Transcriptional Mechanisms Underlying Lymphocyte Tolerance

Home , Myelin regulatory factor

Cell, Vol. 109, 719–731, June 14, 2002, Copyright 2002 by Cell Press Transcriptional Mechanisms Underlying Lymphocyte Tolerance

Fernando Macia´ n,1,4 Francisco Garcı´a-Co´ zar,1,4,5 Activation of the cell-intrinsic mechanism of lympho- Sin-Hyeog Im,1 Heidi F. Horton,2 cyte tolerance is closely linked to the cell-surface stimu- Michael C. Byrne,2 and Anjana Rao1,3 lus received. In both T and B cells, combined activation 1Center for Blood Research of antigen and costimulatory receptors leads to full acti- Department of Pathology vation of all TCR-coupled signaling pathways and culmi- Harvard Medical School nates in a productive immune response. In contrast, 200 Longwood Avenue tolerance is evoked, both ex vivo and in vivo, by unbal- Boston, Massachusetts 02115 anced stimulation through antigen receptors without en- 2 Wyeth Research gagement of costimulatory receptors, or by stimulation 200 CambridgePark Drive with weak agonist antigens in the presence of full costi- Cambridge, Massachusetts 02140 mulation (Schwartz, 1996; Sloan-Lancaster and Allen, 1996; Goodnow, 2001). In each system, the process of tolerance induction may be conceptualized as occurring Summary in two stages. The tolerizing stimulus first elicits partial or suboptimal activation; next, the partially activated In lymphocytes, integration of Ca2؉ and other signaling lymphocytes enter a long-lasting unresponsive state, in pathways results in productive activation, while unop- which they paradoxically become refractory to subse- posed Ca2؉ signaling leads to tolerance or anergy. We quent full stimulation with antigen and costimulatory ,show that the Ca2؉-regulated transcription factor ligands (Schwartz, 1996; Sloan-Lancaster and Allen NFAT has an integral role in both aspects of lympho- 1996). cyte function. Ca2؉/calcineurin signaling induces a lim- The most consistent feature of tolerizing stimuli is 2ϩ ited set of anergy-associated genes, distinct from their ability to induce elevation of intracellular free Ca . genes induced in the productive immune response; One of the simplest methods of inducing tolerance (an- 2ϩ these genes are upregulated in vivo in tolerant T cells ergy) in T cells is treatment with the Ca ionophore and are largely NFAT dependent. T cells lacking NFAT1 ionomycin; conversely, anergy induction is blocked by 2ϩ are resistant to anergy induction; conversely, NFAT1 the extracellular Ca chelator EGTA and by the cal- induces T cell anergy if prevented from interacting cineurin inhibitor cyclosporin A (CsA) (Schwartz, 1996). 2ϩ with its transcriptional partner AP-1 (Fos/Jun). Thus, Ca has also been implicated in a well-established in the absence of AP-1, NFAT imposes a genetic pro- model of B cell tolerance in vivo: B cells bearing an anti- hen egg lysozyme (HEL) Ig transgene that have been gram of lymphocyte anergy that counters the program tolerized to circulating antigen in vivo show a small but of productive activation mediated by the cooperative significant elevation in their basal levels of intracellular NFAT:AP-1 complex. free Ca2ϩ and a concomitant increase in resting nuclear levels of the Ca2ϩ-regulated transcription factor NFAT Introduction (Healy et al., 1997). Ca2ϩ is also implicated in anergy imposed by altered peptide ligands, weakly agonistic The antigen receptors of T and B cells recognize not peptide-MHC complexes that dissociate rapidly from only antigens derived from pathogenic cells and organ- the T cell receptor. Measurements of Ca2ϩ transients in isms, but also self-antigens expressed on the body’s single cells show that these weak agonist peptides elicit own tissues and nonpathogenic antigens responsible for much lower levels of Ca2ϩ mobilization than strong ago- allergic reactions. In healthy individuals, self-antigens do nist peptides, but increased Ca2ϩ levels are maintained not elicit a significant immune response. Self-reactive for much longer times (Rabinowitz et al., 1996; Sloan- lymphocytes are clonally eliminated during develop- Lancaster et al., 1996). ment; cells that survive this process are rendered toler- A major consequence of Ca2ϩ mobilization is activa- ant to self-antigens in the periphery (Kamradt and Mit- tion of the transcription factor NFAT (Rao et al., 1997; chison, 2001). There are at least two mechanisms for Crabtree, 1999). NFAT is a family of highly phosphory- inducing peripheral lymphocyte tolerance. The first is lated proteins residing in the cytoplasm of resting cells; anergy induction, an intracellular process in which anti- when cells are activated, these proteins are dephos- gen receptors become uncoupled from their down- phorylated by the Ca2ϩ/calmodulin-dependent phos- stream signaling pathways (Fields et al., 1996; Li et al., phatase calcineurin, translocate to the nucleus, and be- 1996; Boussiotis et al., 1997). The second involves regu- come transcriptionally active (Kiani et al., 2000; Okamura latory T cells, which limit the responses of other lympho- et al., 2000). In the nucleus, they cooperate with an cytes to self- and environmental antigens, in part by unrelated transcription factor, AP-1 (Fos/Jun), to induce producing immunosuppressive cytokines such as TGF␤ a large number of cytokine genes and other genes that and IL-10 (Maloy and Powrie, 2001). are central to the productive immune response (Rao et al., 1997; Macian et al., 2001). Notably, NFAT activation does not require strong stimulation of antigen receptors 3 Correspondence: [email protected] 4 These authors contributed equally to this work. on B and T cells: substantial nuclear localization of NFAT 2ϩ 5 Present address: Puerto Real University Hospital Research Unit/ can be achieved with low, sustained levels of Ca mobi- School of Medicine, University of Ca´ diz, Ca´ diz, Spain. lization, such as those achieved by low concentrations Cell 720

of Ca2ϩ ionophores, self-antigens, and low-affinity pep- by Th2 cells (lanes 6 and 8). Notably, Th1 and Th2 cells tide-MHC complexes (Dolmetsch et al., 1997). Costimu- differed significantly with respect to IL-10 gene expres- latory receptors are not coupled to Ca2ϩ mobilization sion: Th1 cells showed a striking decrease in IL-10 tran- and contribute only weakly to activation of NFAT (Lyakh script levels following ionomycin pretreatment (lanes 2 et al., 1997). Thus, NFAT activation occurs in response and 4), whereas Th2 cells were unaffected (lanes 6 and to Ca2ϩ signals or TCR stimulation alone, the precise 8). Ionomycin pretreatment also did not affect IL-10 conditions needed to evoke anergy. In contrast, costi- mRNA induction by regulatory Tr1 cells, generated by mulation is critical for optimal activation of NF␬B and culturing stimulated CD4 T cells in media containing IL- AP-1: combined TCR/CD28 stimulation activates cJun 10 (Groux et al., 1997; S.-H.I., unpublished results). Thus, kinase (JNK), p38 MAP kinase, and I␬B kinase (IKK) the net effect of a tolerizing stimulus on T cells is to pathways and increases nuclear levels of NF␬B/Rel and skew the cytokine response toward production of the AP-1 proteins more strongly than TCR stimulation alone immunosuppressive cytokine IL-10 while downregulat- (Su et al., 1994; Harhaj and Sun, 1998). ing production of multiple effector cytokines associated Here we demonstrate that NFAT plays a central role with a productive immune response. in tolerance induction in T cells. Using a simple pharma- As previously noted (Schwartz, 1996), anergy induc- cological method of inducing T cell anergy, we show tion required calcineurin activity (Figure 1F). DO11.10 that anergized T cells express a novel set of anergy- Th1 cells were stimulated overnight with immobilized associated genes, distinct from those characteristic of anti-CD3 in the presence or absence of the calcineurin the productive immune response. These genes are also inhibitor CsA, then detached, washed thoroughly, cul- upregulated in vivo in T cells from orally tolerized mice. tured for 2–3 days to remove both drug and stimulus, T cells lacking a major NFAT protein, NFAT1 (NFATp, and restimulated with antigen/APC (Figure 1F). The CsA NFATc2), are resistant to anergy induction and show washout was successful since it barely affected IL-2 significantly lower expression of many anergy-associ- production at later times (clusters 1 and 3); however, ated genes. Conversely, T cells harboring a constitu- CsA strongly impaired anergy induction by immobilized tively active NFAT1, under conditions where AP-1 is not anti-CD3 (clusters 2 and 4). activated or NFAT: AP-1 cooperation does not occur, show increased expression of anergy-associated genes A Gene Expression Program Activated and display an anergic phenotype of lowered TCR re- by Ca2؉ and Calcineurin sponsiveness. Thus, a single transcription factor, NFAT, Based on these data, we hypothesized that sustained regulates two contrasting aspects of T cell function, Ca2ϩ/calcineurin signaling induced a distinct genetic mediating nonoverlapping genetic programs of produc- program that correlated with the development of a long- tive activation or anergy depending on the presence or lasting anergic state. To test the hypothesis, we evalu- absence of its transcriptional partner AP-1. ated the gene expression profile of D5 T cells stimulated for 2, 6, or 16 hr with ionomycin alone, which results in Results anergy induction; with ionomycin plus CsA, which blocks anergy induction; and with ionomycin plus PMA, Sustained Ca2؉/Calcineurin Signaling Attenuates which pharmacologically mimic complete stimulation Transcription of Effector Cytokine Genes through the TCR and CD28. Using Affymetrix oligonucle- We used the murine antigen-specific Th1 clone D5 to otide arrays, we identified 1358 genes and ESTs whose set up a model of clonal anergy ex vivo (Figure 1). As expression was altered at least 3-fold by any of the previously reported for other T cell clones (Schwartz, treatments at one or more time points. Genes with simi- 1996), pretreatment of D5 T cells with ionomycin greatly lar expression patterns were clustered into 36 panels, diminished their subsequent proliferative response to 20 of which (736 genes and ESTs) are shown in Figure antigen or anti-CD3 (Figures 1A and 1C and data not 2A. Known genes in the 20 panels are listed in Supple- shown) without inducing detectable levels of apoptosis mentary Table S1 at http://www.cell.com/cgi/content/ (Figure 1B). As expected (Schwartz, 1996), anergy was full/109/6/719/DC1. overcome by exposure to IL-2 (Figure 1C). Anergy devel- This analysis established that the effect of ionomycin opment was slow: decreased antigen responses were stimulation on gene expression was distinct from that apparent after 4–6 hr of ionomycin pretreatment, but of combined PMA/ionomycin stimulation, and moreover complete unresponsiveness was only achieved after 16 involved a much smaller number of genes (Figure 2A). hr (data not shown). Ionomycin-treated D5 cells showed Only 205 genes/ESTs could be considered to be iono- were induced more strongly by 165ف :markedly decreased transcription of several inducible mycin induced genes, including IL-2, IFN-␥, TNF-␣, GM-CSF, and MIP- ionomycin than by PMA/ionomycin (panels 15–19), while were equivalently induced by ionomycin and by 40ف /in response to a second stimulation with anti-CD3 ,␣1 anti-CD28 or antigen/antigen presenting cells (APC) PMA/ionomycin (panels 13 and 14 plus a few in panel (Figure 1D; the apparent lack of downregulation of IFN-␥ 6; these are considered to be ionomycin-induced genes transcripts in lane 4 and of MIP-1␣ transcripts in lane on which PMA had no additional effect). In contrast, 585 12 is due to probe saturation). genes/ESTs were upregulated (panels 1–12 and three Ionomycin pretreatment also reduced cytokine gene panels not shown) and 568 were downregulated (panel transcription by primary T cells in response to TCR stim- 20 and 13 panels not shown) in response to PMA/iono- ulation (Figure 1E), essentially abrogating mRNA induc- mycin stimulation, with little or no change in response tion by Th1 cells (lanes 2, 4, 10, and 12) and decreasing to ionomycin alone. As expected (Teague et al., 1999; -the induction of IL-4, IL-5, and IL-13 mRNAs Glynne et al., 2000; Feske et al., 2001), genes upregu %70ف by Transcriptional Regulation of T Cell Tolerance 721

