Onecut Transcription Factor OC2 Is a Direct Target of T-Bet in Type-1 T-Helper Cells

ORIGINAL ARTICLE Onecut transcription factor OC2 is a direct target of T-bet in type-1 T-helper cells

K Furuno1,2, K Ikeda2, S Hamano3, K Fukuyama1, M Sonoda1, T Hara2, T Sasazuki4 and K Yamamoto1 1Department of Molecular Genetics, Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan; 2Department of Pediatrics, Graduate School of Medical Science, Kyushu University, Fukuoka, Japan; 3Department of Parasitology, Graduate School of Medical Science, Kyushu University, Fukuoka, Japan and 4International Medical Center of Japan, Tokyo, Japan

T-box transcription factor, T-bet, has a central role in the differentiation of T-helper (Th) progenitor cells to Th1 or Th2 effector cells, partly by regulating the expression of genes such as interferon-g (IFN-g). However, the direct target genes, especially those mediating the transcriptional network initiated by T-bet, are not yet fully understood. By combining chromatin immunoprecipitation from Th1 cells with human cytosine-phosphate-guanine-island array analysis, Onecut 2 (OC2), which encodes a member of the ONECUT class of transcriptional activators, was identified as a direct target gene of T-bet. OC2 is expressed in Th1 but not Th2 cells and reporter assays showed that T-bet transactivates OC2 transcription through putative T-bet half-sites locating À451 to À347 of OC2 promoter region. Moreover, we found that OC2 binds and transactivates human T-bet promoter. These results suggest that not only cell-extrinsic regulation via the IFN-g/STAT1 pathway, but also cell-intrinsic transcriptional positive feedback loop between T-bet and OC2 could be involved in Th1 development. Genes and Immunity (2008) 9, 302–308; doi:10.1038/gene.2008.18; published online 17 April 2008

Keywords: T-bet; Onecut 2; type-1 T-helper cell; transcription

Introduction T-bet in the cell-extrinsic and -intrinsic pathways

þ involved in Th1 lineage commitment have been well The differentiation of naive CD4 T cells to either established. However, the direct target genes of T-bet are T-helper 1 (Th1) or 2 (Th2) effector cells is a critical not yet fully understood. process during immune responses, with broad implica- Global analysis of the in vivo effects of protein–DNA tions both in responses to pathogens and in autoimmune interactions on gene expression can be achieved by and allergic disorders. Transcription factors including T- hybridizing the DNA fragments isolated by a protein- bet and GATA3 have appeared to play a key role in this specific antibody from the cell against a microarray process by regulating expression of cytokine genes such constructed from sequences containing regulatory g g as interferon- (IFN- ) and interleukin-4 (IL-4) that are elements. Taking advantage of the strong association characterized as dominant factors guiding the develop- 1,2 between CpG islands (CGIs) and gene regulatory ment of Th1 and Th2 cells. T-bet belongs to a family of regions,11 one type of microarray constructed with CGIs transcription factors, the members of which share a has been established to identify transcription factor highly conserved DNA-binding domain T-box, flanked target genes.12–14 To identify genes that are directly by diverse non-DNA-binding domains, which are 3–6 regulated by T-bet in Th1 cells, we employed combina- involved in transactivation or repression. A series of tory analysis of chromatin immunoprecipitation (ChIP) studies in vitro and in vivo showed that T-bet is induced and CGI microarray. We described here that ONECUT by the T-cell receptor and the IFN-gR/STAT1 signaling þ class of transcriptional activator OC2 is a target gene of pathway in antigen-recognizing naive CD4 T cells. The T-bet. T-bet is associated with OC2 promoter and elevated T-bet induces IL-12Rb2 and IFN-g expression, transactivates the expression. Furthermore, there exist allowing IL-12R/STAT4 signaling to optimize IFN-g OC2-binding elements in human T-bet promoter and expression, and thereby amplifying and establishing 7–9 OC2 transactivates T-bet expression. Our finding sug- the effector commitment of Th1 cells. In addition, gests that transcriptional positive feedback loop between T-bet induces the expression of homeobox transcription T-bet and OC2 could be involved in Th1 development. factor, Hlx, and both proteins cooperate to transactivate IFN-g and T-bet expression.10 Thus, the central roles of Results and discussion Correspondence: Dr K Yamamoto, Department of Molecular A microarray containing duplicates of 9216 human CGIs Genetics, Medical Institute of Bioregulation, Kyushu University, was generated and probed with genome fragments 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan. E-mail: [email protected] isolated by ChIP using an antibody specific to T-bet or Received 17 December 2007; revised 21 February 2008; accepted 21 an antibody-free control from Th1-polarized human February 2008; published online 17 April 2008 T cells (Figure 1a). We detected 50 spots at which the T-bet regulates OC2 transcription K Furuno et al 303 Immunoprecipitation of T-bet-bound chromatin Th1 Th2 Purification of DNA and labeling OC2 IFN-γ Hybridization with CpG microarray (9216 spots) IL-4 Pick up 50 positive spots T-bet

