Epigenetic Control of Highly Homologous Killer Ig-Like Alleles Huei-Wei Chan, Jeffrey S. Miller, Mikel B. Moore and Charles T. Lutz This information is current as of September 24, 2021. J Immunol 2005; 175:5966-5974; ; doi: 10.4049/jimmunol.175.9.5966 http://www.jimmunol.org/content/175/9/5966 Downloaded from

References This article cites 40 articles, 23 of which you can access for free at: http://www.jimmunol.org/content/175/9/5966.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

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

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 24, 2021

Subscription Information about subscribing to The Journal of 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 Copyright © 2005 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Epigenetic Control of Highly Homologous Killer Ig-Like Receptor Gene Alleles1

Huei-Wei Chan,*† Jeffrey S. Miller,‡ Mikel B. Moore,*† and Charles T. Lutz2*†

Mature human NK lymphocytes express the highly homologous killer Ig-like receptor (KIR) in a stochastic fashion, and KIR transcription precisely correlates with allele-specific DNA methylation. In this study, we demonstrate that CpG methylation of a minimal KIR promoter inhibited transcription. In human peripheral blood NK cells and long-term cell lines, expressed KIR genes were associated with a moderate level of acetylated histone H3 and H4 and trimethylated histone H3 lysine 4. Histone modifications were preferentially associated with the transcribed allele in NK cell lines with monoallelic KIR expression. Although reduced, a substantial amount of histone acetylation and H3 lysine 4 trimethylation also was associated with nonexpressed KIR genes. DNA hypomethylation correlated with increased chromatin accessibility, both in vitro and in vivo. Treatment of NK cell lines and ؅ developing NK cells with the DNA methyltransferase inhibitor, 5-aza-2 -deoxycytidine, caused a dramatic increase in KIR RNA Downloaded from and expression, but little change in histone modification. Our findings suggest that KIR transcription is primarily con- trolled by DNA methylation. The Journal of Immunology, 2005, 175: 5966–5974.

ranscription is controlled by trans-acting factors that bind Killer Ig-like receptor (KIR)3 molecules are critical for human to specific sites in promoters, enhancers, and other regu- NK and T lymphocyte discrimination of aberrant cells from nor- mal self and influence the outcome of autoimmune diseases, trans- T latory regions (1). Transcription also is controlled by her- http://www.jimmunol.org/ itable epigenetic features, which include DNA methylation and a plantation, infectious diseases, and cancer (9). KIR gene expres- variety of histone protein posttranslational modifications (1–3). sion is controlled by an unusual probabilistic mechanism. Acetylation of histones H3 and H4 and methylation of histone H3 Individual NK cells from the same human subject may express at the lysine 4 position are associated with transcription activation. from one to nine of the highly homologous KIR genes from the In contrast, transcription silencing is associated with methylation densely packed KIR locus (9–11), and expressed KIR genes may of the promoter and 5Ј region of the gene and with methylation of be transcribed from the maternal, paternal, or both alleles (12). Long-term NK clones maintain uniform KIR gene- and allele-spe- histone H3 at the lysine 9 position. DNA methylation and sup- cific expression, indicating that KIR transcription patterns are sta- pressive histone modifications may prevent transcription even ble in mature NK cells through many rounds of DNA replication by guest on September 24, 2021 when trans-acting factors are abundant. In turn, trans-acting fac- (10, 12). Because KIR differ in ligand specificity and tors may cause alterations in DNA methylation and histone mod- function, variegated KIR expression is functionally important. At ifications in response to cell signaling and differentiation. It is clear the KIR3DL1 locus, for example, the *001 and *002 alleles encode that DNA methylation and histone modifications influence one an- inhibitory receptors and the KIR3DS1 allele encodes a stimulatory other, but it is less clear which features operate independently and receptor (9). The KIR alleles are extremely similar in both coding whether changes occur in a defined sequence. Suppressive histone and noncoding regions (11, 12), and differential transcription most modifications may precede DNA methylation by a considerable likely is regulated by epigenetic mechanisms. KIR promoters and time interval (2–6). This suggests that DNA methylation is sec- 5Ј regions are methylated in a nearly all-or-none pattern that ondary to repressive histone modification. However, in several strictly correlates with allele-specific KIR expression, both in vitro models, DNA methylation appears to be a primary event (7, 8). It and in vivo (12). This finding suggests an important role of epi- is likely that each model system provides insight into the available genetic controls in programming KIR . It is not strategies to control transcription. Specific gene expression re- clear whether DNA methylation is a primary control mechanism or quirements may necessitate distinct combinations of epigenetic whether it is secondary to other epigenetic events, such as histone regulatory mechanisms. acetylation and methylation. We addressed the hypothesis that clonally restricted and allele-specific KIR transcription is con- trolled primarily by DNA methylation in NK cells.

*Department of Pathology and Laboratory Medicine and †Department of Microbiol- Materials and Methods ogy, Immunology, and Molecular Genetics, Markey Cancer Center, University of Kentucky, Lexington, KY 40536; and ‡Department of Medicine, University of Min- Cells, Abs, flow cytometry, and cDNA preparation nesota, Minneapolis, MN 55455 NK92.26.5 NK cells were subcloned from 5-aza-2Ј-deoxycytidine (Aza)- Received for publication July 8, 2005. Accepted for publication August 16, 2005. treated NK92.26 cells, and both cell lines were maintained in IL-2-con- The costs of publication of this article were defrayed in part by the payment of page taining medium, as described (12, 13). YT-Indy cells were provided by Z. charges. This article must therefore be hereby marked advertisement in accordance Brahmi (University of Indiana, Indianapolis, IN), and YT-HY cells were with 18 U.S.C. Section 1734 solely to indicate this fact. provided by W. Leonard (National Institutes of Health, Bethesda, MD). 1 This work was supported by National Institutes of Health Grants R01 AI50656 and R01 HL55417. 2 Address correspondence and reprint requests to Dr. Charles T. Lutz, Department of 3 Abbreviations used in this paper: KIR, killer Ig-like receptor; Aza, 5-aza-2Ј-deoxy- Pathology and Laboratory Medicine, University of Kentucky, 800 Rose Street, cytidine; ChIP, chromatin immunoprecipitation; LCL, lymphoblastoid cell line; TSA, MS117, Lexington, KY 40536-0298. E-mail address: [email protected] trichostatin A.

