Leukemia (1999) 13, 877–883  1999 Stockton Press All rights reserved 0887-6924/99 $12.00 http://www.stockton-press.co.uk/leu -specific region of hypermethylation identified within the HIC1 putative tumour suppressor gene in acute myeloid leukaemia JR Melki1, PC Vincent1 and SJ Clark1,2

1Kanematsu Laboratories, Royal Prince Alfred Hospital, Camperdown; and 2CSIRO, Molecular Science, Sydney Laboratory, North Ryde, NSW Australia

Abnormal DNA has been found to be a common kaemia, we have used bisulphite genomic sequencing to feature in cancer cells, although the mechanism of this alter- characterise the methylation state of the 5Ј untranslated region ation remains poorly understood. HIC1 is a putative tumour suppressor gene on chromosome 17p13.3 and is hypermethyl- and the central region of the gene that is heavily CpG rich. ated in a number of including leukaemia. In this study, In particular, we were interested to determine the methylation using bisulphite genomic sequencing, we have identified a pattern in AML patients at diagnosis to see if critical CpG sites ‘boundary’ sequence within the HIC1 CpG island that shows a were methylated that were undetected by previous restriction marked junction between methylated and unmethylated DNA analysis. In control bone marrow cells we identified an in normal haematopoietic cells. Surprisingly, this boundary of intron/exon sequence boundary within the HIC1 gene separat- differential methylation lies exactly between the intron 2 and exon 3 junction. In contrast to normal haematopoietic cells, ing methylated and unmethylated DNA. In contrast, this hypermethylation extends past this boundary at a high fre- defined methylation boundary no longer existed in the leu- quency (83%) in newly diagnosed acute myeloid leukaemias kaemic cells, instead the CpG rich intron 2 was hypermethyl- (AML). Identification of the hypermethylated boundary ated in the majority of leukaemic patients analysed. sequence not only provides the first step in understanding the mechanisms that normally protect CpG islands from de novo methylation but also may prove to be a useful cancer- specific marker. Materials and methods Keywords: HIC1; tumour suppressor gene; DNA methylation; bisulphite genomic sequencing; leukaemia; cancer Tissue samples