Figure 1. Ionomycin Pretreatment Attenuates T Cell Responses to Subsequent TCR/CD28 Stimulation (A) 3 H-thymidine incorporation was measured in D5 T cells cultured with or without ionomycin for 16 hr, then washed and restimulated with antigen/APC. Unless otherwise indicated, all subsequent experiments were performed using 16 hr pretreatment with 500 nM ionomycin. (B) TUNEL assay showing that ionomycin treatment does not result in cell death. (C) Ionomycin-pretreated T cells remain responsive to IL-2. D5 cells were incubated with or without ionomycin, then washed and stimulated with antigen/APC with or without exogenous IL-2. 3 H-thymidine incorporation was measured. Results are mean and range of two experiments. (D) Ionomycin pretreatment attenuates cytokine expression by D5 T cells. Cells were pretreated with ionomycin, then washed and stimulated for 4 hr. Cytokine mRNA levels were determined by RNase protection assay. (E) Ionomycin pretreatment attenuates most cytokine expression by primary Th1 and Th2 cells. Cells were pretreated with ionomycin, then washed and stimulated with antigen/APC. Cytokine expression was determined by RNase protection assay. (F) Anergy induction is inhibited by CsA. Th1 cells were incubated with or without 1 ␮g/ml plate bound anti-CD3 in the presence or absence of 1 ␮M CsA for 16 hr. After washing and resting for 48–72 hr, cells were restimulated with antigen/APC for 24 hr. IL-2 levels were determined by ELISA. Values are average Ϯ S.E.M. of three independent experiments.

lated by PMA/ionomycin included cytokine, chemokine, ations in expression were abolished by CsA, consistent and other inducible genes characteristic of the produc- with our previous findings using human T cells (Feske tive immune response (IL-2, IFN-␥, GM-CSF, etc.; see et al., 2001). The results support the hypothesis that Supplementary Table S1 at http://www.cell.com/cgi/ Ca2ϩ/calcineurin signaling directs a genetic program as- content/full/109/6/719/DC1). For almost all genes, alter- sociated with anergy development in T cells, distinct Cell 722

Figure 2. Stimulation with Ca2ϩ Ionophore Activates a Calcineurin-Dependent Program of Gene Expression Distinct from that Induced by PMA plus Ionomycin (A) RNA was prepared from resting D5 T cells or cells stimulated for 2, 6, or 16 hr as indicated. Gene transcription profiles were evaluated using Affymetrix oligonucleotide arrays. Genes whose expression levels were altered at least 3-fold in response to any of the treatments were selected for clustering analysis using the self-organizing map (SOM) algorithm on the basis of kinetic expression pattern. The number of clustered genes and ESTs is indicated inside each panel. (B) Expression profiles of 18 specific genes chosen on the basis of their strong activation by ionomycin. The genes are grouped into six categories based on function. Numbers within the panels indicate the fold induction of each transcript after stimulation with ionomycin for 2 hr, as confirmed by real time quantitative PCR. n.d. indicates not determined.

,.anergy-associated genes/ESTs (e.g 205ف of the 70ف from that associated with the productive immune re- of sponse (Glynne et al., 2000; Lechner et al., 2001). For see Figure 3C). Of the 37 known genes in this category the remainder of this study, the ionomycin-induced (see Supplementary Table S2 at http://www.cell.com/ genes will be referred to interchangeably as “anergy- cgi/content/full/109/6/719/DC1), we selected 18 genes be- associated” genes. cause of their robust and reproducible induction and By repeating the DNA arrays with RNA prepared from because their encoded gene products fell into poten- primary Th1 cells, we confirmed ionomycin inducibility tially interesting functional classes (Figure 2B). Like the Transcriptional Regulation of T Cell Tolerance 723

Figure 3. NFAT1Ϫ/Ϫ Th1 Cells Show Reduced Expression of Anergy-Associated Genes (A) NFAT1 is the predominant NFAT protein in resting T cells. Nuclear extracts from wild-type and NFAT1Ϫ/Ϫ Th1 cells were tested in EMSAs of total NFAT DNA binding %90ف using an NFAT probe. Comparison of lanes 2 and 3 with lanes 4 and 5 shows that NFAT1 accounts for activity in wild-type T cells. A control EMSA with an octamer probe showed equivalent binding activity in all lanes (not shown). (B) NFAT1 regulates expression of many anergy-associated genes. Expression of 15 of the ionomycin-induced genes shown in Figure 2B was examined by real-time quantitative PCR in ionomycin-stimulated wild-type and NFAT1Ϫ/Ϫ Th1 cells. Results are represented as fold increase over the levels of mRNA present in resting cells (set to 1). The average of two independent experiments is plotted. (C) Gene transcription profiles of eight selected genes in wild-type and NFAT1Ϫ/Ϫ Th1 cells obtained using Affymetrix oligonucleotide arrays. All genes showed NFAT1-dependent induction in response to ionomycin. For jumonji, Rab10, and CD98, PMA/ionomycin-mediated induction was not NFAT1 dependent and may be mediated by inducible isoforms of NFAT2 (Lyakh et al., 1997).

Ca2ϩ-dependent genes described in a separate study The Transcription Factor NFAT1 Participates (Feske et al., 2001), the ionomycin-induced genes dis- in Anergy Induction played diverse expression patterns consistent with dif- Since calcineurin activity is required for anergy induc- ferential regulation by PMA- and ionomycin-induced sig- tion, we tested the role of the calcineurin-regulated tran- naling pathways (Figure 2B). To validate the array data, scription factor NFAT. We first asked whether induction we evaluated expression of 15 of the 18 genes in D5 T of the anergy-associated genes was regulated, directly cells by quantitative real-time PCR, and in every case or indirectly, by NFAT. To do this, we took advantage we were able to confirm ionomycin-mediated induction of the fact that the NFAT family member NFAT1 is the (Figure 2B). predominant NFAT protein in resting T cells (Figure 3A). Cell 724

NFAT1Ϫ/Ϫ T cells do not show compensatory increases in other NFATs (F.M., unpublished); thus, these cells not only lack all NFAT1, but also contain only about 10%–15% of the normal levels of total NFAT. About 35 ionomycin-induced genes/ESTs, and 15 of 70ف of the the 18 selected anergy-associated genes (Figures 3B and 3C), showed significantly lower expression in NFAT1Ϫ/Ϫ relative to wild-type Th1 cells following ionomycin stimulation, consistent with participation in an NFAT1-dependent anergy program. Twenty to twenty- five genes were equivalently induced in wild-type and NFAT1Ϫ/Ϫ T cells, implying participation of transcription factors other than NFAT1 (see Discussion). The remaining genes were upregulated in NFAT1Ϫ/Ϫ relative to wild-type T cells, suggesting loss of an inhibitory signal mediated by NFAT1. The results implicate NFAT proteins, directly or indirectly, in a substantial proportion of ionomycin-induced gene transcription in T cells. We asked whether NFAT1Ϫ/Ϫ T cells were also defec- tive for anergy induction ex vivo (Figure 4). As expected, ionomycin pretreatment of DO11.10 TcR transgenic Th1 cells resulted in markedly decreased induction of IL-2 and IFN-␥ mRNAs (Figure 4A, lanes 1 and 2). Because of their lower levels of total NFAT, Th1 cells from NFAT1Ϫ/Ϫ DO11.10 mice showed somewhat lower induction of cytokine mRNAs compared to wild-type Th1 cells (1.5-fold and 2-fold decrease for IL-2 and for IFN-␥, respectively, lanes 1 and 3), but they were much less susceptible to anergy induction, showing perceptible induction of IL-2 and IFN-␥ mRNA even after ionomycin pretreatment Ϫ/Ϫ (Figure 4A, lane 4). Ionomycin-treated wild-type Th1 Figure 4. NFAT1 T Cells Are More Resistant to Anergy Induction than Wild-Type T Cells cells showed 18-fold and 12-fold decreases in IL-2 and Ϫ/Ϫ IFN-␥ transcripts following stimulation, but NFAT1Ϫ/Ϫ T (A) Wild-type or NFAT1 Th1 cells were incubated with or without ionomycin, then stimulated with antigen/APC or PMA/ionomycin for cells showed only 2.5-fold decrease in either case (Fig- 4 hr. IL-2 and IFN-␥ transcript levels were determined by RNase ure 4A, right). The anergized cells were fully responsive protection assay (left). Unstimulated cells did not express detectable to PMA/ionomycin stimulation, which bypasses the cytokine transcripts. The graph on the right shows the anergy index membrane-proximal steps of TCR signal transduction (fold reduction of cytokine transcript levels following ionomycin pre- Ϫ Ϫ (Figure 4A, lanes 5–8; note that the relative defect of treatment) for IL-2 and IFN-␥ in wild-type and NFAT1 / Th1 cells. the NFAT1Ϫ/Ϫ T cells is overcome under these “strong” The results are representative of two independent experiments. (B) Th1 cells from wild-type or NFAT1Ϫ/Ϫ DO11.10 transgenic mice Ϫ/Ϫ stimulation conditions). NFAT1 Th1 cells did not be- were incubated with or without 1 ␮g/ml plate bound anti-CD3 for come anergic in response to anti-CD3 pretreatment, 16 hr. After resting for 48–72 hr, cells were restimulated with antigen/ compared to wild-type T cells, which were effectively APC, and IL-2 production was determined by ELISA. Values are anergized under these conditions (Figure 4B), again sup- average Ϯ S.E.M. of three independent experiments. porting a role for NFAT1 and possibly other NFAT proteins in anergy. tolerized T cells showed upregulation of 13 of these 14 Orally-Tolerized T Cells Upregulate the Expression genes (Figures 5C and 5D). Thus, in vivo tolerized T cells of Anergy-Associated Genes exhibit a gene expression profile very similar to that To ask whether anergy-associated genes were upregu- observed in ex vivo anergized T cells, downregulating lated in tolerant T cells in vivo, we set up a model of production of IL-2 and other effector cytokines while 2ϩ oral tolerance in which administration of high doses of establishing a distinct Ca /calcineurin-dependent pat- protein antigens induces systemic, antigen-specific T tern of gene expression. In vivo tolerized T cells also 2ϩ cell tolerance in mice (Garside and Mowat, 2001). Oval- show basal elevation of intracellular Ca levels (V. bumin (OVA) was administered to DO11.10 TCR trans- Heissmeyer, S. Feske, S.-H.I., and A.R., unpublished), genic mice in their drinking water for 5 days, after which suggesting strongly that T cell tolerance, like B cell toler- CD4 T cells were isolated from spleen and lymph nodes ance (Healy et al., 1997; Goodnow, 2001), depends on sustained low-level signaling through the Ca2ϩ/cal- and their response to OVA323–339 peptide was measured. As expected, T cells from OVA-fed mice showed pro- cineurin/NFAT pathway. foundly diminished proliferation and IL-2 production compared to T cells from control transgenic mice that Anergy Is Induced by NFAT in the Absence had not been given OVA (Figures 5A and 5B). In parallel of AP-1 (Fos-Jun) we assessed expression of 14 selected ionomycin- NFAT:AP-1 cooperation is critical for transcription of induced genes by quantitative real-time PCR: the in vivo most genes induced during the productive immune re- Transcriptional Regulation of T Cell Tolerance 725

