1 Marco A. Cassatella,2* Gabriela Pereira da Silva,* Ilaria Tinazzi,† Fabio Facchetti,‡ Patrizia Scapini,* Federica Calzetti,* Nicola Tamassia,* Ping Wei,§ Bernardetta Nardelli,¶ Viktor Roschke,储 Annunciata Vecchi,# Alberto Mantovani,# Lisa M. Bambara,† Steven W. Edwards,** and Antonio Carletto†


heumatoid arthritis (RA)3 is a chronic idiopathic disease characterized by persistent inflammation of the synovium, leading to local destruction of bone and cartilage and a variety of systemic manifestations that lead to disability (1). RA involves all elements of the immune response, with autoimmune recognition of self-Ags by repeatedly activated autoreactive Th1 lymphocytes, the functional down-regulation and elimination of which appears defective (2). In addition, the local production of Igs and rheumatoid factors (RF), which are autoantibodies directed against the Fc portion of IgG, along with the local production of

*Department of Pathology, Division of General Pathology, and †Department of Clinical and Experimental Medicine, Division of Rheumatology, University of Verona, Verona, Italy; ‡Department of Pathology, Spedali Civili, University of Brescia, Brescia, Italy; §Amgen, Thousand Oaks, CA 91301; ¶Human Genome Sciences, Rockville, MD 20850; 储CoGenesys, Rockville, MD 20850; #Fondazione Humanitas per la Ricerca, Rozzano, Italy; and **School of Biological Sciences, University of Liverpool, Liverpool, United Kingdom Received for publication July 7, 2006. Accepted for publication March 27, 2007. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by grants (to M.A.C.) from Ministero dell’Istruzione, dell’Universita` e della Ricerca (Progetti di Rilevante Interesse Nazionale 2005, Fondo di Investimento per la Ricerca di Base, and 60%), Fondazione Cassa di Risparmio, and Associazione Italiana per la Ricerca sul Cancro. 2 Address correspondence and reprint requests to Dr. Marco A. Cassatella, Division of General Pathology, Department of Pathology, Strada Le Grazie 4, Verona, Italy. E-mail address: [email protected] 3 Abbreviations used in this paper: RA, rheumatoid arthritis; IC, immune complex; RF, rheumatoid factor; SF, synovial fluid; TL1A, TNF-like cytokine; sTL1A, soluble TL1A; mTL1A, membrane-bound TL1A; DR3, death domain receptor 3; IL-1ra, IL-1 receptor antagonist; PMX, polymyxin B sulfate; RF⫹/RA, RF-seropositive RA patients; RF⫺/RA, RF-seronegative RA patients; OA, osteoarthritis; PEG, polyethylene glycol; CHX, cycloheximide.

Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00

immune complexes (IC) leading to complement activation, appear important in the destructive events associated with the synovitis (3). Although the etiological stimulus has not been identified, a number of inflammatory mediators produced in the inflamed rheumatoid synovial tissue, including arachidonic acid metabolites, vasoactive amines, platelet-activating factor, proteinases, growth factors, and complement cleavage products, contribute to the inflammatory process (1). In addition, many of the local and systemic manifestations of RA appear to result from the production of a variety of cytokines within the inflamed synovium, particularly TNF-␣, IL-1, IL-6, and IL-15, but also CD40/CD40L, B lymphocyte stimulator, receptor activator for NF-␬B ligand, 41BB/41BBL, IL-18, chemokines, and angiogenic factors (4), contribute to the inflammatory process. Indeed, biological agents that specifically inhibit the effects of the cytokines implicated in the pathogenesis of RA (e.g., TNF-␣, IL-1, or IL-15) can lead to a rapid improvement of inflammation, and substantial clinical benefits as compared with more conventional therapeutic approaches (5). TNF-like cytokine (TL1A) is a newly identified member of the TNF superfamily of ligands, which has been identified as a longer variant of TL1 (also called VEGI) that binds the death domain receptor 3 (DR3; Refs. 6 and 7). Expression of DR3, a TNF superfamily receptor member with highest homology to TNFR1, is restricted to lymphocytes and is up-regulated upon T cell activation (8). TL1A can also bind to the soluble decoy receptor called TR6/DcR3, which also interacts with two other TNF ligands, namely Fas ligand and lymphotoxin-related inducible ligand that competes for glycoprotein D binding to herpes virus entry mediator on T cells, and competes with DR3 for TL1A binding (6). Like other TNF family ligands, TL1A contains a predicted transmembrane domain and a bioactive, proteolytically cleaved truncated form that can be released as a soluble factor (6). Recent

Downloaded from by guest on January 12, 2018

TNF-like cytokine (TL1A) is a newly identified member of the TNF superfamily of ligands that is important for T cell costimulation and Th1 polarization. However, despite increasing information about its functions, very little is known about expression of TL1A in normal or pathological states. In this study, we report that mononuclear phagocytes appear to be a major source of TL1A in rheumatoid arthritis (RA), as revealed by their strong TL1A expression in either synovial fluids or synovial tissue of rheumatoid factor (RF)-seropositive RA patients, but not RFⴚ/RA patients. Accordingly, in vitro experiments revealed that human monocytes express and release significant amounts of soluble TL1A when stimulated with insoluble immune complexes (IC), polyethylene glycol precipitates from the serum of RFⴙ/RA patients, or with insoluble ICs purified from RA synovial fluids. Monocyte-derived soluble TL1A was biologically active as determined by its capacity to induce apoptosis of the human erythroleukemic cell line TF-1, as well as to cooperate with IL-12 and IL-18 in inducing the production of IFN-␥ by CD4ⴙ T cells. Because RA is a chronic inflammatory disease with autoimmune etiology, in which ICs, autoantibodies (including RF), and various cytokines contribute to its pathology, our data suggest that TL1A could be involved in its pathogenesis and contribute to the severity of RA disease that is typical of RFⴙ/RA patients. The Journal of Immunology, 2007, 178: 7325–7333.


INDUCIBLE sTL1A RELEASE FROM ACTIVATED MONOCYTES were preincubated for 1 h at 4°C with 10 ␮g/ml anti-Fc␥R Abs and related isotype controls (IgG1 for all of them) in the absence of FCS. Anti-Fc␥R Abs were: anti-Fc␥RI/CD64 (clone 10.1; Ancell); anti-Fc␥RII/CD32 (AT10 clone, which binds to all three forms of Fc␥RII (a, b, and c); Serotec) and anti-Fc␥RIII/CD16 (3G8 clone; BD Pharmingen). IgG1 were purchased from BD Pharmingen. Monocytes were then washed, suspended in medium, plated, and then stimulated. After culture, cells were detached, collected, and spun at 350 ⫻ g for 5 min; the resulting supernatants were immediately frozen in liquid nitrogen and stored at ⫺80°C. All reagents were of the highest available grade and were dissolved in pyrogen-free water suitable for clinical use.

Patients RA patients, selected according to the American Rheumatism Association criteria (15), were enrolled at first diagnosis after their signed informed consent and approval by the local ethical committees. These patients were categorized as having early, disease-modifying anti-rheumatic drug-free, active-phase disease (with disease activity score of ⬎3.5) and grouped into RF-seropositive (RF⫹/RA) or RF-seronegative (RF⫺/RA), according to the detection of threshold levels of both serum IgM RF (ⱖ10 U/ml, as measured by nephelometry) and IgG RF (ⱖ6 U/ml, as measured by QUANTA Lite RF IgG ELISA (Menarini Diagnostics). Age- and sexmatched healthy subjects were enrolled as controls.

SF samples SF samples were collected from RF⫹/RA and RF⫺/RA patients, and, as control, osteoarthritis (OA) patients who underwent therapeutic and/or diagnostic arthrocentesis of a knee effusion. After withdrawal, samples were retained for cytospins and subsequent immunocytochemistry, whereas cellfree SF supernatants were isolated for subsequent analyses.

Polyethylene glycol (PEG) precipitation of serum ICs

Materials and Methods Cell purification and culture Leukocytes were isolated under endotoxin-free conditions from either buffy coats of healthy donors or whole blood of healthy donors and patients, by standard procedures. Briefly, buffy coats or blood were layered on Ficoll-Hypaque density gradient (Amersham Biosciences) and then centrifuged at 400 ⫻ g for 30 min at room temperature. The interface between plasma and Ficoll, corresponding to the mononuclear cell (PBMC) fraction, was then collected, washed five times with PBS to eliminate platelets, suspended in isosmotic (285 mOsm) culture medium (RPMI 1640 containing 10% low endotoxin FCS and 4 mM HEPES), and layered over a 46% isosmotic Percoll solution (Pharmacia). After centrifugation (30 min at room temperature, 650 ⫻ g), the monocyte and lymphocyte fractions were collected, washed in PBS, and suspended in standard culture medium (as described later in this paragraph). Monocyte preparations were ⬎85% pure as evaluated by both CD14 expression by FACS and morphology after May-Gru¨nwald-Giemsa staining of cytospins. In some experiments, monocytes were also isolated by negative selection using MACS separation (Monocyte Isolation kit II; Miltenyi Biotec) to ⬎95% purity. Neutrophils were isolated by Ficoll centrifugation followed by dextran sedimentation and osmotic lysis and were ⬎95% pure (as evaluated by May-Gru¨nwaldGiemsa staining) (11). After purification, cells were suspended in RPMI 1640 supplemented with 10% low-endotoxin FCS (⬍0.5 endotoxin U/ml; BioWhittaker; defined as standard culture medium), plated in six 24-well tissue culture plates (Orange Scientific Cambrex, Techno Plastic Products) at 5 ⫻ 106/ml, and then cultured for up to 48 h in a 5% CO2 atmosphere in the presence or absence of various stimuli, including 100 ng/ml LPS (from Escherichia coli, serotype 026:B6; Sigma-Aldrich), 5 ng/ml TNF-␣, 10 ng/ml GM-CSF, 10 ng/ml IL-4, 10 ng/ml IL-13 (Peprotech), 200 U/ml IFN-␥ (R&D Systems), 1000 U/ml IFN␣ (Roferon; Roche Laboratories), 200 U/ml IL-10 (DNAX and Schering-Plough), 100 nM fMLP (SigmaAldrich), and 50 ␮g/ml insoluble immune complexes (IC). The latter were prepared by using OVA from chicken egg (Sigma-Aldrich) or BSA (SigmaAldrich) as Ags, with, respectively, anti-albumin chicken egg rabbit antiserum (Calbiochem) and anti-BSA rabbit antiserum (Calbiochem) as previously described (12). Endotoxin contamination of IC (at the working concentrations) was ⬍0.06 endotoxin U/ml (corresponding to ⬃6 pg/ml), as determined by the Limulus amebocyte lysate assay (BioWhittaker). For direct Fc␥R engagement, 96- or 24-well plates were coated with purified human IgG (Sigma-Aldrich) at 20 ␮g/0.2 ml/well in sterile PBS for 2 h and washed three times with warm PBS before use (13). In selected experiments, before stimulation, monocytes were preincubated with: 20 ␮g/ml polymyxin B sulfate (PMX); anti-TNF-␣-neutralizing mAbs (B154.2 clone; Ref. 14); 10 ng/ml IL-4; 200 U/ml IFN-␥. Alternatively, monocytes

Sera from RF⫹/RA, RF⫺/RA, and healthy donors were precipitated with 3% (w/v) ice-cold PEG 6000 (Sigma-Aldrich), centrifuged, washed three times in sterile PBS, and finally diluted to the initial serum volume in sterile PBS as described (16). Cell-free SF (from RF⫹/RA patients) was centrifuged at 11,600 ⫻ g for 5 min to pellet insoluble ICs (which were further purified by PEG precipitation as above), whereas the soluble IC retained in the supernatant was discarded (17).

Analysis of mediator concentration sTL1A concentrations in cell-free supernatants and freshly prepared sera and SF were measured by a specific ELISA developed at Human Genome Sciences (Rockville, MD; detection limit, 20 – 40 pg/ml). Briefly, flat-bottom 96-well plates (MaxiSorp 439454; Nunc) were coated with 75 ␮l/well of 2 ␮g/ml anti-TL1A capture mAb (clone 04H08), diluted in PBS (pH 7.6), and incubated overnight at 4°C. After an extensive rinsing in washing buffer (0.05% Tween 20 in PBS, pH 7.2–7.4), the plates were incubated for at least 1 h at room temperature with 225 ␮l/well blocking buffer (1% BSA, 5% sucrose in PBS). After a washing, 75 ␮l of either recombinant TL1A standards or samples were added to the wells and incubated for 2 h at room temperature. Plates were then extensively washed before addition of 75 ␮l/well detection biotin-conjugated mAb (clone 16H02) suspended in reagent diluent (1% BSA in PBS, pH 7.2–7.4, 0.2 ␮m filtered) and incubated for 2 h at room temperature. After extensive washings, 75 ␮l/well of streptavidin-HRP (SNN2004; Biosource International) diluted 1/10,000 in reagent diluent were added, and plates were incubated for 45– 60 min at room temperature in the dark. After a rinsing, 75 ␮l/well substrate solution containing 3,3⬘,5,5⬘-tetramethylbenzidine (Sigma-Aldrich) were added and incubated for 20 min at room T in the dark. The reaction was stopped with 50 ␮l/well of 2 N H2SO4 and optical densities were measured at 450 nm using a microplate reader (Packard Spectracount BSL0001). All assays were conducted in duplicate. A second TL1A ELISA purchased from Peprotech (detection limit, 62 pg/ml) was used according to the manufacturer’s instructions, which fully reproduced the data obtained with the Human Genome Sciences ELISA. IL-1 receptor antagonist (IL-1ra) and TNF-␣ were measured by using ELISA kits purchased from R&D Systems, whereas IFN-␥ was determined by the ELISA kit obtained from ImmunoTools.

Northern blotting and real-time RT-PCR analysis These experiments were conducted as described previously (6, 11). For real-time RT-PCR, ␤2-microglobulin was selected as a normalizing gene, due to its stable expression levels in leukocytes (18). Data were calculated with Q-Gene software ( and are expressed as mean normalized expression units after ␤2-microglobulin normalization.

Downloaded from by guest on January 12, 2018

studies have shown that recombinant soluble TL1A (sTL1A): augments IFN-␥ and GM-CSF production and increases IL-2 responsiveness by anti-CD3/CD28-stimulated T cells (6); synergizes with IL-12 and IL-18 to augment IFN-␥ release in human T and NK cells (8); and biases T cells to differentiate toward the Th1 phenotype (9, 10). It has also been shown that resting T cells, B cells, NK cells, dendritic cells, and monocytes do not express TL1A mRNA, in contrast with HUVECs stimulated by TNF, IL-1␣, or PMA (6) or CD11chigh dendritic cells in mice with inflammation (10). In contrast, membrane-bound TL1A (mTL1A) was reported as expressed by in vitro activated T cells and by a subset of mucosal T cells and macrophages in inflamed intestinal mucosa in Crohn’s disease and ulcerative colitis (9). In this work, we explored whether activated human leukocytes can produce sTL1A and, if so, which leukocyte subtype was responsible. We show that mononuclear phagocytes are potent producers of sTL1A and that insoluble IC are extremely strong activators of this secretion. These observations led us to investigate whether TL1A is expressed in pathological conditions associated with increased levels of circulating or deposited IC, such as RA. Remarkably, we found a marked expression of TL1A associated with mononuclear phagocytes present in synovial membranes and synovial fluids (SF) of medication-free RF⫹/RA patients at first diagnosis with active disease. Our data imply that, in vivo, monocytes and macrophages of RA patients express TL1A in response to IC.