Bone marrow samples with blast counts higher than 70% were Introduction aspirated from 12 patients with AML. Seven control samples of normal bone marrow were aspirated from the sternal cavity DNA methylation patterns are often altered in cancer. These from patients who were undergoing cardiac surgery and who changes comprise increased levels of DNA methyltransferase had given prior informed consent. This study was approved by ,1,2 widespread genomic hypomethylation and simul- the Ethics Committee of RPAH (protocol number X93-0073). taneous regional increases in DNA methylation.3,4 Hyperme- thylated regions often harbour CpG island-containing genes such as calcitonin, p15, p16, Rb, VHL, E-cadherin, ER, and Methylation analysis HIC1, reviewed in Ref. 5. HIC1 is a candidate tumour sup- pressor gene encoding a zinc finger factor DNA was isolated using TriZOL Reagent (Gibco-BRL, Gai- belonging to the ZIN family. The HIC1 gene lies distal to p53 thersburg, MD, USA) from bone marrow cells which had been on 17p13.3 and centromeric to the area of minimal deletion, lysed in hypotonic lysis buffer. Bisulphite genomic sequencing which also harbours two putative tumour suppressor genes was used to analyse the methylation patterns. The bisulphite ° ␮ OVA1 and OVA2.6 The entire HIC1 gene is CpG rich and is reaction was carried out for 16 h at 55 C on 1–2 gofHindIII- frequently hypermethylated in various cancers, including digested patient DNA, under conditions described by Clark et 13 lung, colon,7 breast,8 brain, prostate and leukaemia.9 Hyper- al. After bisulphite conversion, the DNA was ethanol pre- ␮ methylation of HIC1 appears to be an early event in prostate, cipitated, dried, resuspended in 100 l TE (10 mM Tris-HCl − ° kidney and ovary tumours but has been proposed to be a late (pH 8), 1 mM EDTA) and stored at 20 C. Nested PCR ampli- ␮ event in haematopoietic .9 Also, in contrast to other fications were performed in 50 l reaction mixtures contain- ␮ ␮ genes studied in haematopoietic neoplasms such as calci- ing 1 l of bisulphite-treated genomic DNA, 200 M of each tonin,10 ER11 and p15,12 HIC1 methylation appears to display of the four dNTPs, 300 ng of each primer, 1.5 mM MgCl2,2 a relatively disease-specific profile in acute leukaemias. Issa units AmpliTaq DNA polymerase (Perkin-Elmer, Norwalk, CT, et al9 reported that the HIC1 is rarely methylated (10%) in USA), in a reaction buffer consisting of 67 mM Tris, 16.6 mM acute myeloid leukaemias (AML) but frequently methylated ammonium sulphate, 1.7 mg/ml bovine serum albumin and (53%) in acute lymphoblastic leukaemias (ALL). 10 mM B-mercaptoethanol in TE buffer (10 mM Tris/HCl (pH Methylation studies on HIC1 so far have been limited 8.0), 0.1 mM EDTA). The reactions were amplified in a Hybaid mainly to restriction analysis of NotI sites spanning an 11 kb DNA Thermal Cycler (Ashford, UK) under the following con- ° × ° region incorporating the HIC1 gene. In order to study the ditions, 1st round: 96 C/3 min 1 cycle; 95 C/1 min, ° ° × ° ° methylation profile of the HIC1 gene in more detail in leu- 45 C/2 min, 72 C/3 min, 5 cycles; 95 C/1 min, 50 C/2 min, 72°C/2 min, × 23 cycles; 72°C/4 min × 1 cycle, 2nd round: 96°C/3 min × 1 cycle; 95°C/1 min, 50°C/2 min, 72°C/3 min, × 5 cycles; 95°C/1 min, 55°C/2 min, 72°C/2 min, × 23 cycles; Correspondence: SJ Clark, CSIRO Molecular Science, Sydney Labora- ° × tory, PO Box 184, North Ryde, NSW 1670, Australia; Fax: 612 72 C/4 min 1 cycle. 94905005 The outer primers used for amplification of the 5Ј untrans- Received 10 August 1998; accepted 2 February 1999 lated region of HIC1 from bisulphite-treated DNA were: HIC5 Hypermethylation of HIC1 gene in AML JR Melki et al 878 (163–187): 5Ј-TATTTTTTTTAATTGGGGTAATTTT; HIC7 (663– fingers14 (Figure 1). The HIC1 contains a TATAA box 693): 5Ј-ATTAAACTACAACAACAACTACCTAAAATAA and sequence 40 bp upstream to the transcription start site. The inner primers were: HIC8+M13 (237–260): 5′-TGTAAAAC- entire HIC1 gene is G+C rich and has a large CpG island span- GACGGCCAGTAAAGTTTTTTGTTTTGAATGAT; HIC7 (as for ning the body of the gene (Figure 1). This is in contrast to most first round). The outer primers used for amplification of the CpG island genes that typically span the promoter and the first CpG-rich core HIC1 region from bisulphite-treated DNA exon. Previous methylation studies have shown that hyperme- were: HIC1 (1291–1319): 5Ј-TTTTTTGTGGTTTGGATTTG thylation of the five NotI sites spanning the gene are linked TTTAAGAAG; HIC4 (1907–1935): 5Ј-CAACTACTCAAAAC- with HIC1 gene silencing in neoplastic cells.14 Moreover, hyp- TAAAAAAACCCTTAC and inner primers were: HIC2+M13 ermethylation of these sites has been shown to occur fre- (1452–1477): 5Ј-TGTAAAACGACGGCCAGTTTTTTTAGAAG quently in a number of leukaemia subtypes but was only TTGGAGGAGGT; HIC3 (1760–1786): 5Ј-ATCTCCTCACTAC- detected in 10% of AML.9 TACTCTTATAATCA. Co-ordinates of the primers indicates To determine if abnormal methylation of the HIC1 gene is their location on the HIC1 sequence (GenBank accession more frequent in AML patients at CpG sites not detectable No. L41919). by restriction enzyme analysis, we measured, using bisulphite Automated direct sequencing of the HIC1 PCR products genomic sequencing, the