Figure 5. Orally-Tolerized T Cells Upregulate Most Anergy-Associated Genes (A and B) CD4 cells, isolated from spleen (spl) and lymph nodes (LN) of unfed or OVA-fed DO11.10 mice, were stimulated with different 3 concentrations of OVA323–339. Tolerized T cells show greatly diminished H-thymidine incorporation (A) and IL-2 production (B) of antigen responses compared to control (ctrl) T cells. One of four representative experiments is shown. (C) Relative expression of 14 anergy-associated genes in tolerized versus non-tolerized T cells. Transcript levels were analyzed by real-time quantitative PCR in control and OVA-tolerized CD4 T cells. Bars are grouped according to the functional categories of Figure 2B. Results are average Ϯ S.D. of four independent experiments. (D) PCR analysis using [32P]dCTP and 2-fold serial dilutions of cDNA. For the four genes displayed at top, the signal from tolerized T cells is higher than the signal from control T cells. There was no change in levels of FasL mRNA. L32 levels serve as an internal control. sponse (Rao et al., 1997; Kiani et al., 2000; Macian et pression in response to PMA stimulation, as expected al., 2001). We asked whether NFAT:AP-1 cooperation from the lack of activation of endogenous NFAT (lanes was required for anergy induction as well. Electropho- 1 and 2). Cells transfected with the CA-NFAT1 plasmid retic mobility shift assays showed that ionomycin- showed perceptible basal induction of the TNF␣ gene induced anergy correlated with NFAT but not AP-1 or (lane 3), an NFAT-dependent gene that can be tran- NF␬B activation (Figure 6A, lanes 2, 5, and 8), while scribed in the absence of NFAT-AP-1 cooperation (Gold- combined stimulation with PMA and ionomycin induced feld et al., 1993; Macian et al., 2000), as well as strong the cooperative NFAT:AP-1 complex, the AP-1 complex, PMA-stimulated induction of the IL-3, GM-CSF, and and the p50/p65 NF␬B complex as expected (closed MIP-1␣ genes, which require the cooperative interaction symbols, lanes 3, 6, and 9). of NFAT and AP-1 (lane 4; Macian et al., 2000). PMA To explore the role of NFAT:AP-1 cooperation in an- stimulation also further upregulated TNF␣, a gene that ergy induction, we utilized a constitutively active (CA) normally is maximally activated under conditions of version of NFAT1 (Okamura et al., 2000). This protein combined PMA/ionomycin stimulation (Macian et al., 0 0 .eraaiesubstitutions in 12 phosphorylated serines 2000).bearsalanine whose dephosphorylation is required for nuclear local- Despite the ability of CA-NFAT1 to activate gene tran- ization and is constitutively nuclear under conditions scription, both CA-NFAT1 and CA-RIT-NFAT1 paradoxi- where endogenous NFAT proteins are fully localized to cally reduced TCR responsiveness when retrovirally ex- the cytoplasm (Okamura et al., 2000). We also generated pressed in unstimulated NFAT1Ϫ/Ϫ Th1 cells (Figures 6C CA-RIT-NFAT1, a CA-NFAT1 derivative engineered to and 6D). Five to seven days after infection with IRES- be incapable of cooperation with AP-1. In addition to GFP retroviruses, the ability of GFPϩ (infected) and GFPϪ the 12 Ser to Ala substitutions present in CA-NFAT1, (uninfected) cells to produce IL-2 in response to anti- CA-RIT-NFAT1 contains three point mutations in its DNA CD3/anti-CD28 stimulation was assessed by intracellu- binding domain that abrogate Fos-Jun interaction lar cytokine staining. T cells expressing CA-NFAT1 or (R468A/I469A/T535G; Macian et al., 2000). CA-RIT-NFAT1 showed markedly decreased IL-2 pro- We first showed that CA-NFAT1 could activate the duction, compared to control T cells expressing GFP transcription of endogenous inducible genes (Figure alone (Figure 6C, compare top and bottom; the results 6B). Untransfected Jurkat cells showed no cytokine ex- of four independent experiments are presented in Figure Cell 726

Figure 6. NFAT Induces T Cell Anergy in the Absence of AP-1 (Fos-Jun) (A) Ionomycin treatment activates NFAT but not AP-1 or NF␬B. EMSAs were performed with nuclear extracts of unstimulated or stimulated Th1 cells and labeled NFAT, AP-1, or NF␬B probes. DNA-protein complexes are indicated as follows. Left: open circle, NFAT; closed circle, cooperative NFAT:AP-1 complex. Middle: closed circle, AP-1 complex. Right: open triangle, NF␬B1(p50) homodimer; closed circle, NF␬B1(p50)/ RelA(p65) heterodimer. (B) Constitutively active (CA) NFAT1 induces expression of endogenous cytokine genes. Jurkat T cells were cotransfected with expression plasmids encoding mouse CD4 and either GFP or CA-NFAT1. Productively transfected cells expressing mouse CD4 were isolated and stimulated with PMA as indicated, and cytokine transcript levels were analyzed by RNase protection assay. Endogenous (Endog) NFAT is unable to induce cytokine expression under these conditions (lanes 1 and 2). (C) Constitutive expression of CA-RIT-NFAT1 renders T cells unresponsive to TCR/CD28 stimulation. NFAT1Ϫ/Ϫ Th1 cells were infected with GFP or CA-RIT-NFAT1/GFP retroviruses (RV), then cultured for 3 days with and 24 hr without IL-2. The cells were then stimulated for 4 hr with anti-CD3/anti-CD28, and the percentage of IL-2-producing cells in GFPϩ (infected) and GFPϪ (uninfected) cells was determined by intracellular cytokine staining. Left: dot plot showing GFP and IL-2 expression by individual T cells. Two different gates were used: total GFPϩ cells (continuous vertical line) and cells expressing high levels of GFP (dotted vertical line). Middle and right: open and closed histograms show IL-2 staining in unstimulated and stimulated cells gated for total or high GFP expression. (D) Average Ϯ S.D. of four independent experiments similar to that shown in part (C). Cells were infected with GFP, CA-NFAT1, or CA-RIT- NFAT1 retroviruses. Inset, CA-NFAT1 and CA-RIT-NFAT1 expression in infected populations of NFAT1Ϫ/Ϫ cells was estimated by immunoblotting with an anti-NFAT1 antibody. Expression levels are lower than those of endogenous NFAT1 in wild-type T cells. (E) Constitutive expression of CA-RIT-NFAT1 induces expression of some but not all anergy-associated genes. NFAT1Ϫ/Ϫ Th1 cells were infected with GFP and CA-RIT-NFAT1 retroviruses and sorted for GFP expression. RNA was isolated from sorted cells, and levels of anergy- associated genes were determined by quantitative real-time PCR. mRNA expression in CA-RIT-NFAT1-expressing cells is plotted relative to that in control GFP-expressing cells. Results are average Ϯ S.D. of three independent experiments except for GRG-4, Ikaros, and LDHA, for which the mean and range of two independent experiments is shown.

