PEG10 Is a C-MYC Target Gene in Cancer Cells

Research Article

PEG10 Is a c-MYC Target Gene in Cancer Cells

Chi-Ming Li,1 Adam A. Margolin,1,2 Martha Salas,1 Lorenzo Memeo,3 Mahesh Mansukhani,3 Hanina Hibshoosh,3 Matthias Szabolcs,3 Apostolos Klinakis,1,4 and Benjamin Tycko1,3

1Institute for Cancer Genetics, 2Genome Center, and Departments of 3Pathology and 4Genetics and Development, Columbia University Medical Center, New York, New York

Abstract encodes a protein with distant homology to retroviral gag-pol The product of the imprinted gene paternally expressed proteins, suggesting that the gene arose from an ancient retroviral insertion event and then became fixed in mammalian gene-10 (PEG10) has been reported to support proliferation in hepatocellular carcinomas, but how this gene is regulated evolution (1, 3, 4). Presumably related to its viral origin, the has been an open question. We find that MYC knockdown PEG10 mRNA encodes two protein isoforms (RF1 and RF2) via by RNA interference suppresses PEG10 expression in Panc1 translational frameshifting (4). Recently, two groups have reported pancreatic carcinoma and HepG2 hepatocellular carcinoma overexpression of PEG10 in hepatocellular carcinomas and in cells and that knockdown of PEG10 inhibits the proliferation proliferating cells of regenerating normal liver, and using cell of Panc1, HepG2, and Hep3B cells. Conversely, PEG10 was transfections, both concluded that this gene has growth-promoting up-regulated by inducing c-MYC expression in a B-lymphocyte activity (5, 6). PEG10 is also overexpressed in the embryonic cell line. Chromatin immunoprecipitation from Panc1 cells form of biliary atresia, a disease associated with cell proliferation showed c-MYC bound to an E-box-containing region in the (7). Additional reports have suggested that PEG10 protein may act by blocking transforming growth factor h (TGF-h) signaling PEG10 first intron and site-directed mutagenesis showed h that the most proximal E-box is essential for promoter in epithelial cancers via binding to TGF- receptor II (8) or by activity. In a mouse mammary tumor virus (MMTV)-MYC blocking the apoptotic factor SIAH1 (5). It was recently found transgenic mouse model of breast cancer, most but not all that specific chemokines induce PEG10 in normal B-lymphocytes of the mammary carcinomas had strongly increased Peg10 and in B-cell leukemias, correlating with increased cellular mRNA compared with normal mammary gland. By immuno- resistance to apoptosis (9). However, the transcriptional pathways histochemistry, normal human breast and prostate epithelium that activate this gene in proliferating cells have not been pre- was negative for the major isoform [reading frame-1 (RF1)] viously reported. of PEG10 protein, but this cytoplasmic protein was strongly The c-MYC transcription factor heterodimerizes with MAX and expressed in a subset of breast carcinomas in situ and inva- activates a large number of downstream target genes, many of sive ductal carcinomas (f30%) and in a similar percentage of which are essential for cell growth and proliferation (10–12). In prostate cancers. As in the mouse model, we found positive, an expression profiling experiment in which we created a c-MYC but not absolute, correlations between PEG10 and c-MYC in knockdown in human carcinoma cells, we noticed that PEG10 was reproducibly down-modulated. Here we describe this experiment tissue arrays containing 161 human breast cancers (P < 0.002) and 30 prostate cancers (P = 0.014). Immunostaining of and additional analyses implicating c-MYC as an activator of human placenta showed PEG10 and c-MYC proteins coex- PEG10 and showing recurrent acquisition of PEG10 protein pressed in proliferating cytotrophoblast and coordinately lost expression in primary human breast and prostate cancers. in postmitotic syncytiotrophoblast. These findings link cancer genetics and epigenetics by showing that a classic proto- Materials and Methods oncogene, MYC, acts directly upstream of a proliferation- Knockdown of c-MYC and PEG10 by RNA interference. Panc1 positive imprinted gene, PEG10. (Cancer Res 2006; 66(2): 665-72) pancreatic carcinoma cells and HepG2 and Hep3B hepatocellular carcinoma cells were maintained in DMEM with 10% of fetal bovine serum Introduction plus 1.5 g/L sodium bicarbonate, 0.1 mmol/L nonessential amino acids, and 1 mmol/L sodium pyruvate. Two double-stranded short interfering RNAs Cancer genetics and epigenetics have generally been studied as (siRNA) directed to human c-MYC had the DNA target sequences 5V- separate topics and unifying these research areas is an interesting AAGGACTATCCTGCTGCCAAG-3V (Qiagen-Xeragon, Germantown, MD) challenge. It will be important to understand how genetic pathways and 5V-AAGGTCAGAGTCTGGATCACC-3V (Qiagen-Xeragon); two siRNAs intersect with epigenetic gene regulation in the multistage directed to human PEG10 had the DNA target sequences 5V-AAAGCTG- evolution of human cancers. The paternally expressed gene-10 GAGCGCTCCCACTA-3V(Qiagen-Xeragon) and 5V-AAGTCGCTGTCTGCTCT- V (PEG10) gene was identified based on its location in an imprinted GATT-3 (Qiagen-Xeragon). Cells at 60% confluence on 60-mm tissue culture plates were transfected with 2 Ag of the specific siRNAs or control domain on human chromosome 7q21 and characterized as siRNAs with scrambled DNA target sequence 5V-AATTCTCCGAACGTGT- paternally expressed/maternally silenced (1). Peg10 is also CACGT-3V using Transmessenger reagent (Qiagen, Valencia, CA). The imprinted in mice (2). The major open reading frame of PEG10 transient transfection procedure was carried out twice on consecutive days. After another 24 hours of incubation, cellular RNA and protein were isolated and used for further analysis. Gene expression profiling in Panc1 cells. Human genome 133A2.0 Requests for reprints: Benjamin Tycko, Institute for Cancer Genetics, Columbia GeneChips (Affymetrix, Santa Clara, CA), which contain 22,000 oligonucle- University Medical Center, 1150 St. Nicholas Avenue, New York, NY 10032. Phone: 212- otide probe sets, querying 14,500 well-documented genes and 3,900 less 305-1149; Fax: 212-305-5489; E-mail: [email protected]. I2006 American Association for Cancer Research. well-characterized transcripts, were used for analyzing Panc1 RNAs from doi:10.1158/0008-5472.CAN-05-1553 triplicate plates treated with MYC-directed or control siRNAs. The cRNA www.aacrjournals.org 665 Cancer Res 2006; 66: (2). January 15, 2006