methylation pattern in the 5Ј was performed using a PRISM Dye Primer Cycle sequencing untranslated region and in the central CpG rich region span- kit (−21M13 Fwd) with AmpliTaq FS (Perkin-Elmer/ABI, Foster ning intron 2 and exon 3, as shown in Figure 1. The 5Ј City, CA, USA) on an automated 373A DNA Sequencer (ABI, untranslated region contained a single NotI site previously Foster City, CA, USA). Sequencing reactions were performed studied by restriction analysis9 whereas the central CpG rich as recommended by the manufacturer. region does not incorporate any NotI sites and therefore has not been previously analysed. DNA from bone marrow aspir- ates from 12 patients with AML, and from seven normal con- Expression analysis (RT-PCR) trol samples was bisulphite treated and the target regions amplified. The HIC1 primers used were shown to amplify both methylated and unmethylated DNA without notable PCR bias, cDNA was reverse transcribed from 1–2 ␮g of total RNA in since all CpG sites tested from the 50% methylated PCR con- a25␮l reaction using AMV reverse transcriptase (Promega, trol product were methylated at approximately 30–50% Madison, WI, USA) as per the manufacturer’s instructions. The (Figure 2a and b).15 For the 5Ј untranslated region we meas- reaction was primed with 0.5–1 ␮g of random hexamers ured methylation initially by RsaI digestion of the bisulphite- (Boehringer Mannheim, Mannheim, Germany) and 100 ng of treated amplified PCR products. RsaI (GTAC) will only digest primer HIC1-5Ј−3. The PCR reactions were performed in if the CpG site in the DNA is methylated and converted (ie 25 ␮l volumes consisting of 200 ␮M dNTPs, 100 ng primers, GCACG is converted to GTACG). Figure 2a shows that 11/12 1.75 mM MgCl , 1.0 ␮Ci 33P-␣ dATP, and 1 unit AmpliTaq 2 patient samples and seven normal samples were unmethylated DNA polymerase (Perkin-Elmer) with 1 ␮l(Rb)or2␮l(HIC1) at this site as no digestion was detected, whereas digestion cDNA. The Rb PCR was performed in the manufacturer’s was detected in the spiked 50% methylated DNA sample. We buffer (Perkin-Elmer), while the PCR for the 5Ј untranslated confirmed the lack of methylation in the AML and normal region was performed in Boehringer optimisation buffer 5. The samples at 13 other CpG sites (including a NotI site) in the 5Ј primers for the Rb PCR were Rb 1: GGAGGACCTG untranslated region by directly sequencing the PCR fragments CCTCTCGTCAG and Rb 39: TGATTTCGTATGTTTTTC (Figure 2c). This was not totally unexpected as Issa et al9 TGTAGC. The primers for the HIC1 5Ј untranslated region reported only 10% methylation in AML patients surveyed by were HIC1-5Ј–3 (591–612): CACCCGGCCGGTGTGT restriction enzyme analysis. GTCCCC, and HIC1-5Ј–2 (877–898): CCAGGCGGCCGG However, since the 5Ј untranslated region of the HIC1 gene TGTAGATGAA. Reactions were performed in a Hybaid DNA that was analysed was relatively devoid of CpG sites and Thermal Cycler under the following conditions: HIC1: appeared to flank the CpG island sequence, we decided to 96°C/4 min × 1 cycle; 95°C/60 s, 68°C/90 s, 72°C/90 s × 5 expand our methylation studies to include the central CpG- cycles; 95°C/60 s, 70°C/60 s, 72°C/60 s × 25 cycles; rich region of the HIC1 gene because this region represented 72°C/3 min × 1 cycle, or for Rb: 96°C/4 min × 1 cycle; a more typical CpG island sequence. We measured the 95°C/60 s, 62°C/90 s, 72°C/90 s × 5 cycles; 95°C/60 s, methylation of 45 CpG sites spanning the central CpG-rich 64°C/60 s, 72°C/60 s × 25 cycles; 72°C/3 min × 1 cycle. Prod- region and Figure 3 shows examples of the sequence profiles ucts were electrophoresed on a 1.4% agarose gel. HIC1 was from AML patients R59 and R63 and from normal bone mar- detected by Southern analysis using an internal oligonucleo- row N35. The methylation state of each was quantit- tide probe for HIC1 (661–687) (CTGCAGCAGCAGC ated by comparing the peak height of the cytosine signal with TGCCTGGAGTGGCC) that was radiolabelled with ␥32P-ATP the peak height of the cytosine plus signal. Figure 4 and T4-polynucleotide kinase, and visualised after exposure summarises the methylation state in all the samples tested in to a phosphorimager (Molecular Dynamics, Sunnyvale, CA, the central region of the HIC1 gene. The first observation is USA). Rb was detected by incorporation of ␣33P-dATP into that there is significant methylation of the HIC1 gene in all the RT-PCR and exposure to a phosphorimager (Molecular the AML and all the normal DNA samples sequenced. How- Dynamics) after gel electrophoresis. ever, the methylation patterns do vary between DNA samples, for example in AML R74, methylation is sparse throughout the region and the level of methylation is low (ෂ5–25%) at each Results methylated CpG site, whereas AML R125 is heavily methyl- ated at essentially all the CpG sites assayed. The second The HIC1 gene spans 3.6 kb and contains three exons; the observation is that the entire CpG-rich region within the HIC1 first exon is untranslated, the second exon encodes the zinc- gene is methylated in 83% (10 out 12) of AML samples, finger N-terminus (ZIN) and the third exon encodes five zinc whereas in all seven normal control samples methylation is Hypermethylation of HIC1 gene in AML JR Melki et al 879