6D). When total and bright GFPϩ cells were compared, 20-fold lower than its affinity when it forms a cooperative there was a clear correlation between the levels of GFP/ NFAT:AP-1 complexes on the same site (G. Powers and CA-RIT-NFAT1 expression and the extent to which IL-2 P.G. Hogan, personal communication; Jain et al., 1993). production was reduced (Figure 6C). Note that the con- Together, these results establish that continuous, low- stitutively active proteins are not overexpressed (see level activation of NFAT1 induces a state similar to clas- inset in Figure 6D), ruling out a dominant-negative mech- sical anergy, in which T cells are significantly less capa- anism in which CA-RIT-NFAT1 competes with endoge- ble of producing IL-2 in response to TCR stimulation. nous NFAT for binding to the IL-2 promoter. This mecha- Notably, NFAT:AP-1 cooperation is not required for this nism is unlikely in any case, since without AP-1, the process of anergy induction. Stimulation of CA-RIT- affinity of NFAT for composite NFAT:AP-1 sites is at least NFAT1-expressing T cells with PMA and ionomycin re- Transcriptional Regulation of T Cell Tolerance 727

Figure 7. A Model of Anergy Induction For details, see text. stored IL-2 expression substantially (data not shown), unbalanced activation of NFAT relative to its cooperat- indicating that the anergy involved a TCR-proximal ing transcription factor AP-1 (Fos/Jun), thus diverting block in signaling which could be overcome with phar- NFAT toward transcription of an alternate set of anergy- macological agents that bypassed the TCR. associated genes whose products together impose the Constitutive expression of CA-RIT-NFAT1 induced the tolerant state (Figure 7B). This model does not exclude transcription of specific anergy-associated genes (Fig- the participation of nontranscriptional mechanisms de- ure 6E). CA-RIT-NFAT1 was retrovirally expressed in pendent on Ca2ϩ signaling, or participation of Ca2ϩ-regu- NFAT1Ϫ/Ϫ Th1 cells, GFPϩ cells were isolated by cell lated transcriptional modulators other than NFAT. sorting, RNA was prepared from the unstimulated cells, The model is consistent with essentially all previous and expression of 13 anergy-associated genes was as- data on tolerance induction, both in vivo and ex vivo sessed by real-time PCR. At least 4 of the 13 genes (see Introduction). Our supporting experimental data are showed increased expression in CA-RIT-NFAT1- as follows. First, T cells lacking NFAT1, the major NFAT expressing cells relative to cells expressing GFP alone protein expressed in resting cells, are more resistant (Figure 6E), indicating that NFAT1 is sufficient as well than wild-type T cells to anergy induction ex vivo, con- as necessary (Figure 3B) to induce expression of these sistent with previous findings of T and B cell hyperproli- genes. In contrast, NFAT1 did not induce (or only moder- feration in mice lacking NFAT1 (Hodge et al., 1996; Xan- ately induced) expression of 8 of the 13 genes (Figure thoudakis et al., 1996) or both NFAT1 and NFAT2 (Peng 6E), although it clearly participated in their induction et al., 2001). Second, T cells anergized with ionomycin (Figures 3B and 3C). Thus, the transcriptional arm of show selective NFAT activation as well as induction of the anergy program requires not only NFAT1, but also a novel set of anergy-associated genes; these genes additional signaling pathways and transcription factors are distinct from those activated during the productive induced by Ca2ϩ or Ca2ϩ/calcineurin signaling. immune response and encode diverse categories of proteins that could plausibly impose an anergic state (dis- Discussion cussed below). Third, the anergy-associated genes are also upregulated in T cells rendered tolerant to high A Model of Anergy Induction dose oral antigen in vivo. Fourth, a substantial number Based on our data, we propose that NFAT plays a central of anergy-associated genes are direct or indirect targets role, not only in productive activation of lymphocytes of NFAT, since they are expressed at significantly lower but also in lymphocyte tolerance. Our model of tolerance levels in NFAT1Ϫ/Ϫ T cells following ionomycin stimula- induction is depicted in Figure 7. Combined stimulation tion. Fifth, constitutively active versions of NFAT1, which of TCR and costimulatory receptors results in balanced cannot cooperate productively with AP-1, are capable activation of NFAT, AP-1, and NF␬B; the cooperative of downregulating IL-2 production when retrovirally in- NFAT:AP-1 complexes formed under these conditions troduced into NFAT1-deficient Th1 cells. are necessary for transcription of cytokine genes and Anergy induction is likely to require not only NFAT1, other genes critical for the productive immune response but also other NFAT proteins and other nuclear factors (Figure 7A). In contrast, TCR stimulation without costi- induced by Ca2ϩ/calcineurin signaling. Although NFAT1Ϫ/Ϫ mulation results in higher activation of the Ca2ϩ arm of T cells are resistant to anergy induction (Figure 4), they the TCR signal transduction pathway relative to the can be rendered anergic by high concentrations of iono- PKC/IKK/Ras/MAP kinase arm; these conditions lead to mycin or immobilized anti-CD3 (F.G.-C., F.M., unpub- Cell 728

lished). These data suggest that NFAT2 and NFAT4 also 2000); diacylglycerol kinase-␣, which metabolizes the participate in anergy induction, making increasing con- diacylglycerol required to activate protein kinases C tributions as stronger anergizing stimuli are used. More- (Sanjuan et al., 2001); and the cell-surface receptor over, half of the anergy-associated genes are not af- CD98, which is coupled to increased GTP loading of the fected by NFAT1 deficiency, suggesting redundant small G protein Rap1 (Suga et al., 2001). Rap1 activation effects of NFAT2 and NFAT4. It is likely that other Ca2ϩ/ has been linked to impaired activation of the ERK MAP calcineurin-regulated nuclear factors also have a role: kinase pathway in anergic T cells (Boussiotis et al., 1997; ionomycin pretreatment induces T cell anergy more ef- Bos, 1998). fectively than CA-NFAT1 (Figures 1, 6C, and 6D), and Our data also suggest that proteolytic mechanisms CA-NFAT1 does not upregulate the entire panel of iono- contribute to anergy induction. The procaspase 3 gene mycin-inducible genes (Figures 2 and 6E). Plausible can- is robustly induced under conditions of anergy induction didates include Elk-1 and MEF2 family members, which (Figures 2 and 3), and caspase 3 has been implicated are known to be regulated by calcineurin; NFAT-MEF2 in modulating lymphocyte responses under conditions cooperation has already been documented for other where its activation does not appear to be associated cell types and genes (Aramburu et al., 2000; Olson and with cell death (Alam et al., 1999). Thus, caspase 3 might Williams, 2000). implement T cell anergy in a manner unrelated to apoptosis by cleaving specific signaling proteins down- Relation between Anergy Induction and Cell Death stream of the TCR; its reported targets in the T cell Our model also explains the fact that anergy is frequently activation pathway include Vav1, PKC-theta, the adap- associated with activation-induced cell death (AICD) (Li tor protein Gads, and the zeta chain of the TCR/CD3 et al., 2000; Kamradt and Mitchison, 2001). Mice injected complex (Datta et al., 1997; Gastman et al., 1999; Hof- with high doses of soluble antigen or with superantigens mann et al., 2000; Yankee et al., 2001). SOCS-2 and (proteins that interact simultaneously with MHC Class Traf5 are also products of anergy-associated genes; like ␤ II and the V region of the TCR) delete large numbers the related proteins SOCS-1 and Traf6 (Kamizono et al., of reactive cells, but the surviving cells are tolerant to 2001; Wang et al., 2001), these proteins may be E3 li- subsequent stimulation (Garside and Mowat, 2001). gases involved in ubiquitin transfer. Indeed, mice lacking Since all the signaling pathways shown in Figure 7A can the E3 ligases Itch and Cbl-b show a striking autoim- be activated by TCR stimulation alone, we postulate mune phenotype (Perry et al., 1998; Bachmaier et al., that both NFAT and AP-1 are induced early in response 2000; Chiang et al., 2000), suggesting that protein degra- to high circulating concentrations of antigen or superan- dation has a role in lymphocyte tolerance. Directed pro- tigen, but while NFAT activation is relatively uniform, the teolysis of specific signaling components in anergic T extent of AP-1 activation depends on the strength of the cells could explain the long-lasting nature of anergy in stimulus encountered by the individual T cell, yielding a vivo and ex vivo (Schwartz, 1996; Lanoue et al., 1997), population of reactive cells that express a wide range as well as the finding that anergy is dominant in somatic of relative NFAT:AP-1 ratios. NFAT:AP-1 cooperation is cell fusion experiments (Telander et al., 1999). required for AICD (Macian et al., 2000), and thus cells Ex vivo, activation of NFAT without AP-1 blocks all with the highest AP-1 levels would succumb to AICD while those with the lowest levels of AP-1 would become Th1 cytokine production and skews the cytokine profile anergic. Costimulation would affect both aspects of this of Th2 cells toward IL-10 production (Figure 1E). A simi- process: it provides a strong survival stimulus by activat- lar skewing toward IL-10 expression has been reported ing PI-3 kinase and Akt and Bcl family members (Boise et in tolerant T cells in vivo (Buer et al., 1998). Preferential al., 1995; Kane et al., 2001), thus promoting proliferation IL-10 production by anergic T cells provides a link be- rather than cell death, and at the same time it potentiates tween the two current models of how peripheral tolerance MAP kinase/AP-1 and IKK/NF␬B pathways (Su et al., is maintained: the cell-intrinsic mechanism of anergy in- 1994; Harhaj and Sun, 1998), thus diminishing the proba- duction would attenuate the antigen responsiveness of bility that a cell will become anergic. However, we em- differentiated effector T cells, while the bias toward IL- phasize that T cells anergized with ionomycin ex vivo 10 production by Th2 and Tr1 cells would lead to some showed no evidence of cell death in our experiments immunosuppression by itself but would also result, over (Figure 1B), despite increased expression of FasL mRNA the longer term, in generation of IL-10-producing regula- (Figures 2 and 3) and active caspase 3 protein (F.M., tory T cells capable of suppressing any remaining pro- unpublished). This is likely to reflect their lack of AP-1 ductive response (Buer et al., 1998; Maloy and Powrie, activation (Figure 6A) as well as the ability of Ca2ϩ signal- 2001). The cell type- and gene-specific inhibition of cyto- ing to downregulate Fas mRNA (Feske et al., 2001). kine gene transcription observed in anergic T cells is likely to be imposed in the nucleus by transcriptional Mechanisms of Anergy Induction modulators that act on specific genes, rather than in the Our data support the existence of distinct mechanisms cytoplasm by global interference with the TCR signaling of tolerance induction in lymphocytes. The first is simple complex. Candidate transcriptional modulators emerg- interference with signaling pathways coupled to antigen ing from our screens include Ikaros, a family of proteins receptors (Fields et al., 1996; Li et al., 1996; Boussiotis implicated in gene silencing (Brown et al., 1997; Sabbat- et al., 1997; Healy et al., 1997). This process could be tini et al., 2001); the Groucho-related protein Grg4 (Eber- mediated by the protein products of several of the an- hard et al., 2000); and the DNA binding protein jumonji ergy-associated genes we have identified, including sol- that negatively regulates cell proliferation (Toyoda et al., uble and receptor tyrosine phosphatases (Li and Dixon, 2000). Transcriptional Regulation of T Cell Tolerance 729

