Genome-Wide Identification of Human FOXP3 Target in Natural Regulatory T Cells

This information is current as Timothy J. Sadlon, Bridget G. Wilkinson, Stephen Pederson, of September 26, 2021. Cheryl Y. Brown, Suzanne Bresatz, Tessa Gargett, Elizabeth L. Melville, Kaimen Peng, Richard J. D'Andrea, Gary G. Glonek, Gregory J. Goodall, Heddy Zola, M. Frances Shannon and Simon C. Barry

J Immunol published online 16 June 2010 Downloaded from http://www.jimmunol.org/content/early/2010/06/16/jimmun ol.1000082

Supplementary http://www.jimmunol.org/content/suppl/2010/06/16/jimmunol.100008 http://www.jimmunol.org/ Material 2.DC1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 26, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published June 16, 2010, doi:10.4049/jimmunol.1000082 The Journal of Immunology

Genome-Wide Identification of Human FOXP3 Target Genes in Natural Regulatory T Cells

Timothy J. Sadlon,* Bridget G. Wilkinson,* Stephen Pederson,* Cheryl Y. Brown,* Suzanne Bresatz,* Tessa Gargett,* Elizabeth L. Melville,† Kaimen Peng,‡ Richard J. D’Andrea,x,{ Gary G. Glonek,‖ Gregory J. Goodall,{ Heddy Zola,# M. Frances Shannon,‡ and Simon C. Barry*,†

The transcription factor FOXP3 is essential for the formation and function of regulatory T cells (Tregs), and Tregs are essential for maintaining immune homeostasis and tolerance. This is demonstrated by a lethal autoimmune defect in mice lacking Foxp3 and in immunodysregulation polyendocrinopathy enteropathy X-linked syndrome patients. However, little is known about the molecular

basis of human FOXP3 function or the relationship between direct and indirect targets of FOXP3 in human Tregs. To investigate Downloaded from this, we have performed a comprehensive genome-wide analysis for human FOXP3 target genes from cord blood Tregs using chromatin immunoprecipitation array profiling and expression profiling. We have identified 5579 human FOXP3 target genes and derived a core Treg signature conserved across species using mouse chromatin immunoprecipitation data sets. A total of 739 of the 5579 FOXP3 target genes were differentially regulated in Tregs compared with Th cells, thus allowing the identification of a number of pathways and biological functions overrepresented in Tregs. We have identified gene families including cell surface +

molecules and microRNAs that are differentially expressed in FOXP3 Tregs. In particular, we have identified a novel role for http://www.jimmunol.org/ peptidase inhibitor 16, which is expressed on the cell surface of >80% of resting human CD25+FOXP3+ Tregs, suggesting that in conjunction with CD25 peptidase inhibitor 16 may be a surrogate surface marker for Tregs with potential clinical application. The Journal of Immunology, 2010, 185: 000–000.

egulatory T cells (Tregs) are essential “master regulators” appears to require Treg function (8). The ability to monitor and of tolerance (1), allowing effective responses to infection manipulate Tregs may improve diagnosis or therapeutic interven- R and non-self Ags while preventing autoimmunity and tion for autoimmune diseases. To achieve this, sensitive and spe- immune pathology (2). Although there are several subsets of cells cific biomarkers are required, but currently, however, there are with regulatory capacity (reviewed in Ref. 3), it is clear that the limited markers available. This genome-wide study of human Tregs by guest on September 26, 2021 CD4+CD25+FOXP3+ natural Tregs (nTregs) is a key participant in takes an unbiased approach for identification of novel markers the orchestration of tolerance, because without these cells an early and gene regulation networks. onset aggressive autoimmune condition arises in both mouse and The forkhead (FKH) transcription factor FOXP3 is expressed man (4–6). Furthermore, a number of autoimmune diseases in- constitutively in natural CD4+ Tregs and has been shown to be volve a perturbation in Treg numbers or function (reviewed in essential for nTreg function (9). FOXP3 is a member of the P Ref. 7), and control of the local response to acute tissue damage subfamily of the FKH family of transcription factors sharing a number of common features including a C2H2 zinc finger, a leu- cine zipper, and a FKH DNA-binding domain (10). These domains # *Molecular Immunology Laboratory and Leukocyte Biology Laboratory, Women’s are involved in DNA binding, nuclear transport (11), homomeric and Children’s Health Research Institute, Women’s and Children’s Hospital, North Adelaide; †Discipline of Paediatrics and ‖Discipline of Statistics, School of Mathe- and heteromeric complex formation (12), and transcriptional re- x matical Sciences, University of Adelaide; Department of Haematology and Oncol- pressor activity (12, 13). Mechanisms of action of FOXP3 in Tregs ogy, Queen Elizabeth Hospital; {Centre for Cancer Biology, South Australia Pathology, Adelaide, South Australia; and ‡Gene Expression and Epigenomics have been identified in mouse, but less is known about its role Group, Genome Biology Program, Biomolecular Resource Facility, John Curtin in the development and function of human Tregs (reviewed in Ref. School for Medical Research, Australian National University, Canberra, Australian 12). In particular, the existence of two major human-specific Capital Territory, Australia FOXP3 isoforms (14) with different regulatory capacities (15, Received for publication January 12, 2010. Accepted for publication May 3, 2010. 16) and the transient induction of FOXP3 following activation of The microarray data presented in this article have been submitted to Gene Expression CD4+CD252 non-Tregs in humans (17) suggest a degree of spe- Omnibus under accession numbers GSE20995 and GSE20934. cies-specific function by, and regulation of, FOXP3. Address correspondence and reprint requests to Dr. Simon C. Barry, Molecular Immunology Laboratory, Discipline of Paediatrics, University of Adelaide, and In mouse and human Tregs, FOXP3 not only represses but also Women’s and Children’s Health Research Institute, Women’s and Children’s Hospi- can induce gene expression, although the mechanism by which tal, North Adelaide, South Australia 5006, Australia. E-mail address: simon.barry@ FOXP3 can negatively regulate transcription of some genes, such adelaide.edu.au as IL2, but positively regulate others, such as CD25 and CTLA4,is The online version of this article contains supplemental material. still unclear (reviewed in Ref. 10). FOXP3 interacts with multiple Abbreviations used in this paper: ChIP-on-chip, chromatin immunoprecipitation array profiling; DE, differentially expressed; FDR, false discovery rate; FKH, forkhead; transcription factors known to be involved in activation, differ- + miR, microRNA; mRNA, messenger RNA; nTreg, natural regulatory T cell; PI16, entiation, and response of CD4 T cells to TCR stimulation, in- peptidase inhibitor 16; qPCR, quantitative real-time PCR; ROR, retinoic acid recep- cluding NFAT, NF-kB, Runx1, retinoic acid receptor-related tor-related orphan receptor; Treg, regulatory T cell; TSS, transcription start site. orphan receptors (RORs) (RORa and RORgT), IFN regulatory Copyright Ó 2010 by The American Association of Immunologists, Inc. 0022-1767/10/$16.00 factor 4, STAT3, and Jun (15, 16, 18–24). FOXP3 may act as