Figure 1 Map of the HIC1 gene (accession No. L41919). The exons and NotI sites (N) are marked. A CpG island plot23 is shown above the HIC1 line map. The two regions analysed for methylation are labelled 5Ј UT region, and central region on the HIC1 CpG island map. The HIC1 DNA sequence shown, from base 1478 to 1759, is the central CpG region. CpG dinucleotides for this region are shown in bold and are numbered above each site. Putative Sp1 binding sites are underlined in bold. A putative DPE element is underlined by dots. /, between CpG sites 16 and 17 indicates the intron 1 and exon 2 border. The G residue in CpG site 12 is an A residue in the published sequence. restricted to the 3Ј end of the HIC1 gene. In fact, only two Extensive HIC1 methylation, spanning intron 2, is not lim- AML samples, R99 and R128, had methylation profiles similar ited to acute myeloid leukaemias. We also have found intron to the profiles from normal DNA. The third observation, which 2 hypermethylation in two out of two myeoplastic dysplasia is the most provocative, is that there appears to be a distinct syndrome (MDS) bone marrow aspirates, four out of seven sequence boundary between methylated and unmethylated chronic lymphocytic leukaemia (CLL) bone marrow aspirates, DNA in all of the normal control samples. The methylation one out of one chronic myelogenous leukaemia (CML) aspir- boundary lies between CpG sites 16 and 17. Interestingly, ates and two out of three acute lymphocytic leukaemia (ALL) these sites also span the sequence boundary between intron aspirates (data not shown). The fact that extensive methylation 2 and exon 3 of the HIC1 gene (Figures 1 and 4). spanning the intron/exon boundary was observed in a range To determine if hypermethylation of intron 2 affected HIC1 of leukaemia subtypes at presentation and not in normal bone expression we determined the transcription levels using RT- marrow suggests that hypermethylation of intron 2 in the HIC1 PCR on samples which had matched RNA and were shown gene is not disease specific in leukaemia and may indeed be by PCR not to contain DNA contamination (data not shown). an early event as suggested for other cancer types. Rb expression was used to control for RNA integrity (Figure 5). Figure 5 shows that a HIC1 RT-PCR product was amplified in the normal and every AML sample analysed, indicating that Discussion transcription was being initiated from the unmethylated pro- moter region. Therefore, methylation of the central CpG-rich HIC1 is a putative tumour suppressor gene on 17p13.3 that region of the HIC1 gene does not directly inhibit expression has been reported to be hypermethylated in a number of can- of the gene. cers including leukaemia. In this study we have defined a spe- We found no correlation of the HIC1 hypermethylation pro- cific region in the HIC1 gene that shows differential methyl- files with either chromosome abnormalities, age, sex or DNA ation between intron 2 and exon 3 in normal haematopoietic methyltransferase mRNA levels (data not shown). Bone mar- cells but becomes hypermethylated at a high frequency (83%) row aspirates from all of the AML samples assayed had elev- in acute myeloid leukaemic cells. ated levels of DNA methyltransferase mRNA, on average 4- Previous studies using restriction have reported to 5-fold, compared to their normal counterparts.1 In fact, of that the HIC1 CpG-rich sequence is unmethylated in normal the samples tested, AML R99 showed the highest DNA haematopoietic cells but is frequently methylated in advanced methyltransferase level but was one of the few patient samples stages of ALL and CML but rarely methylated in AMLs.9 There- that displayed a methylation profile similar to normal control fore, it was proposed that HIC1 hypermethylation may be dis- DNA. Unlike calcitonin and oestrogen receptor that are ease and stage specific unlike HIC1 methylation in other reported to be hypermethylated with increasing age, the HIC1 tumours characterised. In our studies, we found like Issa et al9 gene in this core CpG region appears to remain unmethylated that the 5Ј end of the HIC1 gene is rarely methylated in AMLs, in all the normal control DNA from N46 age 19 to N35 age however, in contrast we have found that hypermethylation of 74. the CpG-rich sequence in intron 2 appears to be a common Hypermethylation of HIC1 gene in AML JR Melki et al 880 a