Therapeutic Implications NFAT1Ϫ/Ϫ Th1 cells 24 and 48 hr after stimulation with 1 ␮g/ml plate In transplant patients, interference with costimulatory bound anti-CD3⑀ and 5 ␮g/ml anti-CD28 (Pharmingen) in media pathways leads to tolerance induction; paradoxically, supplemented with 20 U/ml of IL-2. Cells were analyzed 72 hr postin- fection and if necessary sorted for GFP expression. Infection effi- this process is blocked by CsA (Li et al., 1998). Our ciencies were similar for all three retroviruses, ranging between 10% model explains both findings and offers an alternate and 40% in different experiments. Protein expression was confirmed strategy that might synergistically induce tolerance in by Western analysis. combination with costimulatory blockade. The predic- tion is that even in the presence of ongoing immune ELISA stimulation, disrupting the interaction of NFAT with Fos Supernatants were collected 24 hr after T cell activation, and IL-2 levels were measured in a sandwich ELISA (Pharmingen). and Jun would induce a long-lasting tolerant state: it would eliminate or severely disrupt transcription of Immunoblotting genes involved in the productive immune response for Whole cell extracts were prepared by boiling cell pellets directly in which NFAT-AP-1 cooperation is essential, while at the SDS to prevent proteolysis during cell lysis. Anti-NFAT1 antibodies same time switching the cell’s genetic program toward have been described (Okamura et al., 2000). transcription of the distinct set of anergy-inducing genes that are activated by NFAT in the absence of AP-1. A Proliferation Assay D5 or primary T cells were stimulated with APC and antigen, 3 H- detailed molecular structure of the NFAT-Fos-Jun-DNA thymidine (10 ␮Ci/ml) was added, and incorporation was measured complex is available (Chen et al., 1998) and should facili- during a 16 hr pulse beginning at 24 hr following stimulation. DNA tate identification of peptide and small molecule inhibi- was collected using a cell harvester, and the amount of radioactivity tors that selectively disrupt cooperative NFAT:Fos:Jun incorporated was measured in a ␤ counter. complexes on composite NFAT-AP-1 sites, without af- fecting independent binding of NFAT or Fos:Jun to non- Tunel Assay Apoptosis was detected by the Tunel method using the In situ Cell composite sites. Death Detection kit (Boehringer). Stained cells were analyzed on a FACSCAN (Beckton-Dickinson). Experimental Procedures RNA Samples and DNA Array Procedures Mice See Supplementary Methods at http://www.cell.com/cgi/content/ Mice were maintained in pathogen-free conditions in a barrier facil- full/109/6/719/DC1. ity. BALB/cJ DO11.10 TCR transgenic mice were bred with NFAT1Ϫ/Ϫ mice (Xanthoudakis et al., 1996) or their isogenic wild-type controls Intracellular Cytokine Staining to obtain NFAT1Ϫ/Ϫ or wild-type DO11.10 TCR transgenic mice. T cells were stimulated for 4 hr with 1 ␮g/ml plate bound anti-CD3⑀ and 5 ␮g/ml anti-CD28. For the last 2 hours, Brefeldin A was added Cell Culture at 10 ␮g/ml to promote intracellular accumulation of IL-2. After The murine Th1 cell clone D5 (Ar-5) was cultured as previously stimulation, cells were fixed in 4% paraformaldehyde and perme- ϩ described (Agarwal and Rao, 1998). Primary CD4 T cells were abilized in PBS/1% BSA/0.5% saponin. Cells were then washed and Ϫ/Ϫ isolated from lymph nodes and spleen of NFAT1 or wild-type incubated for 10 min with Fc-block (Pharmingen) and then for 30 DO11.10 transgenic mice using magnetic beads (Dynal) and differ- more minutes with 10 ␮g/ml phycoerythrin (PE)-conjugated anti- entiated in vitro by stimulating for 1 week with irradiated APC and mouse IL-2 antibody (Pharmingen) to detect intracellular IL-2. ␮ 1 g/ml OVA323–339 as previously described (Agarwal and Rao, 1998). Stained cells were analyzed on a FACSCAN (Beckton-Dickinson). Jurkat and Phoenix Ecotropic cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, Quantitative RT-PCR 10 mM HEPES, and 2 mM glutamine. Total RNA was prepared from resting or stimulated T cells using Ultraspec reagent (Biotecx). cDNA was synthesized using oligo- RNase Protection Assay dT primers and Superscript polymerase (Invitrogen) following the Total cellular RNA was analyzed using the RiboQuant multiprobe manufacturer’s recommendations. Quantitative real-time PCR was RNase protection kit and specific multiset probes (Pharmingen). performed in an I-Cycler (BioRad) using a SYBR Green PCR kit from Jurkat cells were selected after transfection with 10 ␮g/106 cells of Applied Biosystems and specific primers to amplify 100–200 bp a murine CD4 plasmid and pEGFPN1 (Clontech) or pNFAT1-NLS- fragments from the different genes analyzed. A threshold was set (ST2ϩ5ϩ8) (Okamura et al., 2000) as previously described (Macian in the linear part of the amplification curve (fluorescence ϭ f [cycle et al., 2000). number]), and the number of cycles needed to reach it was calcu- lated for every gene. Melting curves and agarose gel electrophoresis Electrophoretic Mobility Shift Assays (EMSAs) established the purity of the amplified band. Normalization was Nuclear extracts were prepared from Th1 cells, unstimulated or achieved by including a sample with primers for L32. stimulated for 1 or 6 hr with 500 nM ionomycin or 20 nM PMA plus 500 nM ionomycin. Binding reactions were performed as previously Induction of Oral Tolerance described (Macian et al., 2000) using probes for NFAT (distal murine OVA (20 mg/ml) was administered to DO11.10 TCR transgenic mice IL-2 promoter), AP-1, and NF-␬B (Goldfeld et al., 1993). in their drinking water for 5 days, after which CD4 T cells were isolated from spleen and lymph nodes of these mice and age- and Retroviral Infections sex-matched controls. The ability of the control and tolerized T cells

Three different retroviral vectors were used: MSCV-containing GFP- to initiate a productive immune response to OVA323–339 was analyzed KV-DV (Wyeth Research) that expresses GFP from an IRES se- by measuring 3 H-thymidine incorporation during a 16 hr pulse begin- quence, GFP-KV-DV-CA-NFAT1, and GFP-KV-DV-CA-RIT-NFAT1. ning 60 hr after stimulation with irradiated splenic APC pulsed with

The latter two vectors were constructed by subcloning DNAs encod- different concentrations of OVA323–339. ing murine CA-NFAT1 [(ST2ϩ5ϩ8) NFAT1; Okamura et al., 2000] with or without a R468A/I469A/T535G mutation (Macian et al., 2000) Acknowledgments into GFP-KV-DV. The Phoenix ecotropic packaging cell line (kindly provided by G.P. Nolan) was transfected with the retroviral vectors We thank members of the Rao and Byrne laboratories for valuable and supernatants were collected 24 and 48 hr later, supplemented discussions, K. Murphy for the original version of the retroviral vec- with polybrene (8 ␮g/ml), and used to spin-infect (1000 ϫ g/90 min) tor, and G. Nolan for the Phoenix Ecotropic packaging cell line. This Cell 730