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1000082 2 FOXP3 TARGET GENES IN HUMAN nTregs a transcriptional corepressor, because FKH domain-independent hybridization to Affymetrix Human Tiling 2.0R arrays was by whole- inhibition of the trans-activation activity of NF-kB, CREB, and genome amplification (Sigma-Aldrich GenomePlex Whole Genome Am- RORa has been reported (16, 21). FOXP3 is also likely to mod- plification Kit) (31) adapted for Affymetrix labeling systems by the inclu- sion of 0.11 mM dUTP in the amplification reaction. Amplified material ulate gene expression through epigenetic mechanisms, such as was purified using a PCR clean-up kit (Qiagen, Valencia, CA). Data from chromatin remodeling and histone deacetylation (25, 26). It is two independent ChIP-on-chip experiments were analyzed as replicates evident that the composition and activity of the FOXP3 regulatory using Model-based Analysis of Tiling-array (32). Labeling and complex is dynamic and is influenced by external stimuli acting hybridization of amplified material to Affymetrix human tiling arrays was carried out at the Biomolecular Resource Facility (John Curtin School of partly through the posttranslational modification of FOXP3 (27, Medical Research, Australian National University). Data files for this 28). These observations demonstrate the complex nature of the experiment are lodged on Gene Expression Omnibus (accession no. interaction of FOXP3 with its target genes. GSE20995; www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20995). In this article, we identify molecular targets of human FOXP3 in ChIP-on-chip data analysis nTregs using a combination of chromatin immunoprecipitation array profiling (ChIP-on-chip) and expression profiling. Defining Data from two independent FOXP3 ChIP whole-genome array sets were the direct binding sites for FOXP3 reveals target genes that may analyzed as replicates using Model-based Analysis of Tiling-array (32). A false discovery rate (FDR) of 0.5% was used to identify significantly be regulated by FOXP3 and allows a detailed analysis of FOXP3- enriched regions in FOXP3-immunoprecipitated material relative to input bound promoter elements, providing important information relat- chromatin. Gene accessions were assigned to an individual ChIP region if ing to FOXP3 transcription factor function. We have also identified the peak of the enriched region was within 20 kb upstream of a transcrip- direct targets of FOXP3 that are differentially expressed in human tion start site (TSS) and 20 kb downstream of the transcription end site. Annotation was performed using gene accessions from both University of nTregs, and analysis of the FOXP3 target genes and differentially Downloaded from California Santa Cruz and National Center for Biotechnology Information, regulated genes has uncovered a number of transcription factors, encompassing RefSeqs, mRNAs, and miRs, and where more than one microRNAs, signaling molecules, and surface molecules that have accession could be associated with a ChIP peak, all possible annotations potentially important roles in Treg formation or function. In the were recorded. Species conservation was analyzed using VISTA software search for positive cell surface markers for human Tregs, we report (http://genome.lbl.gov/vista/index.shtml) (33). for the first time that peptidase inhibitor 16 (PI16) is expressed RNA preparation and expression array on the surface of a significant proportion of human Tregs, pointing http://www.jimmunol.org/ to the potential utility of this molecule as a new biomarker for Differential expression analysis was carried out using Affymetrix Human Exon 1.0 ST arrays. Total RNA was isolated from expanded Th cells and nTregs. nTregs that were rested for ∼60 h prior to treatment for 2 h with either ionomycin or vehicle (DMSO). Total RNA, including small m.w. RNA, was isolated using QIAshredder and a miRNeasy Mini Kit (Qiagen). RNA Materials and Methods quality was assayed using an Agilent Systems Bioanalyzer (Santa Clara, Isolation, in vitro expansion, and characterization of cord CA). Labeling and hybridization were carried out according to the man- blood T cell populations ufacturer’s protocols at the Biomolecular Resource Facility (John Curtin School of Medical Research, Australian National University). Data files for Cord blood was obtained with approval from the donor and the Children’s, this experiment are lodged on Gene Expression Omnibus (accession no.

Youth and Women’s Health Service Research Ethics Committee. Mono- GSE20934; www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20934). by guest on September 26, 2021 nuclear cells were isolated from cord blood postpartum as previously de- 2 scribed (29). Briefly, cord blood CD4+CD25+ (Treg) and CD4+CD25 (Th) RNA expression array data normalization cells were isolated from purified mononuclear cells using a Regulatory CD4+CD25+ T Cell Kit (Dynabeads; Invitrogen, Carlsbad, CA). Ex vivo Microarray data were processed using RMA (34) and version 11.0 of an expansion of isolated T cell populations (1 3 106 cells per well in a 24-well Entrez Gene-centric cdf (35). All of the processing was conducted using plate) were performed in X-Vivo 15 media supplemented with 5% human the statistical software R (R Development Core Team) under the aroma. AB serum (Lonza, Walkersville, MD), 20 mM HEPES (pH 7.4), 2 mM affymetrix framework (36). Log fold change in a four-way comparison L-glutamine, and 500 U/ml recombinant human IL-2 (R&D Systems, between resting and stimulated treatments of nTregs and Th cells was Minneapolis, MN) in the presence of CD3/CD28 T cell expander beads estimated within individual donors after Loess normalization. Final expres- (Dynabeads; Invitrogen; catalog no. 111-41D) at a bead-to-cell ratio of sion analysis was performed using Limma (37), and raw p values for each 3:1. The phenotypes of expanded cells were characterized by surface ex- term were adjusted globally to provide an estimate of the FDR (38). pression of CD4 (Phycoerythrin-CY5; eBioscience, San Diego, CA; clone RPA-T4), CD25 (PE; BD Biosciences, San Jose, CA; clone M-A251), and Differential gene expression group assignment and analysis CD127 (PE; eBioscience; clone ebioRDR73) in combination with intracel- Differentially expressed genes were divided into seven mutually exclusive lular detection of FOXP3 (Foxp3-Alexa Fluor 488; BD Biosciences, San groups (T1–T6 and A1), based upon their behavior across all four com- Jose, CA; clone 259D/C7; Human FOXP3 Buffer Set; BD Biosciences; parisons. For genes with a significant adjusted p value in only one com- catalog no. 560098) by three-color flow cytometry on a Beckman Coulter parison, the accepted FDR was raised to p , 0.1 in all of the other ´ Epics Elite ESP flow cytometer (Beckman Coulter, Fullerton, CA). Func- comparisons to enable more accurate group assignment. Clustering within tional assays using unmatched donor MLR suppressor assays were carried each subgroup was performed using the AGNES algorithm (M. Maelcher, out as described previously (29). unpublished observations), based on log fold-change and whether there ChIP, ChIP PCR, and whole-genome array was an associated FOXP3 binding site. Overrepresentation of ChIP hits within individual groups was assessed by using Fisher’s exact tests (all p Expanded cord blood Tregs were cultured overnight in X-Vivo 15 media with values ,1027). 100 U/ml recombinant human IL-2 prior to a 2 h restimulation with 1 mM ionomycin (Sigma-Aldrich, St. Louis, MO) before cross-linking. Isolated Pathway mapping T cells were subjected to cross-linking for 10 min in 1% formaldehyde solution (50 mM HEPES KOH [pH 7.5], 100 mM NaCl, 1 mM EDTA, FOXP3 ChIP hits and the sets of differentially expressed genes comprising 0.5 mM EGTA, and 11% formaldehyde). Formaldehyde was quenched by groups T1–T6 were assessed for overrepresentation of genes from ca- the addition of glycine to a final concentration of 112 mM. Cell lysis, ChIP, nonical pathways and known biological functions either individually or and DNA isolation steps for human FOXP3 ChIP-on-chip experiments were together using Fisher’s exact tests within the Ingenuity Pathways Analysis carried out essentially as described in Ref. 30, using 5 3 107 cells per software package (Ingenuity, Redwood, CA). immunoprecipitation with either a rabbit anti-Foxp3 IgG (Novus Biochem, Transcription factor binding motif analysis Littleton, CO) or ChIP-grade control rabbit IgG Sera (Abcam, Cambridge, U.K.). Chromatin shearing conditions consisted of 12 repeats of 30 s pulses Transcription factor binding site analysis was carried out on 400-bp sequen- with an amplitude of 90% with 90 s between pulses with a S-4000 ultrasonic ces centered on the ChIP peak associated with either the top 1000 FOXP3- processor (Misonix, Farmingdale, NY) with a microtip. Amplification of bound regions (ranked on p value) or genes differentially regulated in immunoprecipitated and input chromatin (10 ng) for labeling and Tregs. Scans for overrepresented motifs were performed on these regions The Journal of Immunology 3 using WEEDER and MatInspector (Genomatix, Munich, Germany) (39– 42). Motifs identified by WEEDER were matched against known transcrip- tion factor binding motifs in JASPAR (43). For the 739 genes that were ChIP hits and members of groups T1–T6, 400-bp genomic sequences centered at the ChIP peak were scanned for the presence of transcription factor binding motifs using the default settings of MatInspector. Of these sequences, the 681 identified as containing one or more binding motifs belonging to the FKH family were further examined for co-occurrence of other transcription factor family motifs. A reference set of sequences was generated by randomly selecting 1000 genes from the Entrez Gene data- base that were not FOXP3 ChIP hits and selecting 400-bp sequences around the TSS using the observed distribution for the differentially reg- ulated ChIP hits, as described above. A further set of 701 sequences con- taining a FKH motif was extracted as a subset of this reference set. Transcription factor families present in at least 30% of the differentially expressed ChIP genes with an FDR-adjusted p value , 0.05 were consid- ered significantly overrepresented.

Quantitative real-time PCR Validation of differentially regulated target genes identified by expression array analysis was performed on RNA from five donors. Random primed cDNA was prepared from total RNA using a QuantiTect Reverse Tran- Downloaded from scription Kit (Qiagen). Relative gene expression analysis was carried out using SYBR Green quantitative real-time PCR (qPCR) on a Rotor-Gene 6000 PCR machine (Corbet Research, Quiagen, Valencia, CA). Gene- specific primers pairs for qPCR were selected from PrimerBank (44). A primer set specific for RPL13a was used as an internal control (primer sequences listed in Supplemental Table X). Results were analyzed using FIGURE 1. Expanded cord blood nTregs express FOXP3 and have

a Rotor-Gene 6000, QGene software (45), and R. Validation of FOXP3 suppressive activity. A, Flow cytometric analysis of surface CD4 and CD25 http://www.jimmunol.org/ binding to genomic regions was carried out by ChIP qPCR. For ChIP, PCR and intracellular FOXP3 expression. Isotype control of CD4+ and CD25+ primers to amplify target regions were either designed using Primer 3 Plus population (dotted line), FOXP3 expression in CD252 cells (dashed line), (www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) or purchased + + and FOXP3 expression in CD4 and CD25 cells (solid line). B, Functional from SABiosciences, Frederick, MD as validated ChIP PCR primers (Cham- pionChIP PCR assays). Reactions were performed using a RT2 SYBR activity of expanded cells in a MLR. Expanded cord blood nTregs were Green/ROX qPCR Master Mix (SABiosciences). The relative enrichment cultured for 5 d with unmatched T cells, APCs, and soluble anti-CD3 prior 3 of target regions in FOXP3-immunoprecipitated material relative to input to a 16 h pulse of [ H]thymidine. Robust suppression of proliferation of 2 chromatin analysis was carried out using a ChampionChIP data analysis CD4+CD25 responder cells by expanded nTregs was observed from a re- 2ΔΔ spreadsheet (SABiosciences) using the 2 CT method. MicroRNA expres- sponder-to-nTreg ratio of 1:1 to 10:1. sion level comparisons were performed on total RNA (miRNeasy kits; Qiagen) isolated from freshly purified Tregs or Th cells by real-time by guest on September 26, 2021 PCR using specific TaqMan miR assays (Applied Biosystems, Carlsbad, CA). Data were normalized to a housekeeper (RNU6B) and non-FOXP3- Identification of human FOXP3 binding sites using regulated miRS (miR-16/miR-24). RNA from three independent donors ChIP-on-chip assays was used to confirm the differential expression. The human FOXP3 ChIP protocol was established using the Validation of cell surface molecule expression IL7Ralpha (CD127) and IL2 genes (Supplemental Fig. 1), because both promoters have been previously reported to contain binding For surface molecule analysis, freshly isolated adult peripheral blood CD4+ + + 2 sites for human FOXP3 (23, 48). FOXP3-bound DNA isolated by CD25 Tregs or CD4 CD25 Th cells were either assayed immediately or stimulated overnight in the presence of CD3/CD28 beads (bead-to-cell ratio, ChIP was amplified using a modification of the whole-genome 1:1) and 100 U/ml IL-2 prior to four-color flow cytometry using Abs against amplification method (31), and enrichment of the same target CD4, CD25, FOXP3, and test Ag. For PI16 detection, a mouse polyclonal Ab genes postamplification was confirmed (data not shown) prior to was used (Abnova, Walnut, CA; catalog no. H00221476-B01P) labeling and hybridization to Affymetrix Human Tiling 2.0R arrays. With a FDR of 0.5%, a total of 8308 unique genomic regions were determined as significantly enriched in the FOXP3 Results ChIP material (Supplemental Table I). As a first step in validating Isolation and expansion of human cord blood Tregs the human FOXP3 ChIP-on-chip data set, a search for the pres- We chose to use ex vivo expanded cord blood nTregs to dissect the ence of FKH DNA-binding motifs was performed on the top 1000 FOXP3-driven events within human Tregs because large numbers FOXP3-bound genomic regions (ranked on p value) using Mat- can be grown for ChIP-on-chip studies and these cells display Inspector (Genomatix). FOXP3-bound regions were significantly strong FOXP3 expression and a robust suppressive ability in vitro enriched for FKH motifs when compared against the Genomatix (29, 46, 47). We routinely obtained 100- to 200-fold expansion of promoter database, with 922 of 1000 sequences (92%; p = 2.19 3 cord blood CD4+CD25+ Tregs when cultured in vitro using anti- 10233) containing at least one FKH motif. Using the modeling CD3/CD28 beads. These cells maintained a Treg phenotype, with algorithm WEEDER, we identified a motif (consensus [(G/A)(T/ .90% of the expanded cells staining CD4+, CD25hi, and FOXP3+ A)AAA(C/A)AA]) that was significantly overrepresented in these (Fig. 1A). These cells were also CD127dim (data not shown). The human FOXP3 ChIP regions, and comparison with transcription Tregs retained regulatory function because they were able to ro- factor position weight matrices in the JASPAR database indicated bustly suppress the proliferation of CD4+CD252 cells in vitro in that this represented a FKH binding site [(A/G)(C/T)(A/C)AA(C/ an unmatched donor mixed leukocyte suppression assay (Fig. 1B). T)A] (49). Our human FOXP3 target sequence (Fig. 2) is similar to Thus, having demonstrated that we could expand large numbers of the published mouse Foxp3 binding motif [(A/G)(T/C)AAACA] functional Tregs ex vivo, these cells were used for the ChIP-on- (50). This is consistent with localization of human FOXP3 occur- chip and expression profiling experiments. ring primarily through binding to a FKH-like motif. 4 FOXP3 TARGET GENES IN HUMAN nTregs

intrinsic expression pattern, those that responded to stimulation in a cell type-specific manner, and those that display common re- sponses to T cell activation. Genes displaying an nTreg-specific response were then clustered into six groups based on patterns of differential expression by cell type and/or stimulation (Table I). In total, 1851 genes displayed significant differential expression in a Treg compared with a Th cell (T1–T6), with a further 746 genes showing a similar response to activation in both cell types (A1) (Fig. 4B; described in Table I; gene identifications in Supplemen- tal Table IV). A total of 739 genes in these groups were also FOXP3 ChIP targets (gray bar in Fig. 4B). These FOXP3 ChIP hits were statistically overrepresented in all T groups (p , 1 3 1027), but we did not observe any statistically significant differ- FIGURE 2. The human FOXP3 binding motif in the top 1000 human ence in the distribution of FOXP3 ChIP hits between the upregu- FOXP3 ChIP hits derived from sequence information using WEEDER. lated and downregulated genes. Of note, groups T1 and T2, which represent the steady-state hu- man nTreg signature, also contained a higher proportion of FOXP3 Location and functional annotation of FOXP3 binding regions ChIP hits when compared with those of T3–T6 (50% versus 35% 2

9 Downloaded from We mapped FOXP3-bound regions relative to known and predicted p = 3.064 3 10 ), which is consistent with a role of FOXP3 in genes annotated in University of California Santa Cruz and Na- stabilizing and maintaining the Treg phenotype (52). Also, within tional Center for Biotechnology Information to identify potential the group of genes that were differentially expressed upon stimu- FOXP3 target genes. With a window beginning 20 kb upstream lation of the cells (T3–T6), there were clear differences, with of annotated TSSs and ending 20 kb downstream of the annotated a number of genes responding to stimulation in a Th cell but transcript end site of a gene, 7012 of the 8308 unique FOXP3- remaining unchanged in a Treg (T5), whereas a further set of bound regions were associated with a gene. The remaining 1296 genes responded to stimulation in an nTreg but remained un- http://www.jimmunol.org/ regions (13%) fell outside this window and were labeled intergenic. changed in a Th cell (T6). qPCR performed on a group of candi- All of the possible gene annotations for a given ChIP peak were date genes identified in our array confirmed the differential recorded, resulting in 5579 genes associated with at least one expression and clustering analysis (Fig. 4C). Similarly, TaqMan FOXP3-bound region (Supplemental Table I). Within this data microRNA (miR) assays performed on 20 of the miRs predicted to set, FOXP3 binding sites were found in close proximity to 63 loci be FOXP3 targets in nTregs demonstrated that five were differen- encoding miRS, including binding sites near loci encoding mir- tially expressed in nTregs (nine miRs shown in Fig. 4D). To iden- 155, mir-21, mir-22, and mir-142 that were are also mouse Foxp3 tify potential targets regulated by these miRs, a search for miR targets (50, 51) (Supplemental Table II). Examples of FOXP3- binding sites in the 39 UTRs of downregulated FOXP3 target bound regions annotated to the are shown in genes was performed using the microRNA target prediction pro- by guest on September 26, 2021 Fig. 3A. FOXP3 appears to have a preference for engaging in grams in miRGen (www.diana.pcbi.upenn.edu/miRGen.html). Us- short-range promoter interactions to regulate the transcription of ing stringent criteria in which a miR binding site must be found by target genes because we observed a clear peak of FOXP3-bound three separate prediction programs (miRanda, PicTar4, and Tar- regions starting 2 kb upstream and extending 3 kb downstream of getScan), we identified a number of high confidence gene targets the TSS (Fig. 3B). for these validated microRNAs (Supplemental Table V). Many of We selected 17 ChIP regions and four control regions for con- these predicted miR target genes (21 of 33) were also FOXP3 firmation by ChIP PCR (Fig. 3C). Included in this subset are species- ChIP hits, identifying a group of genes that are potentially sub- conserved FOXP3 targets and FOXP3 targets unique to our human jected to multiple FOXP3-dependent mechanisms to ensure their data set. Greater than 2-fold enrichment relative to input chromatin downregulation in nTregs. (normalized for nonspecific chromatin immunoprecipitation) was Transcription factor motifs in differentially expressed FOXP3 detected at the 17 sites tested by qPCR, whereas no enrichment (#1- target genes fold) was observed for four different genomic regions predicted not to bind FOXP3 (intergenic 12p13, AFM intron 1, PDE3B intron 10, The molecular mechanism by which FOXP3-dependent regulation and human IGX1A negative control, SA Biosciences). of gene expression occurs is not well characterized, but a growing list Analysis of the potential FOXP3 target genes using Ingenuity of transcription factors has been shown to interact with FOXP3. To Pathway Analysis software demonstrated a substantial enrichment identify other potential transcription factors that may coregulate for genes involved in the proliferation, activation, and control of gene expression with FOXP3, we analyzed sequences flanking the cell death in T lymphocytes in normal immune responses and in FOXP3 ChIP peak for the 739 genes that were also differentially disease states, such as inflammatory and autoimmune diseases. expressed. Position weight matrices from Genomatix corresponding Molecules involved in the response to TCR signaling and im- to the binding sites for transcription factor families AP-1, BCL6, munoregulatory cytokines/growth factors were significantly HAML (Runx family), HNF1, MYBL, NFAT, OVOL, PAX2, PRDF overrepresented in the human FOXP3 target gene data set (Sup- (PRDM1/Blimp1 family), and STAT transcription factor families plemental Table III). were found to be significantly overrepresented in FOXP3-bound regions,whereas the CAATmotifonlyappearssignificantlyoverrep- Differential gene expression in resting and activated nTregs resented in groups T1 and T2 (FDR-adjusted p = 0.0462) (Fig. 5A). To correlate FOXP3 binding with the transcriptional regulation of In the STAT family (Fig. 5B), the STAT5 consensus binding motif target genes, expression profiling was carried out on expanded cord was the most significantly enriched, suggesting a preferential inter- blood nTregs and Th cells. Expression profiling of donor-matched action of FOXP3 with STAT5 within nTregs. These results have Treg and Th cell populations in both a stimulated and a resting state therefore uncovered several new transcription factors that potentially (Fig. 4A) identified three classes of genes: those with an nTreg directly interact with, or coregulate gene expression by, FOXP3. The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021

FIGURE 3. Annotation, validation, and distribution of FOXP3 target genes identified by ChIP-on-chip. A, Annotation of three genes enriched for FOXP3 binding by ChIP showing the mouse/human conserved regions of the locus, the scale in kilobases of the region, the number and location, the individual FOXP3 binding regions, and the gene identification and the gene structure. B, Distribution of FOXP3 binding sites annotated to genes in relation to the distance from the TSS. C, The fold enrichment of target regions in FOXP3-immunoprecipitated material, normalized to control IgG immunoprecipitations, relative to input chromatin was determined for selected FOXP3 target genes by qPCR. Displayed is the mean fold enrichment 6 SD; n = 3 for each region.

Comparison of human and mouse FOXP3/Foxp3 ChIP data human FOXP3 targets with mouse Foxp3 targets indicated that We next compared our ChIP data set with the previously identified 23% (888 genes) of our human FOXP3 ChIP targets were also murine targets (50, 51). Because substantial differences existed Foxp3 targets in at least one mouse data set (Supplemental Table between the two mouse data sets, the mouse data sets were each VI, Table II) (50, 51), with 107 genes common to all three data 2 compared separately with our human data (Table II). The mouse sets. These overlaps were statistically significant (p , 10 5), and data were then combined for clarity (Fig. 6). Comparison of the a number of genes previously implicated in Treg function were 6 FOXP3 TARGET GENES IN HUMAN nTregs Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021

FIGURE 4. A, Schematic representation of the cells and comparisons used in the expression profiling experiment. B, Heat map representation of sig- nificant differentially expressed genes in all four expression array comparisons. The normalized log2 fold changes for repressed (blue) and induced (red) transcripts across each comparison are shown. FOXP3 target genes identified by ChIP-on-chip analysis are indicated in gray on the right-hand side. C,RT- PCR validation of differentially expressed genes in expanded cord blood nTregs versus Th cells. Displayed is the mean log fold change in the relative level of the genes in expanded nTregs compared with Th cells in stimulated (open bar) and resting (filled bar) cells (n = 6 donors). The log fold change determined from analysis of the exon arrays for the nTreg versus Th cell comparisons in resting (hatched bar) and stimulated (stripped bar) cells is also shown. Dotted lines represent fold change of 62. D, RT-PCR validation of potential FOXP3 target miRs. The relative expression of the miR in matched freshly isolated cord blood nTregs and Th cells was determined on n = 3 donors in triplicate using TaqMan miR assays. Statistical significance (p . 0.05) using a paired t test was achieved for all but those marked with an asterisk. identified in these overlapping groups. On the basis of our gene among these conserved targets. These pathways were also found expression profiling, we found a significantly higher proportion of to be significantly overrepresented among the human-specific differentially expressed genes (20–24%) among the conserved ChIP genes (Supplemental Tables VII, VIII), suggesting that al- FOXP3 targets (Table II), suggesting that these genes may rep- though the overall regulation of essential pathways and biological resent a core group of FOXP3-dependent genes involved in the processes may be conserved in human and mouse Tregs, the genes maintenance of the regulatory phenotype in both species. Pathway within a pathway used to achieve this may vary between species. and biological function analysis of these 888 species-conserved FOXP3 target genes using Ingenuity Pathway Analysis software Analysis of cell surface molecules in nTregs revealed that members of critical T cell signal transduction path- When we applied functional annotation to differentially expressed ways including TCR, CD28, and CTLA4 signaling and the in- genes in the data set, we discovered a highly significant (p = 3.36 3 tracellular MAPK signaling pathway were overrepresented 10216) enrichment for surface molecules in groups T1 and T2, The Journal of Immunology 7

Table I. Group definitions for the differentially expressed genes

Group Inclusion Criteria Number DE ChIP Hits (% DE) T1 Effects in both stimulated and resting Tregs versus Th cells 417 215 (51) T2 Effects only in resting Tregs versus Th cells 135 63 (46) T3 Effects only in stimulated Tregs versus Th cells 176 65 (37) T4 Effects in stimulated Tregs versus Th cells activation effects in both 226 99 (43) T5 Effects in stimulated Tregs versus Th cells and activation effects only in Th cells 335 125 (37) T6 Effects in stimulated Tregs versus Th cells and activation effects only in Tregs 562 172 (30) Total 1851 739 (40) A1 Common stimulation effect in both Tregs and Th cells 746 281 (37) DE, differentially expressed. with 42.6% of genes in these groups being cell surface related We next examined PI16 expression in adult peripheral blood versus 23.8% in groups T3–T6. This included known CD mole- CD4+ T cell subsets. PI16 was readily detectable on the surfaces cules (Supplemental Table IX) as well as less well characterized of resting and stimulated nTregs (Fig. 7A,7B). In particular, on surface molecules. However, FACS-based analysis failed to iden- resting Tregs, a significant portion (∼85%) of the CD4+CD25bright tify a CD molecule uniquely expressed on .50% of the FOXP3+ FOXP3+ cells were PI16+, and the majority (∼70%) of the PI16+ cells (data not shown). We therefore further mined our expression cells coexpress FOXP3, with only ∼15% FOXP3+PI162 (Fig. Downloaded from array data set to identify additional surface not assigned 7Aii). In contrast, although ∼50% of the resting CD252 cells are a CD designation that were selectively upregulated on the nTregs, PI16+, the majority of these (80%) are FOXP32 (Fig. 7Aiii). Upon and from this we focused on PI16. The differential expression of isolation of CD4+CD25+ cells and overnight stimulation, ∼80% of PI16 in resting and stimulated cord blood nTregs versus Th cells the PI16+ cells coexpress PI16 and FOXP3, although ∼40% of the was confirmed by real-time PCR (Fig. 4C). activated FOXP3+ cells are PI162 (Fig. 7Bii). Importantly, stim- ulated CD4+CD252 do not express significant amounts of PI16 or FOXP3, suggesting that PI16 expression is not linked to activation http://www.jimmunol.org/ but rather to the nTreg phenotype. It is interesting to note that although the proportion of FOXP3+ PI16+ cells is similar between resting and stimulated overnight Tregs (Fig. 7Aii,7Bii) the pro- portion of FOXP3+PI162 cells increases upon stimulation, sug- gesting that stimulation alters surface expression of PI16. This is not due to loss of PI16 transcription, as confirmed by the array and quantitative RT-PCR data in resting and activated Tregs. There are + 2

also FOXP3 PI16 cells at a lower frequency, which may by guest on September 26, 2021 represent a distinct Treg subset. In our chip analysis, PI16 was not a direct target of FOXP3 in human or mouse, suggesting that it is regulated by genes downstream of FOXP3. Together, these data indicate that PI16 warrants further investigation as a FOXP3 surrogate marker.

Discussion Using expanded cord blood nTregs to generate FOXP3 ChIP-on- chip and expression profiling data, we show, for the first time, hu- man FOXP3-bound genes and their differential expression profile in human nTregs. We detected 8308 potential FOXP3 binding sites in the human genome, which were associated with 5579 genes. Of these potential FOXP3-regulated genes, 13% (739) displayed dif- ferential expression between donor-matched expanded nTregs and Th cells. We selected an early time point for the expression analysis to identify genes directly downstream of this altered signaling. Ex- pression profiles at later time points may uncover additional FOXP3 target genes whose expression level changes more slowly following stimulation. Nonetheless, our finding that a subset of FOXP3-bound regions can be linked to the regulation of the neigh- boring genes is consistent with other genome-wide ChIP studies where a similar level of transcription factor binding and gene ex- pression changes in nearby genes has been reported (reviewed in Ref. 53). FIGURE 5. Overrepresentation of transcription factor binding sites in The comparison of FOXP3-bound genes from this study in FOXP3 target gene promoters. A, Analysis of differentially expressed genes in the Treg groups that are also ChIP hits for the corepresentation of humans with published studies in mice (50, 51) provided a means transcription factor family binding sites. Red bar, T1 + T2; blue bar, T3– of identifying genes with a conserved function in Tregs. Although T6; white bar, control promoters. B, Analysis of the STAT family of .50% of the FOXP3 targets are bound in a species-specific man- transcription factors demonstrating that STAT5 binding sites are signifi- ner, 23% of FOXP3-bound genes identified in our human study cantly overrepresented in FOXP3 target genes. were present in one of the published mouse Treg studies. 8 FOXP3 TARGET GENES IN HUMAN nTregs

Table II. Differential expression of human and mouse FOXP3 target genes

Number %DEin Raw Comparison Size DE % DE Control p Value Human/either 877 175 20.0 16.2 0.0104 mouse Human/Marson 662 120 18.1 16.2 0.2240 Human/Zheng 322 80 24.8 16.2 1.66 3 1024 Human/mouse 107 25 23.4 16.2 0.2558 Control, nonconserved Foxp3 target genes.

However, upon pooling of the Foxp3-bound genes identified in the two mouse studies, 45% of FOXP3 targets identified in mouse Tregs were also bound by FOXP3 in human Tregs. A caveat to these comparisons is the small overlap of the Foxp3 target genes between the two mouse studies (7% overlap; data not shown). However, this degree of conservation of FOXP3 target genes in

human and mouse cells is consistent with other large scale tran- Downloaded from scription factor binding and expression studies, where a substantial divergence in transcription factor binding events in human and mouse cells has been reported (54–57). These species-conserved FOXP3 targets contained proportionately more differentially reg- ulated genes (20% of conserved targets versus 16.2% in all targets,

p = 0.0104), suggesting that the curation of conserved targets is http://www.jimmunol.org/ a valid approach for identifying proteins associated with the key role of FOXP3 within the Treg. The proportion of differentially regulated genes increased to ∼25% (p = 1.66 3 1024) when only the Foxp3-bound genes from primary mouse Tregs (51) were compared with our human FOXP3 targets. In contrast, when the group of mouse Foxp3-bound genes derived using forced expres- sion in hybridoma cells was compared (50), there was no signif- icant enrichment for differentially regulated genes (18%, p = FIGURE 7. PI16 and FOXP3 expression in nTregs and Th cells. A, 0.224). This may be a consequence of the differences in intracel- by guest on September 26, 2021 Validation of PI16 as a surrogate for FOXP3 in resting adult Tregs. When lular signaling pathways and transcription factor levels in hydri- CD4+ cells were gated on the top 1% CD25+ (i), .85% of the FOXP3+ domas compared with those of primary mouse T cells or a result cells are PI16+ (ii). When gated on CD252, few of the PI16+ cells are of overexpression of FoxP3 and highlights the difficulties in com- FOXP3+ (iii). B, CD25+ and CD252 CD4 cells were isolated according to paring transcription factor binding in different cell systems. In the gates shown (i) and then stimulated overnight with anti-CD3/CD28 support of this, expression profiling performed by Marson et al. beads and stained for PI16 and FOXP3 expression. Sorted stimulated (50) indicated that Foxp3 overexpression almost overwhelmingly CD25+ cells stained robustly with the anti-PI16 Ab, and this population downregulated gene expression in an activation-dependent manner contains .60% of the FOXP3+ cells (ii). The majority of the activated 2 in hybridoma cell lines, whereas 30–50% of the Foxp3-bound CD25 cells do not express PI16 or FOXP3 (iii). Representative data from genes were upregulated in mouse (50) and human (this study) n = 3 donors. primary Tregs. Our ChIP-on-chip assay also identified FOXP3 binding regions in proximity to loci encoding 63 microRNAs (miRs), which is con- [DICER (58–60) or DROSHA (61)]. Deletion of either of these sistent with FOXP3 partly exerting its regulatory program by mod- enzymes led to a spontaneous lethal inflammatory disease in mice ifying the expression of these posttranscriptional rheostats of due to a severe reduction in the stability and function of nTregs. expression. In mice, a critical role of miRs in nTreg bio MicroRNA signatures for mouse and human nTregs have been logy has been demonstrated by the T cell-targeted deletion of either reported (59, 62) and several miRs have been shown to be impor- of the two key enzymes involved in the generation of mature miRs tant in nTreg function, including miR155 in mouse (63, 64) and miR21 in human cells (62). We have confirmed that both of these miRs are located near FOXP3 binding sites by ChIP PCR and that both are differentially upregulated in human nTregs, suggesting that these are conserved targets for FOXP3 regulation. In addition, we have identified three further miRs potentially regulated by Foxp3 (miR-146a, miR-101, and miR-7) that are upregulated in human nTregs. Thus, in addition to the four miRs also identified as potential Foxp3 targets in the mouse (miR-7, miR-21, miR-22, and miR-155) (50, 51) and 10 microRNAs recently identified as FIGURE 6. Overlap between human and mouse genes associated with differentially expressed in human Tregs (62), we find that the a FOXP3-bound region. The numbers represent target genes common FOXP3 targets miR-101 and miR-146a are also significantly up- (center) and unique to each species (outer). All of the comparisons were regulated. These results suggest that other miRs are likely to contrib- statistically significant (p , 0.005). ute to the Treg gene program and that further work is required to The Journal of Immunology 9 establish their identity and role. Knockdown of FOXP3 in human FOXP3 expression and the localization of latent TGF-b on the Tregs to assess the role of individual microRNAs underway. surface of nTregs upon TCR stimulation (83, 84). Significantly, Analysis of the human FOXP3-bound regions that were associ- we have found that PI16, unlike LRRC32/GARP, is differentially ated with differentially regulated genes identified a number of tran- expressed on resting Tregs. Hence, we propose it as a potential scription factor motifs that appeared to co-occur with FKH motifs, biomarker for human nTregs, where in conjunction with CD25 including AP-1, Runx, NFAT, and STAT binding motifs. The co- and LRCC32/GARP it may be feasible to identify both resting occurrence of these motifs is consistent with known roles for these and stimulated Treg subsets. Little is known about the function transcription factors in nTregs (18, 65–69) and, in the case of AP- of PI16, which was initially identified as a serum binding partner 1, NFAT, and Runx family members, with their known physical of prostate secretory protein 94 (85) and more recently has been interaction with FOXP3 (18, 20, 23). Several of the other tran- shown to have growth inhibitory properties (86, 87). The Treg scription factor family motifs that are overrepresented in these expression profile and the potential growth suppressive activity sequences, such as BCL6 and PRDF, also contain members known of PI16 indicate that the contribution of this surface protein to to have a role in T cells, such as BCL6, PRDM1/Blimp-1, and c- Treg function clearly warrants further investigation. myb (70–73). The finding of potential PRDM1 motifs within In conclusion, we have generated a gene profile of human Tregs FOXP3-bound regions is of particular interest given its relatively and identified putative direct targets of FOXP3, which allows us to high level of expression in natural Tregs and its requirement model direct and indirect FOXP3 target genes. Pathway mapping for T cell homeostasis, nTreg function, and self-tolerance in the and transcription factor over representation analysis provide novel mouse (74, 75), suggesting that FOXP3 and PRDM1 may co- information on the mechanisms by which human nTregs function, operate to regulate gene expression in human Tregs. Furthermore, and we can now undertake functional studies of the cell surface, Downloaded from the PRDM1 gene is a conserved FOXP3/Foxp3 target in humans secreted, and other molecules, which may be critical for Treg and mice (Ref. 51 and this study), suggesting a regulatory loop function. This may hence provide new leads for diagnosis or involving these two transcription factors. therapeutic intervention. A major role for FOXP3 is to modulate the response of a T lymphocyte to external stimuli, such as TCR and cytokine sig- Acknowledgments naling, by altering the level and activity of downstream tran- We thank Sylvie Nobbs and Dr. Randall Grose for expert assistance http://www.jimmunol.org/ scription factors (76). Several of the transcription factors that are with flow cytometry and Dr. Doreen Krumbiegel for critical reading of this able to bind to sites enriched in FOXP3-bound regions are them- manuscript. selves potential targets of FOXP3 and are differentially expressed. For example, NFATc2, NFATc3, STAT1, and STAT4 are all po- Disclosures tential ChIP targets of FOXP3 and are downregulated in Tregs. A The authors have no financial conflicts of interest. direct role for FOXP3 downregulating the level of NFAT proteins has recently been supported by the finding that forced FOXP3 expression in human cell lines inhibits the activation-induced ex- References by guest on September 26, 2021 pression of NFAT2 (NFATc/NFATc1) in T cells by competing with 1. Tang, Q., and J. A. Bluestone. 2008. The Foxp3+ regulatory T cell: a jack of all NFAT1 (NFATc2) for binding to a site within the NFAT2 promoter trades, master of regulation. Nat. Immunol. 9: 239–244. (77). Among the binding sites identified as belonging to the STAT 2. Chatenoud, L., B. Salomon, and J. A. Bluestone. 2001. Suppressor T cells— they’re back and critical for regulation of autoimmunity! Immunol. Rev. 182: family of transcription factors, there was a strong overrepresenta- 149–163. tion of STAT5 binding sites at FOXP3-bound regions. This en- 3. Bluestone, J. A., and A. K. Abbas. 2003. Natural versus adaptive regulatory T cells. Nat. Rev. Immunol. 3: 253–257. richment was not observed for the binding sites of other STAT 4. Bennett, C. L., J. Christie, F. Ramsdell, M. E. Brunkow, P. J. Ferguson, family members. Thus, the apparent enrichment of STAT5 sites L. Whitesell, T. E. Kelly, F. T. Saulsbury, P. F. Chance, and H. D. Ochs. 2001. within FOXP3-bound regions, coupled with a decrease in the ex- The immune dysregulation, polyendocrinopathy, enteropathy, X-linked syn- drome (IPEX) is caused by mutations of FOXP3. Nat. Genet. 27: 20–21. pression of other STAT family members, such as STAT1 and 5. Brunkow, M. E., E. W. Jeffery, K. A. Hjerrild, B. Paeper, L. B. Clark, STAT4, suggests an essential requirement for and increased sen- S. A. Yasayko, J. E. Wilkinson, D. Galas, S. F. Ziegler, and F. Ramsdell. 2001. sitivity to the IL-2/STAT5 signaling pathway for Treg develop- Disruption of a new forkhead/winged-helix protein, scurfin, results in the fatal lymphoproliferative disorder of the scurfy mouse. Nat. Genet. 27: 68–73. ment and homeostasis (66, 67, 69). 6. Wildin, R. S., F. Ramsdell, J. Peake, F. Faravelli, J. L. Casanova, N. Buist, Clustering of our gene expression data showed clear differences E. Levy-Lahad, M. Mazzella, O. Goulet, L. Perroni, et al. 2001. X-linked neo- natal diabetes mellitus, enteropathy and endocrinopathy syndrome is the human in the response to stimulation, with over one third of the differ- equivalent of mouse scurfy. Nat. Genet. 27: 18–20. entially expressed genes showing cell type-specific responses to 7. Vila, J., J. D. Isaacs, and A. E. Anderson. 2009. Regulatory T cells and auto- stimulation (groups T5 and T6; 35% = 897/2597). On top of these immunity. Curr. Opin. Hematol. 16: 274–279. 8. Liesz, A., E. Suri-Payer, C. Veltkamp, H. Doerr, C. Sommer, S. Rivest, T. Giese, activation-influenced signatures, another group of genes was and R. Veltkamp. 2009. Regulatory T cells are key cerebroprotective revealed that was differentially expressed in both resting and stim- immunomodulators in acute experimental stroke. Nat. Med. 15: 192–199. ulated Tregs, suggesting that these genes represent a steady-state 9. 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: human nTreg expression profile (T1 + T2; 21% = 552/2597). To- 330–336. gether these data may be helpful for the further development of 10. Hu, H., I. Djuretic, M. S. Sundrud, and A. Rao. 2007. Transcriptional partners in regulatory T cells: Foxp3, Runx and NFAT. Trends Immunol. 28: 329–332. a gene signature to investigate the relative contributions of nTregs 11. Hancock, W. W., and E. Ozkaynak. 2009. Three distinct domains contribute to and other CD4 subsets in disease pathology (78, 79). nuclear transport of murine Foxp3. PLoS One 4: e7890. We detected a significant enrichment of differentially regulated 12. Li, B., A. Samanta, X. Song, K. Furuuchi, K. T. Iacono, S. Kennedy, M. Katsumata, S. J. Saouaf, and M. I. Greene. 2006. FOXP3 ensembles in T-cell cell surface molecules in the steady-state nTreg signature (T1 + T2), regulation. Immunol. Rev. 212: 99–113. consistent with a role for surface molecules in Treg function, includ- 13. Lopes, J. E., T. R. Torgerson, L. A. Schubert, S. D. Anover, E. L. Ocheltree, ing the homing response (80, 81) and suppressive function (82). In H. D. Ochs, and S. F. Ziegler. 2006. Analysis of FOXP3 reveals multiple domains required for its function as a transcriptional repressor. J. Immunol. 177: particular, two molecules, LRRC32/GARP and PI16, were signif- 3133–3142. icantly upregulated in human nTregs. LRRC32/GARP has recently 14. Allan, S. E., L. Passerini, R. Bacchetta, N. Crellin, M. Dai, P. C. Orban, S. F. Ziegler, M. G. Roncarolo, and M. K. Levings. 2005. The role of 2 FOXP3 been reported to be specifically upregulated in human nTregs upon isoforms in the generation of human CD4+ Tregs. J. Clin. Invest. 115: 3276– stimulation where it has been proposed to play a role in stabilizing 3284. 10 FOXP3 TARGET GENES IN HUMAN nTregs

15. Zhou, L., J. E. Lopes, M. M. Chong, I. I. Ivanov, R. Min, G. D. Victora, Y. Shen, 42. Frisch, M., K. Frech, A. Klingenhoff, K. Cartharius, I. Liebich, and T. Werner. J. Du, Y. P. Rubtsov, A. Y. Rudensky, et al. 2008. TGF-beta-induced Foxp3 2002. In silico prediction of scaffold/matrix attachment regions in large genomic inhibits T(H)17 cell differentiation by antagonizing RORgammat function. Na- sequences. Genome Res. 12: 349–354. ture 453: 236–240. 43. Bryne, J. C., E. Valen, M. H. Tang, T. Marstrand, O. Winther, I. da Piedade, 16. Du, J., C. Huang, B. Zhou, and S. F. Ziegler. 2008. Isoform-specific inhibition of A. Krogh, B. Lenhard, and A. Sandelin. 2008. JASPAR, the open access database ROR alpha-mediated transcriptional activation by human FOXP3. J. Immunol. of transcription factor-binding profiles: new content and tools in the 2008 update. 180: 4785–4792. Nucleic Acids Res. 36(Database issue): D102–D106. 17. Allan, S. E., S. Q. Crome, N. K. Crellin, L. Passerini, T. S. Steiner, R. Bacchetta, 44. Wang, X., and B. Seed. 2003. A PCR primer bank for quantitative gene ex- M. G. Roncarolo, and M. K. Levings. 2007. Activation-induced FOXP3 in pression analysis. Nucleic Acids Res. 31: e154. human T effector cells does not suppress proliferation or cytokine production. 45. Muller, P. Y., H. Janovjak, A. R. Miserez, and Z. Dobbie. 2002. Processing of Int. Immunol. 19: 345–354. gene expression data generated by quantitative real-time RT-PCR. Biotechniques 18. Ono, M., H. Yaguchi, N. Ohkura, I. Kitabayashi, Y. Nagamura, T. Nomura, 32: 1372–1374, 1376, 1378–1379. Y. Miyachi, T. Tsukada, and S. Sakaguchi. 2007. Foxp3 controls regulatory 46. Godfrey, W. R., D. J. Spoden, Y. G. Ge, S. R. Baker, B. Liu, B. L. Levine, T-cell function by interacting with AML1/Runx1. Nature 446: 685–689. C. H. June, B. R. Blazar, and S. B. Porter. 2005. Cord blood CD4(+)CD25 19. Zheng, Y., A. Chaudhry, A. Kas, P. deRoos, J. M. Kim, T. T. Chu, L. Corcoran, (+)-derived T regulatory cell lines express FoxP3 protein and manifest potent P. Treuting, U. Klein, and A. Y. Rudensky. 2009. Regulatory T-cell suppressor suppressor function. Blood 105: 750–758. program co-opts transcription factor IRF4 to control T(H)2 responses. Nature 47. Seddiki, N., B. Santner-Nanan, S. G. Tangye, S. I. Alexander, M. Solomon, 458: 351–356. S. Lee, R. Nanan, and B. Fazekas de Saint Groth. 2006. Persistence of naive 20. Lee, S. M., B. Gao, and D. Fang. 2008. FoxP3 maintains Treg unresponsiveness CD45RA+ regulatory T cells in adult life. Blood 107: 2830–2838. by selectively inhibiting the promoter DNA-binding activity of AP-1. Blood 111: 48. Liu, W., A. L. Putnam, Z. Xu-Yu, G. L. Szot, M. R. Lee, S. Zhu, P. A. Gottlieb, 3599–3606. P. Kapranov, T. R. Gingeras, B. Fazekas de St Groth, et al. 2006. CD127 21. Bettelli, E., M. Dastrange, and M. Oukka. 2005. Foxp3 interacts with nuclear expression inversely correlates with FoxP3 and suppressive function of human factor of activated T cells and NF-kappa B to repress cytokine gene expression CD4+ T reg cells. J. Exp. Med. 203: 1701–1711. and effector functions of T helper cells. Proc. Natl. Acad. Sci. USA 102: 5138– 49. Carlsson, P., and M. Mahlapuu. 2002. Forkhead transcription factors: key players in development and metabolism. Dev. Biol. 250: 1–23. 5143. Downloaded from 22. Chaudhry, A., D. Rudra, P. Treuting, R. M. Samstein, Y. Liang, A. Kas, and 50. Marson, A., K. Kretschmer, G. M. Frampton, E. S. Jacobsen, J. K. Polansky, A. Y. Rudensky. 2009. CD4+ regulatory T cells control TH17 responses in K. D. MacIsaac, S. S. Levine, E. Fraenkel, H. von Boehmer, and R. A. Young. a Stat3-dependent manner. Science 326: 986–991. 2007. Foxp3 occupancy and regulation of key target genes during T-cell stimu- 23. Wu, Y., M. Borde, V. Heissmeyer, M. Feuerer, A. D. Lapan, J. C. Stroud, lation. Nature 445: 931–935. D. L. Bates, L. Guo, A. Han, S. F. Ziegler, et al. 2006. FOXP3 controls regu- 51. Zheng, Y., S. Z. Josefowicz, A. Kas, T. T. Chu, M. A. Gavin, and latory T cell function through cooperation with NFAT. Cell 126: 375–387. A. Y. Rudensky. 2007. Genome-wide analysis of Foxp3 target genes in devel- 24. Rudra, D., T. Egawa, M. M. Chong, P. Treuting, D. R. Littman, and oping and mature regulatory T cells. Nature 445: 936–940. A. Y. Rudensky. 2009. Runx-CBFbeta complexes control expression of the tran- 52. Williams, L. M., and A. Y. Rudensky. 2007. Maintenance of the Foxp3- dependent developmental program in mature regulatory T cells requires contin- http://www.jimmunol.org/ scription factor Foxp3 in regulatory T cells. Nat. Immunol. 10: 1170–1177. Nat. Immunol. 25. Chen, C., E. A. Rowell, R. M. Thomas, W. W. Hancock, and A. D. Wells. 2006. ued expression of Foxp3. 8: 277–284. 53. Farnham, P. J. 2009. Insights from genomic profiling of transcription factors. Transcriptional regulation by Foxp3 is associated with direct promoter occu- Nat. Rev. Genet. 10: 605–616. pancy and modulation of histone acetylation. J. Biol. Chem. 281: 36828–36834. 54. Odom, D. T., R. D. Dowell, E. S. Jacobsen, W. Gordon, T. W. Danford, 26. Li, B., A. Samanta, X. Song, K. T. Iacono, K. Bembas, R. Tao, S. Basu, K. D. MacIsaac, P. A. Rolfe, C. M. Conboy, D. K. Gifford, and E. Fraenkel. J. L. Riley, W. W. Hancock, Y. Shen, et al. 2007. FOXP3 interactions with 2007. Tissue-specific transcriptional regulation has diverged significantly be- histone acetyltransferase and class II histone deacetylases are required for re- tween human and mouse. Nat. Genet. 39: 730–732. pression. Proc. Natl. Acad. Sci. USA 104: 4571–4576. 55. Vaquerizas, J. M., S. K. Kummerfeld, S. A. Teichmann, and N. M. Luscombe. 27. Samanta, A., B. Li, X. Song, K. Bembas, G. Zhang, M. Katsumata, S. J. Saouaf, 2009. A census of human transcription factors: function, expression and evo- Q. Wang, W. W. Hancock, Y. Shen, and M. I. Greene. 2008. TGF-beta and lution. Nat. Rev. Genet. 10: 252–263. IL-6 signals modulate chromatin binding and promoter occupancy by acetylated 56. Birney, E., J. A. Stamatoyannopoulos, A. Dutta, R. Guigo´, T. R. Gingeras, FOXP3. Proc. Natl. Acad. Sci. USA 105: 14023–14027. E. H. Margulies, Z. Weng, M. Snyder, E. T. Dermitzakis, R. E. Thurman, et al; by guest on September 26, 2021 28. Tao, R., E. F. de Zoeten, E. Ozkaynak, C. Chen, L. Wang, P. M. Porrett, B. Li, ENCODE Project Consortium; NISC Comparative Sequencing Program; Baylor L. A. Turka, E. N. Olson, M. I. Greene, et al. 2007. Deacetylase inhibition promotes College of Medicine Human Genome Sequencing Center; Washington Univer- the generation and function of regulatory T cells. Nat. Med. 13: 1299–1307. sity Genome Sequencing Center; Broad Institute; Children’s Hospital Oakland 29. Bresatz, S., T. Sadlon, D. Millard, H. Zola, and S. C. Barry. 2007. Isolation, Research Institute. 2007. Identification and analysis of functional elements in 1% propagation and characterization of cord blood derived CD4+ CD25+ regulatory of the human genome by the ENCODE pilot project. Nature 447: 799–816. T cells. J. Immunol. Methods 327: 53–62. 57. Wilson, M. D., N. L. Barbosa-Morais, D. Schmidt, C. M. Conboy, L. Vanes, 30. Lee, T. I., S. E. Johnstone, and R. A. Young.2006. Chromatin immunoprecipitation V. L. Tybulewicz, E. M. Fisher, S. Tavare´, and D. T. Odom. 2008. Species- and microarray-based analysis of protein location. Nat. Protoc. 1: 729–748. specific transcription in mice carrying human chromosome 21. Science 322: 31. O’Geen, H., C. M. Nicolet, K. Blahnik, R. Green, and P. J. Farnham. 2006. 434–438. Comparison of sample preparation methods for ChIP-chip assays. Biotechniques 58. Liston, A., L. F. Lu, D. O’Carroll, A. Tarakhovsky, and A. Y. Rudensky. 2008. 41: 577–580. Dicer-dependent microRNA pathway safeguards regulatory T cell function. J. 32. Johnson, W. E., W. Li, C. A. Meyer, R. Gottardo, J. S. Carroll, M. Brown, and Exp. Med. 205: 1993–2004. X. S. Liu. 2006. Model-based analysis of tiling-arrays for ChIP-chip. Proc. Natl. 59. Cobb, B. S., A. Hertweck, J. Smith, E. O’Connor, D. Graf, T. Cook, S. T. Smale, Acad. Sci. USA 103: 12457–12462. S. Sakaguchi, F. J. Livesey, A. G. Fisher, and M. Merkenschlager. 2006. A role 33. Frazer, K. A., L. Pachter, A. Poliakov, E. M. Rubin, and I. Dubchak. 2004. for Dicer in immune regulation. J. Exp. Med. 203: 2519–2527. VISTA: computational tools for comparative genomics. Nucleic Acids Res. 32 60. Zhou, X., L. T. Jeker, B. T. Fife, S. Zhu, M. S. Anderson, M. T. McManus, and (Web Server issue): W273–W279. J. A. Bluestone. 2008. Selective miRNA disruption in T reg cells leads to un- 34. Irizarry, R. A., B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, controlled autoimmunity. J. Exp. Med. 205: 1983–1991. U. Scherf, and T. P. Speed. 2003. Exploration, normalization, and summaries 61. Chong, M. M., J. P. Rasmussen, A. Y. Rudensky, and D. R. Littman. 2008. The of high density oligonucleotide array probe level data. Biostatistics 4: 249–264. RNAseIII enzyme Drosha is critical in T cells for preventing lethal inflammatory 35. Dai, M., P. Wang, A. D. Boyd, G. Kostov, B. Athey, E. G. Jones, W. E. Bunney, disease. J. Exp. Med. 205: 2005–2017. R. M. Myers, T. P. Speed, H. Akil, et al. 2005. Evolving 62. Rouas, R., H. Fayyad-Kazan, N. El Zein, P. Lewalle, F. Rothe´, A. Simion, gene/transcript definitions significantly alter the interpretation of GeneChip data. H. Akl, M. Mourtada, M. El Rifai, A. Burny, et al. 2009. Human natural Treg Nucleic Acids Res. 33: e175. microRNA signature: role of microRNA-31 and microRNA-21 in FOXP3 ex- 36. Bengtsson, H., K. Simpson, J. Bullard, and K. Hansen. 2008. In Tech Report 745. pression. Eur. J. Immunol. 39: 1608–1618. Department of Statistics, University of California, Berkeley, CA. 63. Lu, L. F., T. H. Thai, D. P. Calado, A. Chaudhry, M. Kubo, K. Tanaka, 37. Smyth, G. 2005. Limma: linear models for microarray data. In Bioinformatics G. B. Loeb, H. Lee, A. Yoshimura, K. Rajewsky, and A. Y. Rudensky. 2009. and Computational Biology Solutions using R and Bioconductor. V. C. R. Foxp3-dependent microRNA155 confers competitive fitness to regulatory T cells Gentleman, W. Huber, R. Irizarry, and S. Dudoit, eds. Springer, New York, p. by targeting SOCS1 protein. Immunity 30: 80–91. 397–420. 64. Kohlhaas, S., O. A. Garden, C. Scudamore, M. Turner, K. Okkenhaug, and 38. Hochberg, Y., and Y. Benjamini. 1990. More powerful procedures for multiple E. Vigorito. 2009. Cutting edge: the Foxp3 target miR-155 contributes to the significance testing. Stat. Med. 9: 811–818. development of regulatory T cells. J. Immunol. 182: 2578–2582. 39. Pavesi, G., P. Mereghetti, G. Mauri, and G. Pesole. 2004. Weeder Web: discovery 65. Mantel, P. Y., N. Ouaked, B. Ru¨ckert, C. Karagiannidis, R. Welz, K. Blaser, and of transcription factor binding sites in a set of sequences from co-regulated C. B. Schmidt-Weber. 2006. Molecular mechanisms underlying FOXP3 genes. Nucleic Acids Res. 32(Web Server issue): W199–W203. induction in human T cells. J. Immunol. 176: 3593–3602. 40. Pavesi, G., and G. Pesole. 2006. Using Weeder for the discovery of conserved 66. Yu, A., L. Zhu, N. H. Altman, and T. R. Malek. 2009. A low transcription factor binding sites. Curr.Protoc.Bioinformatics Chapter 2: Unit 2.11. interleukin-2 receptor signaling threshold supports the development and homeo- 41. Cartharius, K., K. Frech, K. Grote, B. Klocke, M. Haltmeier, A. Klingenhoff, stasis of T regulatory cells. Immunity 30: 204–217. M. Frisch, M. Bayerlein, and T. Werner. 2005. MatInspector and beyond: pro- 67. Yao, Z., Y. Kanno, M. Kerenyi, G. Stephens, L. Durant, W. T. Watford, moter analysis based on transcription factor binding sites. Bioinformatics 21: A. Laurence, G. W. Robinson, E. M. Shevach, R. Moriggl, et al. 2007. Non- 2933–2942. redundant roles for Stat5a/b in directly regulating Foxp3. Blood 109: 4368–4375. The Journal of Immunology 11

68. Strauss, L., M. Czystowska, M. Szajnik, M. Mandapathil, and T. L. Whiteside. 78. Pfoertner, S., A. Jeron, M. Probst-Kepper, C. A. Guzman, W. Hansen, 2009. Differential responses of human regulatory T cells (Treg) and effector A. M. Westendorf, T. Toepfer, A. J. Schrader, A. Franzke, J. Buer, and T cells to rapamycin. PLoS One 4: e5994. R. Geffers. 2006. Signatures of human regulatory T cells: an encounter with 69. O’Gorman, W. E., H. Dooms, S. H. Thorne, W. F. Kuswanto, E. F. Simonds, old friends and new players. Genome Biol. 7: R54. P. O. Krutzik, G. P. Nolan, and A. K. Abbas. 2009. The initial phase of an immune 79. TrehanPati, N., R. Geffers, S. Sukriti, P. Hissar, T. Riese, J. Toepfer, M. Buer, response functions to activate regulatory T cells. J. Immunol. 183: 332–339. C. A. Kumar, Guzman, and S. K. Sarin. 2009. Gene expression signatures of 70. Johnston, R. J., A. C. Poholek, D. DiToro, I. Yusuf, D. Eto, B. Barnett, peripheral CD4+ T cells clearly discriminate between patients with acute and A. L. Dent, J. Craft, and S. Crotty. 2009. Bcl6 and Blimp-1 are reciprocal and chronic hepatitis B infection. Hepatology 49: 781–790. antagonistic regulators of T follicular helper cell differentiation. Science 325: 80. Huehn, J., and A. Hamann. 2005. Homing to suppress: address codes for Treg 1006–1010. migration. Trends Immunol. 26: 632–636. 71. Cimmino, L., G. A. Martins, J. Liao, E. Magnusdottir, G. Grunig, R. K. Perez, 81. Walker, L. S. 2009. Regulatory T cells overturned: the effectors fight back. and K. L. Calame. 2008. Blimp-1 attenuates Th1 differentiation by repression of Immunology 126: 466–474. ifng, tbx21, and bcl6 gene expression. J. Immunol. 181: 2338–2347. 82. Shevach, E. M. 2009. Mechanisms of foxp3+ T regulatory cell-mediated sup- 72. Yu, D., S. Rao, L. M. Tsai, S. K. Lee, Y. He, E. L. Sutcliffe, M. Srivastava, pression. Immunity 30: 636–645. M. Linterman, L. Zheng, N. Simpson, et al. 2009. The transcriptional repressor 83. Wang, R., Q. Wan, L. Kozhaya, H. Fujii, and D. Unutmaz. 2008. Identification of Bcl-6 directs T follicular helper cell lineage commitment. Immunity 31: 457– a regulatory T cell specific cell surface molecule that mediates suppressive 468. signals and induces Foxp3 expression. PLoS One 3: e2705. 73. Peng, S., S. Lalani, J. W. Leavenworth, I. C. Ho, and M. E. Pauza. 2007. c-Maf 84. Tran, D. Q., J. Andersson, R. Wang, H. Ramsey, D. Unutmaz, and interacts with c-Myb to down-regulate Bcl-2 expression and increase apoptosis E. M. Shevach. 2009. GARP (LRRC32) is essential for the surface expression of in peripheral CD4 cells. Eur. J. Immunol. 37: 2868–2880. 74. Kallies, A., E. D. Hawkins, G. T. Belz, D. Metcalf, M. Hommel, L. M. Corcoran, latent TGF-beta on platelets and activated FOXP3+ regulatory T cells. Proc. P. D. Hodgkin, and S. L. Nutt. 2006. Transcriptional repressor Blimp-1 is es- Natl. Acad. Sci. USA 106: 13445–13450. sential for T cell homeostasis and self-tolerance. Nat. Immunol. 7: 466–474. 85. Reeves, J. R., J. W. Xuan, K. Arfanis, C. Morin, S. V. Garde, M. T. Ruiz, 75. Martins, G. A., L. Cimmino, M. Shapiro-Shelef, M. Szabolcs, A. Herron, J. Wisniewski, C. Panchal, and J. E. Tanner. 2005. Identification, purification and E. Magnusdottir, and K. Calame. 2006. Transcriptional repressor characterization of a novel human blood protein with binding affinity for prostate Blimp-1 regulates T cell homeostasis and function. Nat. Immunol. 7: 457–465. secretory protein of 94 amino acids. Biochem. J. 385: 105–114. Downloaded from 76. Gavin, M. A., J. P. Rasmussen, J. D. Fontenot, V. Vasta, V. C. Manganiello, 86. Frost, R. J., and S. Engelhardt. 2007. A secretion trap screen in yeast identifies J. A. Beavo, and A. Y. Rudensky. 2007. Foxp3-dependent programme of protease inhibitor 16 as a novel antihypertrophic protein secreted from the heart. regulatory T-cell differentiation. Nature 445: 771–775. Circulation 116: 1768–1775. 77. Torgerson, T. R., A. Genin, C. Chen, M. Zhang, B. Zhou, S. An˜over-Sombke, 87. Crawford, N. P., R. C. Walker, L. Lukes, J. S. Officewala, R. W. Williams, and M. B. Frank, I. Dozmorov, E. Ocheltree, P. Kulmala, et al. 2009. FOXP3 inhibits K. W. Hunter. 2008. The Diasporin Pathway: a tumor progression-related tran- activation-induced NFAT2 expression in T cells thereby limiting effector cyto- scriptional network that predicts breast cancer survival. Clin. Exp. Metastasis 25: kine expression. J. Immunol. 183: 907–915. 357–369. http://www.jimmunol.org/ by guest on September 26, 2021