Identification of 28 candidate genes GATA3 G3PDH Comparison of expression between Th1 and Th2 cells

OC2

G3PDH

OC2 Th1 β-actin OC2 OC2 G3PDH Th2 β-actin

Figure 1 Identification of Onecut 2 (OC2) as T-bet target gene. (a) A flow chart describing the methodology used to identify T-bet targets. Human naive CD4 þ T cells, which were isolated from cord blood, were stimulated in Th1 condition. A total of four individual chromatin immunoprecipitation (ChIP) assays were performed with anti-T-bet antibody or no antibody, using approximately 107 cells for each experiment. The immunoprecipitates were pooled and then labeled to probe the CpG islands (CGI) island microarray. A total of 28 genes were selected based on the genomic location of the DNA sequences of positive spots, and the expression in Th1 and Th2 cells was analyzed by reverse transcription (RT)-PCR. (b) The expression of OC2 in Th1 cells. cDNAs from Th1 and Th2 cells were amplified by specific primers to the indicated genes. (c, d) The expression pattern of OC2 in adult organs and peripheral blood cells. OC2 transcripts were detected by RT-PCR. (e) T-bet binds the OC2 promoter in Th1 cells. ChIP assays were performed with no antibody (lane 2) and anti-T-bet antibody (lane 3). Data representative of the four independent assays are shown. signal obtained with T-bet-associated chromatin was at (Figure 1d). Hence, OC2 seems to be specifically least twofold higher than that obtained by probing with expressed in activated CD4 þ T cells, especially in Th1 the control chromatin. The end reads of the positive CGI cells (Figure 1b and d). clones were aligned to the human genome sequence We next investigated whether T-bet is physically using the Blast-like alignment tool obtained from the associated with the OC2 promoter in vivo by ChIP University of California Santa Cruz Genome Bioinfor- experiments using PCR primers specific for OC2 matics Web site (http://www.genome.ucsc.edu/). The promoter region from À618 to À270. As shown in size of the chromatin in the ChIP was approximately Figure 1e, genomic DNA fragment including OC2 1–3 kb. As a result, 28 genes located within 5 kb of the promoter was coimmunoprecipitated with T-bet from CGIs were selected for further comparison of their the cell extract of Th1 but not Th2 cells. Thus, OC2 can be expression in Th1 and Th2 cells by reverse transcrip- a target gene of T-bet in Th1 cells. tion-polymerase chain reaction (RT-PCR) excluding To determine whether T-bet directly activates the OC2 false-positive clones due to repeated elements and/or gene, a human genome fragment including the OC2 nonspecific precipitation. promoter region from À1905 to þ 124 was examined to We found higher expression level of the gene encoding induce reporter transcription in the presence of T-bet or the OC2 in Th1 cells compared with Th2 cells (Figure 1b). GATA3 as a control. We found that overexpressed T-bet, The expression patterns of OC2 in hematopoietic but not GATA3, transactivated the reporter through the lineages, including lymphocytes, have remained unclear, OC2 promoter region in a dose-dependent manner although a previous study reported high expression (Figure 2a). A similar level of transactivation by T-bet levels in the brain, testis, liver and skin.15 We examined was observed when the OC2 promoter region from À451 the expression of OC2 in various tissues and found the to þ 124 was tested, whereas transactivation was transcript in the thymus and spleen (Figure 1c). OC2 markedly decreased in the OC2 promoter region from transcripts were not detected in resting CD4 þ , CD8 þ , À347 to þ 124 (Figure 2b). We did not find the typical CD19 þ , CD14 þ or activated CD8 þ cells using RT-PCR consensus-binding sequence for T-box in the region