The Journal of Immunology


Flow cytometry for mTL1A expression

Statistical analysis

Freshly isolated or activated monocytes were stained with mAbs raised against human TL1A (12F11; Human Genome Sciences) or control mouse IgG1 (Sigma-Aldrich), followed by biotinylated secondary sheep antimouse IgG (Sigma-Aldrich). Alternatively, TL1A staining was performed with polyclonal biotinylated Abs from R&D (BAF744) or goat biotinylated IgG (BD Pharmingen) as isotype controls. CR3 staining was performed with OKM1 mAb (Caltag Laboratories). Immunolabeling was achieved after the addition of PE-conjugated streptavidin (BD Biosciences). Cytofluorographic analyses (using at least 104 cells/sample) were performed on a FACScan (BD Biosciences) using CellQuest software (19).

Data are expressed as means ⫾ SEM. Statistical evaluation was performed by Student’s t test for paired data and considered to be significant if p ⬍ 0.05.

Apoptosis of TF-1 cells

IFN-␥ production assay Human CD4⫹ T lymphocytes were isolated from the Percoll lymphocyte fraction using a negative selection enrichment kit for human CD4⫹ T cells (StemCell Technologies), according to the manufacturer’s instructions. Purity was routinely ⬎97%, as determined by FACS. CD4⫹ T cells were cultured at 106/ml in the presence or absence of 2 ng/ml IL-12 (PeproTech) plus 50 ng/ml IL-18 (R&D Systems), either singly or in combination with various concentrations of recombinant TL1A. CD4⫹ T cells were also cultured with or without IL-12 plus IL-18 in supernatants harvested from resting or IC-stimulated monocytes (previously immunodepleted of TNF-␣, as described in “Apoptosis of TF-1 cells”). After 72 h of incubation, supernatants were collected and assayed for IFN-␥ content by ELISA. To assess the specific effect of the TL1A contained in monocyte-derived supernatants, the latter were preincubated with 1 ␮g/ml anti-TL1A or isotype-matched control mAbs for 20 min at 37°C.

TL1A expression in RA tissue Synovial sections explanted from six RF⫹/RA patients, two RF⫺/RA patients, and, as controls, from four patients with synovitis associated with chronic OA were examined by immunohistochemistry. In addition, s.c. nodules from patients with active RA displaying typical rheumatoid granulomas were analyzed. All specimens were formalin fixed and paraffin embedded. Serial sections were stained with mAbs against TL1A (clone 12F11; overnight incubation at 1/3000), CD3 (1/50), CD20 (1/200) and CD68 (1/50) (all from DakoCytomation), followed by a peroxidase-conjugated dextran polymer (ChemMate; DakoCytomation). Sections were developed using diaminobenzidine as chromogen and Mayer’s hematoxylin as counterstaining. Staining for TL1A on cells obtained upon cytocentrifugation of SFs from two RA patients followed the same procedure. For double immunofluorescence for CD14 and TL1A, sections were first incubated with anti-CD14 (clone MM42; IgG2a; Novo Castra; 1/50), followed by Texas red-conjugated anti-IgG2a (SouthernBiotech), anti-TL1A, and FITC-conjugated anti-IgG1 (SouthernBiotech). Negative controls consisted of the omission of primary Abs and by the use of irrelevant isotype matched Abs. Preincubation of 12F11 mAbs with IgM-RF or IgG-RF preparations (508692 and 708685, respectively, from Inova-San Diego) before their addition to tissue specimens did not modify the outcome of the final TL1A staining. Sections were then finally examined with an Olympus BX60 microscope and equipped with relevant fluorescence excitation/emission filters and a DP-70 digital camera (Olympus). Images were acquired using analySIS ImageProcessing software (Soft Imaging System); immunofluorescence images were merged as previously described (23).

Human monocytes incubated with insoluble IC release sTL1A To investigate whether mononuclear phagocytes express TL1A, human monocytes isolated from buffy coats of healthy donors were cultured for up to 48 h with optimal concentrations of various inflammatory mediators, including LPS (n ⫽ 13), TNF-␣ (n ⫽ 6), GM-CSF (n ⫽ 6), fMLP (n ⫽ 6), IFN-␥ (n ⫽ 6), IFN␣ (n ⫽ 3), IL-4 (n ⫽ 6), IL-10 (n ⫽ 3), IL-13 (n ⫽ 3), aggregated IgG (n ⫽ 4), and BSA/anti-BSA or OVA/anti-OVA insoluble IC (n ⫽ 13). Cell-free supernatants were collected to determine the yields of sTL1A, whereas adherent cells were processed for TL1A mRNA accumulation by Northern blot analysis and/or real time RT-PCR. Of all these stimuli, only insoluble IC and aggregated IgG stimulated a remarkable extracellular release of sTL1A by 20-h-cultured monocytes (596 ⫾ 83 pg/ml for OVA/anti-OVA IC, n ⫽ 10; 683 ⫾ 152 pg/ml for BSA/anti-BSA IC, n ⫽ 15; 258.4 ⫾ 62.1 pg/ml, n ⫽ 10, for aggregated IgG). This effect of insoluble IC toward sTL1A release was selective in that their capacity to elicit the production of IL-1ra was substantially similar to that triggered by LPS, which was only a modest activator of sTL1A release (61.2 ⫾ 15.5 pg/ml; n ⫽ 15) compared with untreated monocytes (⬍20 pg/ml, n ⫽ 15). Northern blot experiments revealed that sTL1A production by insoluble IC-activated monocytes is preceded by an up-regulation of TL1A mRNA expression, which progressively increased over time and reached maximal expression between 6 and 20 h of incubation, depending on the donor (Fig. 1A). In contrast, up-regulation of TNF-␣ and IL-1ra mRNA expression in insoluble IC-treated monocytes occurred with faster kinetics (Fig. 1, A and B). Similar results were obtained by real-time RT-PCR studies, which also confirmed the very weak up-regulatory effect of LPS on TL1A mRNA expression compared with that induced by OVA/anti-OVA insoluble IC (Fig. 1B) and BSA/anti-BSA insoluble IC (data not shown). Time-course studies demonstrated that the ability of insoluble IC and LPS to stimulate the release of sTL1A from monocytes is relatively slow (starting 5 h poststimulation and progressively increasing up to 48 h of incubation) (Figs. 1C and 2A), being more delayed compared with the kinetics of TNF-␣ (Fig. 1C) or IL-1ra release (data not shown). In addition, dose-response experiments revealed that concentrations as low as 2.5–5 ␮g/ml and 1–10 ng/ml insoluble IC and LPS, respectively, can trigger a detectable sTL1A production (Fig. 2B). Preincubation of OVA/anti-OVA or BSA/anti-BSA insoluble IC with PMX did not suppress their capacity to induce the production and secretion of sTL1A or IL-1ra, whereas it completely blocked the effect of LPS (Fig. 2C), thus excluding endotoxin contamination of our IC preparations (see also Materials and Methods for Limulus amebocyte lysate assay results). Surprisingly, only activated monocytes, but neither neutrophils nor lymphocytes, were able to release sTL1A under these experimental conditions (Fig. 2D). Activated neutrophils, however, did produce considerable amounts of IL-1ra (data not shown), in line with previous results (24). Experiments in which we added anti-TNF-␣neutralizing Abs to insoluble IC monocytes indicated that monocytederived TNF-␣ does not play a significant role in inducing TL1A (data not shown). All these results have been reproduced with a second, commercially available TL1A ELISA (see Materials and Methods).

Downloaded from by guest on January 12, 2018

The biological activity of sTL1A released by activated monocytes into culture medium was tested by the ability of monocyte-derived supernatants to induce apoptosis of the human erythroleukemic cell line, TF-1, as recently described (6). Briefly, TF-1 cells (8 ⫻ 105 cells/ml) were suspended either directly in supernatants harvested from resting or IC-stimulated monocytes, or, as control, in standard culture medium containing various concentrations of recombinant TL1A and then seeded in 96-well culture plates (Orange Scientific Cambrex). Cell cultures were incubated in the presence or absence of 10 ␮g/ml cycloheximide (CHX), as described (6). After 6 h, TF-1 cells were harvested, and their apoptosis was measured with an Annexin V Fluos staining kit (Roche; Ref. 20). To assess the specificity of the TL1A-mediated proapoptotic activity, monocyte-derived supernatants and recombinant human TL1A were preincubated with 600 ng/ml anti-TL1A mAbs (clone 16H02) or isotype-matched control mAbs (IgG1) for 20 min at 37°C. All monocyte-derived supernatants used for these experiments were previously immunodepleted of TNF-␣ using wellestablished procedures (21, 22). Complete removal of TNF-␣ was verified by ELISA quantification of TNF-␣-immunodepleted supernatants.




Finally, and in contrast with the remarkable release of sTL1A, expression of mTL1A increased only slightly in insoluble ICtreated monocytes, as assessed by flow cytometry using Abs from two different sources (Fig. 3). Collectively, our data un-

equivocally identify insoluble IC as very potent and specific stimuli for the induction of both TL1A mRNA expression and sTL1A release in human monocytes, but not in neutrophils or lymphocytes.

FIGURE 2. Characterization of TL1A production by activated monocytes. A and B, Monocytes (5 ⫻ 106/ml) were incubated either for various times with 50 ␮g/ml OVA/ anti-OVA insoluble IC or 100 ng/ml LPS (A) or with increasing concentrations of stimuli for 20 h (B), before determination of sTL1A levels in cell-free supernatants. C, Monocytes were preincubated for 30 min with 20 ␮g/ml PMX before stimulation with OVA/ anti-OVA insoluble IC or 100 ng/ml LPS. D, PBMC, monocytes, lymphocytes, and neutrophils purified from the same donor were cultured at 5 ⫻ 106/ml for 20 h with or without 50 ␮g/ml OVA/anti-OVA insoluble IC and 100 ng/ml LPS. Culture supernatants were then collected for sTL1A and IL-1ra measurement. Values are means ⫾ SEM of duplicate determinations calculated from three to four independent experiments.

Downloaded from by guest on January 12, 2018

FIGURE 1. TL1A mRNA expression in IC-stimulated monocytes. Purified populations of monocytes were incubated for the times indicated with 50 ␮g/ml BSA/anti-BSA insoluble IC (A) or 50 ␮g/ml OVA/anti-OVA insoluble IC or 100 ng/ml LPS (B). TL1A, TNF-␣, and actin mRNA expression was then detected either by Northern blotting analysis (A) or by real-time RT-PCR (B). B, Gene expression is depicted as mean normalized expression (MNE) units after ␤2-microglobulin normalization of triplicate reactions for each sample. Data are representative of results from three independent experiments for each panel. C, Determination of sTL1A and TNF-␣ levels in supernatants harvested from the experiment shown in A.

The Journal of Immunology


Biological activities of monocyte-derived sTL1A To ascertain whether sTL1A released by IC-stimulated monocytes was biologically active, we tested the ability of monocyte-derived supernatants to induce the apoptosis of the TL1A-sensitive TF-1 cells (6). As shown by annexin V staining (Fig. 4A), sTL1A that accumulated in conditioned medium obtained from IC-stimulated monocytes was able to potentiate the proapoptotic effect of CHX on TF-1 cells by ⬃2-fold. These supernatants contained ⬃0.5–1 ng/ml sTL1A, and at these concentrations recombinant TL1A induced identical effects of apoptosis on CHX-treated TF-1 cells (Fig. 4A). No significant apoptosis of TF-1 cells was observed if TL1A was used in the absence of CHX, or if TF-1 cells were incubated in medium from resting monocytes (Fig. 4A). Importantly, promotion of TF-1 cell apoptosis by conditioned medium obtained from IC-stimulated monocytes was TL1A-specific because it was completely abrogated by neutralizing anti-TL1A mAbs (Fig. 4A), but not by isotype-matched control Abs (data not shown). All monocyte-derived supernatants used in these assays were subjected to TNF-␣ removal by immunoabsorption (21, 22). This was necessary because preliminary experiments revealed that recombinant TNF-␣, used at the same concentrations as those detected in conditioned medium harvested from IC-activated monocytes (ranging from 5 to 25 ng/ml, depending on the donor), also exerted potent proapoptotic activity on CHX-treated TF-1 cells. Subsequently, we examined whether monocyte-derived supernatants could augment the production of IFN-␥ by IL-12 plus IL18-stimulated CD4⫹ T cells, as described by Papadakis et al. (8).

FIGURE 4. Supernatants from IC-stimulated monocytes promote apoptosis of TL1A-sensitive cells and IFN-␥ production by CD4⫹ T cells stimulated with IL-12 plus IL-18. A, TF-1 cells were suspended in supernatants harvested from resting or IC-stimulated monocytes that were previously immunodepleted of TNF-␣ (see Materials and Methods) and, as control, in standard culture medium containing 0.5 ng/ml recombinant TL1A and then cultured in the presence or the absence of 10 ␮g/ml CHX. Aliquots of monocyte-derived supernatants or recombinant TL1A were also preincubated with 600 ng/ml neutralizing antiTL1A mAbs or isotype-matched control mAbs for 30 min to assess the specificity of the TL1A-mediated proapoptotic activities. After 6 h, TF-1 cells were collected, and apoptosis was assessed by annexin V staining by flow cytometry analysis. Means ⫾ SEM of the relative percent of annexin V-positive cells calculated from three independent experiments are shown. B, CD4⫹ T cells were incubated at 106/ml in standard culture medium with or without 2 ng/ml IL-12 plus 50 ng/ml IL-18, singly or in combination with either supernatants harvested from resting or IC-stimulated monocytes (previously immunodepleted of TNF-␣), or 1 ng/ml recombinant TL1A. After 72 h of incubation, supernatants were collected and assayed for IFN-␥ content by ELISA. Aliquots of monocyte-derived supernatants or recombinant TL1A were also preincubated with 1 ␮g/ml neutralizing anti-TL1A mAbs or isotype matched control mAbs for 30 min. Means ⫾ SEM of IFN-␥ release calculated from three independent experiments are shown. ⴱ, Significant effects; ⴱ, p ⬍ 0.05; ⴱⴱ, p ⬍ 0.01.