b

c

Figure 2 Methylation analysis of the 5Ј untranslated region. (a) Ethidium bromide-stained agarose gel showing RsaI digestion of the PCR products of bisulphite converted DNA from AML and normal bone marrow samples. The lanes are marked with the sample number. The last lane is the 50% methylated control (unmethylated DNA spiked with 50% SssI methylated DNA). Arrows mark the uncut and cut PCR products. (b) A portion of the direct PCR sequencing profile of the HIC1 5Ј untranslated region of the 50% spiked control, where all CpG sites are displaying methylated (blue peak) at approximately 30–50%. (c) A portion of the direct PCR sample sequencing profile of the HIC1 5Ј untranslated region of the AML patient R37, where no methylation is evident. The HIC1 5Ј untranslated region shown (co-ordinates 269– 325) incorporates three CpG sites (shaded).

event in all leukaemia subtypes tested, including AMLs. More- methylated in relapsed leukaemia patients. Moreover, the over, we found that intron 2 methylation is cancer specific spreading of methylation from the 3Ј end of HIC1 gene into and does not occur in normal haematopoetic cells whereas intron 2 in the cancer cells does not appear to be simply the downstream sequence in exon 3 is methylated in both related to elevated DNA methyltransferase levels in the indi- normal and leukaemic cells. The methylation patterns we vidual tumours or to the age of patients in normal cells. have identified in intron 2 and exon 3 would not have been One of the main questions in understanding the mechanism identified previously as the CpG-rich region we sequenced of DNA methylation is what normally protects CpG islands did not contain any NotI sites. from methylation and how is this process altered in a cancer Methylation in the body of the gene is not uncommon and cell? The methylation boundary identified in the HIC1 gene, typically does not affect gene transcription.16,17 This is also separating essentially methylated DNA from unmethylated true for the HIC1 gene where intron 2 methylation does not DNA, may provide some useful insights. A number of possible directly silence expression in the AML cells. However, the fact mechanisms can be considered. First, the DNA sequence or that HIC1 hypermethylation is so highly cancer specific and local DNA conformation may provide a directional de novo is localised to the more typical CpG island sequence within methylation or protection signal. In this regard, it is striking the HIC1 gene is intriguing. Equally intriguing is the discovery that in normal haematopoietic cells the unmethylated CpG of the marked boundary of methylation between intron 2 and sites are located in intron 2 and the methylated CpG sites are exon 3 in normal cells. It is possible that the abnormal methyl- in exon 3 with the boundary of differential methylation span- ation of intron 2 is an early marker of disease, signalling ‘leak- ning the intron/exon junction of the HIC1 gene. Both the age’ of methylation across the methylation boundary, and that intron and exon sequences are CpG rich, however, at the the spread of methylation to the 5Ј region of the HIC1 gene intron/exon junction there is a three base hairpin stem. How- may be associated with later disease stages.9 We found little ever, small hairpin loop structures are not an uncommon evidence for spreading of methylation into the 5Ј untranslated structure in DNA sequences, so it seems unlikely that this region, but all our samples were from early presentation, how- structure alone would be sufficient to elicit such a gross ever, Issa et al9 showed that the 5Ј region is more commonly methylation change. A second possible mechanism is that a Hypermethylation of HIC1 gene in AML JR Melki et al a 881