Genes and Immunity T-bet regulates OC2 transcription K Furuno et al 304 (data not shown). However, the introduction of mutations in the first and second half-sites of the OC2 200 promoter region between À451 and þ 124 reduced the T-bet-mediated transactivation to a similar level as that seen in the OC2 promoter region from À347 to þ 124 in reporter assays (Figure 2c). Together with the ChIP experiments showing that OC2 promoter À618 to À270 100 region was coimmunoprecipitated with T-bet (Figure 1e),

luciferase activity our data suggested that OC2 gene was transactivated in conjunction with these sites in cells. We next examined the effects of truncated mutations of 0 T-bet on transcriptional activation of OC2 promoter in 021 (µg) reporter assays. It was found that the transactivation OC2 promoter + T-bet required the T-box DNA-binding domain (Figure 3a). We OC2 promoter + GATA3 also observed that both the N- and C-terminal regions of no promoter + T-bet T-bet were involved in transactivation, although previous reports suggested that the C-terminal alone was an activation domain.10,18 Consistent with this observation, 300 when the heterologous GAL4 DNA-binding domain T-bet (+) (GAL4DBD) was fused to each region of T-bet, both the T-bet (-) N- and C-terminal parts showed transactivation 200 (Figure 3b). These results revealed that T-box is flanked by two potent transactivation domains in T-bet. Taken together, our findings indicate that T-bet binds the OC2 100 promoter via T-box for transactivation in Th1 cells. luciferase activity Members of the ONECUT class of transcription factors are characterized by a bipartite DNA-binding domain 0 composed of a single cut domain associated with a -1905 -451 -347 -92 divergent homeodomain.19 In mammals, three ONECUT OC2 promoter factors, OC1 (HNF6), OC2 and OC3, have been identified so far. The molecular dissection of OC1 revealed that the relative luciferase activity cut domain and the homeodomain act together to 0155 10 determine DNA-binding affinity, specificity and transac- OC2 promoter (-451) tivation.20 The sequence conservation in these domains I(w)-II(w)-III(w) among the three paralogues suggests that OC2 binds a I(m)-II(w)-III(w) similar DNA consensus sequence to OC1. Indeed, many I(w)-II(m)-III(w) of the OC1 targets are also recognized by OC2.21 I(w)-II(w)-III(m) However, the optimal binding sequence for OC2 has I(m)-II(m)-III(w) I(m)-II(w)-III(m) not yet been identified. The identification of the target I(w)-II(m)-III(m) genes of OC2 is essential for elucidating its functional I(m)-II(m)-III(m) roles in Th1 cells. Thus, we determined the consensus- OC2 promoter (-347) binding sequence of OC2 through an in vitro selection experiment using a random sequence oligonucleotide Figure 2 Onecut 2 (OC2) is a direct target gene of T-bet. (a) T-bet and a truncated OC2 protein containing both the cut transactivates the OC2 promoter. The reporter activities of human domain and the homeodomain (Supplementary Figure OC2 promoter (À1905 to þ 124) were examined in the presence of S2a). Five rounds of binding reactions in combination T-bet or GATA3 in COS7 cells. All experiments were performed in ± with PCR enriched the DNA species that might interact triplicate and the mean s.d. is presented. (b) The OC2 promoter with OC2 (Figure 4a). Subsequently, we cloned the PCR region between À451 and À347 is required for transactivation by T-bet. Reporter activities driven by the indicated OC2 promoter products amplified in the final round and analyzed the regions were examined in COS7 cells. (c) Putative T-box half-sites of sequences of 96 clones. We identified 20 different the OC2 promoter region between À451 and À347 affect transacti- sequences, 16 of which were significantly bound to vation by T-bet. Three putative T-box half-sites (designated as I, II OC2 in an in vitro band-shift assay (Supplementary and III; Supplementary Figure S1) were mutated in the reporter Figure S2b). The alignment of these sequences revealed containing OC2 promoter (À451 to þ 124) and the reporter activities the consensus, 50-AATCG(A)ATA(C)-30 (Figure 4b), of each construct, as well as the wild-type (top) and OC2 promoter (À347 to þ 124, bottom), were examined in the presence or absence which overlapped significantly with those of the 0 0 of T-bet in COS7 cells. The relative luciferase activities were cut domain of human CDP (5 -ATCGAT-3 ) and OC1 calculated by dividing the normalized activities in the presence of (50-ATCAAT-30).20,22 T-bet by those in the absence of T-bet. Based on the consensus-binding sequence of OC2, we searched for candidate target genes encoding proteins related to Th1 function by checking the genome between À451 and À347; however, there are at least three sequences of the promoter region. We found tandem putative consensus half-sites (Supplementary Figure S1) repeats of the consensus sequence 50 upstream of the that might be functionally similar to a proximal promoter human T-bet gene, 50-ATCAATAAAGATCGAT-30, in the and a distal enhancer of the IFN-g gene.16,17 We could not region from À490 to À475 (the consensus sequences are detect the in vitro binding of T-bet to these sequences underlined and designated as ‘site 1’ and ‘site 2’,