Copyright © 2005 by The American Association of Immunologists, Inc. 0022-1767/05/$02.00 The Journal of Immunology 5967

These NK cells acted equivalently and are designated YT in the manu- munoprecipitated complexes were washed six times, and DNA was eluted script. Hut78 T cells, U937 monocytes, K562 myeloid cells, and FaDu with 0.3 ml of elution buffer (50 mM NaHCO3 and 1% SDS). DNA was epithelial cells were obtained from the American Type Culture Collection. released from cross-linking by incubation at 65°C for 4–5 h in the presence The EBV-transformed YS-LCL B cell line has been described and is re- of 0.3 M NaCl and 10 ␮g of RNase A (Sigma-Aldrich). Samples were ferred to as lymphoblastoid cell line (LCL) in the text (14). These cells proteinase K treated for2hat45°C and applied to QIAquick columns were grown in medium containing 10% iron-supplemented bovine serum (Qiagen). Eluted DNA was digested with ApaI to improve specificity for (HyClone). KIR3DL1. Three percent of the treated DNA was used for each PCR am- Allophycocyanin-labeled anti-CD56 and PE-labeled or unlabeled p70 plification. The quantity of KIR3DL1 or ␣-globin PCR product was nor- anti-KIR3DL1 mAb were obtained from Beckman Coulter or from BD malized to G6PD levels, after subtracting the background from normal Biosciences. DX31 anti-KIR3DL2 mAb was provided by J. Phillips rabbit IgG control for each cell line. The results shown are averages, with (DNAX, Palo Alto, CA). GL183 and EB6 were obtained from Beckman error bars representing 1 SD. Coulter. Allophycocyanin-labeled anti-CD34, PE-labeled anti-CD38, and a FITC-labeled mixture of lineage depletion markers (CD2, CD3, CD4, Quantitative PCR CD5, CD7, CD8, CD10, and CD19) were all obtained from BD Bio- Taq sciences. Control mouse IgG and PE-conjugated goat anti-mouse IgG were PCR was performed (SybrGreen JumpStart ReadyMix; Sigma-Al- obtained from Southern Biotechnology Associates. drich) on an MJ Opticon instrument (Bio-Rad), and results were quantified All human cells were obtained with approval of the appropriate Insti- using standard curves that were included in each PCR. Primers were ob- ϩ Ϫ Ϫ tained from Integrated DNA Technologies. For cDNA analysis, the primers tutional Review Board committee. CD34 CD38 lineage progenitors were isolated from umbilical cord blood after CD34 enrichment using were: KIR3DL1, TTCTTGGTCCAGAGGGCCGTT and CTGTAGGTC MACS columns (Miltenyi Biotec), as indicated by the manufacturer. Cells CCTGCAAGGGAAA; KIR3DL2, CGGTCCCTTGATGCCTGT and GAC were then cultured as described with AFT024 mouse fetal liver stromal CACACGCAGGGCAG; and GAPDH, AGTCAGCCGCATCTTCTTTT and cells, primitive-acting factors, IL-7, and either IL-2 or IL-15, conditions GGGAAGGTGAAGGTCG. In all cases, primers were located in cDNA that known to induce NK cell differentiation from primitive progenitors (15). was encoded by separate exons. For quantitative RT-PCR, the amount of sam- Downloaded from NK cells were then cultured in various concentrations of Aza and tricho- ple template was adjusted to give similar amplification of control GAPDH KIR3DL1 statin A (Sigma-Aldrich), as indicated. Cells were stained with anti-CD56 cDNA. For ChIP assays, the primers were: F, GTGAAGGACGC KIR3DL1 mAb and for individual anti-KIR mAb or a mixture containing anti-KIR GAGGTGTCAATTCTAGTGACAG and R, ACCTCTAGGC KIR3DL2 mAb, GL183, EB6, and DX9. CCATATCTTTACCTCCAAGT; , AGAATTCAATCACCTCAT G6PD Mature NK cells were obtained as described from donor L, who was GTG and GCTTCCTTGAAATTGTTGTG; and , TAGGGCCGCATC heterozygous at the KIR3DL1 locus (12). Briefly, blood was incubated with CCGCTCCGGAGAGAAGTCT and AGAGACGAGGAGGCGTGTAGGC Ab complexes bispecific for and leukocyte CD3, CD4, GGTGACG. http://www.jimmunol.org/ CD19, CD36, or CD66b (RosetteSep; StemCell Technologies). Erythro- Chromatin accessibility assay cyte-Leukocyte rosettes were removed by centrifugation through a Ficoll density gradient (Sigma-Aldrich). This technique routinely produced 85– A total of 1.5–3 ϫ 106 cultured cells or 1.5 ϫ 106 freshly sorted peripheral ϩ 90% CD56 cells. Enriched NK cells were stained with anti-CD56 and p70 blood NK cells was washed with PBS and resuspended in RSB buffer. ϩ ϩ anti-KIR3DL1 mAb and flow cytometry sorted into CD56 KIR3DL1 Nuclei were washed once with 1ϫ restriction digestion buffer and divided ϩ Ϫ and CD56 KIR3DL1 NK cells. Cells were cultured for 8–15 days with into aliquots for digestion with MspI I (0, 20, or 40 U at 37°C for 15 min; a stimulation mixture of PHA, IL-2, irradiated autologous PBMC, and New England Biolabs). DNA was isolated using QIAamp DNA blood irradiated 721.221 B lymphoblasts. minikit (Qiagen), and 0.5–1 ␮g of DNA was ligated at the MspI overhangs Total RNA was extracted using TRIzol reagent, RNA was treated with with linker oligonucleotides: LK1, CGAGTACTGCACCAGCAAATCC DNase I to remove genomic DNA, and reverse transcription was performed and LK2, GGATTTGCTGGTGCAGTACT. After ApaI digestion to in- using SuperScript II, all according to manufacturer protocols (Invitrogen crease KIR3DL1 specificity, ligation products were PCR amplified using by guest on September 24, 2021 Life Technologies). LK2 and KIR3DL1R, and analyzed by agarose gel electrophoresis. Methylation and transient transfection Results KIR3DL1 DNA containing 256 bp of the core promoter region was iso- KIR expression levels vary in NK and T cell lines lated and gel purified. Purified DNA fragments were either treated with SssI methylase (New England Biolabs) or mock-treated (without enzyme). To determine whether KIR epigenetic features correlate with ex- DNA was ethanol precipitated before undergoing a second round of treat- pression levels, we measured KIR3DL1 RNA and cell surface ex- ment. Completeness of CpG methylation was confirmed by HhaI restric- pression in several model cell lines. Of the cell lines examined, tion digestion of a small portion of DNA samples. An equal amount of NK92.26.5 NK cells had the highest levels of KIR3DL1 RNA and methylated and mock-methylated fragment was then ligated into SstI-XhoI sites of pGL3-basic plasmid (Promega). These constructs were directly cell surface expression (Fig. 1, A and B). NK92.26 had RNA levels transfected into YT cells along with control plasmids containing the ␤-ga- that were ϳ1.5% (range 1–3%) those of NK92.26.5. Some or all of lactosidase gene (16). Cell lysates were assayed for luciferase and ␤-ga- the KIR3DL1 RNA may have come from a small minority of lactosidase activity 40 h later, as described (16). NK92.26 cells (up to 0.8% of cells) that expressed cell surface DNA methylation was analyzed, as described (12). In brief, genomic DNA was treated with sodium bisulfite for6hat50°C and PCR amplified KIR3DL1 at levels comparable to NK92.26.5 cells (Fig. 1C). Al- using 3DL1-specific primers. PCR products were directly subcloned into though YT NK cells have been reported to express KIR3DL1 RNA pCRII-TOPO vector (Invitrogen Life Technologies) for sequencing. (18), the YT cell lines used in these experiments did not express KIR3DL1 RNA or cell surface protein (Fig. 1, A and C). As ex- Chromatin immunoprecipitation (ChIP) analysis pected, LCL B lymphocytes, K562 myeloid cells, U937 monocytic ChIP analysis was modified from a published protocol (17). For each im- cells, and FaDu epithelial cells had no detectable KIR3DL1 RNA ϫ 6 ϫ munoprecipitation, cells (5–10 10 cells for long-term cell lines and 3 or cell surface expression (Fig. 1, A and B, and data not shown). 