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Cancer Research probes were synthesized as described (13). After scanning, mRNA Chromatin immunoprecipitation assays. Proliferating Panc1 cells, four expression values were determined using Affymetrix microarray suite 150-mm plates at 90% confluence, were analyzed using the ChIP-IT kit (Active v. 5.0 and significantly up-regulated and down-regulated genes were Motif, Carlsbad, CA). Briefly, the cells were fixed in 1% formaldehyde at room identified by ANOVA and displayed graphically using the GeneSpring temperature for 10 minutes and scraped in 1 PBS containing software package (Agilent, Palo Alto, CA). 0.5 mmol/L phenylmethylsulfonyl fluoride (PMSF) after treatment with Induction of c-MYC and gene expression profiling in EREB.TMYC glycine-containing stop-fix solution. After centrifugation, the cell pellet cells. EREB.TMYC B cells, which contain an EBV gene (EBNA-2) for was resuspended and incubated for 30 minutes in ice-cold lysis buffer conditional immortalization and a tetracycline-repressible c-MYC con- with PMSF and proteinase inhibitor cocktail. The nuclei were released by struct, were maintained in RPMI with 1 Amol/L h-estradiol and 1 Ag/mL 10 strokes of an ice-cold Dounce homogenizer, centrifuged, and resuspended tetracycline (14). Endogenous c-MYC expression was down-regulated due in shearing buffer. Chromatin was sheared to an average size of 1,000 bp to growth arrest by removing h-estradiol from the culture medium for (verified on an ethidium-stained agarose minigel) by sonication at 4jC. After 24 hours. After washing the cells with PBS, exogenous c-MYC expression preclearing with protein-G beads, in addition to saving an aliquot as a was induced by incubating with medium lacking h-estradiol and positive input control, the sheared chromatin was aliqouted into three tubes tetracycline. After 24 hours of induction, cellular RNA was harvested and and incubated in ice-cold chromatin immunoprecipitation buffer used for expression profiling on U95A GeneChips (Affymetrix). Cells for overnight at 4jC with 3 Ag of polyclonal IgG antibody specific for maintained in medium with tetracycline only for 48 hours were used as c-MYC (N-262, Santa Cruz Biotechnology, Santa Cruz, CA), 3 Ag of rabbit pre- the c-MYC-negative control. The microarray data were analyzed as above. immune serum as a negative control (Santa Cruz Biotechnology), or 1.8 Agof Growth kinetics in RNA interference–treated cells. To measure the transcription factor IIB antibody provided in the kit as a positive control, effect of PEG10 knockdown on cell growth, 2 105 of Panc1 cells, 1.5 105 respectively. The immunoprecipitated chromatin was purified from the of HepG2 cells, and 5 104 of Hep3B cells were plated in 60-mm dishes and, chromatin/antibody mixtures by incubating with protein G beads followed after 24 and 48 hours, were transfected with 1 Ag of the specific siRNAs or by several washing steps and the chromatin immunoprecipitation DNA control siRNAs. Cells from triplicate plates were counted in a hemocytom- was eluted in a solution containing 10 mmol/L NaHCO3 and 1% SDS. eter at 72 hours and RNA harvested from parallel plates for Northern blot After removal of proteins by digestion with proteinase K, the DNA was confirmation of the knockdowns. purified and examined by PCR with primer pairs within and around the Northern blot analysis of RNA and Southern blot analysis of DNA PEG10 promoter region. The primer sequences were A-forward: 5V- methylation. Total RNA was prepared after solubilizing cells or tissues in AAGCCTGAATGGAGTGTGCG-3V, A-reverse: 5V-CAAAATACCCAGCCACC- Trizol reagent (Invitrogen, Carlsbad, CA) according to the protocol of TTCTTC-3V; B-forward: 5V-CAACAGATTGTCAGTTTCCCAAGC-3V, B-reverse: the manufacturer. The RNA was electrophoresed on 1.0% agarose gels 5V-CGATTACGAGGTTTACACAGAGACC-3V;C-forward:5V-CTCTGTTT- containing formaldhehyde and then transferred to Nytran membranes TCTTGGAGCAGGACC-3V, C-reverse: 5V-ACCACAACTACACCGCCACTTC- (Schleicher & Schuell, Keene, NH). The blots were hybridized at 42jC in the 3V; D-forward: 5V-CTCCCTAAGCCTGCCTCTG-3V, D-reverse: 5V-GGGCTCA- Ultrahyb (Ambion, Austin, TX) solution, with cDNA probes specific for GGCAAGGAAGGT-3V;andE-forward:5V-TAACTTGTGCTGCCTCAG- MYC or PEG10 (human and mouse probes for each gene were generated TCGC-3V, E-reverse: 5V-CCTGCCACCAGACATTTCATTC-3V. by reverse transcription-PCR; primers available on request), and washed at Promoter reporter assays and site-directed mutagenesis. The PEG10 high stringency in 0.1% SDS and 0.1 SSC for 1 hour at 64jC. The blots were promoter fragments, corresponding to nucleotides 80,003 to 81,049 of stripped and rehybridized with a probe for glyceraldhehyde-3-phosphate GenBank accession AC069292, were cloned upstream of the firefly luciferase dehydrogenase (GAPDH) as a loading control. Genomic DNA was prepared gene in the pGL3-Basic vector (Promega, Madison WI) to yield the PEG10 by lysing the cells or tissue (pulverized under liquid nitrogen) in SDS/ promoter plasmid pGL3-PEG10. Three E-box elements in the first intron proteinase K, incubating for several hours at 50jC, followed by phenol/ of PEG10, corresponding to nucleotides 81,698 to 82,061, 83,270, or 83,894 chloroform extraction and precipitation in ethanol. The DNA, 2.5 Ag, was of GenBank accession AC069292, were sequentially cloned into the BamHI digested overnight with restriction enzyme RsaI, XbaI, CfoI, HpaII, or MspI and SalI sites downstream of the luciferase cassette in the pGL3-PEG10 (New England Biolabs, Beverly, CA), either singly or in the indicated plasmid to yield the PEG10 promoter-reporter-enhancer plasmids pE1, combinations, and the digested DNA was resolved on 1% agarose gels, pE12, and pE123, respectively. The mutant (cacatg ! cacaGT) construct at denatured, and neutralized under standard conditions and transferred to the first E-box of the PEG10 gene, pPEG10m, was generated by site-directed Nytran membranes. The Southern blots were hybridized with genomic mutagenesis using the QuickChange kit (Stratagene, La Jolla, CA) with probes specific to the human or mouse PEG10/Peg10 promoter regions. primers forward 5V-gcgctgcgaggcacaGTaactgcagaggtaca-3V and reverse 5V-tg- The primers for synthesizing the PEG10 promoter probe by PCR of tacctctgcagttACtgtgcctcgcagcgc-3V. The c-MYC expression construct, pMT2- human genomic DNA were PEG10 promoter US, 5V-CACGCAAAACTTGT- MYC, and the E-box-containing construct, pTERT, and its E-box mutant CACGCC-3V, and PEG10 promoter DS, 5V-GAGTGGGAGCCATTCCAAAAG-3V. construct, pTERTdm, used for the positive and negative controls in our The primers for generating the mouse Peg10 promoter probe were mPeg10 experiments, have been previously described (15). To measure MYC promoter US, 5V-CGAAGCACGCTGGGATTTGG-3V, and mPeg10 promoter responses, 7.5 104 Hep3B cells in 35-mm plates were transfected in DS, 5V-GCTCTCGTCTCAGGATCTGG-3V. triplicate with 0.2 Ag of pTERT, pTERTdm, or each of the PEG10 promoter/ Antibodies and Western blotting. After boiling at 100jC for 10 minutes E-box constructs with or without 0.2 Ag of pMT2-MYC using Lipofectamine in a denaturing solution containing 12 mmol/L Tris (pH 6.8), 5% glycerol, 2000 transfection reagent (Invitrogen). As controls for normalization, 50 ng 0.4% SDS, 3 mmol/L 2-mercaptomethanol, and 0.02% bromophenol blue, of the Renilla luciferase plasmid, pTK-luc (Promega), and 50 ng of the green 50 Ag of total protein lysates from the experimental samples were electro- fluorescent protein plasmid, pEGFP-N1 (Clontech, Mountain View, CA), phoresed on 4% to 20% polyacrylamide gradient/SDS gels (Invitrogen). were cotransfected and dual luciferase activities were measured using After transferring to Immobilon membranes (Millipore, Bedford, MA) and the Dual-Luciferase Reporter Assay System (Promega) after 36 hours. All blocking by 5% milk in 1 TBS, the membrane was hybridized with an transfections were done in triplicate. affinity-purified rabbit polyclonal antibody against an 18-mer peptide at the Immunohistochemistry for PEG10. Formalin-fixed, paraffin-embedded COOH terminus of the major open reading frame (RF1; see ref. 4) of PEG10 sections of human breast carcinomas, including adjacent nonneoplastic (peptide sequence: NV-CPAKASKSSPAGNSPAPL-CV), or a mouse monoclonal glandular structures, atypical ductal hyperplasia and carcinoma in situ antibody against a-actin (Sigma) as a loading control, in 1 TBS containing lesions, and custom breast cancer and prostate cancer tissue arrays 5% dry milk and 0.1% Tween 20 overnight at 4jC. After washing, the signal produced in the Columbia University Pathology Department, were heated in was amplified and detected using a peroxidase-conjugated goat anti-rabbit 10 mmol/L citrate buffer (pH 6.0) in a microwave oven at the maximum or anti-mouse immunoglobulin G (IgG; Amersham Pharmarcia Biotech, power for 10 minutes and then at a reduced power for 15 minutes for Piscataway, NJ) and enhanced chemiluminescence plus detection system antigen recovery, and the sections were blocked in 5% or 10% goat serum (Amersham Pharmarcia Biotech). for 20 minutes and then exposed to affinity-purified antiserum raised

Cancer Res 2006; 66: (2). January 15, 2006 666 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. PEG10 and c-MYC against a COOH-terminal peptide of PEG10 (see Western blotting methods Activation of c-MYC induces PEG10 mRNA in conditionally above) at a 1:200 dilution in 1 PBS containing 0.01% Tween 20 for 1 hour immortalized B cells. To ask whether the converse situation, at room temperature or to an anti-c-MYC antibody (N-262; Santa Cruz activation of PEG10 transcription by inducing c-MYC expression, j Biotechnology) at a dilution of 1: 200 for overnight at 4 C. The sections were could also be shown, we examined oligonucleotide microarray data then incubated with the secondary antibody, goat anti-rabbit IgG, at 1:200 from EREB.TC-MYC cells, a conditionally immortalized B-cell line dilution in 1 PBS containing 0.01% Tween 20 for 30 minutes and developed using the Elite ABC kit (Vector Laboratories, Burlingame, CA). in which cell proliferation and MYC expression could be controlled The sections were counterstained with hematoxylin and microscopic independently. In this cell line, immortalization is dependent on images were captured digitally. The results for c-MYC and PEG10 induction by estrogen because the EBV gene encoding EBNA-2 is immunostaining in breast cancers were scored independently and blindly expressed as a chimeric fusion with the hormone-binding domain by three surgical pathologists (H.H., L.M., and M.S.) who concurred in the of the estrogen receptor and c-MYC protein expression can be scoring. The results for c-MYC and PEG10 in prostate cancers were scored induced by removing tetracycline from the culture medium independently and blindly by two surgical pathologists (M.M. and M.S.). because this line carries a TET-repressible c-MYC expression construct. In these cells, we observed strong activation of PEG10 mRNA at 24 hours after induction of exogenous c-MYC by the Results removal of tetracycline (Fig. 1C). In contrast, SGCE mRNA was not RNA interference–mediated c-MYC knockdown suppresses induced in these cells and in fact was slightly repressed after c-MYC PEG10 expression in carcinoma cells. To identify candidate induction (Fig. 1C), a result paralleling the negative findings for c-MYC target genes in Panc1 pancreatic carcinoma cells, we this flanking gene in our c-MYC knockdown experiments. transfected these cells with MYC-directed or control siRNAs and Knockdown of PEG10 by RNA interference inhibits carci- profiled gene expression using oligonucleotide microarrays. The noma cell proliferation. MYC mRNA is known to turn over RNA interference procedure effectively suppressed both MYC rapidly in the cell, allowing the levels of c-MYC to be regulated mRNA and c-MYC protein expressions (Figs. 1A and 2). This during the cell cycle. When we inhibited new mRNA synthesis in c-MYC knockdown was accompanied by cell growth arrest without Panc1 cells using actinomycin D and measured PEG10 and MYC loss of cell viability. Among the large number of genes scoring as mRNA over time, we found that PEG10 mRNA was stable for up to significantly down-modulated in the microarray data, we noticed 8 hours whereas MYC mRNA declined by 1 hour and disappeared an imprinted gene, PEG10 (Fig. 1B). This gene was consistently from the cells by 2 hours (Fig. 2B). Despite this stability of PEG10 suppressed by MYC siRNA with down-modulation seen not only in mRNA under normal conditions, the levels of this transcript were Panc1 cells but also in the hepatocellular carcinoma line HepG2 strongly reduced by transfecting cells with PEG10-directed siRNA (Fig. 2A). The effect was validated using two double-stranded (Fig. 2A). Because prior data have suggested a proliferation-positive siRNAs, matching different sequences within the MYC transcript function for the PEG10 gene, we asked whether down-modulating (see Materials and Methods), arguing against ‘‘off-target’’ artifacts. PEG10 expression by RNA interference would inhibit the growth of Interestingly, the SGCE gene, which is oriented head-to-head with a series of carcinoma cell lines [i.e. Panc1, HepG2, and Hep3B PEG10, separated by an 800-bp CpG island, and is coordinately (hepatocellular carcinoma)]. We found that cell proliferation was imprinted in the same parental ‘‘direction’’ with PEG10, was not significantly inhibited in each of these lines but that in each cell affected by the c-MYC knockdown. Because of our interest in line the PEG10 knockdown had a weaker inhibitory effect than a pathways involving imprinted genes, we selected the PEG10 gene MYC knockdown (Fig. 2C). By using two different PEG10-directed for further validation as a c-MYC target. siRNAs, we confirmed that the inhibition of cell proliferation was a

Figure 1. Knockdown of c-MYC protein by MYC-directed siRNA down-modulates PEG10 mRNA and activation of c-MYC induces PEG10 mRNA. A, MYC-directed and scrambled (S) control siRNAs were introduced into Panc1 cells by two successive transient transfections (see Materials and Methods). Cellular proteins were immunoblotted with antibody against human c-MYC. Immunoblotting for h-actin is shown as a loading control. B, cRNA probes from four replica plates of Panc1 cells transfected with scrambled siRNA and four plates of Panc1 cells transfected in parallel with MYC-directed siRNA were applied to Affymetrix 133A GeneChips. Genes (probe sets) differentially expressed between these two conditions were identified by three successive filtering steps: first, ANOVA requiring P < 0.01; second, requiring Affymetrix presence calls in at least three of the eight samples; and third, requiring 1.3-fold variation from the experiment mean in at least three of the eight samples. This procedure produced 79 probe sets reporting increased mRNA expression (blue lines) and 248 probe sets reporting decreased mRNA expression (red lines) after MYC RNA interference. Solid, dashed, and stippled white lines, mRNA values for PEG10, MYC, and SGCE, respectively, normalized to the experiment mean. C, cRNA probes from six replica plates of EREB.TC-MYC cells in the presence of tetracycline (c-MYC negative) and six replica plates in the absence of tetracycline (MYC positive) were hybridized to Affymetrix U95A GeneChips. The data were filtered as in (B). www.aacrjournals.org 667 Cancer Res 2006; 66: (2). January 15, 2006

Figure 2. Half-life of MYC and PEG10 mRNAs and effects of PEG10- and MYC-directed RNA interference on cell proliferation. A, MYC-directed RNA interference down-modulates PEG10 mRNA in both Panc1 and HepG2 carcinoma cells whereas PEG10-directed siRNA effectively reduces PEG10, but not MYC mRNA. The lanes correspond to duplicate transfections for each condition. B, half-life determination for MYC and PEG10 mRNAs. Total RNA was analyzed by Northern blotting at the indicated times after the addition of actinomycin D (5 Ag/mL). An ethidium bromide–stained image of the 28S rRNA is shown as a loading control. PEG10 mRNA is seen as a major band at f6 kb and a minor band at a larger size. C, PEG10-directed siRNA slows the growth of pancreatic and hepatocellular carcinoma cells. Cells (triplicate plates) were transfected at 24 and 48 hours with the indicated siRNAs and cell counts were determined at 72 hours. Columns, percent increase in cell number between 24 and 72 hours; bars, SD.

specific effect due to the PEG10 knockdown. These data are of the DNA in and around the PEG10 gene by PCR of the immu- consistent with PEG10 being one of many effector genes influ- noprecipitated DNA. As shown in Fig. 3A, this procedure revealed encing net cell proliferation downstream of MYC. c-MYC protein bound specifically to DNA near the PEG10 E-box c-MYC protein binds to E-box-containing sequences in the sequences but not to more distant upstream or downstream PEG10 first intron. Analysis of the human PEG10 genomic sequences. sequence revealed three consensus E-boxes (CACATG) in the first As noted above, the expression of the SGCE gene, which is intron (Fig. 3A). To determine whether the PEG10 promoter the closest flanking gene to PEG10 and which is co-coordinately region is a direct target for the c-MYC transcription factor, which imprinted with PEG10, was not strongly affected by up- or down- binds E-box sequences as MYC/MAX heterodimers, we carried modulation of c-MYC, and in fact was slightly repressed by out chromatin immunoprecipitation assays. Proliferating Panc1 activation of c-MYC in the B cell line. This gene specificity of the cells were treated with formaldehyde to produce protein-DNA regulation of PEG10 by c-MYC may be facilitated by the fact that cross-links and chromatin immunoprecipitation was done using the binding sites for c-MYC are located in the PEG10 first intron anti-c-MYC or control preimmune serum. We queried five regions and not in the 800-bp CpG island between SGCE and PEG10.

Figure 3. PEG10 is a direct transcriptional target of c-MYC. A, chromatin immunoprecipitation assays showing c-MYC binding to an E-box-containing region in PEG10 intron 1. The genomic DNA around PEG10 is approximately drawn to the scale; rectangles, first exons of SGCE and PEG10. These genes are divergently transcribed (arrows). Closed circles, E-boxes, CAC(G/A)TG. For the chromatin immunoprecipitation assays, genomic DNA from Panc1 cells was cross-linked and sonicated, then immunoprecipitated with a rabbit polyclonal antibody against human c-MYC (MYC) or with control preimmune serum (PI). After reversing the cross-links, PCR was done to map the region bound by c-MYC using the five indicated primer pairs. The resulting 200- to 300-bp PCR products are shown. Regions C and D bind c-MYC. Input DNA (0.5% of the total chromatin DNA; I) was the positive control for each PCR reaction. Similar results were obtained in three experiments. B, site-directed mutagenesis of the PEG10 promoter. The indicated promoter fragment containing the proximal intronic E-box has MYC-responsive activity in the luciferase reporter assay, which is abrogated by a point mutation in the E-box. The MYC construct, pMT2-MYC, used to ascertain E-box-dependent MYC/MAX activity, the E-box-containing construct, pTERT, and its E-box mutant construct, pTERTdm, used as positive and negative controls, have been previously described (15). See Materials and Methods for details of the plasmid constructs.

Cancer Res 2006; 66: (2). January 15, 2006 668 www.aacrjournals.org

The proximal intronic E-box is essential for PEG10 promoter Peg10 mRNA was expressed at very low levels in the normal activity in reporter assays. To further test whether PEG10 is a mammary gland, with Northern blot bands visualized only after direct c-MYC target gene, we cloned a nested series of PEG10 prolonged exposures of the blots. In contrast, of the 20 mammary intron fragments downstream of the firefly luciferase (LUC) gene, carcinomas, 11 expressed easily detectable Peg10 mRNA, substan- placing the PEG10 5V sequences (the region between PEG10 and tially elevated compared with the normal mammary tissue (Fig. 4A SGCE) upstream of LUC. This series of promoter-reporter-enhancer and B), with 7 of these tumors showing a >10-fold increase plasmids allowed us to test which of the three E-box elements in (Fig. 4B). the first intron were crucial for PEG10 promoter activity in Hep3B Although all of the carcinomas in this mouse model are initiated cells. Luciferase assays showed that the smallest construct via overexpression of the MYC transgene, rehybridization of the containing only the most proximal E-box (363 bp of first intron Northern blots with a MYC probe showed that as these tumors sequence) had full activity, with no increase in reporter expression progress, some of them cease to express high levels of MYC mRNA. seen with the longer inserts containing both the proximal E-box Consistent with Peg10 being a target of c-MYC in this system, Peg10 and the two downstream E-boxes (data not shown). We therefore mRNA was expressed very poorly in both of the two carcinomas carried out site-directed mutagenesis of this minimal promoter- that no longer expressed MYC (Fig. 4B). However, we also observed reporter-enhancer construct, changing the E-box CACATG se- a third class of tumor in which Peg10 mRNA was poorly expressed quence to CACAGT (Fig. 3B). This mutation completely abolished despite persistent high expression of MYC. Among these seven basal promoter activity and abrogated the response to exogenous tumors, Southern blotting of genomic DNA digested with c-MYC (Fig. 3C). These findings suggest that the E-box in the methylation-sensitive restriction enzymes showed that the Peg10 PEG10 first intron is an essential promoter/enhancer element. In promoter was hypermethylated at CpG sites in one case whereas conjunction with the chromatin immunoprecipitation data above, the remaining six Peg10-low cases did not show promoter these results implicate PEG10 as a direct transcriptional target hypermethylation (Fig. 4C and data not shown). An additional of c-MYC. example of Peg10 promoter hypermethylation was found in one of The mouse Peg10 gene is frequently activated in c-MYC- the two tumors with low MYC mRNA. The methylation of the Peg10 driven mammary carcinomas. The sequence of PEG10 is promoter was not affected in the tumors with high expression of conserved (66% amino acid identity) between humans and mice, Peg10, which retained a biphasic pattern (fully methylated and as is the gene structure, suggesting that it should be possible to unmethylated bands corresponding to the imprinted and non- study the regulation of this locus in cancer using mouse model imprinted alleles, respectively) identical to the normal mammary systems. Mice carrying an mouse mammary tumor virus (MMTV)- gland. In aggregate, these data indicate that Peg10 is frequently MYC transgene have a highly penetrant phenotype of mammary activated in this c-MYC-driven transgenic model of breast cancer carcinoma, with most of the animals developing palpable tumors by without loss of methylation imprinting, and that in some of around 8 months (16, 17). To ask whether the murine Peg10 gene is the mammary tumors this gene can be a target for epigenetic a target of c-MYC in tumorigenesis, we collected biopsies repression, either with or without a gain of promoter methylation. of 20 mammary carcinomas arising in these mice and measured To ask whether MYC expression is sufficient for inducing Peg10 mRNA in these tumors and in two control biopsies of normal Peg10 in the mouse mammary gland, we examined nonneoplastic mammary glands by Northern blotting and Phosphorimaging. mammary glands from first pregnancy MMTV-MYC mice by

Figure 4. Peg10 expression is activated in breast carcinomas from MMTV-MYC transgenic mice. A, total RNA from primary breast tumors from MMTV-MYC transgenic mice was analyzed by Northern blotting with Peg10 and MYC probes. The tumors overexpress Peg10 mRNA relative to the normal mammary gland. GAPDH is a loading control. B, phophorimaging data from Northern blots analyzing 22 mammary tumors and two normal multiparous (nonlactating) mammary glands. The tumors are grouped based on their MYC mRNA expression, with low indicating the few tumors with transgenic MYC mRNA bands not detectable after 3-day exposure of Northern blots and high indicating the majority of the tumors with transgenic MYC mRNA clearly seen on Northern blots after overnight exposure. Filled diamonds, tumors showing Peg10 promoter hypermethylation by Southern blotting; open diamonds, tumors and normal controls with normal methylation. C, analysis of CpG methylation in the mouse Peg10 promoter by Southern blotting. Genomic DNA from MMTV-MYC tumors (T ) and normal mammary gland (N) was digested with RsaI(R), either alone or with the methylation sensitive restriction enzymes CfoI(C)orHpaII (H). MspI(M) is a non-methylation-sensitive isoschizomer of HpaII. A restriction map of the Peg10 promoter region is shown below with the flanking RsaI sites scored by the bold lines and the internal CfoI and HpaII sites scored by thin lines. Filled circles, E-boxes. The Peg10 CpG island is biallelically methylated (protected from CfoI and HpaII digestion) in tumor T10 but shows a normal imprinted pattern in the other tumor (T2), with one allele methylated and therefore not digested (dash) and the other allele unmethylated and therefore digested (brackets).

www.aacrjournals.org 669 Cancer Res 2006; 66: (2). January 15, 2006

Northern blotting. Peg10 mRNA was not highly expressed in these restricted to the ductal carcinoma in situ or was more strongly nonneoplastic glands although MYC mRNA was expressed at levels expressed in the ductal carcinoma in situ compared with the comparable to those seen in the breast tumors (data not shown). invasive carcinoma (Fig. 5D-F). As shown in Table 1B, results from Overall, these results suggest that additional factors associated with serial sections of the tissue array stained with anti-PEG10 and the transition to neoplasia are necessary to activate Peg10 and that anti-c-MYC antibodies showed that PEG10 protein expression is c-MYC is necessary, but not sufficient, for Peg10 induction in this positively (P < 0.002, Fisher’s exact test), but not absolutely, model system. correlated with c-MYC expression in human breast cancers. PEG10 is activated in subsets of human breast and prostate Similar results were obtained with a tissue microarray containing carcinomas. A substantial subset of human breast carcinomas, 30 prostate carcinomas. The normal prostatic epithelium stained including ductal carcinoma in situ, overexpress MYC either due to only very weakly for the RF1 isoform of PEG10, and only in the basal gene amplification or transcriptional deregulation (18–20). To ask cell layer, whereas 37% of the prostate cancers in this tissue array whether the PEG10 gene is activated in primary human breast were strongly PEG10 positive (Fig. 5G). As was true for the breast cancers, we carried out immunohistochemistry for PEG10 protein, carcinomas, we observed a positive, but not absolute, correlation of using an affinity-purified polyclonal antiserum raised against a PEG10 with c-MYC in these prostate cancers (Table 1C). peptide epitope in the PEG10 major open reading frame (the PEG10 and c-MYC are coexpressed in placental cytotropho- RF1 protein isoform), in a series of 23 paraffin-embedded breast blast. Lastly, we wished to determine if there is a correlation biopsies with standard histologic sections and in a second series between PEG10 and c-MYC protein expressions in a major normal of 161 breast cancer cases in a tissue array. Several of the cases site of PEG10 expression, the developing placenta. We carried out examined in standard sections contained areas of ductal carcino- immunostaining of a normal mid-gestation (16 weeks) placenta, ma in situ and invasive ductal carcinoma, as well as nonneoplastic applying anti-PEG10 and anti-c-MYC to serial sections of the glandular tissue, thus allowing an assessment of PEG10 protein chorionic villi. These villi are organized as three major tissue expression during breast cancer progression. The results indicated layers from inside to outside: a mesenchymal core, a cytotropho- that PEG10 is not detectable in normal quiescent breast epithelial blast layer, and a syncytiotrophoblast layer. The cytotrophoblast cells but that this protein becomes strongly expressed in 55% is a proliferating tissue compartment, which is the direct precursor of ductal carcinoma in situ lesions and 32% of invasive ductal of the postmitotic syncytiotrophoblast. We found that nuclear carcinomas, where it is found predominantly in the cytoplasm c-MYC and cytoplasmic PEG10 were coexpressed in the cytotro- (Table 1A; Fig. 5A-F). In the tissue array, 36% of the breast phoblast layer whereas both gene products were absent from the carcinomas, which were mostly invasive ductal cancers with a overlying syncytiotrophoblast (Fig. 5H and I). Also evident in small number of ductal carcinoma in situ lesions, were PEG10 these tissue sections was the more uniform expression of PEG10 positive (Table 1B). Combining the standard sections and the tissue compared with c-MYC (Fig. 5H and I). This finding is consistent array, we found 11 PEG10-positive cases in which both a ductal with the rapid regulation of the MYC gene in the cycling cells of carcinoma in situ and an invasive component could be evaluated, this growing tissue and the greater stability of PEG10 mRNA than and in 8 of these cases, immunoreactive PEG10 was either MYC mRNA.

Table 1. Immunostaining for PEG10 and c-MYC in human breast and prostate cancers

(A)

Total PEG10 positive %

Ductal carcinoma in situ 11 6 54.5 Invasive carcinoma 22 7 31.8

(B)

Total c-MYC positive PEG10 positive c-MYC + PEG10 positive

161 58 (36%) 39 (24%) 25 (16%)

(C)

Total c-MYC positive PEG10 positive c-MYC + PEG10 positive

30 14 (47%) 11 (37%) 9 (30%)

NOTE: A, PEG10 protein (RF1) is frequently expressed in breast carcinomas, including ductal carcinoma in situ. Data are from 23 cases of breast carcinoma examined in standard histologic sections. B, PEG10 (RF1) and MYC proteins are positively, but not invariably, correlated in primary breast cancers. Data are from 161 cases of breast carcinoma examined as cores in tissue arrays. PEG10 expression dependent on MYC expression; P < 0.002, Fisher’s exact test. C, PEG10 (RF1) and MYC proteins are positively, but not invariably, correlated in primary prostate cancers. Data are from 30 cases of prostate carcinoma examined as cores in tissue arrays. PEG10 expression dependent on MYC expression; P = 0.014, Fisher’s exact test.

Cancer Res 2006; 66: (2). January 15, 2006 670 www.aacrjournals.org

Figure 5. PEG10 protein (RF1) expression in human breast and prostate carcinomas and in placental cytotrophoblast. Histologic sections from one patient with breast cancer are shown in (A-C) and from another patient in (D-F), immunostained with affinity-purified antipeptide antiserum recognizing PEG10 RF1. The nonneoplastic breast epithelium (A and D) is negative for PEG10 in both cases whereas the carcinoma in situ lesions (B and E) are positive. The invasive carcinoma remains positive in the first patient (C) but is negative for PEG10 in the second patient (F). A section of a PEG10-positive prostate carcinoma is shown in (G); the nonneoplastic prostate gland (N) is only faintly positive for PEG10 (in basal epithelial cells); the benign lymphoid infiltrate (Ly)is negative; and the adenocarcinoma cells (CaP) are strongly positive for cytoplasmic PEG10 at the luminal borders of the malignant glands. The numbers of PEG10-positive and PEG10-negative cases and correlations with immunoreactive c-MYC protein are shown in Table 1A to C. H and I, immunohistochemistry for c-MYC (H) and PEG10 (I) in human placenta. Arrows, proliferating villous cytotrophoblast layer, which is PEG10 and c-MYC positive; asterisks, overlying layer of postmitotic syncytiotrophoblast, which is PEG10 and c-MYC negative.

Discussion two candidate pathways—inhibition of apoptosis and inhibition of h Certain imprinted genes play a role in the initiation and TGF- signaling (5, 8, 9). progression of human cancers and this in part reflects a subversion We have also shown evidence for several factors that could of the normal functions of these genes in regulating tissue growth. explain the minority of primary cancers with discordant expression Epigenetic phenomena such as de novo DNA methylation and loss of PEG10 and MYC. Epigenetic silencing of Peg10 by promoter of imprinting have been well documented in preneoplastic and methylation can explain some cases of MMTV-MYC-induced malignant tissues (21) but the problem of how classic genetic mammary carcinomas that fail to express Peg10 mRNA whereas pathways may contribute to the dysregulated expression of other cases must reflect non-methylation-dependent silencing. In imprinted genes has received less attention. In this report, we the future, it will be interesting to study whether the loss of Peg10 describe a link between the classic proto-oncogene MYC and the expression in this group of mouse carcinomas is analogous to the proliferation-positive imprinted gene PEG10. PEG10 is imprinted in loss of expression of this gene that we have described in the invasive humans and mice, with the paternal allele active and maternal allele component of some human breast cancers in which the intraductal silent. PEG10 encodes a protein with distant similarity to retroviral component is PEG10 positive. There are precedents for cancer- gag-pol proteins and is expressed most strongly in the developing associated genes that are subject to different selective pressures in placenta but also in certain normal adult organs including the heart early (in situ) versus late (invasive) tumors, a notable example being h and lung (2, 3). The PEG10 gene is activated in a variety of human genes in the TGF- pathway. In a situation opposite to PEG10, these h cancers and data from manipulating the expression of this gene in TGF- pathway genes are selected for loss of function at early stages cancer cell lines, both in prior work (5, 8) and in the current study, but gain of function at late stages in the evolution of human indicate that it has a positive role in cell proliferation. Given the carcinomas. A factor that we can propose to explain some of the parental ‘‘direction’’ of its imprinting (paternal allele active), this human cancers that are PEG10 positive/c-MYC negative by growth-positive role for PEG10 would be predicted by the genomic immunohistochemistry is the long half-life of PEG10 mRNA in the conflict or kinship theory of imprinting (22). cell, contrasting with the short half-life of MYC mRNA. In addition, With this background, we were intrigued by finding PEG10 as a there are likely to be parallel pathways for activating PEG10, c-MYC-responsive gene in our MYC RNA interference experiments, possibly via other E-box binding proteins. Lastly, the antiserum used and as shown here, we have validated this gene as a direct in this study recognizes only PEG10 RF1, and in the future it will downstream target of c-MYC by additional functional, biochemical, be interesting to characterize the expression in cancer cells of the and correlative criteria (i.e., induction after c-MYC activation in alternative isoform of this protein (RF2), which can be produced EREB.TC-MYC cells, chromatin immunoprecipitation with anti- from the PEG10 mRNA transcript by translational frameshifting (4). c-MYC, E-box-dependent promoter activity in reporter assays, and positive correlations with c-MYC expression in a mouse model Acknowledgments of breast cancer and in human primary breast and prostate Received 5/5/2005; revised 10/13/2005; accepted 11/2/2005. Grant support: Department of Defense grant BC030061 (B. Tycko). carcinomas). The positive correlation of cytoplasmic PEG10 protein The costs of publication of this article were defrayed in part by the payment of page with nuclear c-MYC in a specific tissue compartment of the normal charges. This article must therefore be hereby marked advertisement in accordance human placenta further supports PEG10 as a c-MYC target gene. with 18 U.S.C. Section 1734 solely to indicate this fact. We thank Michelle Wei for affinity purification of anti-PEG10, Rongzhen Chen for How PEG10 protein acts in the cytoplasm to promote or support PEG10 immunohistochemistry, and Katia Basso and Brendan Chen in the Dalla-Favera cell proliferation is not yet certain but data have been reported for laboratory for helpful advice. www.aacrjournals.org 671 Cancer Res 2006; 66: (2). January 15, 2006

References and perinatal forms of biliary atresia. Hepatology 2004; regulation of iron-controlling genes, H-ferritin and 39:954–62. IRP2, by c-MYC. Science 1999;283:676–9. 1. Ono R, Kobayashi S, Wagatsuma H, et al. A retro- 8. Lux A, Beil C, Majety M, et al. Human retroviral gag- 15. Wu KJ, Grandori C, Amacker M, et al. Direct transposon-derived gene, PEG10, is a novel imprinted and gag-pol-like proteins interact with the TGF-h activation of TERT transcription by c-MYC. Nat Genet gene located on human chromosome 7q21. Genomics receptor ALK1. J Biol Chem 2004;280:8482–93. 1999;21:220–4. 2001;73:232–7. 9. Hu CS, Xiong J, Zhang LJ, et al. PEG10 activation by 16. Cardiff RD, Sinn E, Muller W, Leder P. Transgenic 2. Ono R, Shiura H, Aburatani H, Kohda T, Kaneko-Ishino co-stimulation of CXCR5 and CCR7 essentially con- oncogene mice. Tumor phenotype predicts genotype. T, Ishino F. Identification of a large novel imprinted gene tributes to resistance to apoptosis in CD19+CD34+ B Am J Pathol 1991;139:495–501. cluster on mouse proximal chromosome 6. Genome Res cells from patients with B cell lineage acute and chro- 17. Wulf G, Garg P, Liou YC, Iglehart D, Lu KP. Modeling 2003;13:1696–705. nic lymphocytic leukemia. Cell Mol Immunol 2004;1: breast cancer in vivo and ex vivo reveals an essential 3. Smallwood A, Papageorghiou A, Nicolaides K, et al. 280–94. role of Pin1 in tumorigenesis. EMBO J 2004;23:3397–407. Temporal regulation of the expression of syncytin 10. Grandori C, Cowley SM, James LP, Eisenman RN. 18. Aulmann S, Bentz M, Sinn HP. C-myc oncogene (HERV-W), maternally imprinted PEG10, and SGCE in The Myc/Max/Mad network and the transcriptional amplification in ductal carcinoma in situ of the breast. human placenta. Biol Reprod 2003;69:286–93. control of cell behavior. Annu Rev Cell Dev Biol 2000; Breast Cancer Res Treat 2002;74:25–31. 4. Manktelow E, Shigemoto K, Brierley I. Characteriza- 16:653–99. 19. Glockner S, Lehmann U, Wilke N, Kleeberger W, tion of the frameshift signal of Edr, a mammalian 11. Fernandez PC, Frank SR, Wang L, et al. Genomic Langer F, Kreipe H. Amplification of growth regulatory example of programmed 1 ribosomal frameshifting. targets of the human c-Myc protein. Genes Dev 2003;17: genes in intraductal breast cancer is associated with Nucleic Acids Res 2005;33:1553–63. 1115–29. higher nuclear grade but not with the progression to 5. Okabe H, Satoh S, Furukawa Y, et al. Involvement of 12. Zeller KI, Jegga AG, Aronow BJ, O’Donnell KA, Dang invasiveness. Lab Invest 2001;81:565–71. PEG10 in human hepatocellular carcinogenesis through CV. An integrated database of genes responsive to the 20. Courjal F, Cuny M, Simony-Lafontaine J, et al. Mapping interaction with SIAH1. Cancer Res 2003;63:3043–8. Myc oncogenic transcription factor: identification of of DNA amplifications at 15 chromosomal localizations 6. Tsou AP, Chuang YC, Su JY, et al. Overexpression of a direct genomic targets. Genome Biol 2003;4:R69. in 1875 breast tumors: definition of phenotypic groups. novel imprinted gene, PEG10, in human hepatocellular 13. Li CM, Guo M, Borczuk A, et al. Gene expression in Cancer Res 1997;57:4360–7. carcinoma and in regenerating mouse livers. J Biomed Wilms’ tumor mimics the earliest committed stage in 21. Feinberg AP, Tycko B. The history of cancer Sci 2003;10:625–35. the metanephric mesenchymal-epithelial transition. Am epigenetics. Nat Rev Cancer 2004;4:143–53. 7. Zhang DY, Sabla G, Shivakumar P, et al. Coordinate J Pathol 2002;160:2181–90. 22. Haig D. Genomic imprinting and kinship: how good expression of regulatory genes differentiates embryonic 14. Wu KJ, Polack A, Dalla-Favera R. Coordinated is the evidence? Annu Rev Genet 2004;38:553–85.

Cancer Res 2006; 66: (2). January 15, 2006 672 www.aacrjournals.org

Chi-Ming Li, Adam A. Margolin, Martha Salas, et al.

Cancer Res 2006;66:665-672.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/2/665

Cited articles This article cites 22 articles, 5 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/2/665.full#ref-list-1

Citing articles This article has been cited by 9 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/2/665.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/66/2/665. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.