b

c

Figure 3 Direct PCR sample sequencing profiles of the central CpG region of the HIC1 gene from CpG dinucleotides 14–21. (a) AML patient R59, showing the cytosines that remained after bisulphite conversion. CpG site 14 is representative of 75–100% methylated cytosine, CpG site 15 is representative of 25–50% methylation and CpG site 16 is representative of 50–75% methylation. (b) AML patient R63, CpG site 15 is representative of an unmethylated site. (c) Normal bone marrow control N35, CpG sites 17 and 18 are representative of 5–25% methylation. Note unmethylated cytosine residues were bisulphite converted to thymine residues. functional boundary exists between intron 2 and exon 3, methylation in the tumour cells22 or due to the loss of binding where active transcription itself may protect CpG methylation. proteins that may interact with Sp1 sites at island borders.5 Therefore, an intriguing possibility is that an antisense RNA is The results presented here indicate that HIC1 hypermethyl- transcribed from the CpG-rich intron 2, similar to antisense ation of intron 2 may provide an important marker for haema- RNA expression from an intronic CpG island in Igf2r.18 Poss- tological cells that are susceptible to becoming carcinogenic. ible antisense transcription from intron 2 of the HIC1 gene is The reintroduction of a functional HIC1 gene into cell lines supported by two observations. Firstly, we have found two has been shown to retard cell growth and promote differen- putative Sp1 binding sites and a down- tiation of tumour cells.14 Consequently, early detection of stream promoter element (DPE)19 in this region, indicative of HIC1 methylation may be an important step in the treatment a promoter region and secondly, Makos-Wales et al14 using and management of patients with leukaemia. Northern analysis of HIC1 reported a smaller 1.1 kb transcript in normal tissues. In addition, the smaller transcript was not detected using probes from exon 3 or the 3Ј untranslated region. We are now in the process of Acknowledgements determining if there is an antisense transcript and whether its expression is silenced by hypermethylation of intron 2 in can- We thank Cheryl Paul for the excellent automated sequen- cer cells. A third possible mechanism is that Sp1 binding sites cing, Robyn Lukeis for chromosome analyses and Doug Millar located in intron 2 are important for protecting intron 2 from for advice and helpful discussions. This work has been methylation as described for the APRT gene.20,21 In this regard, supported in part by Rothe Memorial Trust and the NSW Sp1 may be prevented from binding due to abnormal CpCpG Cancer Council. Hypermethylation of HIC1 gene in AML JR Melki et al 882

Figure 4 Summary of the HIC1 methylcytosine quantitation for the central CpG region, compiled from direct PCR sequence analysis. The AML patients (R) and normal controls (N) are identified by laboratory number in the left most column. Age and sex is detailed in the next column: M, male; F, female; NA, not available. CpG sites are numbered along the top row from site 1 to site 45. The shaded boxes indicate the degree of methylation as determined by semi-quantitation of all CpG sites described in Figure 3. Absence of a symbol indicates that the methylation status at that site was not determined due to sequencing enzyme stoppage. The bold vertical line marks the border between intron 2 and exon 3. The bold horizontal line separates the patient and normal samples.