106 cells for short-term peripheral blood NK cell lines) were fixed with 1% formaldehyde at room temperature for 10 min. Cross-linking was stopped Hut78 T cells had low-level KIR3DL1 RNA, but no detectable cell by addition of glycine to a final concentration of 0.125 M. After washing surface expression (Fig. 1, A and B). in PBS, cells were lysed in RSB (10 mM Tris (pH 8.0), 3 mM MgCl2,10 mM NaCl, and 0.05% Nonidet P-40), and nuclei were collected and lysed KIR3DL1 methylation and transcription are allele specific in 1% SDS buffer. Nuclear lysates were sonicated on ice in the presence of 0.1 g of glass beads and precleared with fixed protein A-positive Staphy- We and others have shown that KIR transcription is repressed by lococcus aureus cells (both from Sigma-Aldrich). Aliquots of nuclear ly- promoter and 5Ј region DNA methylation (12, 19). We determined sates either were analyzed immediately or frozen at Ϫ70°C. Nuclear ly- the methylation status of the KIR3DL1 gene in NK92.26 NK cells sates were incubated overnight at 4°C with no Abs, 10 ␮g of normal rabbit and Hut78 T cells. The bisulfite technique showed that NK92.26 IgG (Sigma-Aldrich), 10 ␮l of anti-acetylated histone H3 Ab (Upstate Bio- technology), 10 ␮l of anti-acetylated histone H4 Ab (Upstate), or 5 ␮lof had nearly uniform promoter methylation of 21 of 22 sequences Ab specific for trimethylated histone H3 lysine 4 (H3K4; Abcam). Abs examined, including both the *001 and *002 alleles (Fig. 2A). One were precipitated with 10 ␮l of fixed S. aureus cells for1hat4°C. Im- KIR3DL1 sequence was hypomethylated, consistent with bright 5968 PRIMARY CONTROL OF KIR GENES BY DNA METHYLATION Downloaded from FIGURE 1. KIR3DL1 RNA and cell surface protein expression. A, Quantitative RT-PCR was used to measure KIR3DL1 RNA in LCL, YT, Hut78 (Hut), NK92.26 (26), and NK92.26.5 (26.5) cells. The results were calculated as a KIR3DL1:GAPDH ratio, normalized to NK92.26.5 cells, and presented as averages and SDs from four experiments. FaDu, K562, and U937 cells had no detectable KIR3DL1 RNA (data not shown). B, KIR3DL1 cell surface expression was tested using KIRp70 mAb (filled http://www.jimmunol.org/ curve) and mouse IgG isotype control (open curve) and presented on a logarithmic scale. In the same experiment, FaDu, K562, and U937 cells had no staining above background (data not shown). Results shown are typical of three experiments. C, NK92.26 cells contain a small number of KIR3DL1ϩ cells. KIR3DL1 cell surface expression was tested as above and was plotted against forward scatter on a logarithmic scale. The per- centage of events in the upper right quadrant is indicated. Results shown are typical of four experiments. by guest on September 24, 2021 KIR3DL1 cell surface expression by a small minority of NK92.26 cells (Fig. 1C). Hut78 also was heterozygous at the KIR3DL1 lo- cus, with an *002 allele and a 3DS1-like allele. In Hut78, the *002 allele was heavily methylated, but half of the 3DS1 allele se- quences were hypomethylated. These results indicated that about half of Hut78 cells had a hypomethylated 3DS1 allele. 3DS1 is similar to other stimulatory KIR proteins, which require DAP12 protein for cell surface expression (20). Like most T cells (20), Hut78 cells did not express DAP12 RNA (data not shown), which accounts for the apparent lack of 3DS1 cell surface expression. We previously demonstrated that allele-specific KIR3DL1 tran- scription correlated with allele-specific DNA methylation in NK92.26.5 cells and in peripheral blood NK clones (12). To ex- tend this finding, we analyzed KIR3DL1 RNA expression in Hut78 T cells. Restriction endonuclease digestion of KIR3DL1 RT-PCR FIGURE 2. KIR3DL1 DNA methylation and RNA expression are allele products from control heterozygous polyclonal peripheral blood specific. A, Methylation status of KIR3DL1 alleles in NK92.26 and Hut78 cells as determined by bisulfite analysis. The location of exon 1 is indi- NK cells showed two allelic bands with both NlaIII and BsaAI cated. F and E, Represent methylated and nonmethylated CpG sites, re- B lane 3 digestion (Fig. 2 , ). NK92.26.5 RT-PCR products showed spectively. Each row of circles represents an individual DNA subclone. only the *001 allele (Fig. 2B, lane 2), consistent with previous The *001, *002, and 3DS1 KIR3DL1 alleles are indicated. B, KIR3DL1 findings (12). Hut78 RT-PCR products showed a restriction endo- RT-PCR products from Hut78 (Hut), NK92.26.5 (26.5), or control het- nuclease digestion pattern consistent with transcription of only the erozygous peripheral blood NK cells (NK) were digested with NlaIII or 3DS1 allele, without evidence of *002 allele transcription (Fig. 2B, BsaAI and electrophoresed on agarose gels with m.w. standards. The pe- lane 1). This correlates with selective 3DS1 allele DNA hypo- ripheral blood NK cells and NK92.26.5 cells are heterozygous for the *001 methylation in Hut78 (Fig. 2A). Thus, KIR DNA hypomethylation and *002 KIR3DL1 alleles. Hut78 is heterozygous for the 3DS1 and *002 correlated with allele-specific RNA expression in both NK cells KIR3DL1 alleles. The *001 and 3DS1 PCR products share the same en- and T cells, including both inhibitory and stimulatory KIR. donuclease digestion pattern (2 NlaIII sites, 1 BsaAI site), which differs from that of *002 PCR products (1 NlaIII site, 2 BsaAI sites). The poly- DNA methylation inhibits KIR promoter activity morphism leading to differential restriction enzyme digestion is illustrated below each photograph. Results shown are typical of three experiments. The correlation between allele-specific KIR gene expression and promoter hypomethylation may indicate that DNA methylation in- The Journal of Immunology 5969 hibits transcription. Alternatively, DNA methylation changes may 10.6 (histone H3)- and 19.7 (histone H4)-fold higher in K562 cells be secondary to other epigenetic changes that regulate transcrip- than in NK92.26 and NK92.26.5 cells (data not shown). KIR3DL1- tion. To test whether KIR promoter methylation inhibits transcrip- nonexpressing epithelial cells, myeloid cells, B cells, and YT NK tion, we used the cassette methylation assay in which all CpG sites cells had similar low-level KIR3DL1-associated acetylated histone within the 256-bp minimal KIR3DL1 promoter were methylated H3 and H4 levels (Fig. 4A and data not shown). KIR3DL1ϩ without modification of the luciferase reporter plasmid. Plasmids NK92.26.5 cells had KIR3DL1-associated levels of acetylated H3 probably do not fully assemble chromatin within 40 h of transfec- and acetylated H4 that were ϳ5-fold greater than the KIR3DL1Ϫ tion (21) and the cassette methylation assay has been used to test FaDu epithelial cells (Fig. 4A). The level of KIR3DL1-associated the effect of precisely targeted promoter methylation (22). As a histone acetylation in Hut78 cells and NK92.26 cells was inter- control, KIR3DL1 promoter DNA was mock methylated in the mediate between that of KIR3DL1Ϫ FaDu epithelial cells and absence of SssI CpG methylase enzyme. Completeness of meth- KIR3DL1ϩ NK92.26.5 NK cells. The differences between cell ylation was confirmed by lack of digestion with the HhaI methy- lines were surprisingly small. Even though Hut78 cells had easily lation-sensitive restriction endonuclease (data not shown). Equal detectable levels of KIR3DL1 RNA and promoter hypomethy- amounts of mock-methylated and SssI-methylated promoter frag- lation, KIR3DL1-associated histone acetylation was only about ments were ligated to an unmodified luciferase reporter gene, and double that of the FaDu-negative control (Fig. 4A). Likewise, ligation products were directly transfected into YT cells. Although even though NK92.26 cells had almost complete KIR3DL1 pro- YT cells did not express the endogenous KIR3DL1 gene (Fig. 1), moter methylation and only 1–3% as much KIR3DL1 RNA as they did express KIR3DL1 promoter-driven luciferase reporter did NK92.26.5 cells (Fig. 1A), NK92.26 cells had close to half genes in transient transfection experiments (12), indicating that YT the level of KIR3DL1-associated acetylated histone H3 and H4 Downloaded from cells contained the trans-acting factors needed for KIR transcrip- (Fig. 4A). tion. Control, mock-methylated KIR3DL1 promoter DNA directed luciferase expression (Fig. 3). SssI methylase treatment reduced KIR3DL1 promoter function by an average of 92.5% (ranging from 82.2 to 97.7% inhibition in three independent experiments). These