As shown in Fig. 4B, sTL1A that accumulated in medium harvested from IC-stimulated monocytes was able to significantly potentiate the ability of IL-12 plus IL-18 to trigger CD4⫹ T cell production of IFN-␥ by ⬃8-fold (n ⫽ 3). These monocyte-derived

Downloaded from by guest on January 12, 2018

FIGURE 3. TL1A surface expression in activated monocytes. Monocytes were cultured for 20 h with or without 50 ␮g/ml OVA/anti-OVA insoluble IC and 100 ng/ml LPS before analysis of membrane-bound TL1A (mTL1A) and CR3 expression by flow cytometry. Isotype anti-mouse IgG1 mAbs were used as controls. Data are representative of results from five independent experiments.



supernatants were previously immunodepleted of TNF-␣ (as described previously) and contained ⬃1 ng/ml sTL1A, as determined by ELISA. Recombinant TL1A at 1 ng/ml used together with IL12/IL-18 was less potent than monocyte-derived supernatants (Fig. 4B), suggesting that other factor(s) present in supernatants of ICstimulated monocytes synergistically cooperate with endogenous sTL1A in inducing the production of IFN-␥. No significant effects were observed if TL1A was used in the absence of IL-12/IL-18 or if CD4⫹ T cells were incubated in medium from resting monocytes (Fig. 4B). Importantly, stimulation of IFN-␥ production by conditioned medium obtained from IC-stimulated monocytes was TL1A specific because it was completely abrogated by neutralizing antiTL1A mAbs (Fig. 4B), but not by isotype-matched control Abs (data not shown). Effect of Fc␥R blocking and immunoregulatory cytokines on the ability of insoluble IC to trigger sTL1A production by monocytes

TL1A detection in synovial tissue of RA patients The identification of insoluble IC or aggregated IgG as potent stimuli for sTL1A production in vitro led us to hypothesize that sTL1A could be expressed in RA patients. Ig-containing IC or autoantibodies such as RF are abundant in serum and synovial fluid of RA patients (25, 26) and predict a more aggressive, destructive course of disease (1). Because we could not accurately measure the levels of sTL1A in serum or synovial fluid samples due to the interference of RF in our two TL1A ELISAs, we performed an immunohistochemical analysis of both synovial and skin tissue samples and SF cells from RA patients. Fig. 6A shows that synovium from RF⫹ patients with active RA contained high numbers of cells that strongly expressed TL1A in their cytoplasm (Fig. 6A, b and c), whereas synovium from OA patients showed only few TL1A-positive (TL1A⫹) cells (Fig. 6Aa). These TL1A⫹ cells in RA patients displayed a monocyte-macrophage morphology and were dispersed among other inflammatory cells, being particularly abundant beneath the synovium. Remarkably, in rheumatoid nodules of the subcutaneous tissue, the TL1A⫹ cells formed the palisade surrounding the central area of fibrinoid necrosis (Fig. 6Ae). The mononuclear phagocyte origin of the TL1A⫹ cells was supported by the staining of serial sections with

FIGURE 5. Effect of Fc␥R blocking and immunoregulatory cytokines on the ability of insoluble IC to trigger sTL1A production by monocytes. A, Monocytes were preincubated with 10 ␮g/ml anti-Fc␥RI/ CD64, anti-Fc␥RII/CD32, and anti-Fc␥RIII/CD16 and related isotype control Abs in the absence of FCS for 1 h at 4°C. Cells were then washed, suspended in complete medium, plated, and then incubated with 50 ␮g/ml BSA/anti-BSA insoluble IC. After 20 h, culture supernatants were collected for sTL1A measurement and analysis. Bars depict the mean values ⫾ SEM (n ⫽ 3) of the percentage of inhibition exerted by the various anti-Fc␥R Abs on the production of sTL1A triggered by insoluble IC. B and C, Monocytes were preincubated with 10 ng/ml IL-4 (B) or 200 U/ml IFN-␥ (C) for 1 h before stimulation with 50 ␮g/ml BSA/anti-BSA insoluble IC, 50 ␮g/ml OVA/anti-OVA insoluble IC, or 100 ng/ml LPS. After 20 h, culture supernatants were collected for sTL1A measurement and analysis. Bars depict the mean values ⫾ SEM (n ⫽ 4) of the percentage of inhibition or fold increase exerted by IL-4 or IFN-␥, respectively.

anti-CD68 (Fig. 6Ac, anti-TL1A; Fig. 6Ad, anti-CD68). This was confirmed by double immunofluorescence with TL1A and antiCD14 (Fig. 6Af). Similar to the study of Bamias et al. (9), a few TL1A-expressing plasma cells were occasionally found (data not shown), whereas no TL1A⫹ lymphocytes, polymorphonuclear neutrophils, or endothelial cells were observed (Fig. 6Ac). In areas of marked inflammation adjacent to the synovial surface, TL1A positivity appeared to be associated with the synoviocytes, possibly due to release of the cytokine from closely

Downloaded from by guest on January 12, 2018

To establish which Fc␥R(s) was responsible for triggering TL1A expression, monocytes were pretreated for 1 h with anti-Fc␥RI/ CD64, anti-Fc␥RII/CD32, anti-Fc␥RIII/CD16, and related isotype control Abs before their stimulation with BSA/anti-BSA insoluble IC for 20 h. As shown in Fig. 5A, production of sTL1A in IC-stimulated monocytes was reduced by 65 ⫾ 13% and 20 ⫾ 5%, respectively, by anti-Fc␥RII/CD32 and antiFc␥RIII/CD16 Abs but was not affected by anti-Fc␥RI/CD64 (Fig. 5A) or by isotype control Abs (data not shown). AntiFc␥R Abs alone did not trigger, by themselves, any sTL1A expression (data not shown). These data suggest that CD32 appears to be the major Fc␥R required for the induction of sTL1A by insoluble IC in human monocytes. We also investigated whether IL-4 and IFN-␥ might influence the inducible expression of TL1A. Fig. 5B shows that IL-4 almost completely abolished monocyte production of TL1A induced by LPS, OVA/anti-OVA, and BSA/anti-BSA insoluble IC. In contrast, IFN-␥ greatly potentiated the stimulatory effect of all these agonists, particularly LPS and BSA/anti-BSA insoluble IC (Fig. 5C). Consistent with these data, real-time RTPCR experiments mirrored the modulatory actions of both IL-4 and IFN-␥ at the level of TL1A mRNA accumulation (data not shown).

The Journal of Immunology


associated macrophages (Fig. 6Ab). Interestingly, TL1A staining of synovial samples from RF⫺/RA patients was much less intense and characterized by paucicellular infiltrate, other than a TL1A-negative synovium (data not shown). Finally, mononuclear cells with monocyte/macrophage morphology present in the SF of RF⫹/RA patients were intensely stained by the antiTL1A Ab; the staining was confined to the cytoplasm and not detected at the membrane (Fig. 6B). PEG precipitates from serum or insoluble IC purified from SF of RF⫹/RA patients induce the release of sTL1A by monocytes The above findings lead us to speculate that serum or SF IC from RA patients might induce circulating/inflammatory monocytes or synovial macrophages to produce and secrete sTL1A. To verify this hypothesis, monocytes isolated from healthy sub-

Downloaded from by guest on January 12, 2018

FIGURE 6. TL1A expression in synovial tissue and synovial exudates of RA patients. A, TL1A expression in synovial tissue from a patient with osteoarthritis (a, upper left), and in synovial tissue (b– d and f) and skin (e) from patients with RA (b–f). Whereas in OA synovial tissue only few stromal macrophages are TL1A positive, in RA many positive cells are found both in the inflamed stroma, and near the surface of the tissue (b). Serial sections stained with anti-TL1A (c) and antiCD68 (d) show similar distributions of the positive cells, indicating that the TL1A⫹ cells are macrophages. Note the negativity of the lymphocytes within the nodule, as well as of endothelium. Macrophages forming the palisade in a typical rheumatoid granuloma are strongly TL1A⫹ (e). Double immunofluorescence for CD14 (red) and TL1A (green) reveals a colocalization of the two Ags in the vast majority of cells (f). B, Cytospins were prepared by spinning 15 ⫻ 105 SF cells per spot in a Shandon Cytospin 3 centrifuge. Mononuclear cells within the synovial fluid show cytoplasmic TL1A staining, but the neutrophils within this sample fail to stain.

jects were cultured for 20 h in standard medium containing different dilutions of PEG precipitates prepared from the serum of RF⫹/RA patients, RF⫺/RA patients, or healthy subjects. As shown in Fig. 7A, only PEG precipitates prepared from the serum of RF⫹/RA patients were able to stimulate, dose dependently, the release of very high amounts of sTL1A by monocytes, whereas the control RF⫺/RA precipitates did not. Similarly, insoluble IC purified from the SF of two RF⫹/RA patients also potently stimulated sTL1A release by monocytes (Fig. 7B). These insoluble SF IC were shown also to be potent activators of the neutrophil respiratory burst (data not shown), as previously described (12). Collectively, these data are consistent with the notion that, in RF⫹/RA patients, the production of sTL1A by mononuclear phagocytes is triggered by IC present in biological fluids (including SF).



Discussion TL1A (TNFSF15/VEGI) is a recently identified member of the TNF superfamily that interacts with DR3, costimulates T cells augmenting anti-CD3 plus anti-CD28 IFN-␥ and GM-CSF production, and, importantly, biases T cells to differentiate toward the Th1 phenotype (6, 8, 9, 27). However, although much is becoming known about TL1A function, very little is known about its expression in normal or pathological states. In this study, we have demonstrated that, of the various leukocytes examined, only monocytes can be induced to accumulate TL1A mRNA and release significant quantities of sTL1A in response to physiological concentrations (28, 29) of insoluble ICs. sTL1A released into medium by insoluble IC-stimulated monocytes was biologically functional because it specifically synergized with CHX in inducing the apoptosis of the TF-1 cells (6) and also cooperated with IL-12 plus IL-18 in inducing the production of IFN-␥ by CD4⫹ T cells (8). Although LPS also stimulated the release of monocyte-derived sTL1A, its effect was very weak compared with that of insoluble IC. Release of sTL1A by monocytes was slow, delayed in comparison with other inflammatory cytokines (such as TNF-␣), but was sustained over time and dependent on the concentration of the stimuli. The selective action of insoluble IC on sTL1A production was emphasized by the fact that a variety of classical mediators, including TNF-␣, GM-CSF, fMLP, CXCL8, IFN-␥, IFN␣, IL-4, IL-10, and IL-13, all failed to upregulate TL1A gene and protein expression. PBLs and neutrophils cultured with insoluble IC, LPS, or other proinflammatory stimuli were completely unable to release sTL1A or to express mTL1A (P. Scapini, F. Calzetti, and M. A. Cassatella, unpublished observations), further highlighting that monocytes are the major cellular source and insoluble IC the selective stimuli for sTL1A production. In contrast, expression of mTL1A in insoluble IC-stimulated monocytes was minimal, in line with the observations made in LPS-activated monocytes (30). Finally, we have also shown that the molecular interaction between BSA/anti-BSA insoluble IC and monocytes is mainly via Fc␥RII/CD32 (rather than CD64 or CD16), at least under our experimental conditions. Previous find-

Acknowledgment We thank Francesca Gentili for her excellent technical assistance.

Disclosures The authors have no financial conflict of interest.

References 1. Lipsky, P. E. 2005. Rheumatoid arthritis. In Harrison’s Principles of Internal Medicine 16th Ed. D. L. Kasper, E. Braunwald, A. S. Fauci, S. L. Hauser, and D. L. Longo, eds. McGraw Hill, San Diego, CA, pp. 1968 –1977. 2. Firestein, G. S. 2005. Immunologic mechanisms in the pathogenesis of rheumatoid arthritis. J. Clin. Rheumatol. 11: S39 –S44.

Downloaded from by guest on January 12, 2018

FIGURE 7. PEG precipitates from RF⫹/RA sera and SFs induce the production of sTL1A by monocytes. A, Serum samples (250 ␮l) from RF⫹/RA patients and healthy donors were subjected to PEG precipitation and then suspended in 250 ␮l of PBS. Dilutions (1/25, 1/10 and 1/5 v/v, in standard medium) of each precipitate were added to cultures of monocytes isolated from healthy donors. Similarly, insoluble IC purified from SF of two RF⫹/RA patients were added in a 250-␮l volume to cultures of monocytes isolated from healthy donors (B). After 20 h, monocyte-conditioned medium were harvested and centrifuged, and the resulting supernatants were subjected to sTL1A determination. Optical densities of the PEG precipitates were also determined and, if found above the blanks, they were subtracted from the optical densities of the corresponding monocyte-derived conditioned medium. Bars show the net values of sTL1A production by monocytes. Values show means of duplicate determinations calculated from one experiment representative of two for A and from the two SF patients in B.

ing reporting that Fc␥RIII␣/CD16 plays a key role in mediating the induction of both TNF-␣ and IL-1␣ production by human macrophages of RA patients following receptor ligation by small ICs (31) highlight the fact that the mechanisms governing the expression of TL1A and TNF-␣/IL-1␣ might be subjected to distinct regulatory pathways or depend on the cell type. Our in vitro observations prompted us to hypothesize whether sTL1A expression occurred in vivo in conditions such as RA, in which immune complexes and/or autoantibodies are elevated. In support of this hypothesis, we have demonstrated that monocytes incubated either with PEG precipitates prepared from the serum of RF⫹/RA patients, which are enriched for IC (32), or with insoluble IC purified from the SF of RF⫹/RA patients potently stimulated monocytes to release high amounts of sTL1A in vitro. Consistent with a role for monocytes as a TL1A source in RA, synovial tissue samples from active RA patients were characterized by a strong expression of TL1A in macrophages, in keeping with the results of Bamias et al. (9), who also found that the main source of TL1A in inflammatory bowel disease are monocyte-derived cells. Furthermore, synovial monocytes present in the SF of RF⫹/RA patients were intensely TL1A positive. Our data strongly favor the notion that, in RF⫹/RA patients, IC stimulate mononuclear phagocytes to produce and release sTL1A, either into the blood or locally, for example, into inflamed joints. We observed that the capacity of PBMC from RA patients to produce sTL1A when stimulated with insoluble IC in vitro was identical with that of normal PBMC (M. A. Cassatella, G. Pereira da Silva, I. Tinazzi, F. Calzetti, unpublished observations). Although the functional significance of sTL1A in RA remains to be clarified, one possibility is that TL1A contributes to sustain the inflammatory process. For example, in monocytic cell lines, recombinant TL1A used in association with IFN-␥ induces both CXCL8 and matrix metalloproteinase-9 production (30), two mediators implicated in local cell infiltration and joint degradation in RA patients (33, 34). Alternatively, given the ability of TL1A to modulate T lymphocyte functions, as demonstrated by its capacity to maintain mucosal Th1 polarization in patients with Crohn’s disease (9) or in two models of chronic murine ileitis (10), it is possible that the TL1A production by IC-stimulated monocytes might also be involved in directing and sustaining the various Th1-driven immunological responses in RA. In support of this notion, our experiments indicated that IL-4 and IFN-␥ suppress and enhanc, respectively, the production of sTL1A by insoluble IC-treated monocytes, which is exactly what one would expect if TL1A is regulated by the classical Th1-type cytokines. In conclusion, it is well known that there is a short window of opportunity in the early onset of RA, which, if identified and treated in a timely manner, could lead to a better prognosis (35). Therefore, understanding the early events and identifying the precise markers that predict aggressive synovitis would aid in the design of therapeutic strategies. Our data might have identified sTL1A as a novel therapeutic target.