Genes and Immunity T-bet regulates OC2 transcription K Furuno et al 305 luciferase activity 0 40 80 120 160

empty vector

T-bet full N Tbox C

T-bet del-C N Tbox

T-bet del-N Tbox C

T-bet N N

T-bet T-box Tbox

T-bet C C

luciferase activity 0 50 100 150 200

GAL4DBD G

GAL4DBD T-bet full G N Tbox C

GAL4DBD T-bet N G N

GAL4DBD T-bet T-box G Tbox

GAL4DBD T-bet C G C

Figure 3 Activation domains of T-bet and association of T-bet with Onecut 2 (OC2) promoter in cells. (a) T-box is indispensable for T-bet- induced OC2 promoter activity. Reporter activities of OC2 promoter (À451 to þ 124) induced by truncated mutants of T-bet were examined in COS7 cells. The mutants of T-bet express truncated T-bet as follows: T-bet del-C; N-terminal part and T-box (amino acids 1–334), T-bet del-N; T-box and C-terminal part (amino acids 132–535), T-bet N; N-terminal part (amino acids 1–131), T-bet T-box; T-box (amino acids 132–334), T-bet C; C-terminal part (amino acids 335–535). Schematic representation of the deletion mutants has been shown on the left of the data column. (b) T-bet contains transactivation domains in the N- and C-terminal parts. The N-/C-terminal part and T-box were fused to the GAL4 DNA-binding domain (GAL4DBD). The transcriptional activities of each part were examined using pGL3-17m4-tk-luc as a reporter in the 293 cells. Schematic representation of the GAL4DBD-fused deletion mutants has been shown on the left of the data column. respectively) (Figure 4c). OC2 significantly bound this low level of T-bet and could differentiate into Th1 sequence and the interaction was, at best, partially cells,23,24 it is possible that the OC2-mediated transactiva- competed by a cold probe containing mutations in site tion could be involved in the expression of T-bet in a 1 and/or site 2 (Figure 4d), indicating that both sites different pathway from STAT1. Alternatively, OC2 could were involved in binding to OC2 in vitro. contribute to T-bet expression at some later time during To examine whether the OC2 transactivates the T-bet Th1 development, when the stable expression of T-bet is promoter, we performed reporter assays using reporter already driven by the IFN-gR/STAT1 pathway. We constructs containing T-bet or IFN-g promoter. We found should also consider the effects of other target genes of that the reporter gene that contains T-bet promoter (À976 OC2 that were not determined in the present study. The to þ 27), but not the IFN-g promoter, was transactivated studies using OC2-deficient mouse revealed the impor- by OC2 (Figure 5a). Furthermore, a ChIP experiment tant roles of this transcription factor in liver and pancreas using an OC2-specific antibody, which is able to development,25–27 suggesting its broad range of contribu- immunoprecipitate formaldehyde-fixed OC2 protein tion to cellular development. (Supplementary Figure S3), revealed an association In conclusion, the present work revealed a new target between OC2 and the T-bet promoter in Th1 cells gene of T-bet, OC2, in Th1 helper T cell, and implies that (Figure 5b). Thus, these results suggest that T-bet can transcriptional network between T-bet and OC2 coexists be a direct target gene of OC2 and is transactivated by with an exocrine mechanism in the complex cellular this transcription factor in Th1 cells. processes in Th1 development. Our present study suggests a transcriptional positive feedback loop between T-bet and OC2 in Th1 cells. The endogenous expression of T-bet was shown to be Materials and methods induced by the forced expression of ectopic T-bet,8 which implied the autoactivation of T-bet, similar to Plasmid constructs GATA-3. However, a subsequent report revealed that Details of the individual constructs, which were all a cell-extrinsic regulatory circuit involving IFN-gR verified by sequencing, are available upon request. signaling via STAT1 largely maintained the high-level Constructs for mammalian cell transfection were based expression of T-bet in developing Th1 cells, rather than a on pSG5, pCMV-myc (Clontech Laboratories, Mountain cell-intrinsic pathway of T-bet autoactivation.9 Because View, CS, USA), or pG4MpolyII. The human promoter CD4 þ T cells from STAT1-deficient mice still expressed a regions of the T-bet, OC2 and IFN-g gene were cloned to