results indicate that precisely targeted KIR3DL1 promoter CpG http://www.jimmunol.org/ methylation inhibited transcription in transfected NK cells. Based on newly validated criteria (23), the cluster of CpG sites in the KIR3DL1 promoter is not sufficiently dense to qualify as a CpG island. Therefore, our results indicate that relatively low density CpG methylation inhibits transcription.

KIR3DL1 promoter histone acetylation Chromosomal DNA may become methylated secondary to histone modification (3). Therefore, we investigated histone modifications by guest on September 24, 2021 using the ChIP assay. We examined histone H3 acetylation at the lysine 9 and lysine 14 positions and H4 acetylation at the 5th, 8th, 12th, and 16th positions. As a positive control, we measured ␣-glo- bin-associated acetylated histone levels, which were an average of

FIGURE 3. In vitro methylation inhibits KIR3DL1 promoter activity. The 256-bp KIR3DL1 minimal promoter was isolated by restriction endo- nuclease digestion and gel purification. Left panel, Promoter fragments were either methylated with SssI DNA methylase or mock methylated un- der parallel conditions, but without enzyme. Fragments were quantified, FIGURE 4. KIR3DL1-associated histone acetylation was assessed by and equal amounts were ligated to SstI/XhoI-digested pGL3-basic plasmids the ChIP assay. Shown are quantitative PCR values in comparison with the upstream of the luciferase transcription start site. Ligation products were internal control, G6PD. A, The amount of KIR3DL1-associated acetylated transfected into YT cells, and luciferase activity was corrected for trans- histone H3 and H4 is expressed as a fraction of that in NK92.26.5 cells in fection efficiency. The methylated promoter activity is presented as a frac- six experiments. Cell line abbreviations are as in Fig. 1. B, Peripheral blood tion of the mock-methylated promoter activity. Right panel, As comparison NK cells were flow cytometry sorted into CD56ϩKIR3DL1Ϫ and groups, YT cells were transfected with intact pGL3-basic plasmid that CD56ϩKIR3DL1ϩ populations and were cultured for 8–15 days. The contained the 256-bp KIR3DL1 minimal promoter (KIR) or that did not amount of KIR3DL1-associated acetylated H3 and H4 is expressed as a have a promoter (None). The promoterless plasmid activity is presented as fraction of that in KIR3DL1ϩ NK cells. YT NK cells were included as a a fraction of KIR3DL1 promoter plasmid activity. These data are repre- comparison group. The left panel in B is an average of two experiments. sentative of three independent experiments. The right panel is the result of one experiment. 5970 PRIMARY CONTROL OF KIR GENES BY DNA METHYLATION

Because long-term culture may induce epigenetic changes (24), methylation levels correlated well with KIR3DL1 gene expression we wished to examine polyclonal peripheral blood NK cells. Pu- in long-term and short-term cell lines, the differences between rified CD56ϩ NK cells were sorted into KIR3DL1ϩ and KIR3DL1ϩ and KIR3DL1Ϫ polyclonal NK cells were not KIR3DL1Ϫ populations. In two experiments, the KIR3DL1ϩ and dramatic. KIR3DL1Ϫ populations were Ͼ98 and Ͼ95% pure, respectively (data not shown). The cells were expanded with one or two rounds Allele-specific biasing of histone modifications of in vitro stimulation and were analyzed by ChIP. Levels of Numerous DNA-binding transcription factors recruit proteins that KIR3DL1-associated H3 and H4 acetylation were ϳ1.7- and 4.1- remodel chromatin. We have shown that KIR3DL1 can be ex- ϩ Ϫ fold higher in KIR3DL1 NK cells than in KIR3DL1 NK cells, pressed in a monoallelic fashion, despite very high allele sequence respectively (Fig. 4B). In both populations of polyclonal NK cells, similarity in the KIR3DL1 promoter. Under these circumstances, specific anti-acetylated histone Ab produced KIR3DL1 signal that one KIR3DL1 allele remained heavily methylated and not ex- was significantly higher than the background found in the pressed even when trans-acting factors were present to transcribe Ϫ Ϫ KIR3DL1 YT NK cell line (Fig. 4B). Although KIR3DL1 poly- a nearly identical second allele (12). We determined whether his- clonal NK cells had heavily methylated KIR3DL1 promoters (12), tone modifications preferentially associated with the expressed al- KIR3DL1-associated histone acetylation was significant and only lele. We performed ChIP using Abs to acetylated H3 and to meth- ϩ moderately lower than in KIR3DL1 NK cells. ylated H3K4, amplified the immunoprecipitated KIR3DL1 DNA, and determined allele identity by restriction endonuclease diges- Histone methylation at the KIR3DL1 promoter tion of the PCR products (Fig. 6). Because of low signal, two PCR We wished to examine trimethylation of histone H3 lysine 4 rounds were required to amplify ChIP products from Hut78 cells. Downloaded from (H3K4), which correlates precisely with active transcription in A BstUI site is present in the *002 allele, but not in the *001 or other model genes (25, 26). Using an Ab specific for trimethylated 3DS1 alleles. Nonselected DNA gave rise to PCR products with histone H3K4, we found background levels of H3K4 methylation and without a BstUI recognition site (Fig. 6, “Total DNA ϩ”). This in association with the KIR3DL1 promoter and 5Ј region in result is consistent with the heterozygous KIR3DL1 status in KIR3DL1Ϫ non-NK cells (Fig. 5A). In parallel with the hierarchy NK92.26.5 and Hut78 cells and shows that the PCR products were of KIR3DL1 RNA expression, KIR3DL1-associated H3K4 meth- susceptible to BstUI digestion. As an additional control, MspI http://www.jimmunol.org/ ylation levels were highest in NK92.26.5 cells and decreased in completely digested all PCR products (data not shown). DNA that order Hut78 Ͼ NK92.26 Ͼ YT ϭ background (Fig. 5A). The was coprecipitated with Abs to either acetylated histone H3 or to H3K4 methylation level averaged Ͼ7-fold higher in KIR3DL1ϩ methylated H3K4 produced only faint BstUI digestion bands. This NK92.26.5 cells than in KIR3DL1Ϫ NK92.26 cells (Fig. 5A)in finding indicates that the permissive acetylated histone H3 and five experiments. In contrast to the finding of minimal histone H3 methylated H3K4 modifications preferentially associated with the and H4 acetylation, Hut78 cells had a substantial level of active *001 and 3DS1 alleles in NK92.26.5 and Hut78 cells, re- KIR3DL1-associated H3K4 methylation (Fig. 5A) that averaged spectively. Thus, even in the presence of trans-acting factors that close to half of that observed in NK92.26.5 cells. were adequate to drive high-level KIR3DL1 transcription in In both KIR3DL1ϩ and KIR3DL1Ϫ polyclonal NK cells, the NK92.26.5 cells, permissive histone modifications were preferen- by guest on September 24, 2021 KIR3DL1-associated histone H3K4 methylation level was signif- tially associated with the expressed *001 allele, compared with the icantly elevated above the background level characteristic of YT nonexpressed *002 allele. It also is worth noting that even though cells. Histone H3 methylation at lysine 4 was only 2.6-fold higher the level of KIR3DL1-associated histone acetylation was quite in KIR3DL1ϩ polyclonal NK cells than in KIR3DL1Ϫ polyclonal modest in Hut78 cells (Fig. 4A), the acetylated H3 histones pref- NK cells (Fig. 5B). Although KIR3DL1-associated histone H3K4 erentially associated with the expressed 3DS1 allele, compared with the nonexpressed *002 allele.