The Journal of Immunology

18. Hayashi, F., T. K. Means, and A. D. Luster. 2003. Toll-like receptors stimulate human neutrophil function. Blood 102: 2660 –2669. 19. Scapini, P., B. Nardelli, G. Nadali, F. Calzetti, G. Pizzolo, C. Montecucco, and M. A. Cassatella. 2003. G-CSF-stimulated neutrophils are a prominent source of functional BLyS. J. Exp. Med. 197: 297–302. 20. Tecchio, C., V. Huber, P. Scapini, F. Calzetti, D. Margotto, G. Todeschini, L. Pilla, G. Martinelli, G. Pizzolo, L. Rivoltini, and M. A. Cassatella. 2004. IFN␣-stimulated neutrophils and monocytes release a soluble form of TNF-related apoptosis-inducing ligand (TRAIL/Apo-2 ligand) displaying apoptotic activity on leukemic cells. Blood 103: 3837–3844. 21. Yamashiro, S., H. Kamohara, and T. Yoshimura. 2000. Alteration in the responsiveness to tumour necrosis factor-␣ is crucial for maximal expression of monocyte chemoattractant protein-1 in human neutrophils. Immunology 101: 97–103. 22. Leonard, E. J., and A. Skeel. 1985. Disposable microliter immunoabsorbent columns: construction and operation. J. Immunol. Methods 82: 341–348. 23. Marafioti, T., M. Jones, F. Facchetti, T. C. Diss, M. Q. Du, P. G. Isaacson, M. Pozzobon, S. A. Pileri, A. J. Strickson, S. Y. Tan, et al. 2003. Phenotype and genotype of interfollicular large B cells, a subpopulation of lymphocytes often with dendritic morphology. Blood 102: 2868 –2876. 24. Cassatella, M. A. 1999. Neutrophil-derived proteins: selling cytokines by the pound. Adv. Immunol. 73: 369 – 453. 25. Lettesjo, H., E. Nordstrom, H. Strom, and E. Moller. 1998. Autoantibody patterns in synovial fluids from patients with rheumatoid arthritis or other arthritic lesions. Scand. J. Immunol. 48: 293–299. 26. Kageyama, Y., S. Miyamoto, T. Ozeki, M. Hiyohsi, M. Suzuki, and A. Nagano. 2000. Levels of rheumatoid factor isotypes, metalloproteinase-3 and tissue inhibitor of metalloproteinase-1 in synovial fluid from various arthritides. Clin. Rheumatol. 19: 14 –20. 27. Prehn, J. L., S. Mehdizadeh, C. J. Landers, X. Luo, S. C. Cha, P. Wei, and S. R. Targan. 2004. Potential role for TL1A, the new TNF-family member and potent costimulator of IFN-␥, in mucosal inflammation. Clin. Immunol. 112: 66 –77. 28. Theofilopoulos, A. N., C. B. Wilson, and F. J. Dixon. 1976. The Raji cell radioimmune assay for detecting immune complexes in human sera. J. Clin. Invest. 57: 169 –182. 29. Reckel, R. P., J. Harris, E. Botsko, R. Wellerson, and S. Varga. 1984. The detection of circulating immune complexes containing C1q and IgG using a new rapid serum ELISA test system. Diagn. Immunol. 2: 228 –237. 30. Kang, Y. J., W. J. Kim, H. U. Bae, D. I. Kim, Y. B. Park, J. E. Park, B. S. Kwon, and W. H. Lee. 2005. Involvement of TL1A and DR3 in induction of pro-inflammatory cytokines and matrix metalloproteinase-9 in atherogenesis. Cytokine 29: 229 –235. 31. Abrahams, V. M., G. Cambridge, P. M. Lydyard, and J. C. Edwards. 2000. Induction of tumor necrosis factor ␣ production by adhered human monocytes: a key role for Fc␥ receptor type IIIa in rheumatoid arthritis. Arthritis Rheum. 43: 608 – 616. 32. Ferraccioli, G., J. Karsh, and C. K. Osterland. 1983. Immunochemical analyses of components of immune complexes in the sera of patients with autoimmune diseases. J. Rheumatol. 10: 881– 888. 33. Kraan, M. C., D. D. Patel, J. J. Haringman, M. D. Smith, H. Weedon, M. J. Ahern, F. C. Breedveld, and P. P. Tak. 2001. The development of clinical signs of rheumatoid synovial inflammation is associated with increased synthesis of the chemokine CXCL8 (interleukin-8). Arthritis Res. 3: 65–71. 34. Burrage, P. S., K. S. Mix, and C. E. Brinckerhoff. 2006. Matrix metalloproteinases: role in arthritis. Front. Biosci. 11: 529 –543. 35. Westwood, O. M., P. N. Nelson, and F. C. Hay. 2006. Rheumatoid factors: what’s new. Rheumatology 45: 379 –385.

Downloaded from by guest on January 12, 2018

3. Scott, D. L. 2000. Prognostic factors in early rheumatoid arthritis. Rheumatology 39(Suppl. 1): 24 –29. 4. Steiner, G., M. Tohidast-Akrad, G. Witzmann, M. Vesely, A. Studnicka-Benke, A. Gal, M. Kunaver, P. Zenz, and J. S. Smolen. 1999. Cytokine production by synovial T cells in rheumatoid arthritis. Rheumatology 38: 202–213. 5. Maini, R. N., and P. C. Taylor. 2000. Anti-cytokine therapy for rheumatoid arthritis. Annu. Rev. Med. 51: 207–229. 6. Migone, T. S., J. Zhang, X. Luo, L. Zhuang, C. Chen, B. Hu, J. S. Hong, J. W. Perry, S. F. Chen, J. X. Zhou, et al. 2002. TL1A is a TNF-like ligand for DR3 and TR6/DcR3 and functions as a T cell costimulator. Immunity 16: 479 – 492. 7. Chew, L. J., H. Pan, J. Yu, S. Tian, W. Q. Huang, J. Y. Zhang, S. Pang, and L. Y. Li. 2002. A novel secreted splice variant of vascular endothelial cell growth inhibitor. FASEB J. 16: 742–744. 8. Papadakis, K. A., D. Zhu, J. L. Prehn, C. Landers, A. Avanesyan, G. Lafkas, and S. R. Targan. 2005. Dominant role for TL1A/DR3 pathway in IL-12 plus IL-18induced IFN-␥ production by peripheral blood and mucosal CCR9⫹ T lymphocytes. J. Immunol. 174: 4985– 4990. 9. Bamias, G., C. Martin, III, M. Marini, S. Hoang, M. Mishina, W. G. Ross, M. A. Sachedina, C. M. Friel, J. Mize, S. J. Bickston, et al. 2003. Expression, localization, and functional activity of TL1A, a novel Th1-polarizing cytokine in inflammatory bowel disease. J. Immunol. 171: 4868 – 4874. 10. Bamias, G., M. Mishina, M. Nyce, W. G. Ross, G. Kollias, J. Rivera-Nieves, T. T. Pizarro, and F. Cominelli. 2006. Role of TL1A and its receptor DR3 in two models of chronic murine ileitis. Proc. Natl. Acad. Sci. USA 30: 8441– 8446. 11. Cassatella, M. A., F. Bazzoni, R. M. Flynn, S. Dusi, G. Trinchieri, and F. Rossi. 1990. Molecular basis of interferon-␥ and lipopolysaccharide enhancement of phagocyte respiratory burst capability: studies on the gene expression of several NADPH oxidase components. J. Biol. Chem. 265: 20241–20246. 12. Fossati, G., R. C. Bucknall, and S. W. Edwards. 2002. Insoluble and soluble immune complexes activate neutrophils by distinct activation mechanisms: changes in functional responses induced by priming with cytokines. Ann. Rheum. Dis. 61: 13–19. 13. Sironi, M., F. O. Martinez, D. D’Ambrosio, M. Gattorno, N. Polentarutti, M. Locati, A. Gregorio, A. Iellem, M. A. Cassatella, J. Van Damme, et al. 2006. Differential regulation of chemokine production by Fc␥ receptor engagement in human monocytes: association of CCL1 with a distinct form of M2 monocyte activation (M2b, type 2). J. Leukocyte Biol. 80: 342–349. 14. Cassatella, M. A., L. Meda, S. Bonora, M. Ceska, and G. Constantin. 1993. Interleukin 10 (IL-10) inhibits the release of proinflammatory cytokines from human polymorphonuclear leukocytes: evidence for an autocrine role of tumor necrosis factor and IL-1 ␤ in mediating the production of IL-8 triggered by lipopolysaccharide. J. Exp. Med. 178: 2207–2211. 15. Arnett, F. C., S. M. Edworthy, D. A. Bloch, D. J. McShane, J. F. Fries, N. S. Cooper, L. A. Healey, S. R. Kaplan, M. H. Liang, H. S. Luthra, et al. 1988. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum. 31: 315–324. 16. Ronnelid, J., A. Tejde, L. Mathsson, K. Nilsson-Ekdahl, and B. Nilsson. 2003. Immune complexes from SLE sera induce IL10 production from normal peripheral blood mononuclear cells by an Fc␥RII dependent mechanism: implications for a possible vicious cycle maintaining B cell hyperactivity in SLE. Ann. Rheum. Dis. 62: 37– 42. 17. Robinson, J., F. Watson, R. C. Bucknall, and S. W. Edwards. 1992. Activation of neutrophil reactive-oxidant production by synovial fluid from patients with inflammatory joint disease: soluble and insoluble immunoglobulin aggregates activate different pathways in primed and unprimed cells. Biochem. J. 286: 345–351.


The Journal of Immunology

CORRECTIONS Zeini, M., P. G. Trave´s, R. Lo´pez-Fontal, C. Pantoja, A. Matheu, M. Serrano, L. Bosca´, and S. Hortelano. 2006. Specific contribution of p19ARF to nitric oxide-dependent apoptosis. J. Immunol. 177: 3327–3336. In Fig. 3B, the IAP-1 blot for wild-type (WT) cells is duplicated in error in the column for p53⫺/⫺ macrophages. The corrected Fig. 3 is shown below.

Rutjens, E., S. Mazza, R. Biassoni, G. Koopman, L. Radic, M. Fogli, P. Costa, M. C. Mingari, L. Moretta, J. Heeney, and A. De Maria. 2007. Differential NKp30 inducibility in chimpanzee NK cells and conserved NK cell phenotype and function in long-term HIV-1-infected animals. J. Immunol. 178: 1702–1712. The sixth institution in the author affiliations is incorrect. The corrected list is shown below. *Biomedical Primate Research Centre, Rijswijk, The Netherlands; †Centro di Eccellenza per la Ricerca Biomedica, Genoa, Italy; ‡Dipartimento di Medicina Sperimentale, Universita` di Genova, Genoa, Italy; §Istituto Scientifico Giannina Gaslini, Genoa, Italy; ¶Dipartimento di Oncologia Biologia e Ginecologia, Universita` di Genova, Genoa, Italy; 储Istituto Nazionale per la Ricerca sul Cancro, Genoa, Italy; and #Dipartimento di Medicina Interna, Universita` di Genova, Genoa, Italy

Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00



Germain, S. J., G. P. Sacks, S. R. Soorana, I. L. Sargent, and C. W. Redman. 2007. Systemic inflammatory priming in normal pregnancy and preeclampsia: the role of circulating syncytiotrophoblast microparticles. J. Immunol. 178: 5949 –5956. The third author’s last name is incorrect. The correct name is Suren R. Sooranna.

Cassatella, M. A., G. Pereira da Silva, I. Tinazzi, F. Facchetti, P. Scapini, F. Calzetti, N. Tamassia, P. Wei, B. Nardelli, V. Roschke, A. Vecchi, A. Mantovani, L. M. Bambara, S. W. Edwards, and A. Carletto. 2007. Soluble TNF-like cytokine (TL1A) production by immune complexes stimulated monocytes in rheumatoid arthritis. J. Immunol. 178: 7325–7333. The second author’s name is incorrect. The correct name is Gabriela Pereira-da-Silva.

Selvaraj, R. K., and T. L. Geiger. 2007. A kinetic and dynamic analysis of Foxp3 induced in T cells by TGF-␤. J. Immunol. 178: 7667–7677. Changes that the authors did not request were made in production after the authors returned their page proofs resulting in publication of the article with multiple errors. The editors and staff of The Journal of Immunology apologize to the authors and readers for this error. The entire article is reproduced correctly on the following pages in print only. The errors have been corrected in the online version, which now differs from the print version as originally published.

The Journal of Immunology

CORRECTIONS A Kinetic and Dynamic Analysis of Foxp3 Induced in T Cells by TGF-␤1 Ramesh K. Selvaraj and Terrence L. Geiger2 TGF-␤ induces Foxp3 expression in stimulated T cells. These Foxp3ⴙ cells (induced regulatory T cells (iTreg)) share functional and therapeutic properties with thymic-derived Foxp3ⴙ regulatory T cells (natural regulatory T cells (nTreg)). We performed a single-cell analysis to better characterize the regulation of Foxp3 in iTreg in vitro and assess their dynamics after transfer in vivo. TGF-␤ up-regulated Foxp3 in CD4ⴙFoxp3ⴚ T cells only when added within a 2- to 3-day window of CD3/CD28 stimulation. Up to 90% conversion occurred, beginning after 1–2 days of treatment. Foxp3 expression strictly required TCR stimulation but not costimulation and was independent of cell cycling. Removal of TGF-␤ led to a loss of Foxp3 expression after an ⬃4-day lag. Most iTreg transferred into wild-type mice down-regulated Foxp3 within 2 days, and these Foxp3ⴚ cells were concentrated in the blood, spleen, lung, and liver. Few of the Foxp3ⴚ cells were detected by 28 days after transfer. However, some Foxp3ⴙ cells persisted even to this late time point, and these preferentially localized to the lymph nodes and bone marrow. CXCR4 was preferentially expressed on Foxp3ⴙ iTreg within the bone marrow, and CD62L was preferentially expressed on those in the lymph nodes. Like transferred nTreg and in contrast with revertant Foxp3ⴚ cells, Foxp3ⴙ iTreg retained CD25 and glucocorticoid-induced TNFR family-related gene. Thus, Foxp3 expression in naı¨ve-stimulated T cells is transient in vitro, dependent on TGF-␤ activity within a highly restricted window after activation and continuous TGF-␤ presence. In vivo, a subset of transferred iTreg persist long term, potentially providing a lasting source for regulatory activity after therapeutic administration. The Journal of Immunology, 2007, 178: 7667–7677.


mmunological tolerance is achieved developmentally in the thymus as well as through peripheral mechanisms. CD4⫹ regulatory T cells (Treg)3 that express the forkhead transcription factor Foxp3 are critical for maintaining peripheral tolerance; their deficiency leading to early-onset, fatal autoimmune inflammation (1). Foxp3 expression is not only a marker for Treg, but appears to administer a developmental program endowing T cells with regulatory function. Thus, CD4⫹ T cells expressing retrovirally transduced Foxp3 display regulatory properties similar to endogenous Treg (2, 3). Treg are largely produced in the thymus (natural Treg (nTreg)) and constitute ⬃3– 6% of CD4⫹ T cells (4). More recent studies have shown that Foxp3 may also be induced in CD4⫹Foxp3⫺ T cells in vivo during some immune responses, or in vitro after stimulation of Foxp3⫺ cells in the presence of TGF-␤ (induced Treg (iTreg)) (5– 8). TGF-␤ is a critical cytokine for preserving immune homeostasis (9). TGF-␤-deficient mice or mice expressing dominant negative TGF-␤ receptors on T cells develop spontaneous, early-onset autoimmune disease (10, 11). This results both from cell autonomous

Department of Pathology, St. Jude Children’s Research Hospital, Memphis, TN 38105 Received for publication January 12, 2007. Accepted for publication April 9, 2007. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by National Institutes of Health Grant R01 AI056153 (to T.L.G.) and by the American Lebanese Syrian Associated Charities/St. Jude Children’s Research Hospital (to T.L.G. and R.K.S.). 2 Address correspondence and reprint requests to Dr. Terrence L. Geiger, Department of Pathology, St. Jude Children’s Research Hospital, 332 North Lauderdale Street, D-4047, Memphis, TN 38105. E-mail address: [email protected] 3 Abbreviations used in this paper: Treg, regulatory T cell; GITR, glucocorticoidinduced TNFR family-related gene; iTreg, induced Treg; nTreg, natural Treg; LN, lymph node; rhIL-2, recombinant human IL-2; CD62L, CD62 ligand.

Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00

effects of TGF-␤ deficiency on effector T cells and from defects in the Treg compartment. TGF-␤⫺/⫺ mice demonstrate impaired maintenance of Foxp3⫹ Treg, indicating that TGF-␤ plays a critical role in their homeostasis (12, 13). Although deficiency in TGF-␤ dominantly leads to an autoaggressive phenotype, the immunomodulatory role of TGF-␤ is likely complex. TGF-␤ also plays a crucial role in promoting the development of inflammatory Th17 cells and plays a supportive role in Th1 T cell development (14, 15). Several reports have demonstrated utility in manipulating disease states by altering Treg numbers or activity in animal models (16 –20). Increasing the number of Treg by adoptive transfer can diminish pathologic inflammation. Acquiring adequate numbers of Treg for treatment, however, represent a significant challenge due to the small number of nTreg present in vivo and their anergic state in vitro (21). An alternative and potentially simpler approach is to produce Foxp3⫹ iTreg from CD4⫹ Foxp3⫺ T cells by stimulation with TGF-␤, which may then be used as an immunotherapeutic surrogate for nTreg. Indeed, iTreg are able to suppress T cell responses in vitro (22), educate alloreactive CD4⫹CD25⫺ cells to be suppressive (5), and have shown significant potential in the treatment or prevention of graft rejection, colitis, and diabetes in animal models (23–25). The therapeutic applicability of iTreg will not only depend on their expression of Foxp3, but on other cellular characteristics. For instance, one leading hypothesis is that Treg development is guided by a high avidity for self-Ag (26, 27). iTreg, being derived from Foxp3⫺ T cells, lack this high avidity, which may influence their homeostatic or other properties. Differences between thymically derived nTreg and ex vivo-generated iTreg are not well studied. To better characterize iTreg, we have used single-cell analysis to assess the kinetics and sustainability of Foxp3 after induction with TGF-␤ in vitro and the cellular dynamics of iTreg in vivo. We demonstrate that iTreg development requires TGF-␤ exposure

CORRECTIONS within a narrow window after stimulation and that Foxp3 persistence requires continued exposure to TGF-␤. After adoptive transfer, iTreg predominantly and rapidly revert to Foxp3⫺ T cells. A subset of cells, however, retain Foxp3 for a longer time. These cells primarily localize to and possibly expand within the bone marrow and lymph nodes. We conclude that for the majority of iTreg, Foxp3 expression is transient and in vitro is highly dependent on exogenous TGF-␤ exposure. Some cells, however, develop stable expression of Foxp3 in vivo and phenotypically resemble nTreg.

Materials and Methods Animals Mice in which a GFP-Foxp3 fusion has been homologously inserted at the Foxp3 locus have been described (1) and were backcrossed more than five generations onto the C57BL/6J (Thy1.2⫹, CD45.2⫹) background before analysis. Male mice screened for GFP-Foxp3 were used for experimentation. Some GFP-Foxp3 mice were subsequently bred with CD45.1-congenic mice to obtain CD45.1⫹ GFP-Foxp3 mice. C57BL/6J mice and congenic CD45.1 (B6.SJL-Ptprca Pepcb/BoyJ) and Thy1.1 (B6.PL-Thy1a/CyJ) mice were purchased from The Jackson Laboratory. Experimentation was performed in accordance with institutional animal care and use procedures.

Media, reagents, and Abs Media for T cell cultures was prepared as described earlier (28). Unconjugated anti-CD3, anti-CD28, and anti-CD16/CD32 Fc block (2.4G2) and fluorochrome-conjugated anti-CD4 (L3T4), anti-CTLA-4, and anti-CXCR4 were purchased from BD Pharmingen. All other Abs used and recombinant human TGF-␤1 were purchased from eBioscience. Sulfate latex beads (Molecular Probes and Invitrogen Life Technologies) were coated with anti-CD3 (40 ␮g/ ml) or anti-CD28 (40 ␮g/ml) or anti-CD3/CD28 (13.3:26.6 ␮g/ml) as per the manufacturers’ instruction.

Cell purification and cell culture Lymph nodes (LN) and spleen cells were collected as described previously (28). CD4⫹Foxp3⫹ (nTreg) and CD4⫹Foxp3⫺ (non-Treg) cells were isolated by flow cytometric sorting on a MoFlo high-speed sorter (DakoCytomation) gating on CD4 and GFP (Foxp3) expression (28). Sorted cell purity ranged from 97 to 99%. nTreg were grown in medium supplemented with 1 ng/ml PMA, 200 ng/ml ionomycin, and 100 U/ml recombinant human IL-2 (rhIL-2; National Cancer Institute Biological Resources Branch Repository), which we found to optimally preserve Foxp3 expression. CD4⫹Foxp3⫺ cells were stimulated with antiCD3/ CD28 beads at a cell:bead ratio of 1:1 and supplemented with 100 U/ml rhIL-2 with or without TGF-␤ (10 ng/ml) for 7–9 days to obtain iTreg or activated Foxp3⫺ cells, respectively. Cells were split into cytokinecontaining medium, as needed, to prevent overcrowding.

Foxp3 regulation in CD4⫹Foxp3⫺ cells Sorted CD4⫹Foxp3⫺ cells (2.5 ⫻ 104 per well) were stimulated in 96-well plates in medium supplemented with anti-CD3/CD28-coated beads (cell: bead ratio of 1:1) and 100 U/ml rhIL-2, or as indicated. TGF-␤ (10 ng/ml) was added at day 0 or the indicated time after TCR stimulation. For Foxp3 reversion kinetics, cells were stimulated as described above and TGF-␤ was removed at the indicated time by removing the supernatant and washing with PBS three times before adding back IL-2-containing medium. To study Foxp3 up-regulation in memory T cells, CD4⫹CD44highCD45Rblow Foxp3⫺ cells and control CD4⫹Foxp3⫺ cells were isolated by flow cytometric sorting, and 2 ⫻ 104 cells/well cultured in 96-well plates in medium supplemented with anti-CD3/CD28-coated beads (cell:bead ratio of 1:1), 100 U/ml rhIL-2, with or without TGF-␤ (10 ng/ml). To prevent overcrowding in longer-term cultures, wells were examined and split every 3– 4 days into medium supplemented with the same cytokines (IL-2 with or without TGF-␤) as initially present. Quadruplicate samples for each condition were analyzed. Foxp3 analyses were performed every 24 h or as indicated by quantitative flow cytometry by measuring the GFP fluorescence.

PKH26 proliferation analysis Naive flow cytometrically purified CD4⫹Foxp3⫺ cells were dye labeled using the PKH26 red fluorescent cell linker kit (Sigma-Aldrich) as per the manufacturer’s instructions, except that 1 ⫻ 10⫺6 M PKH26 dye was used for 1 ⫻ 107 cells. Cells were cultured in medium supplemented with or

without TGF-␤ (10 ng/ml) and analyzed at the indicated times by flow cytometry.

T cell proliferation suppression assay Flow cytometrically purified CD4⫹CD25⫺ T cells from Thy1.1-congenic mice (responder cells) were mixed with either nTreg or iTreg at the indicated ratios and stimulated with anti-CD3/CD28 beads at a bead:cell ratio of 1:1. Cocultures established in quadruplicate were analyzed by flow cytometry at the indicated times. For analysis of revertant Foxp3⫺ iTreg populations, CD4⫹Foxp3⫺ cells were converted to iTreg and sorted to ⬃99% purity as described above. These iTreg were recultured with or without TGF-␤ for 5 days, at which time the cells were again flow-sorted for Foxp3⫹ or Foxp3⫺ cells, respectively. The iTreg or revertant Foxp3⫺ cells were then evaluated for their ability to suppress proliferation of CFSE-labeled naive T cell responders as previously described (29). Because the iTreg express GFP, the fluorescence emission spectrum of which cannot readily be distinguished from CFSE, Thy1.1-disparate responder cells were used and analysis of CFSE levels in Thy1.1-gated cells was performed.

Cytokine production profile of stimulated nTreg, iTreg, and non-Treg CD4⫹Foxp3⫹ and CD4⫹Foxp3⫺ cells were purified from GFP-Foxp3 knock-in mice and cultured as described above to obtain nTreg, iTreg, and Foxp3⫺ cells. At day 7, the expanded nTreg, iTreg, and non-Treg were flow cytometrically sorted to obtain ⬃99% pure Foxp3⫹ or Foxp3⫺ populations. Five ⫻ 104 cells were added to 96-well plates in 250 ␮l of medium in triplicate and restimulated with anti-CD3/CD28 Ab-coated beads at a 1:1 cell:bead ratio in the presence of 100 IU/ml rhIL-2. At 48 h, murine IL-2, IL-4, IL-5, IL-10, IL-17, IFN-␥, and TNF-␣ were measured in the cell culture supernatant by Bio-Plex according to the manufacturer’s instructions (Bio-Rad).

Migration and survival of iTreg after adoptive transfer CD4⫹Foxp3⫹ and CD4⫹Foxp3⫺ cells were purified from CD45.1⫹ GFPFoxp3 knock-in mice and cultured as described above to obtain CD45.1⫹ nTreg, CD45.1⫹ iTreg, and CD45.1⫹ Foxp3⫺ cells. At days 7–9, the expanded nTreg, iTreg, and non-Treg were flow-cytometrically sorted to obtain ⬃99% pure Foxp3⫹ of Foxp3⫺ populations. Four ⫻ 106 Foxp3⫹ nTreg, iTreg, or non-Treg were mixed with 1 ⫻ 106 Thy1.1⫹ CD4⫹ T cells before adoptive transfer by retro-orbital injection into age- and sexmatched Thy1.1⫺CD45.1⫺ C57BL/6J hosts. The Thy1.1⫹CD4⫹ T cells were purified from a LN and splenic cell suspension from Thy1.1⫹-congenic mice by anti-CD4 magnetic bead separation using the MACS separation system according to the manufacturer’s instructions (Miltenyi Biotec). Two recipient mice for each condition and time point measured were sacrificed at days 2, 5, 12, and 28 after injection, and peripheral blood, LN, spleen, lung, liver, and bone marrow were isolated. Cells were collected by forced passage through a cell strainer. Lymphoid components of liver and lung cell suspensions were further purified by centrifugation over 37.5% Percoll. Cells were stained with the indicated Abs before flow cytometric analysis.

Flow cytometry Cells were stained and analyzed on a FACSCalibur (BD Biosciences) using CellQuest software (BD Biosciences). Quantitative flow cytometry to determine total cell numbers was performed by enumerating all cells in a culture well. The presence of Foxp3 in cells from mice expressing the GFP-Foxp3 fusion was determined by measuring GFP fluorescence.

Statistics Data were analyzed by ANOVA using JMP (SAS).

Results TGF-␤ collaborates with TCR stimulation to up-regulate Foxp3 in CD4⫹ cells Previous studies (6 – 8, 30) have shown that TGF-␤ up-regulates Foxp3 in activated CD4⫹CD25⫺ T cells. To directly visualize this event, we analyzed the induction of Foxp3 in T cells from mice engineered to express a GFP-Foxp3 fusion protein. CD4⫹Foxp3⫺ T cells were purified, stimulated, and analyzed using quantitative flow cytometry. Similar to the results of others, TGF-␤ up-regulated Foxp3 in CD4⫹Foxp3⫺ T cells after CD3/CD28 stimulation,

The Journal of Immunology with conversion efficiencies of 50 –90% routinely observed (Fig. 1, A and B). Costimulation influenced the conversion. Total numbers of Foxp3⫹ cells were increased in the presence of anti-CD28. However, the percentage of Foxp3⫹ cells was equivalent in cultures stimulated with anti-CD3 in the presence or absence of anti-CD28, suggesting that costimulation acted by promoting the expansion of the cells rather than increasing the conversion efficiency (Fig. 1B). Foxp3 up-regulation was dependent on the presence of both TGF-␤- and TCR-specific signaling. As has been reported, in the absence of TGF-␤, little or no up-regulation of Foxp3 was apparent (Fig. 1, A and B, and not plotted; after anti-CD3 and anti-CD3/ CD28 stimulation, day 5 Foxp3⫹/Foxp3⫺ cell counts were 212/ 95,567 and 128/141,065, respectively). In the absence of TCR stimulation, little conversion was also observed. This effect has been previously noted using RT-PCR analysis for Foxp3 (8). However, cell viability is extremely poor in the absence of TCR stimulation and an alternative explanation for this finding is that the Foxp3⫹ cells have impaired survival without TCR stimulation. Indeed, the total number of surviving T cells cultured with IL-2/ TGF-␤ was only 1.1% of that observed in cultures also stimulated with anti-CD3/CD28. We found, however, that culture of CD4⫹ Foxp3⫺ cells in the presence of anti-CD28 but not anti-CD3 greatly improved cell viability, with similar numbers of live cells present at day 5 as at the start of culture. Here too though, few of the cells (mean ⫽ 1.0%) up-regulated Foxp3. Therefore TCR but not CD28 signaling synergizes with TGF-␤ to drive Foxp3 expression. Foxp3 up-regulation was further restricted to the naive T cell population and was not increased in isolated CD4⫹ CD44highCD45RblowFoxp3⫺ memory cells stimulated in the presence of TGF-␤ (Fig. 1C). TGF-␤-induced Treg suppress the proliferation of CD4⫹ target cells Earlier reports (7, 31) have shown iTreg, like nTreg, possess regulatory function and suppress CD4⫹ T cell proliferation in coculture experiments. We verified that our Foxp3⫹ iTreg were similarly capable of suppressing T cell expansion using quantitative flow cytometry (Fig. 2) as well as proliferation analysis of CFSElabeled responder cells (data not shown). In both studies, iTreg showed an efficiency similar to that of nTreg in suppressing T cell proliferation and expansion. Similar cytokine profile of iTreg and nTreg Because iTreg were as efficient as nTreg in suppressing T cell proliferation, we wanted to examine whether their cytokine production profiles were likewise comparable (Table I). Naive CD4⫹ T cells were stimulated and expanded for 7 days without TGF-␤ or converted into iTreg with TGF-␤. nTreg were likewise expanded. Foxp3⫹ (iTreg, nTreg) or Foxp3⫺ (non-Treg) cells were then flow-cytometrically sorted and stimulated. Both iTreg and nTreg demonstrated decreased production of most cytokines when compared with non-Treg, including IL-2, IL-4, IL-5, IFN-␥, and

FIGURE 1. TGF-␤ collaborates with TCR stimulation but not costimulation to up-regulate Foxp3. CD4⫹Foxp3⫺ T cells were flow-cytometrically purified and 3 ⫻ 104 distributed per well of a 96-well plate. The cells were stimulated with anti-CD3-, CD28-, or CD3/CD28 Ab-coated beads at a 1:1 cell:bead ratio. rhIL-2 and TGF-␤ were added at 100 U/ml and 10

ng/ml, respectively. On pretreatment day 0 or posttreatment day 5, wells were harvested and analyzed for expression of Foxp3 by quantitative flow cytometry. Mean pretreatment values for Foxp3⫺ and Foxp3⫹ T cells was 29,534 and 19, respectively, per well. A, Representative Foxp3 histogram plots are shown. B, Percent Foxp3⫹ cells is plotted. Mean absolute cell numbers are shown. C, Foxp3 is not up-regulated in memory cells. Cells were stimulated with anti-CD3/CD28-coated beads and IL-2 with or without TGF-␤. Total cell counts on day 7 for the different populations are listed within parentheses. B and C, Mean ⫾ SEM of quadruplicate samples are plotted.