Genes and Immunity T-bet regulates OC2 transcription K Furuno et al 306 control OC2 (298-485) luciferase activity repeat times: 1234 51 2 3 45 0 400 800 PCR products control T-bet (EtBr staining) OC2 promoter (-976 to +24) γ control IFN- clone 1 TATTAATCAATTAATA promoter clone 2 GTACCTAATCAATAACATTG OC2 (-1049 to +34) clone 4 AAATAATCAATATCATATAG clone 5 CCACATCGATCGAACTATCC clone 6 GTTGCCTTGGGTCGATAATC clone 7 AAAATTAAAAAATCGATAAA clone 8 ATTAAATCAATAATTCCTAT clone 11 AATAGATAGTAATCAATATT clone 12 GCAAACACGAAATCGATAAT clone 13 GTTGAAAAAAATCGATAATA clone 14 CTAAATCAATACTCTAGCTC T-bet clone 16 TCCCATAATCGATCGGTAC clone 17 GAACGGAAATCGATATAAAT β-actin clone 18 GGATAGTCAATCGATAATCT clone 19 TACAGCGGATCGATAACATC clone 20 CAGCCAAATCAATCAATCGA Figure 5 T-bet is a direct target gene of Onecut 2 (OC2). (a) OC2 consensus -----AATCGATA----- transactivates the T-bet promoter. The reporter activities of human A C T-bet and the interferon-g (IFN-g) promoter were examined in the presence or absence of OC2 in COS7 cells. (b) OC2 binds the T-bet promoter in Th1 cells. Chromatin immunoprecipitation (ChIP) human T-bet promoter assays were performed with no antibody (lane 2) and anti-OC2 antibody (lane 3) that was generated against human OC2-derived -497 agtaggtatcaataaagatcgattgaatgt -468 peptide 44-ASPSPHHARR-53 (Supplementary Figure S3). Data site1 site2 representative of the four independent assays are shown.

protein: GST GST-OC2 (296-485) WT mut1 mut2 mut3 Isolation of naive CD4 þ T cells from cord blood competitor: (-) (-) Human naive T cells were prepared from cord blood cells of full-term neonates, who had no hereditary disorders, hematological abnormalities or infectious complications, by density-gradient centrifugation. Written informed consent was obtained from all mothers. This study was approved by the ethics committee of Kyushu University. CD4 þ cells were isolated using anti-CD4 monoclonal antibody-coated magnetic beads and an MACS system labeled probe: WT (Miltenyi Biotec, Bergisch Gladbach, Germany). The site1 site2 expression levels of CD4 and CD45RA were examined WT taggtatcaataaagatcgattgaatg by flow cytometry (EPICS XL, Immunotech Coulter, mut1 taggtctatggatagatcgattgaatg Miami, FL, USA) in each preparation and most cells mut2 taggtatcaataaagctatggagaatg mut3 taggtctatggatagctatggagaatg (495%) were doubly positive.

Figure 4 Identification of Onecut 2 (OC2)-binding element. (a) The T-cell stimulation in vitro selection of an OC2-binding consensus sequence. The PCR products of each round of selection are shown. The PCR products For the CGI microarray, RT-PCR analysis and ChIP assay, þ þ 5 from final round of selection (lane 10) were cloned for sequencing. CD4 CD45RA cells (5 Â 10 ) were stimulated with (b) A total of 16 out of the 20 sequences significantly bound to OC2 1 mgmlÀ1 PHA (Wako Pure Chemical Industries, Osaka, in vitro (Supplementary Figure S2b). These sequences were Japan) and 50 U mlÀ1 recombinant human IL-2 (Shionogi, compiled and the consensus sequences are shown. (c) The OC2 Osaka, Japan) in the presence of either 10 ng mlÀ1 consensus sequences in the human T-bet promoter region. The two recombinant human IL-12 (R&D Systems, Minneapolis, boxed consensus sequences were designated as site 1 and site 2. À1 (d) OC2 binds the fragment of the T-bet promoter in vitro. MN, USA) and 10 mgml anti-human IL-4 (R&D À1 Glutathione S-transferase (GST) or GST-OC2 (296–485) were Systems), or 20 ng ml human IL-4 (Pepro Tech, Rocky subjected to band-shift assay with a labeled probe of the T-bet Hill, NJ, USA) and 10 mgmlÀ1 anti-human IL-12 (R&D promoter region from À497 to À468 (wild-type, WT) in the presence Systems), for Th1 or Th2 differentiation, respectively. or absence of the indicated nonlabeled competitor. The mut1, mut2 After 3 days of culture, the cells were restimulated with and mut3 probes contained mutations in site 1, site 2 and both sites, 20 nM of PMA (Sigma-Aldrich, St Louis, MO, USA) and respectively. 1 mM of ionomycin (Sigma) for 4 h, and then harvested for the experiments. pGL3-basic (Clontech). The pSG5 and pG4MpolyII were kindly provided by P Chambon (Institut de ChIP assay Ge´ne´tique et de Biologie Mole´culaire et Cellulaire, The ChIP assay was performed as described previously28 Strasbourg, France). The pGEX-2T (Amersham, using a ChIP Assay Kit (Upstate, Lake Placid, NY, USA). Buckinghamshire, UK) and pGAD-T7 (Clontech) were Approximately 107 cells were used for each assay. Anti- used to express glutathione S-transferase (GST) fusion T-bet (4B10, Santa Cruz Biotechnology, Santa Cruz, CA, and in vitro translated hemagglutinin (HA)-tagged USA) and anti-OC2 antibodies produced against human proteins, respectively. OC2-derived peptide, 44-ASPSPHHARR-53, were used.