work was supported by National Institutes of Health grants CA42471 Mack, D.H. (2000). How self-tolerance and the immunosuppressive and AI48213 (to A.R.) drug FK506 prevent B-cell mitogenesis. Nature 403, 672–676. Goldfeld, A.E., McCaffrey, P.G., Strominger, J.L., and Rao, A. (1993). Received: September 26, 2001 Identification of a novel cyclosporin-sensitive element in the human Revised: May 7, 2002 tumor necrosis factor alpha gene promoter. J. Exp. Med. 178, 1365– 1379. References Goodnow, C.C. (2001). Pathways for self-tolerance and the treatment of autoimmune diseases. Lancet 357, 2115–2121. Agarwal, S., and Rao, A. (1998). Modulation of chromatin structure Groux, H., Ogarra, A., Bigler, M., Rouleau, M., Antonenko, S., De- regulates cytokine gene expression during T cell differentiation. Im- vries, J.E., and Roncarolo, M.G. (1997). A CD4ϩ T-cell subset inhibits munity 9, 765–775. antigen-specific T-cell responses and prevents colitis. Nature 389, Alam, A., Cohen, L.Y., Aouad, S., and Sekaly, R.P. (1999). Early 737–742. activation of caspases during T lymphocyte stimulation results in Harhaj, E.W., and Sun, S.C. (1998). IkappaB kinases serve as a target selective substrate cleavage in nonapoptotic cells. J. Exp. Med. of CD28 signaling. J. Biol. Chem. 273, 25185–25190. 190, 1879–1890. Healy, J.I., Dolmetsch, R.E., Timmerman, L.A., Cyster, J.G., Thomas, Aramburu, J., Rao, A., and Klee, C.B. (2000). Calcineurin: from struc- M.L., Crabtree, G.R., Lewis, R.S., and Goodnow, C.C. (1997). Differ- ture to function. Curr. Top. Cell. Regul. 36, 237–295. ent nuclear signals are activated by the B cell receptor during posi- Bachmaier, K., Krawczyk, C., Kozieradzki, I., Kong, Y.Y., Sasaki, T., tive versus negative signaling. Immunity 6, 419–428. Oliveira-dos-Santos, A., Mariathasan, S., Bouchard, D., Wakeham, Hodge, M.R., Ranger, A.M., Charles de la Brousse, F., Hoey, T., A., Itie, A., et al. (2000). Negative regulation of lymphocyte activation Grusby, M.J., and Glimcher, L.H. (1996). Hyperproliferation and dys- and autoimmunity by the molecular adaptor Cbl-b. Nature 403, regulation of IL-4 expression in NF-ATp-deficient mice. Immunity 4, 211–216. 397–405. Boise, L.H., Minn, A.J., Noel, P.J., June, C.H., Accavitti, M.A., Lind- Hofmann, T.G., Hehner, S.P., Droge, W., and Schmitz, M.L. (2000). sten, T., and Thompson, C.B. (1995). CD28 costimulation can pro- Caspase-dependent cleavage and inactivation of the Vav1 proto- mote T cell survival by enhancing the expression of Bcl-X(L). Immu- oncogene product during apoptosis prevents IL-2 transcription. On- nity 3, 87–98. cogene 19, 1153–1163. Bos, J.L. (1998). All in the family? New insights and questions regard- Jain, J., Miner, Z., and Rao, A. (1993). Analysis of the preexisting ing interconnectivity of Ras, Rap1 and Ral. EMBO J. 17, 6776–6782. and nuclear forms of nuclear factor of activated T cells. J. Immunol. Boussiotis, V.A., Freeman, G.J., Berezovskaya, A., Barber, D.L., and 151, 837–848. Nadler, L.M. (1997). Maintenance of human T cell anergy: blocking Kamizono, S., Hanada, T., Yasukawa, H., Minoguchi, S., Kato, R., of IL-2 gene transcription by activated Rap1. Science 278, 124–128. Minoguchi, M., Hattori, K., Hatakeyama, S., Yada, M., Morita, S., et al. Brown, K.E., Guest, S.S., Smale, S.T., Hahm, K., Merkenschlager, (2001). The SOCS box of SOCS-1 accelerates ubiquitin-dependent M., and Fisher, A.G. (1997). Association of transcriptionally silent proteolysis of TEL-JAK2. J. Biol. Chem. 276, 12530–12538. genes with Ikaros complexes at centromeric heterochromatin. Cell Kamradt, T., and Mitchison, N.A. (2001). Tolerance and autoimmu- 91, 845–854. nity. N. Engl. J. Med. 344, 655–664. Buer, J., Lanoue, A., Franzke, A., Garcia, C., von Boehmer, H., and Kane, L.P., Andres, P.G., Howland, K.C., Abbas, A.K., and Weiss, Sarukhan, A. (1998). Interleukin 10 secretion and impaired effector A. (2001). Akt provides the CD28 costimulatory signal for up-regula- function of major histocompatibility complex class II-restricted T tion of IL-2 and IFN-gamma but not T(H)2 cytokines. Nat. Immunol. cells anergized in vivo. J. Exp. Med. 187, 177–183. 2, 37–44. Chen, L., Glover, J.N., Hogan, P.G., Rao, A., and Harrison, S.C. Kiani, A., Rao, A., and Aramburu, J. (2000). Manipulating immune (1998). Structure of the DNA-binding domains from NFAT, Fos and responses with immunosuppressive agents that target NFAT. Immu- Jun bound specifically to DNA. Nature 392, 42–48. nity 12, 359–372. Chiang, Y.J., Kole, H.K., Brown, K., Naramura, M., Fukuhara, S., Hu, Lanoue, A., Bona, C., von Boehmer, H., and Sarukhan, A. (1997). ϩ R.J., Jang, I.K., Gutkind, J.S., Shevach, E., and Gu, H. (2000). Cbl-b Conditions that induce tolerance in mature CD4 T cells. J. Exp. regulates the CD28 dependence of T-cell activation. Nature 403, Med. 185, 405–414. 216–220. Lechner, O., Lauber, J., Franzke, A., Sarukhan, A., von Boehmer, Crabtree, G.R. (1999). Generic signals and specific outcomes: sig- H., and Buer, J. (2001). Fingerprints of anergic T cells. Curr. Biol. naling through Ca2ϩ, calcineurin, and NF-AT. Cell 96, 611–614. 11, 587–595. Datta, R., Kojima, H., Yoshida, K., and Kufe, D. (1997). Caspase-3- Li, L., and Dixon, J. (2000). Form, function, and regulation of protein mediated cleavage of protein kinase C theta in induction of apopto- tyrosine phosphatases and their involvement in human diseases. sis. J. Biol. Chem. 272, 20317–20320. Semin. Immunol. 12, 75–84. Li, W., Whaley, C.D., Mondino, A., and Mueller, D.L. (1996). Blocked Dolmetsch, R.E., Lewis, R.S., Goodnow, C.C., and Healy, J.I. (1997). signal transduction to the ERK and JNK protein kinases in anergic Differential activation of transcription factors induced by Ca2ϩ re- CD4ϩ T cells. Science 271, 1272–1276. sponse amplitude and duration. Nature 386, 855–858. Li, Y., Zheng, X.X., Li, X.C., Zand, M.S., and Strom, T.B. (1998). Eberhard, D., Jimenez, G., Heavey, B., and Busslinger, M. (2000). Combined costimulation blockade plus rapamycin but not cyclospo- Transcriptional repression by Pax5 (BSAP) through interaction with rine produces permanent engraftment. Transplantation 66, 1387– corepressors of the Groucho family. EMBO J. 19, 2292–2303. 1388. Feske, S., Giltnane, J., Dolmetsch, R., Staudt, L.M., and Rao, A. Li, X.C., Wells, A.D., Strom, T.B., and Turka, L.A. (2000). The role of (2001). Gene regulation mediated by calcium signals in T lympho- T cell apoptosis in transplantation tolerance. Curr. Opin. Immunol. cytes. Nat. Immunol. 2, 316–324. 12, 522–527. Fields, P.E., Gajewski, T.F., and Fitch, F.W. (1996). Blocked Ras ϩ Lyakh, L., Ghosh, P., and Rice, N.R. (1997). Expression of NFAT- activation in anergic CD4 T cells. Science 271, 1276–1278. family proteins in normal human T cells. Mol. Cell. Biol. 17, 2475– Garside, P., and Mowat, A.M. (2001). Oral tolerance. Semin. Immu- 2484. nol. 13, 177–185. Macian, F., Garcia-Rodriguez, C., and Rao, A. (2000). Gene expres- Gastman, B.R., Johnson, D.E., Whiteside, T.L., and Rabinowich, sion elicited by NFAT in the presence or absence of cooperative H. (1999). Caspase-mediated degradation of T-cell receptor zeta- recruitment of Fos and Jun. EMBO J. 19, 4783–4795. chain. Cancer Res. 59, 1422–1427. Macian, F., Lopez-Rodriguez, C., and Rao, A. (2001). Partners in Glynne, R., Akkaraju, S., Healy, J.I., Rayner, J., Goodnow, C.C., and transcription: NFAT and AP-1. Oncogene 20, 2476–2489. Transcriptional Regulation of T Cell Tolerance 731

Maloy, K.J., and Powrie, F. (2001). Regulatory T cells in the control of immune pathology. Nat. Immunol. 2, 816–822. Okamura, H., Aramburu, J., Garcia-Rodriguez, C., Viola, J.P.B., Raghavan, A., Tahiliani, M., Zhan, X.L., Qin, J., Hogan, P.G., and Rao, A. (2000). Concerted dephosphorylation of the transcription factor NFAT1 induces a conformational switch that regulates transcriptional activity. Mol. Cell 6, 539–550. Olson, E.N., and Williams, R.S. (2000). Remodeling muscles with calcineurin. Bioessays 22, 510–519. Peng, L.P., Gerth, A.J., Ranger, A.N., and Glimcher, L.H. (2001). NFATc1 and NFATc2 together control both T and B cell activation and differentiation. Immunity 14, 13–20. Perry, W.L., Hustad, C.M., Swing, D.A., O’Sullivan, T.N., Jenkins, N.A., and Copeland, N.G. (1998). The itchy locus encodes a novel ubiquitin protein ligase that is disrupted in a18H mice. Nat. Genet. 18, 143–146. Rabinowitz, J.D., Beeson, C., Wulfing, C., Tate, K., Allen, P.M., Davis, M.M., and McConnell, H.M. (1996). Altered T cell receptor ligands trigger a subset of early T cell signals. Immunity 5, 125–135. Rao, A., Luo, C., and Hogan, P.G. (1997). Transcription factors of the NFAT family: regulation and function. Annu. Rev. Immunol. 15, 707–747. Sabbattini, P., Lundgren, M., Georgiou, A., Chow, C., Warnes, G., and Dillon, N. (2001). Binding of Ikaros to the lambda5 promoter silences transcription through a mechanism that does not require heterochromatin formation. EMBO J. 20, 2812–2822. Sanjuan, M.A., Jones, D.R., Izquierdo, M., and Merida, I. (2001). Role of diacylglycerol kinase alpha in the attenuation of receptor signaling. J. Cell Biol. 153, 207–220. Schwartz, R.H. (1996). Models of T cell anergy—is there a common molecular mechanism? J. Exp. Med. 184, 1–8. Sloan-Lancaster, J., and Allen, P.M. (1996). Altered peptide ligand- induced partial T cell activation: molecular mechanisms and role in T cell biology. Annu. Rev. Immunol. 14, 1–27. Sloan-Lancaster, J., Steinberg, T.H., and Allen, P.M. (1996). Selec- tive activation of the calcium signaling pathway by altered peptide ligands. J. Exp. Med. 184, 1525–1530. Su, B., Jacinto, E., Hibi, M., Kallunki, T., Karin, M., and Ben-Neriah, Y. (1994). JNK is involved in signal integration during costimulation of T lymphocytes. Cell 77, 727–736. Suga, K., Katagiri, K., Kinashi, T., Harazaki, M., Iizuka, T., Hattori, M., and Minato, N. (2001). CD98 induces LFA-1-mediated cell adhe- sion in lymphoid cells via activation of Rap1. FEBS Lett. 489, 249–253. Teague, T.K., Hildeman, D., Kedl, R.M., Mitchell, T., Rees, W., Schaefer, B.C., Bender, J., Kappler, J., and Marrack, P. (1999). Acti- vation changes the spectrum but not the diversity of genes expressed by T cells. Proc. Natl. Acad. Sci. USA 96, 12691–12696. Telander, D.G., Malvey, E.N., and Mueller, D.L. (1999). Evidence for repression of IL-2 gene activation in anergic T cells. J. Immunol. 162, 1460–1465. Toyoda, M., Kojima, M., and Takeuchi, T. (2000). Jumonji is a nuclear protein that participates in the negative regulation of cell growth. Biochem. Biophys. Res. Commun. 274, 332–336. Wang, C., Deng, L., Hong, M., Akkaraju, G.R., Inoue, J., and Chen, Z.J. (2001). TAK1 is a ubiquitin-dependent kinase of MKK and IKK. Nature 412, 346–351. Xanthoudakis, S., Viola, J.P., Shaw, K.T., Luo, C., Wallace, J.D., Bozza, P.T., Luk, D.C., Curran, T., and Rao, A. (1996). An enhanced immune response in mice lacking the transcription factor NFAT1. Science 272, 892–895. Yankee, T.M., Draves, K.E., Ewings, M.K., Clark, E.A., and Graves, J.D. (2001). CD95/Fas induces cleavage of the GrpL/Gads adaptor and desensitization of antigen receptor signaling. Proc. Natl. Acad. Sci. USA 98, 6789–6793. Supplementary Methods accomplished as follows. (i) Compute the mean and standard deviation of the Absolute mRNA Frequency RNA samples and DNA array procedures: T cells for a gene across all timepoints (0, 2, 6, 16 hours). (ii) were stimulated for 2, 6 or 16 hours with 500 nM Transform the data for that gene by subtracting the ionomycin, 20 nM PMA plus 500 nM ionomycin, or 1 mean, then dividing by the standard deviation. To mM CsA plus 500nM ionomycin. Total RNA was summarize in a single equation: Normalized isolated with an RNeasy kit (QIAGEN). 10 mg of total Frequency = (Absolute Frequency – (Mean of RNA was quantitatively amplified and biotin-labeled Absolute Frequency)) / (Standard Deviation of as described (Byrne et al., 2000). Hybridization to Absolute Frequency). The method used for Genechips (Affymetrix) displaying probes for 11,000 constructing the Self-Organizing Map has been mouse genes/ESTs was performed at 40oC overnight described (Tamayo et al., 1999). in a mix that included 10 mg fragmented RNA, 6X SSPE, 0.005% Triton X-100 and 100 mg/ml herring sperm DNA in a total volume of 200 ml. Chips were References for Supplementary Methods washed, stained with SA-PE and read using an Affymetrix Genechip scanner and accompanying Byrne, M. C., Whitley, M. Z., and Follettie, M. T. (2000). gene expression software. 11 labeled bacterial RNAs Preparation of mRNA for expression monitoring. In Current of known concentration were spiked into each chip Protocols in Molecular Biology (New York, John Wiley and hybridization mix to generate an internal standard Sons, Inc.), pp. 22.22.21-22.22.13. curve, allowing normalization between chips and Hill, A. A., Brown, E. L., Whitley, M. Z., Tucker-Kellogg, G., conversion of raw hybridization intensity values to Hunter, C. P., and Slonim, D. K. (2001). Evaluation of absolute mRNA frequency (mRNA molecules per normalization procedures for oligonucleotide array data million) (see Figure 2B, 3C) (Lockhart et al., 1996; Hill based on spiked cRNA controls. Genome Biol. 2, et al., 2001). Two arrays were performed with RNA R0055.0051-0055.0013. from D5 T cells, and one array each with wildtype and Lockhart, D. J., Dong, H. L., Byrne, M. C., Follettie, M. T., NFAT1-/- Th1 cells. The y-axis of the Self-Organizing Gallo, M. V., Chee, M. S., Mittmann, M., Wang, C. W., Map (SOM) in Figure 2a is normalized mRNA Kobayashi, M., Horton, H., and Brown, E. L. (1996). Frequency. Genes are clustered on the basis of the Expression monitoring by hybridization to high-density kinetic pattern of expression over the 16 hour period oligonucleotide arrays. Nature Biotechnol. 14, 1675-1680. of treatment, independent of expression magnitude, Tamayo, P., Slonim, D., Mesirov, J., Zhu, Q., Kitareewan, S., on the assumption that some of the genes displaying Dmitrovsky, E., Lander, E. S., and Golub, T. R. (1999). similar expression kinetics will be involved in the Interpreting patterns of gene expression with self-organizing same biological process. The normalization was a maps: Methods and application to hematopoietic “mean zero, variance one” normalization, and is differentiation. Proc. Natl. Acad. Sci. USA. 96, 2907-2912.

