LOT1 (PLAGL1/ZAC1), the candidate tumor suppressor gene at chromosome 6q24-25, is epigenetically regulated in cancer.

LOT1 is a zinc-finger nuclear transcription factor, which possesses anti-proliferative effects and is frequently silenced in ovarian and breast cancer cells. The LOT1 gene is localized at chromosome 6q24-25, a chromosomal region maternally imprinted and linked to growth retardation in several organs and progression of disease states such as transient neonatal diabetes mellitus. Toward understanding the molecular mechanism underlying the loss of LOT1 expression in cancer, we have characterized the genomic structure and analyzed its epigenetic regulation. Genome mapping of LOT1 in comparison with the other splice variants, namely ZAC1 and PLAGL1, revealed that its mRNA ( approximately 4.7 kb; GenBank accession number U76261) is potentially spliced using six exons spanning at least 70 kb of the human genome. 5'-RACE (rapid amplification of cDNA ends) data indicate the presence of at least two transcription start sites. We found that in vitro methylation of the LOT1 promoter causes a significant loss in its ability to drive luciferase transcription. To determine the nature of in vivo methylation of LOT1, we used bisulfite-sequencing strategies on genomic DNA. We show that in the ovarian and breast cancer cell lines and/or tumors the 5'-CpG island of LOT1 is a differentially methylated region. In these cell lines the ratio of methylated to unmethylated CpG dinucleotides in this region ranged from 31 to 99% and the ovarian tumors have relatively higher cytosine methylation than normal tissues. Furthermore, we show that trichostatin A, a specific inhibitor of histone deacetylase, relieves transcriptional silencing of LOT1 mRNA in malignantly transformed cells. It appears that, unlike DNA methylation, histone deacetylation does not target the promoter, and rather it is indirect and may be elicited by a mechanism upstream of the LOT1 regulatory pathway. Taken together, the data suggest that expression of LOT1 is under the control of two epigenetic modifications and that, in the absence of loss of heterozygosity, the biallelic (two-hit) or maximal silencing of LOT1 requires both processes.

LOT1 is a zinc-finger nuclear transcription factor, which possesses anti-proliferative effects and is frequently silenced in ovarian and breast cancer cells. The LOT1 gene is localized at chromosome 6q24 -25, a chromosomal region maternally imprinted and linked to growth retardation in several organs and progression of disease states such as transient neonatal diabetes mellitus. Toward understanding the molecular mechanism underlying the loss of LOT1 expression in cancer, we have characterized the genomic structure and analyzed its epigenetic regulation. Genome mapping of LOT1 in comparison with the other splice variants, namely ZAC1 and PLAGL1, revealed that its mRNA (ϳ4.7 kb; Gen-Bank TM accession number U72621) is potentially spliced using six exons spanning at least 70 kb of the human genome. 5-RACE (rapid amplification of cDNA ends) data indicate the presence of at least two transcription start sites. We found that in vitro methylation of the LOT1 promoter causes a significant loss in its ability to drive luciferase transcription. To determine the nature of in vivo methylation of LOT1, we used bisulfite-sequencing strategies on genomic DNA. We show that in the ovarian and breast cancer cell lines and/or tumors the 5-CpG island of LOT1 is a differentially methylated region. In these cell lines the ratio of methylated to unmethylated CpG dinucleotides in this region ranged from 31 to 99% and the ovarian tumors have relatively higher cytosine methylation than normal tissues. Furthermore, we show that trichostatin A, a specific inhibitor of histone deacetylase, relieves transcriptional silencing of LOT1 mRNA in malignantly transformed cells. It appears that, unlike DNA methylation, histone deacetylation does not target the promoter, and rather it is indirect and may be elicited by a mechanism upstream of the LOT1 regulatory pathway. Taken together, the data suggest that expression of LOT1 is under the control of two epigenetic modifications and that, in the absence of loss of heterozygosity, the biallelic (two-hit) or maximal silencing of LOT1 requires both processes.
LOT1 (lost-on-transformation 1) 1 is a growth suppressor gene (1, 2) localized on chromosome 6 at band q24 -25, which is a frequent site for loss of heterozygosity in many solid tumors including ovarian cancer (2,3). The mouse ortholog of this gene was independently identified by Spengler et al. (4) and designated as Zac1, which is highly homologous to the hLOT1 and rLot1 genes. A splice variant of LOT1/ZAC1 was identified by Kas et al. (5) and named as PLAGL1 based on its homology with PLAG1 protein encoded by the gene PLAG1 localized on chromosome 8q12 (6). The presence of other genes with high homology to LOT1/ZAC1/PLAGL1 indicates that this gene is a member of new family of zinc-finger proteins. Our studies and those of others have shown that LOT1 gene expression is frequently down-regulated in the ovarian and breast carcinoma cells (1,2,7). Functional analysis of LOT1 demonstrated that it may play a significant role as a transcription factor modulating growth suppression through mitogenic signaling pathways (8).
It is increasingly evident that epigenetic modification of genomic DNA by methylation and/or histone deacetylation plays an important role in transcriptional silencing and loss of gene function of certain tumor suppressor genes. Methylation of DNA normally occurs at cytosine residues within CpG dinucleotides in almost all higher eukaryotic organisms (9 -11). This type of modification can suppress gene expression directly by interfering with the binding of transcription factors or indirectly by initiating a complex consisting of methyl-CpG-binding proteins (e.g. MeCP2, MBDs) and histone deacetylases (12)(13)(14)(15)(16). These complexes mediate transcriptional repression through chromatin hypoacetylation, which can be partially reversed by TSA. In human cancers, aberrant methylation of CpG islands silencing the promoters of some genes that function in the suppression of malignant phenotype has been found. For example hypermethylation of CpG islands in different cancer cells has been implicated in the transcriptional inactivation of the Rb, p16, estrogen receptor, INK4B/MTS2, VHL, and H19 genes (17)(18)(19)(20)(21)(22)(23)(24). Similarly, CpG island methylation may play a significant role in the regulation of imprinted genes (25) and genes located on the inactive X chromosome (26,27).
Recently, the LOT1/ZAC1 locus at chromosome 6q24 -25 was identified as a maternally imprinted region (2, 4, 28 -30). These studies have established that the paternal duplication or loss of maternal imprinting of the LOT1 locus is associated with transient neonatal diabetes mellitus, an imprinted disease charac-terized by intrauterine fetal growth retardation and insulin dependence. However, little is known about the molecular mechanisms that down-regulate expression of LOT1 in cancer. In this study, we have characterized the LOT1 gene structure with relation to different splice variants and identified the gene's minimal promoter and a CpG island. Our results indicate that LOT1 is subject to two epigenetic processes, methylation of CpG islands and histone deacetylation, which may synergistically act to regulate the transcriptional silencing of the gene.
Cloning of the Full-length hLOT1 and 5Ј-RACE-To obtain additional sequence information of the 5Ј-untranslated region (5Ј-UTR) of hLOT1 cDNA, which we previously reported (2), a normal human fetal brain cDNA library in ZAPII vector (Stratagene, La Jolla, CA) was used. A positive clone, FB2, which hybridized to the hLOT1 probe, gave extended 5Ј-UTR sequences. Insert sequences of all plasmids were determined by automated ABI PRISM dye terminator cycle sequencing with FS AmpliTaq DNA polymerase and run through an Applied Biosystem 373/377 sequencer (ABI/PerkinElmer Life Sciences). The data were analyzed using FASTA and PILEUP DNA sequence homology searches in the Wisconsin Genetics Computer Group package (Madison, WI).
We predicted the LOT1 transcription start site using GeneRacer cDNA Kit (Invitrogen). The cDNA for 5Ј-RACE was derived from human ovary and generated based on oligo-capping and RNA ligasemediated RACE methods (33,34). The PCR was performed using GeneRacer 5Ј1 adaptor forward primer (5Ј-GCACGAGGACACUGA CAUGGACUGA-3Ј) and gene-specific reverse primer (5Ј-TTCTAAGT-GAGGTACAGATGAGTTTCAGATG-3Ј). The gene-specific primer was designed from the 3Ј end of the LOT1 exon II and at the junction of this exon and exon III. The PCR reaction contained the cDNA, primers, buffer, dNTPs, Me 2 SO, and Taq polymerase in a volume of 50 l. The amplification reaction consisted of denaturation at 94°C for 3 min, 35 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 30 s, followed by extending at 72°C for 7 min. We analyzed the PCR products by cloning into pGEMT vector using the TA cloning kit (Promega, Madison, WI). The clones were sequenced using both T7 and Sp6 primers. All the primers used in this study were synthesized by the Fox Chase Cancer Center DNA synthesis facility.
Genomic Structure and Cloning of hLOT1 Promoter Region-To obtain the human LOT1 promoter, a human placental genomic DNA library (EMBL-3-SP6/T7, Clontech, Palo Alto, CA) was screened using the following fragments in the hLOT1 cDNA (GenBank accession number U72621.3) as probes: a) a fragment containing the complete open reading frame; b) a 271-bp fragment upstream of the LOT1 translation start codon; and c) a fragment containing bases Ϫ1547 to Ϫ567 with respect to the LOT1 translation start codon. Phage inserts were sequenced using both the polylinker-specific and LOT1-specific primers. These primers as well as cDNA-specific primers were used in PCR to amplify regions of the phage insert and to gain additional insight into the genomic structure of LOT1. We also screened a human P1 library (Genome Systems, St. Louis, MO), which resulted in identification of a new genomic clone designated as clone 17513. This P1 clone was purified by Qiagen Plasmid Purification Kit (Qiagen Inc., Valencia, CA) and used in subsequent promoter analyses. The forward and reverse primer, p63 (5Ј-TGGGTGTGAGGAGTGTGGGAAGA-3Ј) and p69 (5Ј-G-CAGTTTGATTACAGAACACGCG-3Ј) were used to sequence most of the LOT1 open reading frame using genomic DNA as template.
Construction of Luciferase Expression Vectors-Reporter constructs containing luciferase gene under the control of the LOT1 promoter were created by amplifying and subcloning the 5Ј-flanking regions of the LOT1 promoter at the nucleotide positions as depicted in Fig. 4, using DNA from the phage or P1 clone 17513 as a template. Primers used to amplify the 5Ј-flanking region of the LOT1 promoter included: TH1645, 5Ј-CTAGTGGGGTAGGGATAGCATT-3Ј; TH1647, 5Ј-CAGGAGGTAA-GTTAGTTTGGCC-3Ј; and TH1648, 5Ј-CTGCCCCGTCCGTCCGTCCG-T-3Ј. The primers TH1645 and TH1647 were linked to the XhoI recognition sequence on their 5Ј ends, and the primer TH1648 contained a HindIII recognition site at its 5Ј end. The plasmids constructed with these PCR products were designated as pGL3-45 and pGL3-47, respectively. Another plasmid designated as PGL3-60 was constructed using a PCR fragment obtained by amplifying human normal ovarian genomic DNA as template with the primers IMP4, 5Ј-GGCCCGTTGG-CGAGGTTAGAGCGC-3Ј and IMP5, 5Ј-ACGGCATCTGCCATTTGTCA-3Ј. The PCR products were subcloned into pGEMT vector using a TA cloning kit (Promega, Madison, WI), and the insert fragments were purified. The DNA fragments obtained by above procedures were ligated into pGL3-Basic reporter vector (Promega) containing the luciferase gene, which was previously digested with either XhoI alone (pGL3-60) or XhoI and HindIII (pGL3-45, pGL3-47) restriction enzymes. The ligation reactions were used to transform DH5␣ Escherichia coli competent cells and the clones were confirmed by sequencing their insert. All plasmids used for the transfections were purified using the Qiagen kit.
In Vitro DNA Methylation-The promoter construct pGL3-45 was in vitro methylated by the bacterial methylase, SssI (New England Biolabs), or mock-treated in the absence of enzyme. The DNA was incubated for 24 h at 37°C with SssI methylase at 1 unit/g DNA in 50 mM NaCl, 10 mM Tris-HCl, 10 mM MgCl 2 , and 1 mM dithiothreitol (pH 7.9) supplemented with 160 M S-adenosylmethionine.
Construction of GFP Fusion Proteins, Fluorescence Microscopy, and FACS Analysis-GFP-tagged LOT1 mammalian expression vector was constructed as described previously (8). Briefly, the GFP coding region was amplified using forward (5Ј-CGCCATGGCTAGCAAGGGCCAG-3Ј) and reverse (5Ј-CTTGTACAGCTCGTCCATGCC-3Ј) primers. The PCR product was ligated with XbaI linker to the LOT1 coding region, and the resulting fragment (GFL) was introduced downstream of the cytomegalovirus promoter into the XhoI site of pcDNA3 plasmid. A2780 cells were cultured in RPMI 1 day before transfection. Transient transfections were performed using TransIT-LT1 transfection reagent (Pan-Vera, Madison, WI) according to the manufacturer's instructions. Cells (2 ϫ 10 5 /well) were seeded in 6-well plates and transfected with 4 g/well of pcDNA/GFL vector carrying GFP-LOT1 fusion protein. The cells were then monitored 18 h later and imaged with a Quantix 12-bit cooled charge-coupled device camera (Roper, Inc., Tuscon, AZ). Fluorescent GFP and phase contrast images were generated simultaneously using Isee software (Inovision Corp., Durham, NC) to drive the chargecoupled device, a Ludl filter wheel, and a shutter attached to a Nikon TE300 inverted microscope. The cells were observed at near-physiological conditions using a forced-air incubator, which encompasses the microscope. For FACS analysis the transfected cells were collected by trypsinization, washed, and resuspended in PBS. The cell suspensions or GFP-transfectants were sorted using a Becton Dickinson FACS Vin-tageSE flow cytometer (Becton Dickinson, Inc., San Jose, CA). 10,000 cells were subjected to FACS analysis.
Transient Transfection and Luciferase Assay-Ovarian cancer cells at a density of 4 ϫ 10 5 /well were cultured in 6-well plates in RPMI 1640 medium 1 day prior to transfection with the reporter plasmid as indicated. Transient transfections were performed using TransIT-LT1 polyamine transfection reagent (PanVera, Madison, WI) according to the manufacturer's instructions. Briefly, the cells were washed once with serum-free medium and incubated with the DNA (2 l)/(6 l) reagent mixture for 5 h in serum-free medium followed by additional incubation for 17 h (overnight) in fresh medium containing 10% fetal bovine serum. Luciferase enzyme was assayed using the Dual-Luciferase Reporter Assay System (Promega). The cells were washed with 1ϫ PBS, harvested with the lysis buffer, and centrifuged, and 5-20 l of the cell lysate (supernatant) was added to 100 l of the luciferase substrate. The luciferase enzyme activity was measured immediately or within 15 s of adding the substrate, luciferin, using the Monolight (R) 2010 Luminometer (Analytical Luminescence Laboratory, San Diego, CA).
Northern and Western Blot Analysis-Total RNA was isolated from the cells by the guanidinium isothiocyanate extraction method (35) or Trizol (Invitrogen) and was separated on 1% agarose gels containing 2.2 M formaldehyde. The RNA was transferred to Nylon membranes (Micron Separations, Inc.) by capillary action and hybridized using a procedure described before (1,2). The probes for visualization of rat Lot1 and human LOT1 transcripts were the same as described previously (1, 2). The probe for ␤-actin was from Clontech, Inc. (Palo Alto, CA). Western blot analysis was performed with lysates obtained from the cells plated on 6-well dishes in RPMI 1640 containing 10% FBS. After incubation with the transfection reagent supplemented with the appropriate DNAs, whole-cell lysates were prepared using 250 l/well of M-PER Protein Extraction Reagent (PIERCE, Rockford, IL) containing phenylmethylsulfonyl fluoride and dithiothreitol. Equal volumes of protein (15 l) from each sample were electrophoresed on SDS-PAGE (10%) followed by transfer to a Hybond nylon membrane (Amersham Life Science, UK) for Western blot analysis. Immunoblots were blocked overnight in 5% nonfat dry milk (w/v) in TBST containing 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, and 0.1% Tween 20. Blots were incubated for 1 h with anti-GFP monoclonal antibody (Clontech, Palo Alto, CA) diluted with 1% nonfat dry milk in TBST. The blots were washed with TBST three times, and incubated for 1 h with the second antibody (anti-mouse IgG horseradish peroxidase, Amersham Biosciences). Anti-glyceraldehyde-3-phosphate dehydrogenase antibody was obtained from NeoMarkers (Fremont, CA). The protein bands were detected by the Western Lightning Chemiluminescence Reagent (PerkinElmer Life Sciences) and exposure to x-ray films (Kodak Co, Rochester, NY).
DNA Isolation, Bisulfite Sequencing, and Methylation-sensitive Restriction Digest-PCR-Genomic DNA was isolated by digestion with proteinase K (Invitrogen) followed by phenol/chloroform extraction (36). After precipitation by the addition of 2 volumes of ethanol and 0.5 volumes of ammonium acetate (7.5 M), the DNA was washed twice with ethanol (80%, v/v), and dissolved in water. Normal and tumor DNA samples were obtained from Biosample Repository Core Facility (Andrew Godwin and Joellen Dangel) and Tumor Bank Facility (Andre Klein-Szanto and Robert Page) at Fox Chase Cancer Center. Genomic DNA (1 g) was treated with sodium bisulfite and used for methylationsensitive PCR as previously described (37,38). Briefly, the DNA in a volume of 50 l was denatured by 0.2 M NaOH and then treated with hydroquinone (Sigma) and sodium bisulfite (Sigma) at 50°C for 16 h. Bisulfite-modified DNA was purified using Wizard DNA Clean-Up system (Promega). Modification was completed by NaOH (0.3 M) for 5 min at room temperature.
The bisulfite-treated DNA was precipitated by glycogen, ammonium acetate, and 3 volumes of ethanol, and the pellets were washed once with 70% ethanol and dissolved in 20 l of water. The modified DNA (4 l) was amplified by PCR under the following reaction conditions: 1ϫ buffer, 125 M dNTPs, 7.5 mM MgCl 2 , and 1 M of each primer in a 50-l reaction. The reaction was carried out with the cycling condition of denaturation at 95°C for 10 min followed by the addition of 0.5 unit of Gold Taq polymerase (Invitrogen) and 32 cycles of PCR (95°C, 30 s; 58°C, 30 s; 72°C, 30 s) with a final extension of 4 min at 72°C. Second round of PCR was performed using 2 l of the first PCR and the same reagents and conditions as above. Primers were designed from the interpolated sequence after bisulfite conversion assuming DNA was either methylated or unmethylated at 14 CpG sites or the imprinting control region (28,29). Primers used for first PCR were UM-S1: 5Ј-G-GGGTAGTTGTGTTTATAGTTTAGTA-3Ј, UM-AS1: 5Ј-CAAACACCC-AAACACCTACCCTA-3Ј and M-S1: 5Ј-GGGGTAGTCGTGTTTATAGT-TTAGTA-3Ј, M-AS1: 5Ј-CGAACACCCAAACACCTACCCTA-3. Primers used for the second PCR were UM-S2: 5Ј-ATAGTTTAGTAGTGTGGG-GT-3Ј, UM-AS2: 5Ј-CCTACCCTACAAAACAACAA-3Ј, M-S2: 5Ј-ATAG-TTTAGTAGCGCGGGGT-3Ј, and M-AS2: 5Ј-CCTACCCTACGAAACG-ACGA-3. Reaction products were separated by electrophoresis on a 2% gel, stained with ethidium bromide, and photographed. The resulting amplification pools were cloned into the pGEMT vector using the TA cloning kit (Promega, Madison, WI). Four to 10 individual clones per PCR reaction were isolated and sequenced. The clones were sequenced using both T7 and Sp6 primers.
The methylation-sensitive restriction digest-PCR was performed on 200 ng of genomic DNA samples. The DNA was digested overnight at 37°C with 20 units of HpaII or MspI in a total volume of 20 l. These restriction enzymes both cleave at CCGG sites. However, HpaII will not cut the DNA when the internal C is methylated. The DNA was purified with Wizard column (Promega) and eluted in 50 l water, and 5 l of the product was subjected to PCR using the primers IMP4 and IMP5 (1 M) and reagents as mentioned above. The PCR condition consisted of denaturation at 95°C for 2 min followed by the addition of 0.5 unit of Taq polymerase and 29 cycles of PCR (96°C, 30 s; 60°C, 30 s; 72°C, 1 min) with a final extension of 4 min at 72°C. The reaction contained the same reagents as mentioned above.
Chromatin Immunoprecipitation (ChIP)-ChIP assay was performed according to the manufacturer's instructions (Upstate Biotech, Lake Placid, NY). Briefly, 1 ϫ 10 6 cells in 100-mm dish were directly treated with 1% formaldehyde (Fisher Scientific) to form protein-DNA crosslinks. The cells were then collected in ice-cold 1ϫ PBS containing protease inhibitors and subjected to centrifugation at 4°C. The cell pellets were resuspended in SDS lysis buffer and incubated on ice for 10 min. The samples were sonicated on ice with an Ultrasonics sonicator at setting 10 for 10-s pulses and then microcentrifuged. The supernatant or chromatin solution from each sample was diluted 10ϫ with ChIP dilution buffer, and an aliquot was removed and used as input after reversing cross-link. The diluted chromatin solution was precleared for 30 min with Salmon Sperm DNA/protein A-Agarose-50% slurry and incubated overnight at 4°C with or without acetyl-histone H4 antibody. The samples were then incubated for 1 h at 4°C with Salmon Sperm DNA/protein A-Agarose-50% Slurry with rotation and washed with different wash buffers supplied in the kit. The chromatin-antibody complexes were eluted, and cross-links were reversed and treated with proteinase K. The DNA was recovered by phenol/chloroform extractions, precipitated, and used in PCR amplification. The primers used for PCR were IMP3: 5Ј-GCGAGGAGGGTGTGCCTTTG-3Ј and IMP4 as shown above. The amplification conditions for the ChIP assay were the same described above for reactions using IMP4 and IMP5 primer set.

RESULTS
Cloning of the Full-length LOT1 cDNA and 5Ј-RACE-As a first step toward analysis of the relationship between the LOT1 gene structure, promoter activity, and the presence of differential splicing and regulation of expression, we began to obtain additional sequence information of the gene's 5Ј-UTR cDNA region, which we previously reported (2). A normal human fetal brain cDNA library in the ZAPII vector (Stratagene) was used for screening, and a positive clone, FB2, hybridized to the hLOT1 probe and produced an extension of the 5Ј-UTR sequence. This newly identified nucleotide sequence of the LOT1 gene was added to the previously published sequence that is stored in GenBank TM with the accession number U72621.
We identified the transcription start site (TSS) of LOT1 using 5Ј-RACE technique. The cDNA for 5Ј-RACE was derived from human ovary and generated based on oligo-capping and RNA ligase-mediated RACE methods (33,34) as described in the GeneRacer cDNA Kit (Invitrogen). We detected a strong PCR band corresponding to about 390 nucleotides in size and weaker bands that are smaller or larger (Fig. 1). Two of the smaller bands are shown with arrows (about 280 and 200 bp). The RACE product was cloned and sequenced. Nine of 35 clones contained the LOT1 sequence, and the others were unrelated. Analysis of the amplified 5Ј ends predicted two potential transcription start sites, TSS1 and TSS2, as are shown in Fig. 1. Two of the positive clones had the 5Ј ends at nucleotide 53080 (TSS1) and six clones started at nucleotide 52919 (TSS2) of clone RP3-340H11 on chromosome 6q (GenBank accession number AL109755), resulting in an alternative size of exon 1 (326 and 165 nucleotides, respectively). Interestingly, an expressed sequence tag clone (accession number BI551851) constructed using the Cap-trapper method also showed the 5Ј end at TSS2. These data suggest that the LOT1 message is preferentially transcribed from these start sites, at least in the ovary.
Analysis of the RACE-PCR sequences also revealed that all the positive clones (100%) were extended 5Ј upstream from the gene-specific reverse primer into the LOT1 splice variant at the junction of exons 1 and 2, as opposed to the ZAC1 splice variant, which diverges from LOT1 at the 5Ј end of exon 2 in LOT1 (exon 3, ZAC1) ( Fig. 1) (see below). In addition, sizes of the bands on the gel (Fig. 1), as shown by arrows, correspond to the fragments (234 and 396 bp) from the LOT1 splice variant at this 5Ј upstream region. Taken together, the sequence analysis and hybridization results using different regions of hLOT1 or rLot1 as probes (data not shown) suggest that human LOT1 (accession number U72621) and rat Lot1 (accession number U72620) represent the major transcripts for this gene in the human and rat, respectively (1,2,8).
Analysis of the LOT1 Splice Variants-To obtain the genomic structure of hLOT1 we first began analyzing the open reading frame (ORF). Several sets of primers were designed in the coding region and used to amplify normal genomic DNA. The primers p63 and p69 generated a PCR product that encompassed the majority of the ORF. The sequence data did not reveal any evidence of introns from zinc-finger four to the 3Ј end of the ORF (data not shown). Screening a placental DNA library indicated that the sequence between ϩ153 (relative to ATG site) and the last polyadenylation signal is devoid of introns, confirming the PCR results. However, analysis of different phage-derived clones and their subclones showed presence of an intron in zinc-finger two and identified additional exon/intron boundaries upstream. In addition, we screened a human genomic P1 library with the FB2-5Ј fragment in the 5Ј UTR of LOT1, and a clone (accession number 17512) was identified containing the LOT1 gene. Several subclones were prepared using PCR or restriction fragments generated from this clone and sequenced. We determined the genomic organization by comparing the LOT1 cDNA with corresponding genomic DNA and by intron-spanning PCR using primers specific to cDNA sequences. These experiments as well as BLAST search of sequences in the GenBank data base allowed us to determine the intron/exon boundaries of the LOT1 gene (Fig. 2), which generally follow the GT-AG rule (39).
As depicted in Fig. 2, the gene consists of six exons and spans at least 70 kb. The introns range from about 2.6 to 39 kb in size. The size of exon 1 was predicted by determination of the transcription start site. Noticeably, exons 1 and 2 are separated from each other by a large intron of ϳ39 kb in size. A CpG island is present in exon 1 and extends into the 5Ј flanking region of the gene, which does not display a TATA box element. The LOT1 peptide is encoded by exons 5 and 6, resulting in a large 5Ј UTR. The translation initiation codon (ATG) is mapped at the 326-bp from the beginning (5Ј end) of exon 5, and the stop codon (TAA) is located at 1238 bp from the 5Ј end of exon 6 followed by about 1.2 kb of nucleotides including the poly(A) tail. Sequence analysis of the coding region in different cancer cell lines and specimens by PCR did not reveal any nucleotide mutation.
Two variants of the LOT1 gene (U72621) (2) have been identified and deposited in the GenBank database as ZAC1 (AJ311395) (40) and PLAGL1 (U81992) (5). Therefore, to better understand the genomic structure of LOT1 in relation with the other variants, we determined precisely where these variants diverge. The LOT1 cDNA was compared with the cDNA sequences from ZAC1 and PLAGL1 (5, 40 -42) and subsequently with the genomic clones in the GenBank (AL109755 and AL049844). The data are shown in Fig. 3, which depicts the relative location and size of exons. These variants have a common 3Ј end on exon 1; however, the subsequent exons are differentially spliced, resulting in six, nine, and five exons for LOT1, ZAC1, and PLAGL1, respectively (Fig. 3). Interestingly, LOT1 and ZAC1 variants have an identical open reading frame. However, the open reading frame of PLAGL1 lacks all the amino acid sequences upstream of the second methionine in the LOT1 and ZAC1 sequence (5).
Characterization of the Human LOT1 Promoter-We then cloned the 5Ј flanking segments of the LOT1 gene into the pGL3-Basic vector (Promega, Madison, WI) upstream of the luciferase gene to determine whether this region possesses transcription activity. The structures of the different constructs are schematically shown in Fig. 4A, and the results of a representative experiment from at least two similar experiments using these plasmids are presented in Fig. 4B. As is shown in Fig. 4B(a), insertion of the ϳ1.5-kb human LOT1 genomic fragment (nucleotides Ϫ1456 to Ϫ1 with respect to the 5Ј end of LOT1 sequence in GenBank accession number U72621) into pGL3, designated as pGL3-45, resulted in a strong increase (up to ϳ35-fold) in the activation of the luciferase gene in the A2780 ovarian cancer cell line. The transcription activity was abolished following the 5Ј truncation of the promoter, as is shown for pGL3-47 (nucleotides Ϫ525 to Ϫ1) plasmid. Interestingly, however, a second but weaker promoter activity (ϳ10-fold compared with ϳ35-fold for pGL3-45) was observed with the exon 1 proximal region (nucleotides Ϫ123 to ϩ450; vector pGL3-60) as is apparent from Fig. 4B(b), suggesting the presence of at least two potentially active promoter region in the LOT1 gene.
We also confirmed the promoter activity in other cell lines including the human ovarian cell line OVCAR3 and the ROSE cell line NuTu 26. Transfection of these cells with the expression vectors pGL3-45 and pGL3-60 showed a considerable level of reporter activity (Fig. 4C), although the transcription activity was lower in the ROSE cells than OVCAR3 cells. Activation of the human LOT1 promoter in ROSE cells was expected since we found the hLOT1 promoter sequence to be highly homologous to the mouse Lot1 promoter (and therefore likely to the rat Lot1 promoter) by aligning different segments of the hLOT1 promoter with that of the mouse  FIG. 2. Genomic structure of LOT1. A, the six exons in LOT1 and the introns are shown (q ϭ long arm of chromosome 6; TEL ϭ telomere; CEN ϭ centromere). B, exon/intron boundaries were determined by subcloning different fragments obtained from a phage DNA, P1 genomic clone, and total genomic DNA using PCR with LOT1-specific primers or restriction digest analysis. The nucleotide sequences of these subclones were aligned to the LOT1 cDNA sequence (GenBank accession number U72621). The numbers on each intron indicate the approximate size of nucleotide sequence. sequences in GenBank (accession number AF314094) (data not shown).
Analysis of LOT1 Gene Methylation in Ovarian and Breast Cancer-It is increasingly evident that methylation of CpGdense islands in the promoter of specific tumor suppressor genes is associated with their transcriptional silencing in human cancer. Therefore, we tested whether in vitro modification of the LOT1 promoter by SssI methylase is capable of transcriptional repression. As shown in Fig. 4D, the methylated pGL3-45 construct has significantly less promoter activity. The 5Ј-upstream region of the LOT1 gene contains a CpG island that spans the nucleotides 54333-52440 of clone RP3-340H11 on chromosome 6q (GenBank accession number AL109755) (Fig. 5). The sequences between nucleotides 53160 and 52440 or the region on exon 1 and part of the first intron have the highest content of CpG dinucleotides (Figs. 1 and 5). Therefore, we examined the methylation status of specific sites within this CpG island or the imprinting control region (28,29) near the transcription start site (Fig. 5) by bisulfite-PCR-sequencing analysis. The genomic DNA isolated from cancer cell lines and tissues were bisulfite-treated, which converts unmethylated but not methylated cytosine into uracil, as is shown in Fig. 6. The bisulfite-treated DNA was subjected to PCR and subcloned into a TA-cloning vector, and independent clones were sequenced. Completion of the bisulfite reaction was judged by the efficient conversion of all the cytosine residues that do not precede guanines. We then determined the ratio of methylated sites versus unmethylated sites, and the data were recorded as percent. As shown in Fig. 7A, the ovarian cancer cell lines (seven cell lines) had between 31 and 52% methylation, and the cell line CAOV3 had 99% methylation of the CpG sites on the LOT1 locus. In the two breast cancer cell lines, MDA-MB-453 and -468, the methylation levels were 93 and 58% (Fig. 7B), respectively. Hence, in these ovarian and breast cancer cell lines the alleles containing one or more methylated CpG dinucleotides in the differentially methylated region ranged from 50 to 99%. It is worth mentioning that we also subjected the DNA from MDA-MB-453 cells to bisulfite/PCR/ restriction enzyme (Taq␣I enzyme, BioLab) analysis and found that the DNA was hypermethylated, confirming the sequencing data. We have previously shown that the ovarian cancer cell lines express differential levels of LOT1 message (1). However, we were not able to find a direct association between the expression data and the methylation status in these cell lines. Furthermore, we treated MDA-MB-453, CAOV3, and NuTu 26 cancer cell lines with the demethylating agent 5-aza-2Ј deoxycytidine to determine whether the expression of LOT1/ZAC1 could be reactivated. The results showed that in these cell lines LOT1 is unresponsive to 5-aza-2Ј deoxycytidine treatment (data not shown). These results suggest that additional mech-anism(s) other than methylation may be involved in the regulation of LOT1. In other words, we hypothesize that in some cases constitutive down-regulation of LOT1 may be due to alteration elsewhere upstream in the pathway and that direct demethylation of CpG dinucleotides may not be sufficient to restore the normal expression pattern of this gene.
We examined the methylation status in paired cancerous and non-cancerous ovarian tissues and found that in three sample pairs (59, 151, 001) the tumor DNA had higher methylation level than in the normal DNA and in another pair (1002582) the difference was insignificant (Fig. 7C). We also examined the methylation status of normal ovary DNA with methylation-sensitive restriction enzyme-PCR. The DNA was digested with HpaII or MspI. These restriction enzymes both cleave at CCGG sites. However, HpaII does not cut the DNA when the internal C is methylated, and MspI digests the DNA regardless of methylation status. The DNA was then subjected  4. Analysis of the LOT1 promoter. A, the strategy for cloning the LOT1 promoter-luciferase constructs used in transient transfection assays. The constructs that were PCR-amplified from the human normal ovary genomic DNA (pGL3-60) or P1 clone 17513 (pGL3-45; pGL3-47) are indicated. Position of each construct is shown relative to the nucleotides (54333-52440) in clone RP3-340 HII (GenBank accession number AL109755). B and C, transcriptional activity of the LOT1 promoter region. The ovarian cells were transfected with the luciferase constructs using TransIT-LT1 reagent (PanVera, Madison, WI), and fresh medium was added to the cultures 5 h after transfection and incubated for an additional 18 h. Duplicate wells were transiently transfected with the DNA, and the luciferase activity for each sample was determined. The relative activity of each promoter-pGL3 vector is presented as a percent of pGL3-Basic luciferase activity (arbitrarily assigned as 100%). Each bar graph represents the mean Ϯ S.E. D, transcriptional silencing of pGL3-45 promoter construct after treatment with SssI methylase enzyme; C, control, M, methylated.
to PCR using the primers IMP4 and IMP5 spanning the nucleotides 53002-52441 in the genomic clone RP3-340H11 on chromosome 6q (GenBank accession number AL109755). A PCR product (561 bp) was visible for the HpaII digest but not for the MspI digest, indicating that one allele is methylated in normal ovary DNA (data not shown). These results and the evidence of methylated cytosines in normal ovary DNA (see above) suggest that imprinting or methylation of the maternal allele is maintained in this tissue, and possibly in breast or other somatic cells, during development.
Inhibition of Histone Deacetylase Activates the Transcription of LOT1-Recent studies have revealed that histone deacetylation is a major mechanism of DNA silencing due to its action on core histones and alterations of chromatin structure (43). Therefore, we tested the effect of TSA, a potent and specific inhibitor of histone deacetylase (HDAC) (44), on LOT1 gene expression. We have previously shown that the NuTu 26 cancer cell line has lost or decreased Lot1 expression in comparison with its normal progenitor cells (1) (Fig. 8A). Therefore, we tested whether deacetylation mechanism plays a role in the repression of Lot1 in this malignantly transformed cell line and found that TSA treatment reverses the gene silencing (Fig. 8). The transcription is induced in an early fashion (2 h) upon treatment with the HDAC inhibitor (Fig. 8B). These results suggest that HDAC activity plays a significant role, at least in part, in regulating Lot1 gene. However, it appears that upregulation of the gene by TSA is transient, suggesting involvement of additional regulatory mechanism(s) (Fig. 8).
We then asked whether reactivation of LOT1 by the inhibition of histone deacetylation is due to direct activation of the promoter. We tested the active 1.5-kb LOT1 promoter and found that it was not responsive to TSA and the histone deacetylase inhibition failed to relieve methylation-mediated repression of transcription from this promoter fragment (data not shown). In addition, we performed ChIP assay of crosslinked chromatin isolated from the control and TSA-treated A2780 (non-transfected) cells with antibodies against acetylated H4 histone and PCR amplified the LOT1 promoter. The data as shown in Fig. 9A indicate that the association of acetylated histone with the promoter is not significantly changed in response to TSA treatment. Therefore, we argued that the TSA treatment may target the LOT1 message due to an indirect effect originated upstream of the LOT1 gene. To test this hypothesis, we fused the LOT1 cDNA downstream of the green fluorescent gene and used the construct to transfect A2780 ovarian cancer cells. The data presented in Fig. 9B show that the number of positive cells expressing GFP-LOT1 fusion protein is amplified following the inhibition of HDAC, as determined by fluorescence image analysis. Similar results were obtained with the NuTu 26 cell line, which has lost or decreased expression of the gene (Fig. 8) (data not shown). Western blot analysis indicated that the GFP-LOT1 protein level, compared with GFP alone, is significantly increased in the cells treated with TSA (Fig. 9C).
The effect of inhibition of histone deacetylase observed in Fig. 9, B and C was also confirmed by FACS analysis and cell sorting of the GFP-expressing cells. As is shown in Fig. 10, the number of positive GFP-LOT1-expressing cells was lower than that of GFP-expressing cells and that the number increased significantly upon the stimulation with TSA. Taken together, these data and the Northern analysis (above) show clear correlation between the transcriptional responses of the LOT1 gene to TSA, exogenously and endogenously. DISCUSSION The candidate tumor suppressor gene LOT1 encodes a nuclear transcription factor and is strongly regulated by the activation of the epidermal growth factor receptor signaling pathway (1, 2, 8). In this paper, we have reported cloning of the full-length LOT1 cDNA, characterized the genomic structure, and analyzed its epigenetic regulation. We have also determined the exon/intron splice junctions of LOT1 and its splice variants, namely ZAC1 and PLAGL1. The LOT1 gene consists of six exons and spans at least 70 kb. The introns range from about 2.6 to 39 kb in size. The LOT1 peptide is encoded by exons 5 and 6, resulting in a large 5Ј UTR. A CpG-rich island is present in exon 1 and extends into the 5Ј flanking region of the gene, which does not display a TATA box element. We predicted the TSS of LOT1 using the 5Ј-RACE technique. Analysis of the amplified 5Ј ends using cDNA obtained from whole normal ovary predicted two potential transcription start sites, TSS1 and TSS2. However, the absence of a TATA box within the 5Ј-upstream sequences of LOT1 suggests that the actual TSS of transcription may be heterogeneous. We have identified a putative promoter region within 1.5 kb 5Ј-upstream of LOT1. Insertion of the ϳ1.5-kb fragment into pGL3 resulted in a strong induction (up to ϳ35-fold) of the luciferase gene in the ovarian cancer cell lines. Interestingly, however, a second but weaker promoter activity (ϳ10-fold) was observed with the exon 1 proximal region (Fig. 5), suggesting the presence of at least two potentially active promoter region in the LOT1 gene.
We previously hypothesized that the LOT1 gene expression may be regulated by mechanism(s) that may involve histone acetylation/deacetylation, DNA methylation, and/or mRNA stability (8). We have identified CpG islands in the upstream sequences of exon 1 and into the promoter of LOT1, which suggest that DNA methylation of CpG nucleotides may play a role in silencing the gene expression. In addition, recent reports suggested that LOT1/ZAC1 is part of a maternally imprinted chromosomal region (28 -30). Therefore, we have analyzed the imprinting based on methylation of the CpG-region and found that DNA from both normal ovary and ovarian cancer cell lines exhibit methylation of the cytosine residues. It appears that ovarian tumors and cancer cells demonstrate relatively increased levels of methylation in CpG dinucleotides. The breast cancer cell lines also had high levels of methylation. The presence of methylated alleles in normal ovary DNA suggests that imprinting of the maternal allele of the actual LOT1 gene is maintained in this tissue, and possibly in breast or other somatic cells, during mammalian development. Thus, based on these results, it is rational to believe that the LOT1 gene may be subject to two-hit inactivation or silencing by loss of heterozygosity and methylation.
Furthermore, in this study we have presented evidence for the possible role of histone deacetylation in repressing the LOT1 gene expression. Treatment of the cells with trichostatin A, a potent inhibitor of histone deacetylase, reverses the gene's FIG. 7. Methylation analysis of the CpG island of the LOT1 gene in ovarian and breast cancer. Genomic DNA was isolated from the indicated cell lines or tumors and analyzed for methylation by direct bisulfite sequencing. Bisulfite-PCR products were subcloned into pGEMT vector and sequenced by Automated Sequencer. The ratio of total methylated sites to unmethylated sites was determined and recorded as percent. Each circle indicates a CpG site in the primary DNA sequence, and each line of circles represents analysis of a single cloned allele. The numbers of the CpG site are indicated at the top and are identical to the numbers in Fig. 6. Open and closed circles represent unmethylated and methylated CpG dinucleotides, respectively. A, ovarian cancer cell lines; B, breast cancer cell lines; C, paired cancerous and non-cancerous ovarian tissues. silencing in NuTu 26 and A2780 ovarian cancer cell lines. These results suggest that HDAC activity plays a role, at least in part, in regulating the LOT1 gene. Direct interaction of HDACs with either the acetylated histones (H2A, H2B, H3, and H4) or non-histone proteins may result in deacetylation of these proteins and regulation of gene transcription. Deacetylation of histones has been found to stabilize the genomic DNA/ histone complex, hindering accessibility of the promoter to the transcription machinery (43). Apparently, this process is further enhanced by DNA methylation, which may promote his-tone deacetylation and/or may assist in recruiting methylated DNA-binding protein MeCP2 to the regions of DNA by the HDACs/co-repressor complex (45)(46)(47)(48). However, in this study we found that TSA treatment does not reactivate the in vitro methylated LOT1 promoter and does not significantly change the promoter DNA-acetylated histone association, suggesting that there exists a histone deacetylase-independent mechanism(s) for gene silencing of this locus by DNA methylation, as has also been described for other loci (49). Consistent with our findings, it has been recently reported that treatment of maternal uniparental mouse embryonic fibroblasts with TSA does not result in the inheritable expression of the imprinted Lot1/ Zac1 gene (50). This notion is supported by other reports that in cancer the hypermethylated genes such as MLH1, TIMP3, CDKN2B/INK4B/p15, and CDKN2A/INK4/p16 can not be transcriptionally reactivated with TSA alone (51). Therefore, it is possible that re-expression of LOT1 by the inhibition of histone deacetylase may be due to the effect on the unmethylated or hypomethylated allele of the gene and be elicited by a process originating elsewhere upstream in the LOT1 pathway. We conclude that distinct mechanisms other than imprinting FIG. 8. The effect of inhibition of HDAC activity on LOT1 expression. TSA treatment relieves the Lot1 transcriptional silencing in NuTu 26 ROSE cells. A-C, normal and NuTu 26 cells were cultured to subconfluence and treated with TSA (100 ng/ml) for the indicated time-points. C, images of the hybridization signals shown in B were determined by AMBIS image analyzer, and the relative intensity was plotted. RNA from the normal cells in A was used as comparison with RNA from NuTu 26 cells hybridized with the Lot1 probe. Total RNA was hybridized with the isotope-labeled Lot1 probe. FIG. 9. A, PCR amplification of the LOT1 promoter from the input (non-precipitated control), mock-precipitated (ϪIP), and immunoprecipitated (ϩIP) DNAs using the IMP3 and IMP4 primer set (see "Materials and Methods"). B, the fluorescence microscopy of A2780 cells transfected with GFP-LOT1 fusion protein. The cells were transfected and then treated with TSA for 18 h. The fluorescence images were examined with a fluorescent microscope. C, Western blot analysis of the proteins extracted from the transfected cells, comparing GFP-LOT1 protein (ϳ78 kDa) expression level with GFP (ϳ27 kDa) expression level in cells treated with or without TSA. Fifteen microliters of wholecell lysate was loaded on each lane, blotted, and exposed to the GFP antibody as described under "Materials and Methods." or methylation may play major roles in regulating the LOT1 expression, particularly when the expression needs to be completely blocked, e.g. expression from the paternal allele or from the maternal allele when imprinting may be erased during development. Perhaps, promotion of cell survival and positive selection for development and maturation in a cell-type and stage-specific manner require repressed activity of LOT1. Studies are currently underway to elucidate in detail the mechanism of LOT1 gene regulation by different genetic and epigenetic factors, which may be useful in determining precisely the role of this gene in cancer and/or development.  Fig. 9 and treated with or without TSA for 18 h. The treated cells were collected, and 10,000 cells were subjected to FACS analysis. B, the percent of the fluorescent cells was plotted.