Genes and Immunity T-bet regulates OC2 transcription K Furuno et al 307 The PCR primers are as follows: OC2, 50-GGCATCTTT protein G. Double-stranded oligonucleotides containing CACCGAATCTC-30 and 50-CGTCTTCTGTTTGGGTGA 20-nt random sequences were generated based on oligo GC-30; T-bet, 50-TTGTCCATCAGGTTCCAGGT-30 and N20 extended by primer B and Klenow DNA polymerase 50-ATGGCCCAAGTGGGACTC-30; b-actin, 50-GGGACT (Supplementary Figure S2a). The immunoprecipitates CAAGGCGCTAACT-30 and 50-GGCCGCGTTATTACCA were preincubated with 0.1 mgmlÀ1 of poly(dI-dC) in 0 TAAA-3 . binding buffer (50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 5% glycerol and 1 mM DTT), and approximately 1 mgof CGI microarray double-stranded oligo N20 was added for further The human CGI library was purchased from the UK incubation. The protein G beads were washed and the Human Genome Mapping Project Resource Centre. A bound DNA was purified by phenol/chloroform extrac- total of 9216 clones were organized individually in 96- tion. The purified DNA was amplified by 15 cycles of well plates. The CGI inserts were amplified by PCR PCR using primers A and B (Supplementary Figure S2a). using the 50-CGGCCGCCTGCAGGTCGACCTTAA-30 The PCR products were used for the next round of and 50-AACGCGTTGGGAGCTCTCCCTTAA-30 pri- incubation. This reaction was repeated five times and the mers.29 The PCR products were purified using Magnesil final PCR products were cloned for sequencing. (Promega, Madison, WI, USA) and then mixed with dimethyl sulfoxide (DMSO) to a final concentration of Band-shift assay 10%. Array Spotter Generation III (Molecular Dynamics, Purification of the GST and GST-OC2 (296–485) proteins Sunnyvale, CA, USA) was used to assemble the PCR/ were performed as previously described.30 Approxi- dimethyl sulfoxide samples on microarray slides (Type 7 mately 50 ng of protein was preincubated with Star; Amersham) as microdots (spot diameter ¼ 250 ml 0.1 mgmlÀ1 of poly(dI-dC) for 5 min in binding buffer, and spot buffer ¼ 30 ml). A total of four individual ChIP and 0.3 pmol of the indicated oligonucleotide probe assays were performed with anti-T-bet antibody or no labeled with biotin at the 50 end of one of strand was antibody, using approximately 107 cells for each experi- added for further incubation. For the competition assay, ment. The combined test or control samples were labeled 0.3 or 1.5 pmol of nonlabeled competitor was added to according to the protocols developed by DeRisi et al. the reaction mixture before incubation with the labeled (Microarrays.org, http://www.derisilab.ucsf.edu/data/ probe. The samples were loaded onto a 10% polyacry- microarray/index.html) and hybridized with microarray lamide gel, transferred onto a nylon membrane slides using an automated slide processor (Amersham). (Hybond-N þ ; Amersham) and visualized using avidin The slides were scanned with the GenePix 