genes on chromosome 17p13.3. Cancer Res 1996; 56: 1997– 2002. 7 Makos M, Nelkin BD, Lerman MI, Latif F, Zhar B, Baylin SB. Dis- tinct hypermethylation patterns occur at altered chromosome loci in human lung and colon cancer. Proc Natl Acad Sci USA 1992; 89: 1929–1933. 8 Fujii H, Biel MA, Zhou W, Weitzman SA, Baylin SB, Gabrielson E. Methylation of the HIC-1 candidate tumor suppressor gene in human breast cancer. Oncogene 1998; 16: 2159–2164. Figure 5 Expression analysis of the HIC1 gene. RT-PCR products 9 Issa JP, Zehnbauer BA, Kaufmann SH, Biel MA, Baylin SB. HICI for the HIC1 gene and the Rb gene are shown by arrows. The sample hypermethylation is a late event in hematopoietic neoplasms. number is labelled above each lane (R, AML patient; N, normal sam- Cancer Res 1997; 57: 1678–1681. ple. +, HIC1 PCR product (308 bp). −, RT-PCR minus cDNA. 10 Baylin SB, Fearon ER, Vogelstein B, de Bustros A, Sharkis SJ, Burke PJ, Staal SP, Nelkin BD. Hypermethylation of the 5Ј region of the calcitonin gene is a property of human lymphoid and acute myeloid malignancies. Blood 1987; 70: 412–417. References 11 Issa JP, Zehnbauer BA, Civin CI, Collector MI, Sharkis SJ, Davidson NE, Kaufmann SH, Baylin SB. The estrogen receptor CpG island 1 Melki J, Warnecke P, Vincent P, Clark S. Increased DNA methyl- is methylated in most hematopoietic neoplasms. Cancer Res 1996; transferase expression in leukemia. Leukemia 1998; 12: 311–316. 56: 973–977. 2 el-Deiry WS, Nelkin BD, Celano P, Yen RW, Falco JP, Hamilton 12 Herman JG, Jen J, Merlo A, Baylin SB. Hypermethylation-associa- SR, Baylin SB. High expression of the DNA methyltransferase gene ted inactivation indicates a tumor suppressor role for p15INK4B. characterizes human neoplastic cells and progression stages of Cancer Res 1996; 56: 722–727. colon cancer. Proc Natl Acad Sci USA 1991; 88: 3470–3474. 13 Clark SJ, Harrison J, Paul CL, Frommer M. High sensitivity map- 3 Jones PA, Rideout WM, Shen J-C, Spruck CH, Tsai C. Methylation, ping of methylated cytosines. Nucleic Acids Res 1994; 22: and cancer. Bioessays 1992; 14: 33–36. 2990–2997. 4 Schmutte C, Yang AS, Nguyen TT, Beart RW, Jones PA. Mech- 14 Makos-Wales M, Biel MA, el Deiry W, Nelkin BD, Issa JP, Cav- anisms for the involvement of DNA methylation in colon carcino- enee WK, Kuerbitz SJ, Baylin SB. p53 activates expression of HIC- genesis. Cancer Res 1996; 56: 2375–2381. 1, a new candidate tumour suppressor gene on 17p13.3. Nature 5 Baylin SB, Herman JG, Graff JR, Vertino PM, Issa JP. Alterations Med 1995; 1: 570–577. in DNA methylation: a fundamental aspect of neoplasia. Adv 15 Warnecke PM, Stirzaker C, Melki JR, Millar DS, Paul CL, Clark Cancer Res 1998; 72: 141–196. SL. Detection and measurement of PCR bias in quantitative 6 Schultz DC, Vanderveer L, Berman DB, Hamilton TC, Wong AJ, methylation analysis of bisulphite-treated DNA. Nucleic Acids Res Godwin AK. Identification of two candidate tumor suppressor 1997; 25: 4422–4426. Hypermethylation of HIC1 gene in AML JR Melki et al 883 16 Gonzalgo ML, Hayashida T, Bender CM, Pao MM, Tsai YC, Gon- 20 Brandeis M, Frank D, Keshet I, Siegfried Z, Mendelsohn M, Nemes zales FA, Nguyen HD, Nguyen TT, Jones PA. The role of DNA A, Temper V, Razin A, Cedar H. Sp1 elements protect a CpG methylation in expression of the p19/p16 locus in human bladder island from de novo methylation. Nature 1994; 371: 435–438. cancer cell lines. Cancer Res 1998; 58: 1245–1252. 21 Macleod D, Charlton J, Mullins J, Bird AP. Sp1 sites in the mouse 17 Millar DS, Ow KK, Paul CL, Russell PJ, Molloy PL, Clark SJ. aprt gene promoter are required to prevent methylation of the CpG Detailed methylation analysis of the glurathione S-transferase ␲ island. Gene Dev 1994; 8: 2282–2292. (GSTPI) gene in prostate cancer. Oncogene 1999; 18: 1313–1324. 22 Clark SJ, Harrison J, Molloy PL. Sp1 binding is inhibited by 18 Wutz A, Smrzka OW, Schweifer N, Schellander K, Wagner EF, (m)Cp(m)CpG methylation. Gene 1997; 195: 67–71. Barlow DP. Imprinted expression of the Igf2r gene depends on an 23 Gardiner-Garden M, Frommer M. CpG islands in gen- intronic CpG island. Nature 1997; 389: 745–749. omes. J Mol Biol 1987; 196: 261–282. 19 Burke TW, Kadonaga JT. The downstream core promoter element. DPE, is conserved from Drosophila to humans and is recognized by TAFII60 of Drosophila. Genes Dev 1997; 11: 3020–3031.