FIGURE 6. Allele-specific histone modifications correlate with allele- specific KIR3DL1 expression. ChIP was performed on Hut78 and NK92.26.5 cells using Abs specific for acetylated histone H3 (H3Ac) and histone H3 trimethylated at lysine 4 (H3Me), as indicated. Hut78 ChIP products were amplified in two sequential rounds of PCR. KIR3DL1 PCR products were digested with BstUI and electrophoresed on agarose gels with m.w. standards. Although BstUI is sensitive to DNA methylation, the PCR products do not contain methylated cytosines, and promoter methyl- ation status does not affect this assay. Heterozygous Hut78 and NK92.26.5 cells express the 3DS1 and *001 KIR3DL1 alleles, respectively. A BstUI FIGURE 5. KIR3DL1 histone methylation. Long-term cell lines (A) and site is present in the *002 allele, but not in the 3DS1 and *001 alleles. As short-term peripheral blood NK cells (B) were analyzed for trimethylation a control, KIR3DL1 PCR products amplified from total cellular DNA were at the histone H3 residue 4 lysine (H3Me), as described in Fig. 4. Results BstUI digested (ϩ) or not digested (Ϫ). Results shown are typical of four shown are an average of five (A) and two (B) experiments. experiments. The Journal of Immunology 5971

KIR3DL1 promoter accessibility correlates with transcription duced cell surface KIR3DL1 expression on 28 and 97% of YT and Covalent modifications on histone N-terminal ends modulate chro- NK92.26 cells, respectively (Fig. 8A). In the same cells, drug treat- matin structure and dictate access to trans-acting factors. Because ment increased KIR3DL1 RNA by 51- and 167-fold, respectively the differences in KIR3DL1-associated histone changes were slight to moderate, we tested chromatin accessibility. The MspI accessi- bility assay was validated in previous studies (27). Inaccessibility in this assay could be due to chromatin compaction or blocking by bound repressive proteins. We measured the ability of MspI re- striction endonuclease to digest chromatin at a site 26 bp down- stream of the transcriptional start site (Fig. 7). Isolated nuclei were incubated with MspI; DNA was extracted and ligated to MspI adapters. The ligation products were PCR amplified using an adapter-specific primer and a KIR3DL1-specific primer. The KIR3DL1 transcription start site was readily accessible in NK92.26.5 and Hut78 cells (Fig. 7A). In NK92.26 cells, the KIR3DL1 transcription start site was weakly accessible in some experiments and not accessible in others (Fig. 7). Accessibility could not be detected in cells that did not express KIR3DL1 mRNA (Fig. 7A). Downloaded from We also performed chromatin accessibility assays on the nuclei of peripheral blood NK cells that were analyzed immediately after sorting. The KIR3DL1 promoter was accessible in polyclonal KIR3DL1ϩ NK cells, but not in polyclonal KIR3DL1Ϫ NK cells (Fig. 7B). Thus, KIR3DL1 chromatin accessibility correlated with transcription in cell lines cultured in vitro and in fresh peripheral http://www.jimmunol.org/ blood NK cells in vivo.

DNA methyltransferase inhibition induces KIR3DL1 transcription and chromatin changes Because DNA methyltransferase inhibition caused increased KIR expression (12, 19), we wished to study how DNA methyltrans- ferase inhibition would affect histone modification. We examined

three cell lines that showed a range of responsiveness to Aza, a by guest on September 24, 2021 powerful inhibitor of DNA methyltransferase. Aza treatment in-

FIGURE 8. Effect of DNA methyltransferase inhibition on KIR3DL1 gene expression and histone modification. LCL, YT, and NK92.26 cells were treated with DMSO solvent or with 1 ␮M Aza for 72 h. A and B are each typical of four experiments. C, Shown as averages of either three (H4Ac) or four (H3Ac and H3Me) experiments. A, Cell surface expression. Shown is staining with control mouse IgG (open curves) and anti-KIR3DL1 mAb (filled curves). B, KIR3DL1 RNA levels normalized to the amount of GAPDH RNA. FIGURE 7. KIR3DL1 promoter accessibility. A, Nuclei isolated from Levels are expressed as a fraction of the value in Aza-treated cells. No treat- long-term cell lines were incubated with 0, 20, or 40 U of MspI restriction ment (Ϫ) and treatment with DMSO solvent or with Aza are indicated for YT endonuclease. DNA was extracted, ligated to MspI adapters, and PCR am- cells and for NK92.26 cells. NK92.26 cells reached a higher absolute level plified with an adaptor-specific and a KIR3DL1-specific primer. Cell line than did YT cells (Ͼ200-fold). C, KIR3DL1-associated histone H3 acetylation abbreviations are as in Fig. 1. B, Sorted CD56ϩKIR3DL1Ϫ (3DL1Ϫ NK) (H3Ac), H4 acetylation (H4Ac), and H3K4 methylation (H3Me). Treatment and CD56ϩKIR3DL1ϩ (3DL1ϩ NK) peripheral blood NK cells were an- with DMSO solvent alone (Ϫ) or with Aza (ϩ) is indicated for YT and for alyzed without in vitro culture and were compared with long-term cell NK92.26 (26) cells. Results are normalized with G6PD values and expressed lines. Results shown are typical of three experiments. as a fraction of the Aza-treated NK92.26 cells. 5972 PRIMARY CONTROL OF KIR GENES BY DNA METHYLATION

(Fig. 8B). Cell surface protein and RNA levels in drug-treated Table II. Inhibition of DNA methyltransferase induces expression of NK92.26 cells were comparable to those of KIR3DL1ϩ multiple KIR proteins in developing NK cellsa NK92.26.5 cells (data not shown), indicating that high-level tran- scription was induced in the great majority of Aza-treated Protocol Anti-KIR mAb No Aza Aza (1 ␮M) p Value NK92.26 cells. DNA methyltransferase inhibition did not induce 1 KIR mixture 18.2 Ϯ 4.8 42.1 Ϯ 5.3 0.012 KIR cell surface expression or RNA in LCL B cells (Fig. 8A and 2 DX9 7.8 Ϯ 1.6 22.3 Ϯ 3.4 0.005 data not shown). In LCL B cells, Aza treatment did not alter GL183 17.0 Ϯ 3.5 48.6 Ϯ 5.5 Ͻ0.001 KIR3DL1-associated histone acetylation or methylation (data not EB6 4.5 Ϯ 0.9 30.7 Ϯ 3.0 Ͻ0.001 shown). Aza-treated NK92.26 NK cells increased KIR3DL1-asso- a CD34ϩCD38ϪlineageϪ progenitor cells were sorted from umbilical cord blood ciated histone acetylation by an average of 1.6-fold and methylated and cultured with murine fetal stroma cells and cytokines. After 28 days, 1 ␮M Aza H3K4 by an average of 2.5-fold (Fig. 8C). In YT NK cells, drug- was added to some wells. The cultures were maintained for an additional 4 wk, with weekly replacement of half the medium volume without additional Aza. In two ex- induced changes in histones were small and inconsistent, reflecting periments (protocol 1), NK cells were stained with an anti-KIR mixture. In two cell surface KIR3DL1 induction on a minority of cells (Fig. 8C). separate experiments (protocol 2), NK cells were stained with individual anti- ϩ Showing the generality of these findings, essentially identical re- KIR mAb, as indicted. The data shown are the percentage of KIR NK cells (mean Ϯ SEM) under each condition. Each data set represents an average of five sults (H.-W. Chan, personal observation) were obtained in all three replicates from each of two independent experiments. Significance was determined by cell lines for KIR3DL2, a locus that is separated from KIR3DL1 by Student’s t test. one to four other KIR genes in various haplotypes (9, 11). Because YT cells had little or no KIR3DL1-associated histone acetylation ϩ or H3K4 trimethylation (Figs. 4 and 5), we tested whether the age of KIR cells. Aza treatment up-regulated the expression of Downloaded from histone deacetylase inhibitor, trichostatin A (TSA), could further several KIR, as detected by three anti-KIR mAb with nonoverlap- increase KIR3DL1 expression in YT and other cells. TSA treat- ping specificities, including DX9 anti-KIR3DL1 (Table II). These ment did not increase KIR expression on NK92.26, YT, or LCL data show that DNA demethylation initiates KIR transcription, and cells, at several doses, alone or with a range of Aza concentrations, suggest that DNA methylation controls KIR transcription during although TSA did up-regulate cell surface CD56 gene expression NK cell development.

on YT cells (data not shown). Collectively, our results indicate that http://www.jimmunol.org/ DNA methyltransferase inhibition dramatically increased KIR Discussion transcription, despite little to no increase in histone acetylation or Gene silencing involves histone posttranslational modifications, methylation. DNA methylation, and association of regulatory proteins that sup- In many developmental systems, change in DNA methylation is press transcription through multiple rounds of DNA replication. a late event that follows changes in histone modification. DNA The study of gene silencing has led to two competing models of methylation controls KIR transcription in mature NK cells, but it is epigenetic control. In several vertebrate developmental systems, not clear how locus- and allele-specific expression is established in DNA methylation is a late event and becomes fully established developing NK cells. To study whether DNA methylation influ- long after repressive histone modifications have been enforced. In ences KIR expression in immature NK cells, we used a culture a recent detailed study of Dntt silencing in thymus cells, Su et al. by guest on September 24, 2021 system that recapitulates several aspects of normal NK cell devel- (4) demonstrated a sequential loss of acetylation at histone H3 opment, including variegated KIR expression (15). Primitive lysine 9 (H3K9), loss of methylation at histone H3 lysine 4 CD34ϩCD38Ϫ hemopoietic progenitor cells that did not express (H3K4), gain of H3K9 methylation, and gradual gain of DNA mature lineage markers were cultured in the presence of mouse methylation. In transgene silencing, early loss of histone acetyla- AFT024 stroma cells and cytokines. After 54 days, Ͼ95% of the tion and H3K4 methylation was followed by a slow, gradual gain cells had matured into CD56ϩ NK cells. The cells were harvested of H3K9 methylation and DNA methylation (28). In X-chromo- and cultured in medium containing various concentrations of Aza some inactivation, loss of H3K9 acetylation and H3K4 methyl- and TSA. As shown in Table I, Aza caused a dose-related increase ation and gain of H3K9 methylation were followed by loss of in the percentage of KIRϩ NK cells after 3 days of culture, which histone H4 acetylation, and followed later by DNA methylation (5, continued to increase over time. Several doses of TSA, up to toxic 6). Although the order of histone modification can be variable, the levels, did not further increase KIR expression at any dose of Aza available studies suggest that H3 histone acetylation is lost early tested (data not shown). To extend these findings, we added Aza and DNA methylation is applied late in gene silencing. These and once at 28 days of culture, soon after CD56ϩ NK cells had begun similar studies led to the concept that DNA methylation reinforces to appear (Table II). KIR cell surface expression was analyzed 28 gene silencing that is already established by other epigenetic days later. Aza caused a highly significant increase in the percent- mechanisms (3). In contrast to the models mentioned above, the study of gene silencing in cancer cells suggests that DNA methylation may pro- Table I. Inhibition of DNA methyltransferases induces KIR expression vide the primary control of transcription, with histone modifica- in developing NK cellsa tions playing a secondary role. This conclusion was supported by observations that silenced tumor suppressor genes were reactivated % KIRϩ NK Cells % KIRϩ NK Cells either by Aza treatment or by combined elimination of dnmt1 and Aza (␮M) (3 days) (7 days) dnmt3b DNA methyltransferase genes, but not by TSA inhibition 0 4.5 7.4 of histone deacetylase activity (7, 29). Aza inhibition of DNA 0.5 15.1 25.3 methyltransferases caused sequential loss of DNA methylation, ac- 1 19.6 30.0 tivation of gene transcription, gain of histone acetylation and 2 21.0 33.8 H3K4 methylation, and loss of H3K9 methylation (7). DNA meth- a CD34ϩCD38Ϫ lineage-progenitor cells were sorted from umbilical cord blood ylation attracts DNMT1 and methyl-CpG-binding proteins, both of and cultured with murine fetal stroma cells and cytokines. After 54 days, cells (Ͼ95% which associate with histone deacetylases (30–32), thus providing CD56ϩ NK cells) were cultured in medium containing Aza at the indicated concen- trations for 3 and 7 days. Cells were then stained with a mixture of anti-KIR mAbs. a mechanism by which DNA methylation directs histone The data are representative of two independent experiments. modification. The Journal of Immunology 5973

In all of the model systems discussed above, heavy DNA meth- closed chromatin configurations are presumed to be quite local, ylation was not observed to coexist with significant amounts of because transcribed and quiescent KIR genes are often adjacent histone acetylation. However, this is what we observed in nonex- (10, 11). pressed KIR genes. Our previous study showed that KIR promoter As proposed for tumor suppressor genes, we hypothesize that and 5Ј regions were methylated in a nearly all-or-none pattern that DNA methylation is the key regulator controlling KIR transcrip- strictly correlated with allele-specific KIR expression, both in vitro tion, both in mature NK cells and in development. In support of and in vivo (12). In this study, we show that histone H3 and H4 this hypothesis, histone acetylation and methylation differed only proteins were substantially acetylated in both KIRϩ and KIRϪ moderately between KIR3DL1ϩ and KIR3DL1Ϫ peripheral blood peripheral blood NK cells. The level of KIR-associated histone NK cells. Hut78 T cells expressed KIR despite levels of histone H3 acetylation was only 1.7- to 4.1-fold higher in KIRϩ NK cells than and H4 acetylation that were only about twice that of KIR3DL1Ϫ in KIRϪ NK cells. We observed similar small differences in his- FaDu epithelial cells. Furthermore, KIR gene expression in NK tone H3K4 methylation levels. Thus, heavily methylated nonex- cell lines was increased by Aza inhibition of DNA methyltrans- pressed KIR genes were associated with histone proteins with per- ferases, but not by TSA inhibition of histone deacetylases. Similar missive posttranslational modifications, showing that heavy DNA observations were made in a culture system that mimicked several methylation did not preclude a substantial level of histone aspects of normal NK cell development, including variegated KIR acetylation. expression. Short-term Aza treatment of NK cell lines caused only How is KIR methylation applied (or removed) selectively? Most small to moderate increases in histone acetylation and H3K4 meth- CpG islands become demethylated in early embryogenesis (3). ylation. These observations support the hypothesis that KIR tran-

However, this is not likely for the KIR locus. The progeny of a scription is primarily controlled by DNA methylation, with histone Downloaded from single hemopoietic progenitor express heterogeneous KIR genes modifications playing a secondary role. (15), which effectively rules out inheritance of gene-specific or It is becoming clear that gene silencing and gene activation can allele-specific epigenetic marks that are set in an early stage of proceed via multiple pathways. The numerous mechanisms that development. KIR gene- and allele-specific expression is activated enforce silencing and activation allow individual control of tran- over an extended period during NK development, but stabilizes scription that fits the unique needs of the cell. For example, most

upon reaching a mature NK cell stage and is maintained through CpG islands are nonmethylated regardless of transcription (3). http://www.jimmunol.org/ subsequent rounds of cell division (10, 12, 15). Our preliminary CpG island hypomethylation keeps genes in a state of readiness, studies indicate that KIR genes are heavily methylated in hemo- allowing rapid activation. Such a control strategy would not work poietic progenitor cells (H.-W. Chan, personal observations). The for the KIR locus in which only some genes and alleles are ex- relative paucity of cis-acting promoter elements reported to be es- pressed, despite very high sequence similarity (11, 12). KIR gene sential for KIR transcription further reinforces the importance of expression may play a role in determining the outcome of serious epigenetic control (33–35). Given the extremely high KIR allele infection and in autoimmune diseases (9, 40, 41). The heteroge- sequence similarity (12), the trans-acting factors that direct high- neity of gene expression control strategies presents an opportunity level allele-specific KIR transcription in mature NK cells are not to manipulate normal and aberrant KIR gene expression without sufficient to demethylate silent KIR genes and initiate new KIR globally dysregulating transcription control of other genes. by guest on September 24, 2021 transcription. DNA methylation maintains stable expression pat- terns of KIR in mature NK cells. Acknowledgments We propose that sequence-specific factors recruit DNA demeth- We thank Drs. Andy Russo, Julie Wells, and Peggy Farnham for their ylase enzymes to KIR genes during NK development. We propose generous advice, and Z. Brahmi, W. Leonard, and J. Phillips for gifts of that one or more components of the resulting demethylation com- cells and mAb. We are grateful to Drs. Mark F. Stinski, Bret Spear, and plex is expressed in limited amounts in immature NK cells and Steve Anderson for their critical reading of the manuscript. disappears in mature NK cells. Because the demethylation com- plex is proposed to be present in limited amounts, KIR genes and Disclosures alleles are demethylated in a largely stochastic fashion. Some KIR The authors have no financial conflict of interest. promoters may be inherently better able to recruit elements of the complex, resulting in unequal KIR gene activation. Consistent with References this possibility, Shilling et al. (36) mapped quantitative differences 1. Fry, C. J., and C. L. Peterson. 2002. Transcription: unlocking the gates to gene expression. Science 295: 1847–1848. in KIR gene expression to the KIR locus. Nonetheless, the selection 2. Smale, S. T. 2003. The establishment and maintenance of lymphocyte identity of KIR genes for activation is largely a stochastic process. The through gene silencing. Nat. Immunol. 4: 607–615. probability of recruitment of the proposed sequence-specific fac- 3. Bird, A. 2002. DNA methylation patterns and epigenetic memory. Genes Dev. 16: 6–21. tors or demethylation complex to KIR proximal promoters may be 4. Su, R. C., K. E. Brown, S. Saaber, A. G. Fisher, M. Merkenschlager, and made more or less likely by antegrade and retrograde transcription S. T. Smale. 2004. Dynamic assembly of silent chromatin during thymocyte maturation. Nat. Genet. 36: 502–506. from bidirectional promoters, such as have been identified up- 5. Heard, E., C. Rougeulle, D. Arnaud, P. Avner, C. D. Allis, and D. L. Spector. stream of Ly-49 genes (37). 2001. Methylation of histone H3 at Lys-9 is an early mark on the X The biochemical identification of DNA demethylase complexes during X inactivation. Cell 107: 727–738. 6. Lock, L. F., N. Takagi, and G. R. Martin. 1987. Methylation of the Hprt gene on is controversial, but it is clear that genes can be rapidly demeth- the inactive X occurs after chromosome inactivation. Cell 48: 39–46. ylated in the absence of DNA replication (38). We suggest that loss 7. Fahrner, J. A., S. Eguchi, J. G. Herman, and S. B. Baylin. 2002. Dependence of of DNA methylation leads to loss of histone deacetylase com- histone modifications and gene expression on DNA hypermethylation in cancer. Cancer Res. 62: 7213–7218. plexes, gain of acetylated histones, and opening of the chromatin 8. Stirzaker, C., J. Z. Song, B. Davidson, and S. J. Clark. 2004. Transcriptional gene to transcription factors. The DNA-binding proteins that direct KIR silencing promotes DNA hypermethylation through a sequential change in chro- transcription may prevent remethylation in a mechanism that in- matin modifications in cancer cells. Cancer Res. 64: 3871–3877. 9. Carrington, M., and P. Norman. 2003. The KIR Gene Cluster. National Library of volves blocking access by DNA methyltransferases, as demon- Medicine (U.S.), National Center for Biotechnology Information, Bethesda. strated for other genes (39). In contrast to transcribed KIR genes, 10. Valiante, N. M., M. Uhrberg, H. G. Shilling, K. Lienert-Weidenbach, K. L. Arnett, A. D’Andrea, J. H. Phillips, L. L. Lanier, and P. Parham. 1997. Functionally and methylated KIR genes and alleles remain in closed chromatin and structurally distinct NK cell receptor repertoires in the peripheral blood of two are quiescent, despite measurable histone acetylation. Open and human donors. Immunity 7: 739–751. 5974 PRIMARY CONTROL OF KIR GENES BY DNA METHYLATION

11. Wilson, M. J., M. Torkar, A. Haude, S. Milne, T. Jones, D. Sheer, S. Beck, and 26. Schneider, R., A. J. Bannister, F. A. Myers, A. W. Thorne, C. Crane Robinson, J. Trowsdale. 2000. Plasticity in the organization and sequences of human KIR/ and T. Kouzarides. 2004. Histone H3 lysine 4 methylation patterns in higher ILT gene families. Proc. Natl. Acad. Sci. USA 97: 4778–4783. eukaryotic genes. Nat. Cell Biol. 6: 73–77. 12. Chan, H. W., Z. B. Kurago, C. A. Stewart, M. J. Wilson, M. P. Martin, 27. Futscher, B. W., M. M. Oshiro, R. J. Wozniak, N. Holtan, C. L. Hanigan, B. E. Mace, M. Carrington, J. Trowsdale, and C. T. Lutz. 2003. DNA methylation H. Duan, and F. E. Domann. 2002. Role for DNA methylation in the control of maintains allele-specific KIR gene expression in human natural killer cells. cell type specific maspin expression. Nat. Genet. 31: 175–179. J. Exp. Med. 197: 245–255. 28. Mutskov, V., and G. Felsenfeld. 2004. Silencing of transgene transcription pre- 13. Lutz, C. T., and Z. B. Kurago. 1999. Human leukocyte antigen class I expression cedes methylation of promoter DNA and histone H3 lysine 9. EMBO J. 23: on squamous cell carcinoma cells regulates natural killer cell activity. Cancer 138–149. Res. 59: 5793–5799. ϩ 29. Bachman, K. E., B. H. Park, I. Rhee, H. Rajagopalan, J. G. Herman, S. B. Baylin, 14. Shi, Y., K. D. Smith, M. G. Kurilla, and C. T. Lutz. 1997. Cytotoxic CD8 T K. W. Kinzler, and B. Vogelstein. 2003. Histone modifications and silencing cells recognize EBV antigen but poorly kill autologous EBV-infected B lympho- prior to DNA methylation of a tumor suppressor gene. Cancer Cell 3: 89–95. blasts: immunodominance is elicited by a peptide epitope that is presented at low 30. Rountree, M. R., K. E. Bachman, and S. B. Baylin. 2000. DNMT1 binds HDAC2 levels in vitro. J. Immunol. 159: 1844–1852. and a new co-repressor, DMAP1, to form a complex at replication foci. Nat. 15. Miller, J. S., and V. McCullar. 2001. Human natural killer cells with polyclonal Genet. 25: 269–277. lectin and immunoglobulinlike receptors develop from single hematopoietic stem 31. Fuks, F., W. A. Burgers, A. Brehm, L. Hughes-Davies, and T. Kouzarides. 2000. cells with preferential expression of NKG2A and KIR2DL2/L3/S2. Blood 98: DNA methyltransferase Dnmt1 associates with histone deacetylase activity. Nat. 705–713. Genet. 24: 88–91. 16. Lanigan, T. M., and A. F. Russo. 1997. Binding of upstream stimulatory factor 32. Nan, X., H.-H. Ng, C. A. Johnson, C. D. Laherty, B. M. Turner, R. N. Eisenman, and a cell-specific activator to the calcitonin/calcitonin gene-related peptide en- and A. Bird. 1998. Transcriptional repression by the methyl-CpG-binding protein hancer. J. Biol. Chem. 272: 18316–18324. MeCP2 involves a histone deacetylase complex. Nature 393: 386–389. 17. Weinmann, A. S., S. M. Bartley, T. Zhang, M. Q. Zhang, and P. J. Farnham. 33. Trompeter, H.-I., N. Go´mez-Lozano, S. Santourlidis, B. Eisermann, P. Wernet, 2001. Use of chromatin immunoprecipitation to clone novel E2F target promot- C. Vilches, and M. Uhrberg. 2005. Three structurally and functionally divergent ers. Mol. Cell. Biol. 21: 6820–6832. kinds of promoters regulate expression of clonally distributed killer cell Ig-like 18. Stewart, C. A., J. Van Bergen, and J. Trowsdale. 2003. Different and divergent

receptors (KIR), of KIR2DL4, and of KIR3DL3. J. Immunol. 174: 4135–4143. Downloaded from regulation of the KIR2DL4 and KIR3DL1 promoters. J. Immunol. 170: 34. Van Bergen, J., C. A. Stewart, P. J. van den Elsen, and J. Trowsdale. 2005. 6073–6081. Structural and functional differences between the promoters of independently 19. Santourlidis, S., H. I. Trompeter, S. Weinhold, B. Eisermann, K. L. Meyer, expressed killer cell Ig-like receptors. Eur. J. Immunol. 35: 2191–2199. P. Wernet, and M. Uhrberg. 2002. Crucial role of DNA methylation in determi- nation of clonally distributed killer cell Ig-like receptor expression patterns in NK 35. Xu, J., A. N. Vallejo, Y. Jiang, C. M. Weyand, and J. J. Goronzy. 2005. Distinct cells. J. Immunol. 169: 4253–4261. transcriptional control mechanisms of killer immunoglobulin-like receptors in 20. Lanier, L. L., B. C. Corliss, J. Wu, C. Leong, and J. H. Phillips. 1998. Immu- natural killer (NK) and in T cells. J. Biol. Chem. 280: 24277–24285. noreceptor DAP-12 bearing a tyrosine-based activation motif is involved in ac- 36. Shilling, H. G., N. Young, L. A. Guethlein, N. W. Cheng, C. M. Gardiner, D. Tyan, and P. Parham. 2002. Genetic control of human NK cell repertoire.

tivating NK cells. Nature 391: 703–707. http://www.jimmunol.org/ 21. Smith, C. L., and G. L. Hager. 1997. Transcriptional regulation of mammalian J. Immunol. 169: 239–247. genes in vivo: a tale of two templates. J. Biol. Chem. 272: 27493–27496. 37. Saleh, A., G. E. Davies, V. Pascal, P. W. Wright, D. L. Hodge, E. H. Cho, 22. Robertson, K. D., and R. F. Ambinder. 1997. Mapping promoter regions that are S. J. Lockett, M. Abshari, and S. K. Anderson. 2004. Identification of probabi- hypersensitive to methylation-mediated inhibition of transcription: application of listic transcriptional switches in the Ly49 gene cluster: a eukaryotic mechanism the methylation cassette assay to the Epstein-Barr virus major latency promoter. for selective gene activation. Immunity 21: 55–66. J. Virol. 71: 6445–6454. 38. Bruniquel, D., and R. H. Schwartz. 2003. Selective, stable demethylation of the 23. Takai, D., and P. A. Jones. 2002. Comprehensive analysis of CpG islands in interleukin-2 gene enhances transcription by an active process. Nat. Immunol. 4: human 21 and 22. Proc. Natl. Acad. Sci. USA 99: 3740–3745. 235–240. 24. Flatau, E., E. Bogenmann, and P. A. Jones. 1983. Variable 5-methylcytosine 39. Hsieh, C. L. 1999. Evidence that protein binding specifies sites of DNA demeth- levels in human tumor cell lines and fresh pediatric tumor explants. Cancer Res. ylation. Mol. Cell. Biol. 19: 46–56. 43: 4901–4905. 40. Rajagopalan, S., and E. O. Long. 2005. Understanding how combinations of HLA

25. Santos Rosa, H., R. Schneider, A. J. Bannister, J. Sherriff, B. E. Bernstein, and KIR genes influence disease. J. Exp. Med. 201: 1025–1029. by guest on September 24, 2021 N. C. Emre, S. L. Schreiber, J. Mellor, and T. Kouzarides. 2002. Active genes are 41. Snyder, M. R., C. M. Weyand, and J. J. Goronzy. 2004. The double life of NK tri-methylated at K4 of histone H3. Nature 419: 407–411. receptors: stimulation or co-stimulation? Trends Immunol. 25: 25–32.