4000A (Axon) peroxidase by standard protocol for western blotting. and the images were analyzed with GenePix Pro 3.0 software. Spots showing a Cy5/Cy3 ratio in duplicate42 were selected as positive.12 Acknowledgements RT-PCR The BD Premium RNAs (Clontech) were used to We thank T Tanaka, T Akinaga, M Goto and M Ohnish examine the expression of OC2 in adult organs and for technical help. This work was supported by a Grant- peripheral blood cells. The primers for RT-PCR are in-Aid for Scientific Research in the priority area of as follows: OC2, 50-ATGATGTCGCACCTGAACG-30 and Genome Biology, and by a grant from the 21st Center of 50-TGGGAATTGTTCCTGTCTTTG-30; IFN-g,50-TGCAG Excellence Program from the Ministry of Education, GTCATTCAGATGTAGC-30 and 50-CAGTTCAGCCATC Culture, Sports, Science and Technology of Japan. ACTTGGA-30;IL-4,50-ACAAGTGCGATATCACCTTAC-30 and 50-CAACGTACTCTGGTTGGCT-30; T-bet, 50-CCAA CAATGTGACCCAGATG-30 and 50-ATCTCCCCCAAG References GAATTGAC-30; GATA3, 50-AACTGTCAGACCACCAC AACCACAC-30 and 50-GGATGCCTTCCTTCTTCATAG 1 Szabo SJ, Sullivan BM, Peng SL, Glimcher LH. Molecular TCAGG-30;G3PDH,50-ACCACAGTCCATGCCATCAC-30 mechanisms regulating Th1 immune responses. Annu Rev and 50-TCCACCACCCTGTTGCTGTA-30. Immunol 2003; 21: 713–758. 2 Murphy KM, Reiner SL. The lineage decisions of helper Reporter assay T cells. Nat Rev Immunol 2002; 2: 933–944. 3 Reporter plasmids (1 mg), phRL-CMV (50 ng, internal 3 Wilson V, Conlon FL. The T-box family. Genome Biol 2002; : reviews 3008.1–3008.7. control for transfection efficiency) and effector plasmids 4 Szabo SJ, Kim ST, Costa GL, Zhang X, Fathman CG, Glimcher (2 mg) were used in all reporter assays except that LH. A novel transcription factor, T-bet, directs Th1 lineage indicated amounts of effector were transfected in the commitment. Cell 2000; 100: 655–669. assay shown in Figure 2a. Firefly and Renilla luciferase 5 Finotto S, Neurath MF, Glickman JN, Qin S, Lehr HA, Green activities were measured sequentially using a dual FH et al. Development of spontaneous airway changes luciferase reporter-assay system (Promega). Transfection consistent with human asthma in mice lacking T-bet. Science efficiencies were normalized to the Renilla luciferase 2002; 295: 336–338. activity. 6 Szabo SJ, Sullivan BM, Stemmann C, Satoskar AR, Sleckman BP, Glimcher LH. Distinct effects of T-bet in TH1 lineage commitment and IFN-gamma production in CD4 and CD8 In vitro binding site selection T cells. Science 2002; 295: 338–342. HA-tagged OC2 protein containing the cut domain and 7 Lighvani AA, Frucht DM, Jankovic D, Yamane H, Aliberti J, the homeodomain was generated by an in vitro trans- Hissong BD et al. T-bet is rapidly induced by interferon- cription/translation system (Promega) and purified by gamma in lymphoid and myeloid cells. Proc Natl Acad Sci USA immunoprecipitation with anti-HA antibody coupled to 2001; 98: 15137–15142.