FIGURE 2. iTreg suppress the proliferation of CD4⫹ target cells. Thy1.1⫹ CD4⫹CD25⫺ responder cells were mixed with congenic Thy1.1⫺ nTreg or iTreg in 96-well plates at the indicated ratios. Cells were stimulated with anti-CD3/CD28-coated beads at a cell:bead ratio of 1:1 for 3 days, and wells were then analyzed by quantitative flow cytometry for viable Thy1.1⫹ responder numbers. Mean ⫾ SEM of quadruplicate samples are plotted. Data are representative of two independent experiments.

TNF-␣. Although more IL-2 and IFN-␥ was produced by the iTreg than nTreg, this was significantly diminished when compared with non-Treg. In contrast, IL-10 was strongly produced by both the iTreg and nTreg. Therefore, TGF-␤-induced iTreg have a cytokine profile similar to that of nTreg, with strong expression of IL-10 and diminished expression of other effector cytokines. Kinetics of Foxp3 expression after anti-CD3/CD28 and TGF-␤ treatment Considering that memory T cells did not up-regulate Foxp3 in response to TGF-␤, (Fig. 1C), we were interested in defining the window period after activation during which T cells were susceptible to TGF-␤. Indeed, a time dependence for the generation of regulatory cells using TGF-␤ has been previously reported (32, 33). To test for Foxp3 induction, we stimulated CD4⫹Foxp3⫺ cells with anti-CD3/CD28 and IL-2, supplementing with TGF-␤ at different time points after stimulation (Fig. 3A). Two effects were notable. First, when TGF-␤ was provided at the time of TCR stimulation, up-regulation of Foxp3 protein only began after an ⬃2-day delay. Interestingly, if TGF-␤ supplementation was provided at later time points after TCR stimulation, Foxp3 up-regulation was delayed by a similar ⬃2-day period from the time TGF-␤ was administered. Second, treatment with TGF-␤ beginning up to 2 days after TCR stimulation had little impact on the ultimate percentage or number of Foxp3⫹ T cells in the culture. In contrast, cells treated with TGF-␤ on or after day 3 showed significantly ( p ⬍ 0.01) diminished conversion into Foxp3⫹ cells. When treatment began on day 3, a peak conversion of only ⬃20% of cells was Table I. Cytokine production pattern of nTreg, iTreg, and non-Treg cellsa Cytokine (pg/ml)




IL-2 IL-4 IL-5 IL-10 IL-17 IFN-␥ TNF-␣

3.9 0.8 0.5 578.5 2.3 3.6 2.7

249.9 7.1 3.1 2,135.0 19.2 487.9 4.8

11,542.4 3,614.6 46.8 501.4 27.6 2,544.7 36.2

a Five ⫻ 104 flow-cytometrically sorted cells of the indicated type from 7-day cultures were added to 96-well plates in 250 ␮l of medium in triplicate and stimulated with anti-CD3/CD28 Ab-coated beads at a 1:1 cell:bead ratio. Mean cytokine production measured 48 h after stimulation is shown.

FIGURE 3. Kinetics of Foxp3 expression after TGF-␤ induction. CD4⫹ Foxp3⫺ cells (2.5 ⫻ 104) were stimulated on day 0 with anti-CD3/CD28coated beads in a 96-well plate at a cell:bead ratio of 1:1 in the presence of 100 U/ml rhIL-2. TGF-␤ (10 ng/ml) was added either at day 0 or at the time denoted. Sample wells were harvested on different days and analyzed by quantitative flow cytometry for Foxp3 expression by measuring GFP fluorescence. Wells were split every 3– 4 days in medium with cytokine to prevent overcrowding. Mean ⫾ SEM of percentage of Foxp3⫹ cells (A) and total cell numbers (B) of quadruplicate samples are plotted. Data are representative of three independent experiments.

observed compared with ⬃80% with day 0 treatment. This difference did not result from an outgrowth of Foxp3⫺ T cells because total cell numbers were similar in the different treatment groups (Fig. 3B). Indeed, quantitative analysis demonstrated that absolute numbers of Foxp3⫹ cells were significantly ( p ⬍ 0.01) higher in the cells treated with TGF-␤ starting days 0 –2 compared with the cells treated after day 2. Therefore, TGF-␤/TCR stimulation has a limited window during which it can up-regulate Foxp3, and TGF-␤ supplementation leads to up-regulation of Foxp3 protein only after a significant (⬃2-day) lag period. TGF-␤- and CD3-induced Foxp3 up-regulation is independent of cell cycling Differentiation of naive T cells into Th1 and Th2 cell types occurs only after multiple rounds of cell cycling, an event believed to be required to relieve epigenetic repression of lineage-specific genes (34). After a T cell is stimulated through the TCR, it begins to cycle ⬃2 days after stimulation, consistent with the time frame for TGF-␤-mediated up-regulation of Foxp3 (Fig. 3A). We were therefore interested whether Foxp3 up-regulation only occurred in T cells that had divided. To test this, we labeled CD4⫹Foxp3⫺ cells with the cell membrane-associated red fluorescent dye PKH26 and stimulated them with anti-CD3/CD28 with or without TGF-␤ (Fig. 4). The fluorescence of PKH-26 is diminished with each cell division. In the absence of TGF-␤, Foxp3 expression was not observed in divided or undivided cells. In contrast, Foxp3 was upregulated in all cell populations treated with TGF-␤, including

The Journal of Immunology

FIGURE 4. TGF-␤-induced Foxp3 up-regulation is independent of cell cycling. Purified CD4⫹Foxp3⫺ cells were labeled with PKH26, washed, and stimulated with anti-CD3/CD28-coated beads at a cell:bead ratio of 1:1 in the presence of 100 U/ml rhIL-2 with or without 10 ng/ml TGF-␤. Flow cytometric analysis of cells immediately after culture (day 0) and at days 2 and 3 are shown. Each condition was analyzed in triplicate with essentially identical results. Data are representative of two independent experiments.

those that had not divided. This demonstrates that Foxp3 up-regulation and iTreg development does not require cell cycling. Limited persistence of Foxp3 in iTreg after TGF-␤ withdrawal To determine whether iTreg require continuous exposure to TGF-␤ to preserve Foxp3 expression, we stimulated CD4⫹ Foxp3⫺ cells with anti-CD3/CD28 and TGF-␤, and withdrew TGF-␤ at various time points after stimulation (Fig. 5A). Foxp3 expression persisted at high levels in cells that received continuous TGF-␤ treatment for the 13 days of culture. In contrast, Foxp3 expression returned to baseline by day 13 in cells that had TGF-␤ removed after initial treatment. After TGF-␤ was withdrawn, Foxp3 persisted for another 3– 4 days, at which point expression began to decrease ( p ⫽ 0.013). Expression was virtually undetectable in the T cells by 9 days after TGF-␤ withdrawal. Cells that were exposed to ⬍3 days of TGF-␤ lost Foxp3 at a much higher rate than those exposed for at least 3 days. Because total cell numbers in the wells were similar with or without withdrawal of TGF-␤ and total numbers of cells steadily increased with culture time (Fig. 5B), the loss of Foxp3 seemed to be primarily due to the conversion of the Foxp3⫹ cells to Foxp3⫺ cells rather than cell death. This conversion was confirmed in separate experiments in which iTreg purified by flow cytometric sorting for GFP-Foxp3 and then cultured without exogenous TGF-␤ displayed a similar loss of Foxp3 (data not shown). When sorted Foxp3⫹ iTreg were allowed to lose Foxp3 expression, Foxp3 could not be reinduced by restimulation, even in the presence of TGF-␤ (data not shown). Therefore, Foxp3 can only be induced in a brief window as naive T cells differentiate into effector/memory cells. Foxp3 persistence is dependent on sustained TGF-␤ signaling. Loss of suppressive potency in iTreg that have down-regulated Foxp3 Considering that removal of TGF-␤ resulted in loss of Foxp3 in iTreg (Fig. 5A), we were interested in whether loss of Foxp3 also

FIGURE 5. Limited persistence of Foxp3 in iTreg after TGF-␤ withdrawal. Purified CD4⫹Foxp3⫺ cells were stimulated with anti-CD3/CD28coated beads at a cell:bead ratio of 1:1 in the presence of 100 U/ml rhIL-2. No or 10 ng/ml TGF-␤ was added at day 0. TGF-␤ was removed by washing the cells at the indicated time. To prevent overcrowding during the extended culture, cells were split every 3– 4 days into medium with IL-2 and with or without TGF-␤. Cells were analyzed by quantitative flow cytometry at the indicated time points. Mean ⫾ SEM of percentage of Foxp3⫹ cells (A) and total cell numbers (B) of quadruplicate samples are plotted. Data are representative of three independent experiments.

leads to a loss of suppressive activity. To test this, TGF-␤ was either added to or excluded from cultures of flow-cytometrically purified Foxp3⫹ iTreg. Five days later, Foxp3⫹ or Foxp3⫺ cells from the respective cultures were flow-cytometrically isolated. The iTreg or revertant Foxp3⫺ cells were then added to naive, Thy1.1disparate, CFSE-labeled T cells, and the proliferation of the naive population to anti-CD3/CD28 was measured by loss of CFSE (Fig. 6). As in Fig. 2, iTreg that retained Foxp3 strongly suppressed naive cell proliferation. In contrast, iTreg that lost Foxp3 showed a substantially reduced ability to suppress T cell proliferation. Thus, loss of Foxp3 expression is accompanied by a loss of suppressive activity. Migration and survival of iTreg after adoptive transfer Our in vitro studies suggested that the persistence of iTreg is dependent upon exogenous TGF-␤ and that Foxp3 is lost within a several-day period after TGF-␤ removal. The transient nature of Foxp3 expression in iTreg implies that iTreg would not be suitable for immunotherapeutic application. Yet, studies have now documented that iTreg are effective in treating model alloimmune and autoimmune diseases (23–25). To determine whether iTreg persist in vivo, we adoptively transferred flow-cytometrically purified GFP-Foxp3⫹ iTreg derived from CD45.1⫹ mice into CD45.1⫺ congenic recipients and followed their migration and survival (Fig. 7). As controls, equivalent numbers of either Foxp3⫺ cells from


FIGURE 6. Suppression of naive T cell proliferation by Foxp3⫺ revertant iTreg. Thy1.1⫺ iTreg were flow-cytometrically sorted for Foxp3⫹ cells 7 days after induction with TGF-␤. These cells were recultured for 5 days with or without TGF-␤. The cells grown with TGF-␤ were then sorted again for expression of Foxp3, whereas the revertant Foxp3⫺ cells were sorted from the population grown in the absence of TGF-␤. Thy1.1⫹ naive T cells were CFSE labeled, mixed with the Foxp3⫹ or Foxp3⫺ populations at the indicated ratios, and stimulated with anti-CD3/CD28-coated beads. At 72 h, cultures were stained for Thy1.1, and CFSE expression on the Thy1.1⫹ population was analyzed by flow cytometry.

cultures stimulated in the absence of TGF-␤ or Foxp3⫹ nTreg from similarly expanded cultures were also transferred into naive mice. The CD45.1-congenic background permitted identification of the transferred cells regardless of their expression of GFP-Foxp3. To control for adoptive transfer efficiency among mice, control freshly purified Thy1.1⫹CD4⫹ cells were admixed at a 1:4 ratio with each of the adoptively transferred cell populations before transfer into congenic Thy1.1⫺CD45.1⫺ C57BL/6 hosts (Fig. 7C). At selected time points mice were sacrificed, cell suspensions were prepared from different organs, and the cells were analyzed by flow cytometry for CD4, Thy1.1, CD45.1, and Foxp3 expression. Similar numbers of Thy1.1⫹ cells were routinely observed in the recipient mice, indicating equivalent adoptive transfer efficiencies (data not shown). CD4⫹ cells were gated to distinguish the CD45.1⫹ transferred population and CD45.1⫺ host cells, and the transferred cells were then further segregated into Foxp3⫹ and Foxp3⫺ populations (Fig. 8, A and B). Numbers of CD45.1⫹ Foxp3⫹ or Foxp3⫺ cells were normalized to endogenous CD4⫹ CD45.1⫺Thy1.1⫺ cell numbers and plotted (Fig. 8C).

FIGURE 7. Migration and survival of iTreg after adoptive transfer. A, Scheme for the preparation and adoptive transfer of iTreg, nTreg, or nonTreg. B, Flow cytometric analysis of expanded iTreg, nTreg, and non-Treg populations before flow purification. C, Flow cytometric analysis of adoptively transferred cell populations, which include purified CD4⫹Thy1.1⫹ injection controls and CD4⫹CD45.1⫹Thy1.1⫺Foxp3⫹ or ⫺ experimental populations. Data are representative of three independent experiments.

Transferred flow-cytometrically purified Foxp3⫹ iTreg largely disappeared within the first 2 days after adoptive transfer (Fig. 8C). This seemed to result from down-modulation of Foxp3 since large numbers of CD4⫹CD45.1⫹Foxp3⫺ cells were simultaneously observed in several organs, including the spleen, liver, blood, and lung. Indeed, in support of this interpretation, numbers of Foxp3⫺ cells detected after transfer of purified Foxp3⫹ iTreg were similar to those observed after adoptive transfer of equal numbers of purified CD4⫹Foxp3⫺ non-Treg T cells. Transferred CD4⫹Foxp3⫺ cells were also found in similar locations as Foxp3⫺ former iTreg, specifically the spleen, liver, blood, and lung, within the first week after adoptive transfer. By 4 wk after transfer, virtually all of the Foxp3⫺ cells from either the iTreg or control non-Treg transfer had disappeared. One organ where Foxp3 was retained on the transferred iTreg was the LN. Indeed, of the organs analyzed at day 2 after iTreg transfer, greater numbers of transferred Foxp3⫹ cells then Foxp3⫺ cells were only detected within the LN. By 2– 4 wk after iTreg transfer, when the revertant Foxp3⫺ cells had largely disappeared, increasing numbers of transferred Foxp3⫹ cells were observed, particularly in the bone marrow and LN. Interestingly, only small numbers of Foxp3⫹ cells remained in the spleen, suggesting differential localization or inadequate support of Foxp3 expression at this site. Because the transferred

The Journal of Immunology

FIGURE 8. Migration and survival of iTreg, nTreg, and non-Treg after adoptive transfer. A, iTreg, nTreg, and non-Treg prepared as in Fig. 7 were adoptively transferred by retro-orbital injection into C57BL/6J mice. After 2 days, the mice were sacrificed, and cells were isolated and stained with CD45.1, Thy1.1, and CD4. Sample flow cytometry plots from forward scatter/side scatter-gated LN cells are shown, demonstrating identification of the transferred CD4⫹CD45.1⫹ cells among endogenous CD45.1⫺ cells. Numbers in the right side quadrants indicate percentage of CD45.1⫹ and CD45.1⫺ cells among the total CD4⫹ population. B, Histogram plots of gated CD4⫹CD45.1⫹ cells (oval gates in A) analyzed for the expression of Foxp3 are shown. C, Mice were sacrificed at the indicated time points and analyzed for the presence of adoptively transferred cells. Numbers of adoptively transferred CD4⫹CD45.1⫹ Foxp3⫹ or Foxp3⫺ cells were normalized to 25,000 host CD4⫹CD45.1⫺Thy1.1⫺ cells in the indicated organ and plotted. Mean ⫾ SEM of two mice per condition are plotted. Data are representative of three independent experiments.

CD45.1⫹ cells were ⬃99% Foxp3⫹ (Fig. 7C), this indicates that a subset of iTreg persisted and possibly expanded, surviving for at least 4 wk and primarily localizing within the bone marrow and LN. The survival and localization of iTreg showed significant differences though also similarities with that of adoptively transferred preactivated nTreg. nTreg retained their Foxp3 to a much greater extent than iTreg. All organs analyzed showed a predominance of Foxp3⫹ cells after nTreg transfer, yet this was only observed for the LN after iTreg transfer. Similar to the Foxp3⫹ cohort after iTreg transfer, the nTreg persisted 4 wk after transfer and at late time points also resided predominantly in the LN and bone marrow. One additional feature of nTreg transfers was that despite the transfer of equivalent numbers of nTreg as non-Treg, fewer nTreg were detected in the different organs than non-Treg. For example ⬃3,000 and 4,000 non-Treg were identified in the LN and spleen per 25,000 endogenous CD4⫹ T cells on day 2 after transfer, whereas ⬃400 and 130 nTreg were detected, respectively. This may reflect an increased rate of death or decreased proliferation of the preactivated nTreg vs non-Treg after transfer, or alternatively differential migration to other organs not analyzed. In summary, transferred Foxp3⫹ iTreg yielded large numbers of Foxp3⫺ cells that entered the blood, lungs, and liver and then disappeared by 2– 4 wk after transfer. A portion of iTreg, however, retained Foxp3. These persevered initially in the LN and then, as for nTreg, increasingly in the bone marrow. Thus, although many iTreg act in

vivo as in vitro, losing Foxp3 after removal from TGF-␤, a subset of these cells persist longer as Foxp3⫹ iTreg. Transferred iTreg phenotypically resemble nTreg nTreg constitutively express several T cell activation markers, including CD25 and glucocorticoid-induced TNFR family-related gene (GITR) (35). These markers were also nearly uniformly upregulated on both iTreg and activated non-Treg before adoptive transfer (Table II).

Table II. Surface marker expression on iTreg, nTreg, and non-Treg cellsa Percentage of Cells Naive

In vitro stimulated

Surface Marker







89.9 88.6 38.2 2.2

2.2 5.1 55.2 1.3

100 99.3 83.2 2.9

98.3 96.6 26.0 2.8

99.9 98.8 29.8 1.9

a Freshly isolated CD4⫹Foxp3⫺ or Foxp3⫹ cells were stimulated as described for 7– 8 days to obtain Foxp3⫹ iTreg, Foxp3⫺ non-Treg, and Foxp3⫹ nTreg. Expression of the indicated markers before and after culture is shown.


FIGURE 9. Transferred iTreg phenotypically resemble nTreg. CD4⫹ CD45.1⫹ adoptively transferred iTreg, nTreg, and non-Treg were identified within the LN of recipient mice and analyzed for Foxp3 expression as in Figs. 7 and 8. These were additionally analyzed for surface expression of CD25 (A) or GITR (B). Percentage of CD4⫹CD45.1⫹ cells of the indicated cell type positive for CD25 or GITR is plotted. Data points are missing for some cell types/conditions at some time points due to an inadequate number of cells in the organs at those times for definitive expression analysis. Sample histogram plots from the day 2 time point of transferred iTreg populations in LN that lost Foxp3 (left) or retained Foxp3 (right) are also shown. Mean ⫾ SEM of two mice per condition are plotted. Data are representative of two independent experiments.

To determine whether expression of CD25 and GITR persisted after adoptive transfer, cells from mice receiving activated and expanded iTreg, nTreg, or non-Treg were analyzed for CD25 and GITR expression. Foxp3⫹ iTreg, like nTreg, retained CD25 and GITR expression in the LN (Fig. 9, A and B) and all other organs analyzed (data not shown) throughout the 28-day time course of the experiment. In contrast, the majority

FIGURE 10. CXCR4 is preferentially expressed on adoptively transferred cells in the bone marrow and CD62L on cells in the LN. Data were acquired and analyzed as in Figs. 7 and 8. A, Percentage of indicated adoptively transferred populations positive for CXCR4 on day 5 after transfer is plotted. Sample histogram plots demonstrating CXCR4 expression on Foxp3⫹ iTreg in the LN or bone marrow (BM) are also shown. B, Percentage of the indicated adoptively transferred population positive for CD62L on day 5 or day 12 after transfer is plotted. Sample histogram plots demonstrating CD62L expression on Foxp3⫹ iTreg in the LN or bone marrow are also shown. Mean ⫾ SEM of two mice per condition are plotted. Data are representative of two independent experiments. ND, No data.

of either transferred non-Treg or revertant iTreg-derived Foxp3⫺ cells lost CD25 and GITR expression within the first 2–5 days after transfer. The small fraction of nTreg that lost

The Journal of Immunology Foxp3 (Fig. 8) also lost surface expression of CD25 and GITR (data not shown). This indicates that Foxp3⫹ iTreg phenotypically resemble nTreg in vivo, while the revertant Foxp3⫺ iTreg resemble Foxp3⫺ cells. Foxp3 protein expression is associated with the sustained expression of CD25 and GITR in vivo. It was interesting that the greatest concentration of transferred activated Foxp3⫹ iTreg or nTreg were found in the bone marrow and LN. Bone marrow localization of transferred nTreg has been previously shown to be associated with expression of the CXCR4 chemokine receptor, which is also expressed on other cell types localizing within the marrow (36). Indeed, blockade of CXCR4 can mobilize cells from the marrow into the blood (37). Analogously, nTreg expression of CD62L, a receptor for glycan ligands expressed on high endothelial venules, is associated with LN homing of nTreg as well as other T cells (38). To determine whether receptor expression correlates with tissue localization of iTreg, as it does for nTreg, we analyzed CXCR4 and CD62L expression on the transferred cells. At the time of transfer, ⬍3% of iTreg expressed CXCR4 and only 26% expressed CD62L (Table II). By 5 days after transfer, ⬃90% of Foxp3⫹ iTreg or nTreg in bone marrow expressed CXCR4, while the same population in other organs showed very little CXCR4 expression (Fig. 10A). Similar results were seen at later time points (data not shown). This is consistent with CXCR4 being the bone marrow homing receptor for iTreg, as for nTreg. At day 5, CD62L was discriminately expressed on cells within the LN; however, the differential expression was not as dramatic as for CXCR4 (Fig. 10B). By day 12, segregation of cells based on CD62L expression was more prominent. Approximately 80% of LN Foxp3⫹ cells (both nTreg and iTreg) expressed CD62L at day 12 (Fig. 10C), whereas only ⬃10% of Foxp3⫹ cells in the bone marrow expressed CD62L. This indicates that the CD62L⫹ Foxp3⫹ iTreg population preferentially localizes to LN. Thus, adoptively transferred iTreg segregated into populations with distinct localization patterns that correlate with the expression of specific chemokine receptors or adhesion molecules.

Discussion It is now well established that TGF-␤ can up-regulate Foxp3 in CD4⫹Foxp3⫺ T cells. Using mice in which a GFP-Foxp3 fusion protein has been homologously inserted into the Foxp3 locus, we have been able to assess on a single-cell basis the kinetics of upregulation of Foxp3 in vitro, as well as the maintenance of Foxp3 expression in vitro and in vivo. Our in vitro data are consistent with and adds to previous results. Foxp3 protein was up-regulated only after ⬃48 h of TGF-␤ treatment. This lag period in single-cell protein expression corresponds to a similar ⬃48-h lag in Foxp3 mRNA expression observed in human T cells (7). TGF-␤-induced Smad activation and nuclear translocation occurs within minutes, and SMAD-DNA complexes can be observed as early as 10 min after TGF-␤ addition (39). This ⬃2-day delay suggests that Foxp3 up-regulation involves a complex developmental program rather than an immediate effect of SMAD-DNA interactions. Indeed, recent identification of the involvement of CTLA-4 and cbl-b in TGF-␤-induced Foxp3 up-regulation (40, 41) supports the idea that complex combinatorial or sequential signals are involved in Foxp3 induction. We further find that the lag in Foxp3 expression is independent of the time from initial TCR stimulation as when TGF-␤ is added 24 or 48 h after stimulation, Foxp3 up-regulation shows an identical ⬃2-day delay. This implies that a TCR-dependent factor is not rate-limiting in inducing Foxp3 protein expression, but rather TGF-␤ induces a sequence of events requiring this time frame to up-regulate Foxp3. Up-regulation does not require T cell cycling

and is therefore unlikely to involve cell cycle-dependent modifications. We find that TCR stimulation is needed for Foxp3 expression. Viability of naive T cells cultured in the absence of TCR stimulation is poor and this viability loss may prevent examination of Foxp3 up-regulation in unstimulated populations. By stimulating cells with anti-CD28 in the absence of anti-TCR Ab, we were able to preserve viability and thereby clearly demonstrate this requirement. At 48 –72 h after TCR stimulation, however, T cells become refractory to the effects of TGF-␤. Therefore, a TCR-induced program is initially required for, although later restricts the ability of TGF-␤ to up-regulate Foxp3. Previous studies have led to different conclusions on the role of CD28 costimulation in Foxp3 up-regulation. One study showed a marked diminishment in Foxp3 mRNA up-regulation after TGF-␤ induction of peripheral T cells stimulated in the presence of antiCD28 Abs (7). In contrast, a second study found that CD28 costimulation was important for Foxp3 expression in TGF-␤-treated CD4⫹CD25⫺ thymocytes, primarily by enhancing cell survival (42). In multiple experiments using highly purified Ab-stimulated T lymphocytes, we failed to observe either a beneficial or deleterious role for CD28 costimulation in Foxp3 up-regulation. The percentage conversion of Foxp3⫺ T cells to Foxp3⫹ cells was similar regardless of the presence of costimulation. However, costimulation increased total cell numbers of both Foxp3⫺ and Foxp3⫹ cells, suggesting that in our system it primarily promotes cell survival or proliferation rather than conversion per se. One important role of CD28 costimulation is the induction of IL-2 production in activated T cells (43). Since our cultures contained exogenous IL-2, which is an important facilitator of Foxp3 induction, this role of costimulation on Foxp3 up-regulation may have been masked in our system. Indeed, recent studies have emphasized the requisite role for IL-2 in Foxp3 up-regulation and in one case supported a role for CD28-induced IL-2 in this regard (44, 45). We further analyzed the sustainability of Foxp3 after removal of TGF-␤. Interestingly, continued Foxp3 expression in vitro requires continued exposure of iTreg to exogenous TGF-␤. Foxp3 is lost from cells beginning several days after TGF-␤ removal and is eventually fully lost. Whether endogenous TGF-␤ expression can replace this exogenous TGF-␤ in sustaining Foxp3 is unclear. In our system, this was not the case. Data showing that TGF-␤-induced T cells can educate naive T cells to develop suppressive properties in a TGF-␤- and IL-10-dependent manner, however, suggest that under some conditions a self-sustaining process promoting continued Treg development may take place (5). In those experiments, persistent Ag stimulation was required to sustain regulatory function and Foxp3 expression, and it may be hypothesized that this difference in stimulation conditions results in the induction of TGF-␤. The down-regulation of Foxp3 in iTreg was accelerated when the iTreg were adoptively transferred in vivo. By day 2 after transfer, few of the cells retained Foxp3. Interestingly, the tissue localization of Foxp3⫺ former-iTreg was similar to that of transferred control Foxp3⫺ T cells and different from that of transferred Foxp3⫹ nTreg (Fig. 8). The reversion of iTreg into Foxp3⫺ T cells has potentially significant implications for their therapeutic utility because it may limit the intrinsic regulatory activity of the transferred cells. Indeed, we observed that iTreg that lost Foxp3 had diminished regulatory activity in a T cell proliferation suppression assay. In contrast to transferred iTreg, transferred activated nTreg showed limited loss of Foxp3 expression. Therefore iTreg and nTreg, although similar in suppressive function in vitro, differentially preserve Foxp3 after adoptive transfer.

CORRECTIONS Maintenance of nTreg requires several factors, including TGF-␤ and IL-2, in vivo (13, 46, 47). nTreg that lack IL-2R or that are from TGF-␤⫺/⫺ mice have a diminished peripheral life span. The differential maintenance of Foxp3 in nTreg and iTreg may suggest that transferred nTreg and iTreg have different access to sites where these cytokines are produced. An alternative or complementary possibility may lie in the distinct TCR repertoire of iTreg and nTreg. nTreg contain a skewed representation of TCR that is biased toward high affinity for self when compared with Foxp3⫺ T cells (27, 48). This self-specificity may alter the interactions of nTreg with resident APC compared with Foxp3⫺ T cells, leading to the distinct homeostatic properties of the different cell types. Because iTreg are derived from Foxp3⫺ T cells, we conjecture that they will lack the inherent self-specificity present in nTreg. Further studies analyzing the repertoire of stable Foxp3⫹ iTreg vs revertant Foxp3⫺ cells are ongoing to explore this possibility. Despite the loss of Foxp3 in the majority of transferred iTreg, a proportion of cells maintain Foxp3. Interestingly these cells concentrated primarily in two sites, the bone marrow and the LN. The number of transferred Foxp3⫹ cells increases in these locations over a 4-wk time period, suggesting that they expand there. Indeed, by day 28 after transfer, the numbers of Foxp3⫹ iTreg in the bone marrow and LN are equivalent to those of similarly transferred nTreg. Cellular localization of Foxp3⫹ T cells follows adhesion molecule expression. CXCR4, which is associated with bone marrow localization in a variety of cell types, is prominently expressed in the bone marrow resident population. Indeed, recent data have demonstrated that purified and transferred CXCR4⫹ nTreg primarily localize to the bone marrow (49). Likewise, we find that the majority of Foxp3⫹ iTreg inhabiting the LN are CD62Lhigh, a marker with a well-established role in LN homing (38). It would be anticipated that the distribution of adoptively transferred iTreg should mimic that of endogenous Treg. Endogenous Treg freely distribute, although modestly different proportions may be detected in different lymphoid compartments. However, even 1 mo after transfer, we observed a skewed distribution of transferred cells to the LN and bone marrow. Our data on CXCR4 and CD62L expression suggest that this is due to preferential homing and localization to these organs; however, preferential survival or expansion of Foxp3⫹ cells in these sites cannot be excluded. Interestingly, few nTreg or iTreg expressed CXCR4 before transfer (Table II). Whether de novo up-regulation of this receptor occurred in vivo or the localization of the transferred cells was predetermined by their receptor expression after in vitro stimulation will be important to examine. The homing and localization of adoptively transferred therapeutic populations may impact their therapeutic qualities and understanding these processes will be important for any future application of in vitro-expanded regulatory cells. Like nTreg, iTreg that retained Foxp3 after adoptive transfer also maintained CD25 and GITR surface expression. Our results suggest that these Foxp3⫹ cells, which were primarily found in the LN and bone marrow, play an important role in the amelioration of immunopathologic diseases after iTreg transfer. However, defining the role of the different cell populations in regulating disease will be important and it remains to be determined whether revertant Foxp3⫺ cells have regulatory properties as well. Analyzing this may be complex since the transferred cells may not uniquely possess regulatory function. In one study, transferred iTreg were able to induce regulatory activity in endogenous populations of T cells (23). We have observed a similar infectious tolerance with transferred nTreg (50). In summary, we demonstrate that iTreg are dependent on TGF-␤ signaling for the preservation of Foxp3 expression. Loss of Foxp3 occurs rapidly after in vivo transfer, with a predominance of

Foxp3⫺ cells appearing within 2 days. The brief life span of Foxp3 in most transferred iTreg contrasts with that of transferred nTreg, which largely preserve Foxp3 expression. Although a subset of Foxp3⫹ iTreg is maintained, the distinct cellular dynamic properties of the iTreg and nTreg populations may lead to differential effects on immune responses, and this, in addition to the impact of the large numbers of Foxp3⫺ cells forming after iTreg transfer, must be considered in potential applications of the different cell types. iTreg, although phenotypically homogeneous after expansion in vitro, further segregate into different populations based on their expression of different homing/adhesion molecules, including CXCR4 and CD62L. Deciphering the role of the multiple cell types generated after iTreg transfer in the induction and preservation of immune tolerance will be important for understanding their therapeutic potential.

Acknowledgments We thank Richard Cross and Jennifer Smith for assistance with flow cytometry, cell sorting, and Bio-Plex assays; Phuong Nguyen and Rajshekhar Alli for technical support; and Sasha Rudensky for providing the GFPFoxp3 knock-in mice.

Disclosures The authors have no financial conflict of interest.

References 1. Fontenot, J. D., J. P. Rasmussen, L. M. Williams, J. L. Dooley, A. G. Farr, and A. Y. Rudensky. 2005. Regulatory T cell lineage specification by the forkhead transcription factor Foxp3. Immunity 22: 329 –341. 2. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell development by the transcription factor Foxp3. Science 299: 1057–1061. 3. Fontenot, J. D., M. A. Gavin, and A. Y. Rudensky. 2003. Foxp3 programs the development and function of CD4⫹CD25⫹ regulatory T cells. Nat. Immunol. 4: 330 –336. 4. Nelson, B. H. 2004. IL-2, Regulatory T cells, and tolerance. J. Immunol. 172: 3983–3988. 5. Zheng, S. G., J. H. Wang, J. D. Gray, H. Soucier, and D. A. Horwitz. 2004. Natural and induced CD4⫹CD25⫹ cells educate CD4⫹CD25⫺ cells to develop suppressive activity: the role of IL-2, TGF-␤, and IL-10. J. Immunol. 172: 5213–5221. 6. Chen, W., W. Jin, N. Hardegen, K. j. Lei, L. Li, N. Marinos, G. McGrady, and S. M. Wahl. 2003. Conversion of peripheral CD4⫹CD25⫺ naive T cells to CD4⫹ CD25⫹ regulatory T cells by TGF-␤ induction of transcription factor Foxp3. J. Exp. Med. 198: 1875–1886. 7. Fu, S., N. Zhang, A. C. Yopp, D. Chen, M. Mao, D. Chen, H. Zhang, Y. Ding, and J. S. Bromberg. 2004. TGF-␤ induces Foxp3 T-regulatory cells from CD4⫹ CD25⫺ precursors. Am. J. Transplant. 4: 1614 –1627. 8. Fantini, M. C., C. Becker, G. Monteleone, F. Pallone, P. R. Galle, and M. F. Neurath. 2004. Cutting edge: TGF-␤ induces a regulatory phenotype in CD4⫹CD25⫺ T cells through Foxp3 induction and down-regulation of Smad7. J. Immunol. 172: 5149 –5153. 9. Li, M. O., Y. Y. Wan, S. Sanjabi, A. K. Robertson, and R. A. Flavell. 2006. Transforming growth factor-␤ regulation of immune responses. Annu. Rev. Immunol. 24: 99 –146. 10. Gorelik, L., and R. A. Flavell. 2000. Abrogation of TGF␤ signaling in T cells leads to spontaneous T cell differentiation and autoimmune disease. Immunity 12: 171–181. 11. Kulkarni, A. B., C. Huh, D. Becker, A. Geiser, M. Lyght, K. C. Flanders, A. B. Roberts, M. B. Sporn, J. M. Ward, and S. Karlsson. 1993. Transforming growth factor ␤1 null mutation in mice causes excessive inflammatory response and early death. Proc. Natl. Acad. Sci. USA 90: 770 –774. 12. Huber, S., C. Schramm, H. A. Lehr, A. Mann, S. Schmitt, C. Becker, M. Protschka, P. R. Galle, M. F. Neurath, and M. Blessing. 2004. Cutting edge: TGF-␤ signaling is required for the in vivo expansion and immunosuppressive capacity of regulatory CD4⫹CD25⫹ T cells. J. Immunol. 173: 6526 – 6531. 13. Marie, J. C., J. J. Letterio, M. Gavin, and A. Y. Rudensky. 2005. TGF-␤1 maintains suppressor function and Foxp3 expression in CD4⫹CD25⫹ regulatory T cells. J. Exp. Med. 201: 1061–1067. 14. Mangan, P. R., L. E. Harrington, D. B. O’Quinn, W. S. Helms, D. C. Bullard, C. O. Elson, R. D. Hatton, S. M. Wahl, T. R. Schoeb, and C. T. Weaver. 2006. Transforming growth factor-␤ induces development of the TH17 lineage. Nature 441: 231–234. 15. Laouar, Y., F. S. Sutterwala, L. Gorelik, and R. A. Flavell. 2005. Transforming growth factor-␤ controls T helper type 1 cell development through regulation of natural killer cell interferon-␥. Nat. Immunol. 6: 600 – 607. 16. Li, W., K. Carper, X. X. Zheng, C. S. Kuhr, J. D. Reyes, Y. Liang, D. L. Perkins, A. W. Thomson, and J. D. Perkins. 2006. The role of Foxp3⫹ regulatory T cells in liver transplant tolerance. Transplant. Proc. 38: 3205–3206.

The Journal of Immunology 17. Green, E. A., L. Gorelik, C. M. McGregor, E. H. Tran, and R. A. Flavell. 2003. CD4⫹CD25⫹ T regulatory cells control anti-islet CD8⫹ T cells through TGF␤-TGF-␤ receptor interactions in type 1 diabetes. Proc. Natl. Acad. Sci. USA 100: 10878 –10883. 18. Jaeckel, E., H. von Boehmer, and M. P. Manns. 2005. Antigen-specific Foxp3transduced T-cells can control established type 1 diabetes. Diabetes 54: 306 –310. 19. Masteller, E. L., Q. Tang, and J. A. Bluestone. 2006. Antigen-specific regulatory T cells: ex vivo expansion and therapeutic potential. Semin. Immunol. 18: 103–110. 20. Mekala, D. J., and T. L. Geiger. 2005. Immunotherapy of autoimmune encephalomyelitis with redirected CD4⫹D25⫹ T lymphocytes. Blood 105: 2090 –2092. 21. Thornton, A. M., and E. M. Shevach. 1998. CD4⫹CD25⫹ immunoregulatory T cells suppress polyclonal T cell activation in vitro by inhibiting interleukin 2 production. J. Exp. Med. 188: 287–296. 22. Yamagiwa, S., J. D. Gray, S. Hashimoto, and D. A. Horwitz. 2001. A role for TGF-␤ in the generation and expansion of CD4⫹CD25⫹ regulatory T cells from human peripheral blood. J. Immunol. 166: 7282–7289. 23. Zheng, S. G., L. Meng, J. H. Wang, M. Watanabe, M. L. Barr, D. V. Cramer, J. D. Gray, and D. A. Horwitz. 2006. Transfer of regulatory T cells generated ex vivo modifies graft rejection through induction of tolerogenic CD4⫹CD25⫹ cells in the recipient. Int. Immunol. 18: 279 –289. 24. Weber, S. E., J. Harbertson, E. Godebu, G. A. Mros, R. C. Padrick, B. D. Carson, S. F. Ziegler, and L. M. Bradley. 2006. Adaptive islet-specific regulatory CD4 T cells control autoimmune diabetes and mediate the disappearance of pathogenic Th1 cells in vivo. J. Immunol. 176: 4730 – 4739. 25. Fantini, M. C., C. Becker, I. Tubbe, A. Nikolaev, H. A. Lehr, P. Galle, and M. F. Neurath. 2006. Transforming growth factor ␤ induced Foxp3⫹ regulatory T cells suppress Th1 mediated experimental colitis. Gut 55: 671– 680. 26. Cabarrocas, J., C. Cassan, F. Magnusson, E. Piaggio, L. Mars, J. Derbinski, B. Kyewski, D. A. Gross, B. L. Salomon, K. Khazaie, et al. 2006. Foxp3⫹CD25⫹ regulatory T cells specific for a neo-self-antigen develop at the double-positive thymic stage. Proc. Natl. Acad. Sci. USA 103: 8453– 8458. 27. Hsieh, C. S., Y. Zheng, Y. Liang, J. D. Fontenot, and A. Y. Rudensky. 2006. An intersection between the self-reactive regulatory and nonregulatory T cell receptor repertoires. Nat. Immunol. 7: 401– 410. 28. Inaba, H., and T. L. Geiger. 2006. Defective cell cycle induction by IL-2 in naive T-cells antigen stimulated in the presence of refractory T-lymphocytes. Int. Immunol. 18: 1043–1054. 29. Duthoit, C. T., D. J. Mekala, R. S. Alli, and T. L. Geiger. 2005. Uncoupling of IL-2 signaling from cell cycle progression in naive CD4⫹ T cells by regulatory CD4⫹CD25⫹ T lymphocytes. J. Immunol. 174: 155–163. 30. Peng, Y., Y. Laouar, M. O. Li, E. A. Green, and R. A. Flavell. 2004. TGF-␤ regulates in vivo expansion of Foxp3-expressing CD4⫹CD25⫹ regulatory T cells responsible for protection against diabetes. Proc. Natl. Acad. Sci. USA 101: 4572– 4577. 31. Marie, J. C., J. J. Letterio, M. Gavin, and A. Y. Rudensky. 2005. TGF-␤1 maintains suppressor function and Foxp3 expression in CD4⫹CD25⫹ regulatory T cells. J. Exp. Med. 201: 1061–1067. 32. Zheng, S. G., J. D. Gray, K. Ohtsuka, S. Yamagiwa, and D. A. Horwitz. 2002. Generation ex vivo of TGF-␤-producing regulatory T cells from CD4⫹CD25⫺ precursors. J. Immunol. 169: 4183– 4189. 33. Gray, J. D., M. Hirokawa, and D. A. Horwitz. 1994. The role of transforming growth factor ␤ in the generation of suppression: an interaction between CD8⫹ T and NK cells. J. Exp. Med. 180: 1937–1942.

34. Sanders, V. M. 2006. Epigenetic regulation of Th1 and Th2 cell development. Brain Behav. Immun. 20: 317–324. 35. Shimizu, J., S. Yamazaki, T. Takahashi, Y. Ishida, and S. Sakaguchi. 2002. Stimulation of CD25⫹CD4⫹ regulatory T cells through GITR breaks immunological self-tolerance. Nat. Immunol. 3: 135–142. 36. Wei, S., I. Kryczek, and W. Zou. 2006. Regulatory T-cell compartmentalization and trafficking. Blood 108: 426 – 431. 37. Flomenberg, N., J. DiPersio, and G. Calandra. 2005. Role of CXCR4 chemokine receptor blockade using AMD 3100 for mobilization of autologous hematopoietic progenitor cells. Acta Haematol. 114: 198 –205. 38. Kannagi, R. 2002. Regulatory roles of carbohydrate ligands for selectins in the homing of lymphocytes. Curr. Opin. Struct. Biol. 12: 599 – 608. 39. Vindevoghel, L., A. Kon, R. J. Lechleider, J. Uitto, A. B. Roberts, and A. Mauviel. 1998. Smad-dependent transcriptional activation of human type VII collagen gene (COL7A1) promoter by transforming growth factor-␤. J. Biol. Chem. 273: 13053–13057. 40. Zheng, S. G., J. H. Wang, W. Stohl, K. S. Kim, J. D. Gray, and D. A. Horwitz. 2006. TGF-␤ requires CTLA-4 early after T cell activation to induce Foxp3 and generate adaptive CD4⫹CD25⫹ regulatory cells. J. Immunol. 176: 3321–3329. 41. Wohlfert, E. A., L. Gorelik, R. Mittler, R. A. Flavell, and R. B. Clark. 2006. Cutting edge: deficiency in the E3 ubiquitin ligase Cbl-b results in a multifunctional defect in T cell TGF-␤ sensitivity in vitro and in vivo. J. Immunol. 176: 1316 –1320. 42. Liu, Y., S. Amarnath, and W. Chen. 2006. Requirement of CD28 signaling in homeostasis/survival of TGF-␤ converted CD4⫹CD25⫹ Tregs from thymic CD4⫹CD25⫺ single positive T cells. Transplantation 82: 953–964. 43. Fraser, J. D., and A. Weiss. 1992. Regulation of T-cell lymphokine gene transcription by the accessory molecule CD28. Mol. Cell. Biol. 12: 4357– 4363. 44. Zheng, S. G., J. Wang, P. Wang, J. D. Gray, and D. A. Horwitz. 2007. IL-2 is essential for TGF-␤ to convert naive CD4⫹. J. Immunol. 178: 2018 –2027. 45. Davidson, T. S., R. J. Dipaolo, J. Andersson, and E. M. Shevach. 2007. Cutting edge: IL-2 is essential for TGF-␤-mediated induction of Foxp3⫹ T regulatory cells. J. Immunol. 178: 4022– 4026. 46. Li, M. O., S. Sanjabi, and R. A. Flavell. 2006. Transforming growth factor-␤ controls development, homeostasis, and tolerance of T cells by regulatory T cell-dependent and -independent mechanisms. Immunity 25: 455– 471. 47. Fontenot, J. D., J. P. Rasmussen, M. A. Gavin, and A. Y. Rudensky. 2005. A function for interleukin 2 in Foxp3-expressing regulatory T cells. Nat. Immunol. 6: 1142–1151. 48. Ribot, J., P. Romagnoli, and J. P. M. van Meerwijk. 2006. Agonist ligands expressed by thymic epithelium enhance positive selection of regulatory T lymphocytes from precursors with a normally diverse TCR repertoire. J. Immunol. 177: 1101–1107. 49. Zou, L., B. Barnett, H. Safah, V. F. LaRussa, M. Evdemon-Hogan, P. Mottram, S. Wei, O. David, T. J. Curiel, and W. Zou. 2004. Bone marrow is a reservoir for CD4⫹CD25⫹ regulatory T cells that traffic through CXCL12/CXCR4 signals. Cancer Res. 64: 8451– 8455. 50. Mekala, D. J., R. S. Alli, and T. L. Geiger. 2005. IL-10-dependent infectious tolerance after the treatment of experimental allergic encephalomyelitis with redirected CD4⫹CD25⫹ T lymphocytes. Proc. Natl. Acad. Sci. USA 102: 11817–11822.