1 Supplementary Table 1: Accession numbers and identities of the genes whose expression profiles are shown in Figure 2A.

Accession number Description

Panel 1 D44464 Uridine phosphorylase d83999 RNA polymerase II subunit 3 j02870 Laminin receptor j04627 NAD-depend. methylenetetrahydrofolate dehydrog.- methenyltetrahydrofolate cyclohydrolase l02333 Bilirubin/phenol family UDP glucuronosyltransferase (ugtBr) l21027 D-3-phosphoglycerate dehydrogenase M32484 Placenta and embryonic expression gene (pem) m35523 Cellular retinoic acid-binding protein (CRABP-II) u05809 LAF1 transketolase U23922 IL-12 receptor beta component u28217 Protein tyrosine phosphatase STEP61 u28493 Lymphotactin u32469 Reduced folate carrier (RFC1) . u35370 CD6 u67837 Multiple exostosis protein (EXT2) u85511 Nucleoside diphosphate kinase A long form x04663 Beta-tubulin (isotype Mbeta 5). X07414 Excision repair protein ERCC-1. X15963 Cytochrome c oxidase subunit Va X53753 Nonmuscle tropomyosin 5. X78682 B-cell receptor associated protein (BAP) 32.

Panel 2 AF013262 Lumican (Ldc) AJ001101 C1q binding protein (gC1qBP) d86729 Heterogeneous nuclear ribonucleoprotein A1 (HNRNP Core protein A1) d90344 T-complex polypeptide 1A (Tcp-1a) l19311 Spermidine synthase l39357 Migration inhibitory factor (Mif) m12302 Granzyme B L19311 Spermidine synthase X97227 CD53 X70847 Adenine nucleotide translocase L25274 Transmembrane glycoprotein (DM-GRASP) u09659 Chaperonin 10 u12273 Apurinic/apyrimidinic endonuclease (Apex) u27316 Adenine nucleotide translocase-2 (Ant2) X01756 Cytochrome C gene (MC1). x02732 Interleukin-3

2 x53584 HSP60 X68193 Nucleoside diphoshate kinase B. X70100 Mal1 mRNA for keratinocyte lipid-binding protein. x97047 M2-type pyruvate kinase. Y13071 26S proteasome non-ATPase subunit. Z16078 CD53 Z31399 CCT eta subunit (chaperonin containing TCP-1). Z31553 CCT (chaperonin containing TCP-1) beta subunit. Z31555 CCT (chaperonin containing TCP-1) epsilon subunit. Z31557 CCT (chaperonin containing TCP-1) zeta subunit. z49085 Mouse developmental kinase 2 (MDK2).

Panel 3 af031486 Spermidine aminopropyltransferase (Mspmsy) D42048 Squalene epoxidase l34290 Guanine nucleotide regulatory protein beta 5 m13445 Alpha-tubulin isotype M-alpha-1 m60456 Cyclophilin M96823 Nucleobindin X04663 Tubulin, beta 5 X78683 BCR-associated protein 37 x06453 Protein disulfide isomerase X70296 PN-1 (protease-nexin 1) X90582 Signal sequence receptor delta

Panel 4 AB000733 AF1q D50589 Prostaglandin E2 Receptor, EP2 Subtype d85570 Proteasome subunit Z (PSMB7) D87691 Eukaryotic translation termination factor 1 (eRF1) d87911 Proteaseome 28 subunit, 3 d90151 CArG-binding factor-A j04633 Heat shock protein 86 K02236 Metallothionein II (MT-II) L09754 Cd30 ligand l32751 GTPase (Ran) m17015 Lymphotoxin (LT) U28404 Macrophage inflammatory protein-1 alpha receptor X16857 HSP86 heat-shock protein U16162 Prolyl 4-hydroxylase alpha(I)-subunit U63648 Myb-binding protein (P160) u69135 Uncoupling protein 2 (UCP2) x06143 CD2 X54511 Myc basic motif homologue-1 (mbh1). Z22593 Fibrillarin

3 Panel 5 AF030559 ATP synthase beta-subunit (beta-F1 ATPase) d55720 Importin alpha2 d82019 Basigin D85904 Apg-2 L15447 Small nuclear RNA (Rnu1a-1) L19737 H+ ATP synthase subunit c L25274 Transmembrane glycoprotein (DM-GRASP) u13393 Delta proteasome subunit (lmp19) U27830 Stress induced hosphoprotein 1 (mSTI1) U31758 Histone deacetylase 2 u43548 Transcription factor like protein 4 TCFL4 u70315 U1 snRNP-specific protein C U96700 Proteinase inhibitor 6 (SPI6) x04480 Preproinsulin-like growth factor IA. x53526 OX45 X59047 CD81

Panel 6 aa161799 Phosphoglycerate mutase (Pgam1) D10024 Calpactin 1 (annexin II) D50031 TGN38A l42115 Insulin-activated amino acid transporter mRNA m14044 Calpactin I heavy chain (p36) m18186 84 kD heat shock protein X16151 Osteopontin X53333 Triosephosphate isomerase X01756 Cytochrome C gene D10024 Protein-tyrosine kinase substrate p36 (calpactin I heavy chain) M18186 84 kD heat shock protein U77040 LIM protein 3 (mSLIM3) X16151 Eta-1, osteopontin x51834 Osteopontin. X67914 PD-1

Panel 7 d78645 78 kDa glucose-regulated protein (grp78) D87990 UDP-galactose transporter related isozyme 1 m94087 ATF4 U17343 Signal recognition particle receptor beta subunit L15447 Small nuclear RNA (Rnu1a-1) U11027 Sec61 protein complex gamma subunit U12922 CD1 geranylgeranyl transferase beta subunit u84211 Defender against death 1 protein (DAD1) Y11666 Hexokinase II

4 Panel 8 ET63083 splicing factor PR264 L20294 GTP-binding protein (mSara) homologue m13227 Enkephalin M58558 Sm D small nuclear ribonucleoprotein sequence. M64085 Spi2 proteinase inhibitor (spi2/eb1) u02298 Rantes (Scya5) U14648 Putative myelin regulatory factor 1 u19482 macrophage inflammatory protein-1gamma u21103 Stat5a x14045 IL-9

Panel 9 GMCSF GMCSF k00083 Interferon-gamma l07264 Heparin-binding EGF-like growth factor m17957 T-cell activation gene 3 (TCA3) m35590 Macrophage inflammatory protein 1-beta (MIP-1) m83749 D-type cyclin (CYL2) MIP1-B MIP1-B M23501 Secreted T cell protein P500 L36611 Protein synthesis initiation factor 4A (elF-4A) X54511 Myc basic motif homologue-1 (mbh1) U72881 Regulator of G-protein signaling-retinal specific u05265 gp49B U19799 IkB-beta U20159 SLP-76 U39827 Putative G protein-coupled receptor TDAG8 (TDAG8) U92972 Protease-activated receptor 3 (PAR3) x03019 Granulocyte-macrophage colony stimulating factor (GM-CSF). x03040 Initiation factor eIF-4A x12531 Macrophage inflammatory protein (MIP).

Panel 10 AA003458 Serca 2 AF032128 Ornithine decarboxylase antizyme inhibitor d67016 Heat shock protein 105 kDa alpha d78354 Phospholipid scramblase 1 M58566 TIS11, butyrate response factor 1 M65027 gp49 u94828 Retinally abundant regulator of G-protein signaling mRGS-r (RGS-r) x06746 EGR-2 X53068 Proliferating cell nuclear antigen.

Panel 11 af019048 Receptor activator of nuclear factor kappa B ligand (RANKL)

5 d31943 Cytokine inducible SH2-containing protein ET61251 Putative serine/threonine kinase (Fnk) ET61450 Interleukin-2 (IL-2) ET63426 Tumor necrosis factor alpha IL2 IL-2 k02891 Interleukin 2 receptor l00039 c-myc L16462 BCL-2 related protein (A1) L28117 NF-kappa-B (p105) U35142 Retinoblastoma-binding protein (mRbAp46) X76850 MAP kinase-activated protein kinase 2 U39192 Heparin-binding epidermal growth factor-like growth factor TNFa TNFalpha u10551 Gem GTPase (gem) u11692 Lymphoid-specific interferon regulator factor (LSIRF) u19118 ATF3 U44088 TDAG51 u59463 Caspase 11 U70139 Nocturnin U88328 Suppressor of cytokine signalling-3 (SOCS-3) x02611 Tumour necrosis factor (TNF). X67644 gly96 x85214 ox40 z72000 BTG3

Panel 12 ab000677 SOCS1 AF006492 Friend of GATA-1 (FOG) d10061 DNA topoisomerase I D13759 Mitogen activated protein kinase kinase kinase 8 D87115 MAP kinase kinase 3b L16956 Jak2 L16956 Jak2 l26489 Furin (FUR) M32489 Interferon consensus sequence binding protein m59821 Ier2, pip92 M83380 Transcription factor relB M95200 Vascular endothelial growth factor M13945 Pim-1 protein kinase L42462 TGF-1 U10102 Bcl-x long U29056 Src-like adapter protein SLAP L48687 Voltage-dependent Na+ channel beta-1 subunit L41352 Amphiregulin u09507 p21 (Waf1) u20735 junB

6 u33840 TRAF-3 U35142 Retinoblastoma-binding protein (mRbAp46) U47543 Mus musculus NGFI-A binding protein 2 (NAB2) u73478 Acidic nuclear phosphoprotein pp32 u78031 Apoptosis inhibitor bcl-x (bcl-x) gene u85786 Mus musculus sodium channel beta-1 subunit mRNA, complete cds. u86783 Steroid/thyroid hormone orphan nuclear receptor (Nurr1) x16995 NUR77 X54149 MyD118

Panel 13 D31967 Jumonji ET62419 p56lck-associated adapter protein Lad j02930 Laminin B2 chain L12120 Interleukin-10 receptor (Il10r) U09507 Cyclin dependent kinase inhibitor 1A (P21) U19463 Zinc finger protein A20 (murine A20) U04379 Protein tyrosine kinase ZAP-70 D90242 nPKC-eta L15435 Mus musculus 4-1BB ligand L15435 Mus musculus 4-1BB ligand U88327 Suppressor of cytokine signalling-2 (SOCS-2) X51829 MyD116

Panel 14 M64292 TIS21 X89749 TGIF y07711 Zyxin. z30940 Histone H2A

Panel 15 AA119194 Rab10 AA061283 DAGkinase alpha (Dagk1) AA066032 DAGkinase alpha (Dagk1) aa710204 Acetyl-CoA acetyl transferase ET63241 Caspase-3 Ikaros Ikaros L03547 Ikaros Y00309 LDH-A u23921 Osmotic stress protein 94 (Osp94) U44731 GBP-3 U67187 G protein signaling regulator RGS2 (rgs2)

Panel 16 af003695 Hypoxia-inducible factor 1 alpha (Hif1a) d28530 Protein tyrosine phosphatase PTPT9 (rPTPps)

7 d38023 PIMT D88315 Tetracycline transporter-like protein L10106 R-PTP-kappa l33768 JAK3 m21952 Macrophage colony-stimulating factor M60474 Myristoylated alanine-rich C-kinase substrate (MARCKS) X57024 Glutamate dehydrogenase U05252 Nuclear matrix attachment DNA-binding protein SATB1 u10871 MAP kinase p38 u30464 Wnt-10B u54803 Caspase 3 U61363 Groucho-related gene 4 protein (Grg4) U83176 ROSA 26 X57024 GLUD (glutamate dehydrogenase)

Panel 17 AA089339 Leukocystatin aa215251 GEG-154, RNF19 or Xybp aa221219 Myoinositol 1-phosphate synthase A1 (IsynA1) AF034610 Nuclear autoantigenic sperm protein d78141 TRAF5 U25708 CD98 heavy chain mRNA, complete cds X71642 GEG-154 u06948 Fas ligand U83148 NFIL3/E4BP4 X14425 Profilin x64068 Cation-dependent mannose-6-phosphate receptor. x66081 Pgp-1 mRNA for CD44 X71642 GEG-154, RNF19 or Xybp W90837 Chloride intracellular channel (Clic1)

Panel 18 ac002397 PTPN6 AF011644 Oral tumor suppressor homolog (Doc-1) D10712 Nedd-1 protein D11091 Protein kinase C theta j02990 CD3-epsilon L02526 MAPKK1 L10244 Spermidine/spermine N1-acetyltransferase (SSAT) m64279 BMI-1) m81477 Protein tyrosine phosphatase, non-receptor type 2 m97636 Helix-loop-helix transcription factor ME1 U19617 Ets-family transcription factor Elf-1 U41341 Calgizzarin U72941 Annexin IV mRNA, complete cds.

8 Panel 19 AA125310 Elastse Inhibitor AA145127 Serpinb1 aa242017 Enoyl coenzyme A hydratase (Auh) AF013632 FAN protein d14566 Lmp-2 ET62534 CD4 l25069 Catalase m37761 Calcyclin m55637 HAM1 m59470 Cystatin C L25885 beta 1,4N-acetylgalactosaminyltransferase L11613 Proteasome Lmp2 U07950 GDP-dissociation inhibitor (GDI-1) U12283 USF2 X66449 Calcyclin u04268 Sca-2 u06923 Signal transducer and activator of transcription (Stat4) U07890 Flotillin 2 u12283 USF2 U19119 Interferon inducible protein 1 U19596 Cdk4 and Cdk6 inhibitor p18 protein u24674 Max interacting protein 1 (Mxi1) u57325 Presenilin 2 X66449 Calcyclin x75129 Xanthine dehydrogenase z14252 Acidic sphingomyelinase (ASM-1)

Panel 20 AF001863 FYN binding protein AF001863 FYN binding protein m89956 Calcium-binding protein (pp52) J05261 Cathepsin A M38381 Serine threonine tyrosine kinase (STY) Z25524 Integrin associated protein CD47 u16985 Lymphotoxin-beta U43673 IL-1Rrp U49351 Lysosomal alpha-glucosidase U69535 Semaphorin M-sema G x03533 Tyrosine protein kinase p56-lck. x73359 AES-1

9 140 18 35 30 AF003695 D31967 L03547 U05252

Hif-1(Hypoxia- SATB1 (Nuclear inducible factor 1) Jumonji matrix attachment Ikaros DNA-binding protein) 0 0 0 0 Transcription 30 15 25 35 U19617 U61363 U83148 X71642

Elf-1(ETS-family GRG-4 (Groucho- RNF19 (Ring transcription factor) related gene 4 NFIL3/E4BP4 finger protein 19, protein) GEG-154, Xybp)

0 0 0 0

80 30 60 D28530 L10106 U24700 Receptor and RPTP-sigma RPTP-kappa PTP-1B non-receptor (Receptor (Receptor protein tyrosine (Protein tyrosine tyrosine protein tyrosine phosphatase 1B) phosphatases phosphatase sigma) phosphatase kappa) 0 0 0

15 75 60 AA119194 U44731 U67187

GBP-3 (Guanylate RGS2 (G protein G-protein Rab10 binding protein 3) signaling regulator-2) signalling

0 0 0

14 15 25 14 AA245242 ET62419 DAGKa (Diacyl AA066032 U04379 Other signalling Lad/TSAd ZAP70 (protein glycerol Mlp (Marcks- (p56lck-associated kinase alpha) like protein) tyrosine kinase adapter protein) ZAP-70)

0 0 0 0 0 2 6 16 Time (hours)

35 30 180 AA125310 D78141 M59470

Traf5 (TNF receptor- Serpinb 1b Cystatin C (Serpin proteinase associated protein 5) inhibitor 1b) Proteolytic 0 0 0 pathways 40 80 U54803 U88327 SOCS-2 Caspase 3 (Suppressor of cytokine signalling-2)

0 0

100 80 300 AA161799 X57024 Y00309

Metabolic Pgam1 Glutamate LDH-A (Lactate (Phosphoglycerate dehydrogenase dehydrogenaseAa) enzymes mutase)

0 0 0

160 140 80 J02990 L15435 U25708 CD3 epsilon (T cell receptor 4-1BB-L CD98 Heavy chain CD3 epsilon chain) (4-1BB ligand)

Cell-surface 0 0 receptors & 0 200 250 180 associated U06948 X64068 Z31137 proteins Gamma-aminobutyric FasL (Fas ligand) Cation-dependent mannose-6-phosphate acid receptor- receptor associated protein- like protein 1 0 0 0

25 240 800 D88315 M21952 M37761

Tetracycline MCSF (Macrophage transporter-like stimulating factor) Calcyclin protein (cytokine)

0 0 0 0 2 6 16 Miscellaneous Time (hours) 50 25 Z31202 U23921

Heme oxygenase 2a Osp94 (Osmotic stress protein 94)

0 0 0 2 6 16 0 2 6 16 PMA+ Ionomycin Ionomycin Alone Ionomycin +CsA Time (hours) Time (hours)

Supplementary Figure 1: Expression profiles of the 37 genes activated by ionomycin in D5 and primary Th1 cells. The genes are grouped into eight categories based on function.