Genes and Immunity T-bet regulates OC2 transcription K Furuno et al 308 8 Mullen AC, High FA, Hutchins AS, Lee HW, Villarino AV, 20 Lannoy VJ, Burglin TR, Rousseau GG, Lemaigre FP. Isoforms Livingston DM et al. Role of T-bet in commitment of TH1 cells of hepatocyte nuclear factor-6 differ in DNA-binding proper- before IL-12-dependent selection. Science 2001; 292: 1907–1910. ties, contain a bifunctional homeodomain, and define the new 9 Afkarian M, Sedy JR, Yang J, Jacobson NG, Cereb N, Yang SY ONECUT class of homeodomain proteins. J Biol Chem 1998; et al. T-bet is a STAT1-induced regulator of IL-12R expression 273: 13552–13562. in naive CD4+ T cells. Nat Immunol 2002; 3: 549–557. 21 Vanhorenbeeck V, Jacquemin P, Lemaigre FP, Rousseau GG. 10 Mullen AC, Hutchins AS, High FA, Lee HW, Sykes KJ, OC-3, a novel mammalian member of the ONECUT class of Chodosh LA et al. Hlx is induced by and genetically interacts transcription factors. Biochem Biophys Res Commune 2002; 292: with T-bet to promote heritable T(H)1 gene induction. Nat 848–854. Immunol 2002; 3: 652–658. 22 Harada R, Berube G, Tamplin OJ, Denis-Larose C, Nepveu A. 11 Murakami K, Kojima T, Sakaki Y. Assessment of clusters of DNA-binding specificity of the cut repeats from the human transcription factor binding sites in relationship to human cut-like protein. Mol Cell Biol 1995; 15: 129–140. promoter, CpG islands and gene expression. BMC Genomics 23 Bettelli E, Sullivan B, Szabo SJ, Sobel RA, Glimcher LH, 2004; 5: 16. Kuchroo VK. Loss of T-bet, but not STAT1, prevents the 12 Weinmann AS, Yan PS, Oberley MJ, Huang TH, Farnham PJ. development of experimental autoimmune encephalomyelitis. Isolating human transcription factor targets by coupling J Exp Med 2004; 200: 79–87. chromatin immunoprecipitation and CpG island microarray 24 Lieberman LA, Banica M, Reiner SL, Hunter CA. STAT1 analysis. Genes Dev 2002; 16: 235–244. plays a critical role in the regulation of antimicrobial 13 Mao DY, Watson JD, Yan PS, Barsyte-Lovejoy D, Khosravi F, effector mechanisms, but not in the development of Wong WW et al. Analysis of Myc bound loci identified by CpG Th1-type responses during toxoplasmosis. J Immunol 2004; island arrays shows that Max is essential for Myc-dependent 172: 457–463. repression. Curr Biol 2003; 13: 882–886. 25 Clotman F, Jacquemin P, Plumb-Rudewiez N, Pierreux CE, 14 Wells J, Yan PS, Cechvala M, Huang T, Farnham PJ. Van der Smissen P, Dietz HC et al. Control of liver cell fate Identification of novel pRb binding sites using CpG micro- decision by a gradient of TGF beta signaling modulated by arrays suggests that E2F recruits pRb to specific genomic sites Onecut transcription factors. Genes Dev 2005; 19: 1849–1854. during S phase. Oncogene 2003; 22: 1445–1460. 26 Vanhorenbeeck V, Jenny M, Cornut JF, Gradwohl G, Lemaigre 15 Jacquemin P, Lannoy VJ, Rousseau GG, Lemaigre FP. OC-2, a FP, Rousseau GG et al. Role of the Onecut transcription factors novel mammalian member of the ONECUT class of home- in pancreas morphogenesis and in pancreatic and enteric odomain transcription factors whose function in liver partially endocrine differentiation. Dev Biol 2007; 305: 685–694. overlaps with that of hepatocyte nuclear factor-6. J Biol Chem 27 Margagliotti S, Clotman F, Pierreux CE, Beaudry JB, Jacque- 1999; 274: 2665–2671. min P, Rousseau GG et al. The Onecut transcription 16 Cho JY, Grigura V, Murphy TL, Murphy K. Identification of factors HNF-6/OC-1 and OC-2 regulate early liver expansion cooperative monomeric Brachyury sites conferring T-bet by controlling hepatoblast migration. Dev Biol 2007; 311: responsiveness to the proximal IFN-gamma promoter. Int 579–589. Immunol 2003; 15: 1149–1160. 28 Remboutsika E, Yamamoto K, Harbers M, Schmutz M. The 17 Lee DU, Avni O, Chen L, Rao A. A distal enhancer in the bromodomain mediates transcriptional intermediary factor interferon-gamma (IFN-gamma) locus revealed by genome 1alpha–nucleosome interactions. J Biol Chem 2002; 272: 50318– sequence comparison. J Biol Chem 2004; 279: 4802–4810. 50325. 18 Kispert A, Koschorz B, Herrmann BG. The T protein encoded 29 Huang TH, Perry MR, Laux DE. Methylation profiling of CpG by Brachyury is a tissue-specific transcription factor. EMBO J islands in human breast cancer cells. Hum Mol Genet 1999; 8: 1995; 14: 4763–4772. 459–470. 19 Lemaigre FP, Durviaux SM, Truong O, Lannoy VJ, Hsuan JJ, 30 Yamamoto K, Sonoda M, Inokuchi J, Shirasawa S, Sasazuki T. Rousseau GG. Hepatocyte nuclear factor 6, a transcription Polycomb group suppressor of zeste 12 links heterochromatin factor that contains a novel type of homeodomain and a single protein 1alpha and enhancer of zeste 2. J Biol Chem 2004; 279: cut domain. Proc Natl Acad Sci USA 1996; 93: 9460–9464. 401–406.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity