Open Full Page

Molecular Cancer Signaling and Regulation Research

Expression Regulation of the Metastasis-Promoting Protein

InsP3-Kinase-A in Tumor Cells

Lydia Chang1, Heidi Schwarzenbach2,Sonke€ Meyer-Staeckling2, Burkard Brandt2, Georg W. Mayr1, Joachim M. Weitzel3, and Sabine Windhorst1

Abstract – Under physiologic conditions, the inositol-1,4,5-trisphosphate (InsP3)-metabolizing, F-actin bundling InsP3- kinase-A (ITPKA) is expressed only in neurons. Tumor cells that have gained the ability to express ITPKA show an increased metastatic potential due to the migration-promoting properties of ITPKA. Here we investigated the mechanism how tumor cells have gained the ability to reexpress ITPKA by using a breast cancer cell line (T47D) with no expression and a lung carcinoma cell line (H1299) with ectopic ITPKA expression. Cloning of a 1,250-bp ITPKA promoter fragment revealed that methylation of CpG islands was reduced in H1299 as compared with T47D cells, but DNA demethylation did not alter the expression of ITPKA. Instead, we showed that the repressor-element-1–silencing transcription factor (REST)/neuron-restrictive silencer factor (NRSF), which suppresses expression of neuronal genes in nonneuronal tissues, regulates expression of ITPKA. Knockdown of REST/NRSF induced expression of ITPKA in T47D cells, whereas its overexpression in H1299 cells strongly reduced the level of ITPKA. In T47D cells, REST/NRSF was bound to the RE-1 site of the ITPKA promoter and strongly reduced its activity. In H1299 cells, in contrast, expressing comparable REST/NRSF levels as T47D cells, REST/NRSF only slightly reduced ITPKA promoter activity. This reduced suppressor activity most likely results from expression of a dominant-negative isoform of REST/NRSF, REST4, which impairs binding of REST/NRSF to the RE-1 site. Thus, ITPKA may belong to the neuronal metastasis-promoting proteins whose ectopic reexpression in tumor cells is associated with impaired REST/NRSF activity. Mol Cancer Res; 9(4); 497–506. 2011 AACR.

Introduction mers, ITPKA bundles actin filaments, resulting in increased size and/or number of dendritic spines (4, 5). In tumor cells InsP3-kinase-A (ITPKA) activity phosphorylates the cal- that have gained the ability to express ITPKA, the F-actin cium-mobilizing second messenger D-myo-inositol-1,4,5- bundling activity of ITPKA, and its interaction with the F- trisphosphate [Ins(1,4,5)P ]toD-myo-inositol-1,3,4,5-tet- actin cross-linking protein filamin C, induces the formation 3 – rakisphosphate [Ins(1,3,4,5)P4] and thus is involved in the of new filopodia and lamellipodia (6 8). As formation of regulation of calcium signaling (1). Under physiologic these cellular protrusions is a prerequisite for cells to conditions, ITPKA is expressed only in neurons of the migrate, the actin-modulating activity of ITPKA increases hippocampus, neocortex, and cerebellum (GeneAtlas, the migratory and the metastatic potential of tumor cells (6). V133A gcma, BioGPS; ref. 2) where it accumulates in In addition, ITPKA promotes migration by its ITPK dendritic spines via its association with F-actin (3). Because activity mediating the induction of store-operated calcium of this association and its characteristic to form homodim- entry and with this the stimulation of calcium-dependent actin-modulating proteins (6). Current studies on lung cancer patients show that adenocarcinoma cells express high Authors' Affiliations: 1Institut fur€ Biochemie und Molekularbiologie I: Zellulare€ Signaltransduktion; 2Institut fur€ Tumorbiologie, Universitatsklini-€ levels of ITPKA in the primary tumors and in metastasis, kum Hamburg-Eppendorf, Hamburg; and 3Leibniz-Institut fur€ Nutztierbio- whereas ITPKA expression was mainly found in metastasis logie, Dummerstorf, Dummerstorf, Mecklenburg-Vorpommern, Germany in small cell and squamous carcinoma (7). Furthermore, it Note: Supplementary data for this article are available at Molecular Cancer has been shown that only particular tumor cell lines could Research Online (http://mcr.aacrjournals.org/). express ITPKA (6, 7). However, the mechanism underlying Corresponding Author: Sabine Windhorst, Institut fur€ Biochemie this tumor cell and stage-specific regulation of ITPKA und Molekularbiologie I: Zellulare€ Signaltransduktion, Universitatskli-€ nikum Hamburg-Eppendorf, Martinistr. 52, D-20246 Hamburg, expression has not been elucidated yet. Germany. Phone: 49-40-7410-56341; Fax: 49-40-7410-56818. E-mail: Because the knowledge of regulation of ITPKA expres- [email protected] sion in tumor cells may provide the possibility to inhibit doi: 10.1158/1541-7786.MCR-10-0556 reexpression of ITPKA, here we examined the mechanism 2011 American Association for Cancer Research. underlying expression regulation of ITPKA in tumor cells.

www.aacrjournals.org 497

Materials and Methods Transient transfections pCMV6-vector (as a control) and pCMV6-XL4 REST Cell culture (for REST overexpression) were gifts from Pranela Ramesh- NCI-H1299 (H1299) cells were kindly provided by war (Newark, NJ). Transfections were done by using Cagatay Gunes€ (Hamburg, Germany) and T47D cells by Lipofectamine reagent (Invitrogen), according to the man- Udo Schumacher (Hamburg, Germany). T47D cells are ufacturer's instructions. Transfected cells were cultured for mammary epithelial gland ductal carcinoma cells derived 48 hours, lysed with M-PER Buffer (Thermo Scientific), or from pleural effusion. They are estrogen receptor positive fixed with 3% paraformaldehyde. and express mutated p53 gene. MCF-7 cells are mammary epithelial gland adenocarcinoma cells derived from pleural Western blot effusion. They are estrogen receptor positive but lack To detect proteins, standard SDS-PAGE and immuno- expression of Her2/neu. NCI-H1299 cells are epithelial blot techniques were applied. Blots were developed using non–small cell lung cancer (NSCLC) cells derived from a the ECL Plus system (GE Healthcare), according to the lymph node metastasis. They have a homozygous partial manufacturer's instructions. Equal gel loading was exam- deletion of the p53 gene and lack expression of p53 (source: ined by glyceraldehyde 3-phosphate dehydrogenase American Tissue Culture Collection). The cell line NCI- (GAPDH) expression. Antibodies were purchased from H1299 was cultured in DMEM, and T47D and MCF-7 Santa Cruz: GAPDH (Sc69778), cyclin A (Sc-596), and cells were grown in RPMI; both media were supplemented ITPKA (Sc69778). Specifity of the ITPKA antibody was with 10% (v/v) fetal calf serum, 3.97 mmol/L L-glutamine, evaluated in a former study (7). In addition, brain lysates of 100 mg/mL streptomycin, and 100 U/mL penicillin. Media ITPKA wild-type and knockout mice were analyzed by were purchased from Invitrogen. Western blotting (Supplementary Fig. S1).

Cloning, transfections, and luciferase assays RNA preparation, RT-PCR, and real-time PCR To amplify DNA fragments upstream of the ITPKA Total RNA was prepared from cell lines using the gene, standard PCRs were carried out in the presence of NucleoSpin RNAII Kit (Machery & Nagel). The recovery sequence-specific primers surrounded by XhoI/HindIII of RNA was quantified spectrophotometrically. One micro- sites for directional cloning. The resulting DNA frag- gram of RNA was synthesized into cDNA with Superscript ments were inserted into the reporter firefly luciferase III, reverse transcriptase, and oligo(dT) primers (all from vector pGL3-basic vector (Promega), in which the back- Invitrogen), according to the manufacturer's instructions. bone was previously digested with XhoI/HindIII, to The cDNA was used as template for semiquantitative or delete the coding sequence for the SV40 promoter. Thus, real-time PCR with gene-specific primers, according to the the ITPKA promoter fragments were cloned into a manufacturer's instructions (Roche). Primer sequences used promoter-less vector. The same promoter-less vector were as follows: served as a negative control to measure background Hsc70: upstream (GCT GCT GCT ATT GCT TAC activity. The altered pGL3 constructs were confirmed GGC); downstream (TGC TGG AAGGAG GGT ACG by sequencing. Transfections of H1299 and T47D cells CT) were carried out by using Lipofectamine reagent (Invi- ITPKA: upstream (AGC TGC AGG ACC TGC TCG trogen), according to the manufacturer's instructions. AT); downstream (TCC GTG GGA GCT TCA GGA After transfection, cells were cultured for 16 hours in TC) serum-free medium and lysed with Lysis Buffer (Pro- REST1: upstream (GCT ACA GTT ATG GCC ACC mega). Luciferase activity was determined with the Dual- CAG GTG AT); downstream (GGC TTC TCA CCC Glo Luciferase Assay Reagent (Promega), according to ATC TAG ATC CAC T) the manufacturer's instructions by using a Berthold REST4: upstream (CTACAT GGC ACA CCT GAA luminometer (Berthold). Luciferase activity values were GCA CAC); downstream (GGC TTT CAC CCA TCT corrected by internal normalization performing cotrans- AGA TCA CAC T) fection of the pGL4.74 vector (Promega) containing Cyclin D: upstream (CCG TCC ATG CGG AAG Renilla luciferase. Luciferase assays were carried out in ATC), downstream (GAA GAC CTC CTC CTC duplicate, and each experiment of transfection was per- GCA CT) formed at least 3 times. The coding region of REST4 was amplified by PCR from pCMV6-XL4 REST (see the following text), using the Determination of mRNA and protein half-life oligonucleotides GGT ACC ATG GCC ACC CAG For examination of mRNA half-life, H1299 cells were GTA ATG GGG (sense) and GGATCC TCA CCC seeded in 24-well plates and grown to 80% confluence. AAC TAG ATC ACA ACC TGA ATG AGT ACG Then, the cells were treated with 10 mg/mL actinomycin for CAT ATG (antisense) surrounded by restriction sites for 0, 1, 3, 6, 16, and 24 hours. The mRNA level was KpnI and BamHI for directional cloning into the plasmid determined by real-time PCR as described earlier. For each vector pEGFP-C1 (Clontech TakaraBio) to generate sample, experiments were done in duplicates. To analyze EGFP–REST4. protein half-life, cells grown in full medium were treated

498 Mol Cancer Res; 9(4) April 2011 Molecular Cancer Research

with 750 mmol/L cycloheximide for 0, 4, 10, and 24 hours. bated with 2 mL of antibody [specific for repressor-element- Cell extracts were prepared by using M-PER Buffer 1–silencing transcription factor (REST)/neuron-restrictive (Thermo Scientific). The protein concentration was quan- silencer factor (NRSF)] overnight. Then, 50 mL of protein tified by spectrophotometry by using a standard curve (Bio- G agarose (Roche Diagnostics) was added to obtain pro- Rad Protein Assay Dye Reagent Concentrate; Bio-Rad), and tein–DNA complexes of interest. IgG-coupled protein G 20 mg protein were applied to Western blot analysis. agarose was used as a control. The amount of specific immunoprecipitated DNA was analyzed by PCR amplifica- Electrophoretic mobility shift and EMSA supershift tion by using gene-specific primers. assay Nuclear extracts from H1299 cells were prepared using Downregulation of REST/NRSF by shRNA the CelLytic NuCLEAR Extraction Kit (Sigma), and pro- For stable knockdown of REST/NRSF expression in tein concentration was quantified by spectrophotometry by T47D cells a lentiviral transduction approach was using a standard curve (Bio-Rad Protein Assay Dye Reagent employed. Five different vectors (pLKO.1 shRNA; Sigma) Concentrate, Bio-Rad). Oligonucleotides were synthesized expressing short hairpin RNAs (shRNA) against REST/ as complementary single strands (Eurofins MWG Operon), NRSF were tested. The most effective downregulation of annealed in 180 mmol/L NaCl by equimolar mixing, boiled REST/NRSF was achieved with one of them (target for 5 minutes, and slowly cooled down to room temperature sequence: 50- CCG GCG AGT CTA CAA GTG TAT overnight. We used Klenow polymerase to catalyze the CAT TCT CGA GAA TGA TAC ACT TGT AGA CTC addition of nucleotides to the 30 end and added GTT TTT -30), and this vector was used for all further [a-32P]d-CTP (50 mCi) to nucleotide mixtures. Oligonu- experiments. As a control, cells were transduced with the cleotides were purified by centrifugation over a Sephadex G- control vectors (expressing GFP or scrambled shRNA). 50 QuickSpin column (Roche Molecular Biochemicals). Infection of cells was carried out as described (6). Sequences (only sense strand shown): [a-32P]d-CTP Sp1 probe, GAG GAG GAT CCA GGC CCC GCC CCC Sequencing of the REST/NRSF gene TGA CCT; [a-32P]d-CTP GSG probe, TGG AAG GAG Genomic DNA from H1299 cells was isolated by using TGG GCG GGG GGC GCG GCC GAG; [a-32P]d-CTP the DNAeasy Blood and Tissue Kit from Qiagen, according STAT3 probe, GGG GGG AAG GCA ATC CTT CTG to the manufacturer's protocol. To amplify exon 2, exon 3, GGG CTG CTC; [a-32P]d-CTP GATA3 probe, GGG and exon 4 of the REST/NRSF gene, 40 ng of genomic GAC ACT GAT AAG CAC TTT GCA GTG TCT; DNA were used as a template for standard PCRs. Exon 2 [a-32P]d-CTP NF-kB probe, GGG GGG GGG ACC (907 bp) was amplified by 3, exon 3 (84 bp) by 1, and exon CTC ACT CCT GTG GCA CTA; [a-32P]d-CTP REST/ 4 (6,004 bp) by 5 sequential PCR steps by using over- NRSF probe, GGG GTG CGC GGG GAC AGG GCC lapping primer pairs. The PCR products were gel extracted TGT GGA GCG; [a-32P]d-CTP NGFI-C probe GGG using the QiaQuick Extraction Kit (Qiagen). The purified TGG GCG GGG GGC GCG GCC GAG GCG GGG. DNA (100 ng) was labeled with fluorescence oligonucleo- For electrophoretic mobility shift assay (EMSA), 0.500 tides by PCR, using the Bigdye Terminator v3.1 Cycle mg poly(dI/dC) from Sigma-Aldrich and binding buffer Sequencing Kit with 5 Bigdye Terminator Seq buffers (500 mmol/L NaCl, 1 mmol/L EDTA, 10 mmol/L dithio- (Applied Biosystems) and sequenced on the 3130 Genetic threitol, 5 mmol/L MgCl2 1 mg/mL bovine serum albumin, Analyzer from Applied Biosystems. Finally, the sequence 50% glycerol, 100 mmol/L Tris-HCl, pH 7.5) were com- was aligned to the reference human REST/NRSF sequence bined with nuclear extracts and incubated for 15 minutes at by using the databank http://genome.ucsc.edu/. Each room temperature. To verify specificity of oligonucleotide– sequencing reaction was repeated at least 3 times. protein complex, unlabeled oligonucleotides of the same sequence were used as a specific competitor. Mutated Identification of DNA methylation sites by bisulfite oligonucleotides were generated by scrambling the sequence treatment of DNA and inhibition of methylation by 5- and employed as a further specificity control. Furthermore, aza-CdR antibodies were added to verify the DNA-bound protein by Bisulfite conversion of 500 ng DNA was done using a inducing a supershift. The incubation mixtures were then commercially available kit (Qiagen). After treatment, 1 mL applied to a 5% polyacrylamide Tris borate EDTA gel and of the converted DNA was amplified in 20 mL reaction run at 240 V for 2 hours at 4C. The gels were vacuum heat mixtures containing 10 mL Taq PCR Master Mix (Qiagen), dried, and bands were visualized by exposure to autoradio- 0.5 mL of each primer (10 mmol/L), and 8 mL nuclease- graphic film (GE Healthcare) for 12 to 48 hours at 80C. free water. Primers were designed on the basis of the nucleotide sequence of the 50 upstream region of the ITPKA Chromatin immunoprecipitation assay gene submitted by the human ensemble database H1299 cells (1 106) were cross-linked with 37% (ENSG00000137825) upstream (GGA GGG ATA AAG formaldehyde solution (1% final concentration). The GGA TTG TTT ATA) downstream (CTA TCC CCA cross-linked protein–chromatin material was then sonicated TAA CCC AAC CTA AT). 5 times by a 10-second burst, with 1-minute cooling, to For inhibition of DNA methylation, cells grown in obtain 500-bp DNA fragments, which were then incu- full medium were treated with 1 mmol/L 5-aza-CdR

www.aacrjournals.org Mol Cancer Res; 9(4) April 2011 499

(5-aza-20-deoxycytidine) for 72 hours. Cell extracts were prepared by using the MPER (mammalian protein extraction reagent; Thermo Scientific). The protein concentration was quantified by spectrophotometry by using a standard curve (Bio-Rad Protein Assay Dye Reagent Concentrate; Bio-Rad), and 20 mg proteins were applied to Western blot analysis.

Results

Half-life of ITPKA protein and mRNA We found ectopic expression of the neuron-specific isoform A of ITPK in particular tumor cell lines and revealed that this upregulated expression increased their metastatic potential (6–8).Therefore,weexaminedthe mechanism underlying regulation of the ITPKA level. For this approach, we used the tumor cell line H1299 which has been shown to ectopically express ITPKA (6, 8). In addition, a cell line (T47D) that does not express ITPKA wasexamined,asunderphysiologicconditionsITPKAis not expressed in nonneuronal cells (GeneAtlas, V133A gcma; ref. 2). The characteristics of both cell lines are described in Materials and Methods. First, we examined whether the expression level of ITPKA in H1299 cells is comparable with the physiologic expression level in brain. Figure 1 shows that the ITPKA-specific band does not co- migrate in lysates from tissue and cell lines which most likely result from posttranslational modifications of ITPKA in cell lines. In addition, the data depicted in this figure reveal that the protein level in H1299 cells is upregulated in comparison with T47D cells but still about 6-foldlowerthanthatobservedinbrain.Onthebasisof these data, we conclude that in H1299 cells as well, the level of ITPKA is more restrictively regulated than under physiologic conditions. To examine a potential regulation of the ITPKA level by degradation of the protein, the half-life of ITPKA was analyzed in comparison to a protein with short half-life (cyclin A) and with long half-life (GAPDH). Figure 1B Figure 1. Regulation of ITPKA protein and mRNA levels. A, expression shows that ITPKA has a long half-life similar to the levels of ITPKA in brain and tumor cell lines. Protein extracts (10 mg) from proteinofthehousekeepinggeneGAPDH,makingit mouse brain, T47D, and H1299 cells were analyzed by Western blot for unlikely that the ITPKA protein concentration in H1299 expression of ITPKA. GAPDH signals served as a loading control. B, half- life of ITPKA protein. Top, Western blot analysis of ITPKA expression cells is regulated by degradation. Next, half-life of ITPKA before () and after (þ) treatment of H1299 cells with 750 mmol/L mRNA in H1299 cells was analyzed relative to the house- cycloheximide. Bottom, cyclin A and GAPDH expression before and after keeping gene Hsc70 and the short half-life mRNA of treatment with cycloheximide was analyzed as controls by Western cyclin A (Fig. 1C). This analysis revealed that the half- blotting. C, half-life of ITPKA mRNA. H1299 and T47D cells were treated m life of ITPKA mRNA was comparable with that of the with actinomycin D (10 g/mL), and after different incubation times, total Hsc70 t > RNA was prepared and transcribed into cDNA. Half-life of ITPKA, and housekeeping gene (with 1/2 24 hours). Hsc70 and cyclin A as controls, was measured by real-time PCR using Thus, both protein and mRNA of ITPKA have relative gene-specific primers. Three independent experiments were done in long half-lives. duplicates, and data represent mean values SD. D, cDNAs derived from From these results, we conclude that the protein con- H1299 and T47D cells were prepared, and the mRNA level of ITPKA was determined by semiquantitative RT-PCR. Three independent experiments centration of ITPKA is not primarily regulated by degrada- were done in duplicates, and 1 representative analysis is shown. tion of either the protein or the mRNA. Therefore, a potential transcriptional regulation of ITPKA was investigated by comparing the ITPKA mRNA level of H1299 cells tase PCR (RT-PCR). A clear signal was detected in H1299 with that of T47D cells (Fig. 1D). The mRNAs extracted cells, whereas no signal was visible in T47D cells. This result from H1299 and T47D cells were transcribed into cDNAs strongly indicates that the ITPKA level in these tumor cells and their ITPKA levels were analyzed by reverse transcrip- is regulated by gene expression.

500 Mol Cancer Res; 9(4) April 2011 Molecular Cancer Research

Figure 2. Cloning of a promoter fragment of the ITPKA gene. Serum-starved H1299 cells were transiently transfected with the pGL3-ITPKA-1250 construct, the original pGL3-basic vector (which includes the SV40 promoter), and with the promoter less pGL3-basic vector. For normalization, each construct was cotransfected with the pGL4 vector, coding for Renilla luciferase. Promoter activity was expressed as a ratio relative to the promoter-less construct (set to 1) and normalized to Renilla luciferase expression. Data represent mean values SD of 3 independent experiments.

Cloning of the human ITPKA promoter Negative regulation of ITPKA promoter activity The result showing that regulation of ITPKA is con- As no mutations in ITPKA-1250 were detected, we trolled at the transcriptional level let us to examine whether analyzed potential dysfunction of suppression of ITPKA upregulated expression of ITPKA in H1299 cells may result promoter activity in H1299 cells. Because methylation of from modulation of promoter activity. First, potential CpG dinucleotides is a common mechanism to silence gene mutations of promoter sequences were examined. We expression (9), we analyzed whether ITPKA promoter cloned a 1,250-bp DNA fragment of the hitherto unchar- activity may be regulated by DNA methylation. Therefore, acterized promoter of ITPKA (designated further on as the abundance of CpG dinucleotides in ITPKA-1250 was ITPKA-1250) upstream of the ATG start codon into a examined (Fig. 3A). A high density of CpG islands were promoter-less pGL3-basic vector. After transfection of detected at nucleotide (nt) 55 to 228, 323 to 388, H1299 cells’ promoter activity of the vector construct 392 to 531, and 671 to 919. Subsequently, the (pGL3-ITPKA-1250) was compared with activity of the methylation pattern of the ITPKA promoter in H1299 and SV40 promoter (the original pGL3-control vector) and the T47D cells was analyzed. As a positive control, MCF-7 cells promoter-less pGL3-basic vector. The 1,250-bp fragment (for characterization of this cell line, see Materials and showed a clear promoter activity, which, was, however, 10- Methods) were used because it has been shown that gene fold lower than that of the SV40 promoter (Fig. 2). Sequen- expression of these cells is strongly regulated by DNA cing and alignment by the human ensemble database methylation (10). Twenty-six methylated CpGs in MCF- revealed no mutations in the 1,250-bp DNA fragment. 7 cells, 19 in T47D cells, and 12 in H1299 cells were

Figure 3. Effect of DNA methylation on expression of ITPKA. A, schematic representation of the ITPKA-1250, illustrating the predicted putative CpG dinucleotides. B, after treatment with bisulfite, the predicted CpG dinucleotides were amplified by PCR and sequenced to determine the methylation pattern. C, analysis of ITPKA expression before and after treatment with 5-aza-CdR (Aza). Cells were treated with 1 mmol/L 5- aza-CdR for 72 hour, and protein levels of ITPKA and GAPDH were analyzed by Western blotting.

www.aacrjournals.org Mol Cancer Res; 9(4) April 2011 501

identified in the ITPKA promoter. In addition to this cance level 0.95) motif search. Potential binding sites for observation, we found conserved methylated CpGs present Sp1 (nt 89 to 144), NGFI-C (nerve growth factor- at nt 38, 41, 88, 94, 336, and 341 in all 3 cell induced clone C, nt 134 to 144), REST/NRSF (nt lines (indicated by triangles in Fig. 3B). The result that the 385 to 404), GATA3 (trans-acting T-cell–specific tran- lowest DNA methylation frequency was found in H1299 scription factor, nt 1,030 to 1,038), NF-kB (nt 1,152 cells points to upregulation of ITPKA expression by reduced to 1,163), and STAT3 (nt 1,208 to 1,125) were DNA methylation in this cell line. predicted (Fig. 4A). To determine whether these transcrip- To test this hypothesis, DNA methylation was inhibited tion factors may bind to the predicted binding sites, the by 5-aza-CdR in H1299, T47D, and MCF-7 cells and EMSA technique was used followed by a supershift analysis ITPKA expression was examined by analyzing its protein (for details, see Materials and Methods). The supershift levels (Fig. 3C). If expression of ITPKA is controlled by analysis revealed that only the antibodies directed against DNA methylation, 5-aza-CdR will induce ITPKA expres- Sp1 and REST/NRSF formed complexes (Fig. 4B, arrow B) sion in MCF-7 and T47D cells and expression in H1299 with DNA–protein complexes (Fig. 4B, arrow A) obtained cells should increase. However, we found that ITPKA by EMSA. Thus, the transcription factors Sp1 and REST/ expression was induced only in MCF-7 cells in response NRSF, but not STAT3, NGFI-C, GATA3, or NF-kB, may to 5-aza-CdR–mediated inhibition of DNA methylation be involved in the regulation of the ITPKA gene. but no detectable differences between treated and untreated Sp1 is a very well-characterized transcription factor cells were detected in H1299 and T47D cells. From these (11) known to positively control activity of TATA-box- data, we conclude that only in MCF-7, but not in H1299 less promoters by meditating the binding of TATA-box and T47D cells, transcription of ITPKA is mainly regulated binding protein to DNA and RNA polymerase II. The by hypermethylation of the ITPKA promoter. result that Sp1 binds to the ITPKA promoter, together with our observation that the only TATA-box-like Binding of transcription factors to the ITPKA promoter sequence found in the ITPKA promoter lies very distant On the basis of the result that expression of ITPKA is not from the ATG start codon (Fig. 4A), strongly indicates exclusively regulated by DNA methylation in T47D and that the ITPKA promoter belongs to the TATA-box-less H1299 cells, regulation of ITPKA promoter activity by promoters. In contrast to Sp1, REST/NRSF is a tran- transcription factors was investigated. Therefore, transcrip- scription suppressor. It suppresses transcription of neu- tion factor binding sites in ITPKA-1250 were analyzed by a ronal genes in nonneuronal tissues by recruitment of Softberry NSITE motif search (core similarity 0.8, signifi- histone deacetylase (HDAC) complexes. REST/NRSF

Figure 4. Binding of transcription factors to the ITPKA promoter. A, schematic representation of ITPKA-1250 illustrating the predicted motifs for transcription factors. B, EMSA of DNA–protein complexes formed with H1299 cell nuclear extract proteins (1 mg) and the predicted REST/NRSF and Sp1 binding sites derived from the ITPKA gene promoter. The variable components of each binding reaction are listed below the respective lane in the gel. A 20- fold molar excess of competitor DNA [wild-type (wt) and mutated (mut)] was included in the binding reaction. The specific A (DNA– protein complex) and B (DNA– protein–antibody complex) complexes are indicated by arrows.

502 Mol Cancer Res; 9(4) April 2011 Molecular Cancer Research

contains a DNA-binding domain, which binds to the background promoter activity and showed that pGL3- conserved DNA response element RE-1, and 2 repressor ITPKA-1035 and pGL3-ITPKA-813 exhibited similar luci- domains (12). The N-terminal domain recruits mSin3A/ ferase activities. On the other hand, isolation of the Sp1 HDAC, and the C-terminal domain the CoREST com- binding fragment (pGL3-ITPKA-168) increased promoter plex. These complexes suppress activation of genes by activity by 63 17% in H1299 and by 30 9.3% in histone deacetylation (13). T47D cells as compared with pGL3-ITPKA-1035. Isolation of the REST/NRSF binding fragment (pGL3-ITPKA-587) Comparison of functional ITPKA promoter fragments decreased activity by 36 4.5% in H1299 and by 75 between H1299 and T47D cells 6% in T47D cells. In T47D cells also the region flanking To analyze promoter activity of the identified Sp1 and the REST/NRSF binding fragment (pGL3-ITPKA-437) REST/NRSF binding fragments between H1299 and showed significant reduced promoter activity relative to T47D cells, these fragments were isolated. Subsequently, pGL3-ITPKA-1035. To control the specificity of pGL3- their luciferase activity was compared with larger (pGL3- ITPKA-587, point mutations in the RE-1 site were con- ITPKA-1035, pGL3-ITPKA-813) and shorter (pGL3- ducted to prevent binding of REST/NRSF. The luciferase ITPKA-50) fragments of the ITPKA promoter, whereby activity of this fragment was measured in H1299 and T47D luciferase activity of pGL3-ITPKA-1035 was set to 100%. A cells in comparison with pGL3-ITPKA-587 (Supplemen- comparison of relative luciferase activity of pGL3-ITPKA- tary Fig. S2). We found that mutation of the RE-1 site in 1035 between H1299 and T47D cells (Fig. 5A) shows that pGL3-ITPKA-587 increased promoter activity by 42 4% in T47D cells pGL3-ITPKA-1035 has only 27% of the in H1299 and by 72 6% in T47D cells, showing the ITPKA activity as that measured in H1299 cells, which strongly specificity of REST/NRSF binding to pGL3- -587. suggest that the low basal expression level of ITPKA in In summary our data show that the Sp1 binding fragment T47D cells (see Fig. 1) results from the reduction of pGL3-ITPKA-168 significantly enhances whereas the promoter activity. REST/NRSF binding fragment pGL3-ITPKA-587 lowers Analysis of the isolated promoter fragments (Fig. 5B) activity of pGL3-ITPKA-1035. Noteworthy, in T47D cells, revealed that in both cell lines the first 50 bp had only pGL3-ITPKA-587 showed a significant lower luciferase activity than that (P < 0.001) in H1299 cells.

Impact of REST/NRSF on expression of ITPKA Because we showed that isolation of the REST/NRSF binding fragment significantly reduced activity of the ITPKA promoter, we examined whether REST/NRSF may be sufficient to affect expression of ITPKA. There- fore, the transcription factor was overexpressed in H1299 and downregulated in T47D cells. For stable downregulation of REST/NRSF, 5 different shRNAs were tested and the most effective shRNA, mediating a downregulation of about 80%, was used in this study. Subsequently, T47D cells with downregulated and H1299 cells with upregulated REST/NRSF expression were analyzed for ITPKA protein levels by Western blotting. As shown in Figure 6A, overexpression of REST/NRSF in H1299 cells reduced expression of ITPKA by about 90% whereas its downregulation induced ITPKA expression in T47D cells (Fig. 6B). These results clearly show that REST/NRSF negatively regulates expression of ITPKA in H1299 and T47D cells. Figure 5. Promoter activity of ITPKA in T47D and H1299 cells. A, luciferase activity of the ITPKA promoter fragment pGL3-ITPKA-1035 was compared However, it was unclear why in T47D cells promoter between T47D and H1299 cells. pGL3-ITPKA-1035 was cotransfected activity and expression of ITPKA was nearly completely with the pGL4 vector, coding for Renilla luciferase. Promoter activity was suppressed by endogenous REST/NRSF but only partly in normalized to Renilla luciferase expression and expressed as a ratio H1299 cells. An explanation for the reduced activity of relative to the promoter-less construct (set to 1). Data represent mean 0 REST/NRSF in H1299 cells could be impaired binding of values SD of 4 independent experiments. B, 5 -deletional constructs of ITPKA the ITPKA promoter are illustrated. The numbers on the left of each REST/NRSF to the RE-1 site of the promoter. To promoter deletion construct refer to the beginning of the promoter test this presumption, a chromatin immunoprecipitation fragments. Each construct was transiently cotransfected with the pGL4 (ChIP) assay was done by using a REST/NRSF-specific vector into H1299 and T47D cells. Luciferase activity of the promoter antibody and binding of the RE-1 site of the ITPKA fragments was normalized to Renilla luciferase activity. To compare the normalized luciferase activities between T47D and H1299 cells, activity of promoter was analyzed by PCR. A clear signal was observed pGL3-ITPKA-1035 was set to 100%. Data represent mean values SD of in T47D cells (which was validated by DNA sequencing), at least 4 independent experiments. *, P < 0.05. whereas no signal in H1299 cells was detected (Fig. 6C).

www.aacrjournals.org Mol Cancer Res; 9(4) April 2011 503

antibody against REST4 is available, expression was examined at the mRNA level in comparison to the mRNA level of REST/NRSF. Our data show similar mRNA expression levels of REST/NRSF in T47D and H1299 cells, but expression of REST4 was detected only in H1299 cells (Fig. 7A). In a next step, we examined whether REST4 may influence expression of ITPKA in T47D cells by overexpressing an EGFP–REST4 fusion protein. Because of low transfection efficiency of T47D cells (10%–20%), expression of ITPKA was analyzed in single, paraformaldehyde-fixed cells by immunocytology (Fig. 7B). We detected ITPKA-specific signals only in cells overexpressing EGFP– REST4, which strongly indicates that REST4 is sufficient to induce expression of ITPKA.

Discussion

Recently, we detected ectopic expression of the neuron- Figure 6. REST/NRSF was overexpressed in H1299 (A) and specific ITPKA in certain tumor cell lines and types and downregulated in T47D cells (B), and lysates of these cells were analyzed showed that this pathologic expression of ITPKA increased for REST/NRSF and ITPKA expression by Western blotting. GAPDH – expression served as a loading control. o/e, overexpression. C, ChIP assay their metastatic potential (6 8). To provide the possibility so conducted with primers encompassing nt 240 to 587 of the ITPKA as to reduce the level of ITPKA in tumor cells, here we promoter. Control beads (IgG coupled) or anti-REST/NRSF antibody– examined the molecular mechanism underlying regulation of coupled beads were used to immunoprecipitate-associated chromatin. D, ITPKA. Therefore, a breast cancer cell line with no expres- analyses of REST/NRSF expression in H1299 and T47D cells by Western blotting. sion (T47D) and a lung carcinoma cell line (H1299) with

This result confirms that REST/NRSF binds to the ITPKA promoter in T47D cells and indicates disturbed binding of REST/NRSF in H1299 cells. As this finding could simply be due to reduced protein levels of REST/NRSF in H1299 cells, its level was compared between T47D and H1299 cells by Western blot analysis (Fig. 6D). However, no significant differences of the REST/NRSF levels between H1299 and T47D cells were found. To show whether mutations may cause impaired binding of REST/NRSF to the ITPKA promoter in H1299 cells, the coding exons (exons 2–4) of the REST/NRSF gene were sequenced using genomic DNA extracted from H1299 cells. We determined the sequence from the first base of exon 2 (nt 339, the ATG start codon is located at nt 348) up to the last base of exon 4. An alignment with the reference DNA sequence did not reveal any mutations within exons 2 to 4. These data show impaired binding of REST/NRSF to the ITPKA promoter in H1299 cells and that this dysfunction of REST/NRSF does not result from downregulation of REST/NRSF expression or mutation of the REST/NRSF gene.

Regulation of REST/NRSF-mediated suppression of ITPKA expression On the basis of our aforementioned data, we looked for Figure 7. Regulation of REST/NRSF suppressor activity. A, amplification alternative mechanisms that might lead to regulation of of REST/NRSF (REST1), REST4, and Hsc70 from cDNAs of H1299 and REST/NRSF activity. Shimojo and colleagues (14) reported T47D cells. Three independent experiments were done in duplicates, and 1 that an alternative splicing form of REST/NRSF, REST4, representative analysis is shown. B, REST4 was transiently overexpressed as EGFP fusion protein in T47D cells. ITPKA expression (red; lower and interacts with REST/NRSF and thereby prevents its bind- middle) was analyzed in single, paraformaldehyde-fixed cells by the use ing to the RE-1 site, leading us to analyze expression of of a primary anti-ITPKA and a secondary Alexa Flour568–coupled REST4 in H1299 and T47D cells. As no commercial antibody. o/e, overexpression.

504 Mol Cancer Res; 9(4) April 2011 Molecular Cancer Research

ectopic expression of ITPKA were examined. We rule out neuronal genes (16), the mechanisms leading to impaired that the level of ITPKA is mainly controlled by directed REST/NRSF activity in tumor cells are well characterized. In degradation of mRNA or protein, as both had long half-lives, colon cancer, deletions of the REST/NRSF gene locus are comparable with those of housekeeping genes. Instead, we frequent events (17); in breast and prostate cancer, decreased showed that ITPKA is regulated at the transcriptional level levels of REST/NRSF correlate with an aggressive cellular and cloned a fragment of the ITPKA promoter to analyze phenotype (17, 18). Furthermore, in NSCLC, loss of the regulation of ITPKA transcription. Because under physio- REST/NRSF corepressor complex BRM–BRG1 has been logic conditions expression of ITPKA in nonneuronal cells is described (19), and in small cell lung carcinoma, expression suppressed (2), we focused on the analysis of negative of REST4 was detected (16). REST4 exhibits an insertion of regulation of promoter activity. As DNA methylation of 16 bp in the region encoding the spacer between the zinc promoter elements is a widespread mechanism underlying finger motifs 5 and 6, leading to a translational frame shift. downregulation of tissue-specific gene expression (9), we This translated truncated form of REST/NRSF lacks a part investigated a potential negative regulation of ITPKA pro- of the DNA binding domain and the C-terminal suppressor moter activity by methylation of the identified CpG dinu- domain (12). Expression of this isoform inhibits binding of cleotides (see Fig. 3A). However, we revealed that 5-aza- REST/NRSF to RE-1 sites by interacting with the functional CdR–mediated inhibition of DNA methylation did not suppressor and thus reduces suppressor activity of REST/ increase expression of ITPKA in H1299 and T47D cells. NRSF (14). Here we show that REST4 is expressed in Only in the control cell line MCF-7, whose promoter regions H1299 but not in T47D cells and reveal that overexpression are highly methylated (10), expression of ITPKA was of REST4 in T47D cells induces expression of ITPKA. induced in response to treatment with 5-aza-CdR. This Thus, we conclude that expression of REST4 is sufficient to result could be explained by our finding that in MCF-7 induce reexpression of ITPKA in tumor cells. cells, the highest methylation frequency of the ITPKA In summary, we show that REST/NRSF substantially promoter was observed (26 methylated CpGs), with methy- contributes to negative regulation of ITPKA expression in lation clusters inside the Sp1 binding site. As we showed that tumor cells and reveal that ectopic expression of ITPKA in the activity of the TATA-box-less ITPKA promoter is posi- tumor cells is associated with impaired suppressor function tively controlled by Sp1, we assume that hypermethylation of of REST/NRSF due to expression of the dominant-negative the Sp1 binding site (15) is sufficient to inhibit the activity of REST/NRSF isoform REST4. On the basis of these data, the ITPKA promoter. These data clearly show that expression we conclude that expression of ITPKA may contribute to of ITPKA can be controlled by DNA methylation. On the the malignant phenotype observed in tumor cells with other hand, they reveal that silencing of the ITPKA gene impaired REST/NRSF function and suggest that reconsti- requires a high DNA methylation frequency as in T47D and tution of the REST/NRSF suppressor system may reduce H1299 cells, exhibiting a lower methylation frequency (19 malignancy of tumor cells. and 12 methylated CpGs, respectively) than MCF-7 cells, and ITPKA expression was not exclusively regulated by DNA Disclosure of Potential Conflicts of Interest methylation. On the basis of this observation, we concluded that in these cell lines, a further mechanism must be involved No potential conflicts of interest were disclosed. ITPKA in the transcriptional control of the gene. Acknowledgments Indeed, we found that REST/NRSF, which binds to the RE-1 site identified in the ITPKA promoter, negatively The authors thank Angelika Harneit for advice on EMSA assays and Andreas regulates ITPKA expression in T47D and H1299 cells, Bauche for excellent technical assistance. whereby only in T47D cells endogenous REST/NRSF suppressed promoter activity and expression of ITPKA. In Grant Support H1299 cells, in contrast, ITPKA expression and promoter This study was funded by the Gertrud und Erich Roggenbuck-Stiftung-0157/ activity were incompletely suppressed and only overexpres- 101, Hamburg, Germany. sion of functional REST/NRSF was sufficient to repress The costs of publication of this article were defrayed in part by the payment of page expression of ITPKA. Thus, expression of ITPKA in H1299 charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cells is associated with impaired REST/NRSF function. Because loss or dysfunction of REST/NRSF increases the Received December 7, 2010; revised February 16, 2011; accepted February 23, malignancy of tumor cells, due to increased expression of 2011; published OnlineFirst February 25, 2011.

References 1. Schell MJ. Inositol trisphosphate 3-kinases: focus on im- 3. Schell MJ, Erneux C, Irvine RF. Inositol 1,4,5-trisphosphate 3-kinase A mune and neuronal signaling. Cell Mol Life Sci 2010;67: associates with F-actin and dendritic spines via its N terminus. J Biol 1755–78. Chem 2001;276:37537–46. 2. Vanweyenberg V, Communi D, D’Santos CS, Erneux C. Tissue- and 4. Kim IH, Park SK, Hong ST, Jo YS, Kim EJ, Park EH, et al. Inositol 1,4,5- cell-specific expression of Ins(1,4,5)p3 3-kinase isoenzymes. Bio- trisphosphate 3-kinase A functions as a scaffold for synaptic Rac chem J 1995;306:429–35. signaling. J Neurosci 2009;29:14039–49.

www.aacrjournals.org Mol Cancer Res; 9(4) April 2011 505

5. Johnson HW, Schell MJ. Neuronal IP3 3-kinase is an F-actin-bundling 12. Palm K, Belluardo N, Metsis M, Timmusk T. Neuronal expression of protein: role in dendritic targeting and regulation of spine morphology. zinc finger transcription factor REST/NRSF/XBR gene. J Neurosci Mol Biol Cell 2009;20:5166–80. 1998;18:1280–96. 6. Windhorst S, Fliegert R, Blechner C, Mollmann€ K, Hosseini Z, 13. Lunyak VV, Burgess R, Prefontaine GG, Nelson C, Sze SH, Cheno- Gunther€ T, et al. Inositol 1,4,5-trisphosphate 3-kinase-A is a weth J, et al. Corepressor-dependent silencing of chromosomal new cell motility-promoting protein that increases the metastatic regions encoding neuronal genes. Science 2002;298:1747–52. potential of tumor cells by two functional activities. J Biol Chem 14. Shimojo M, Paquette AJ, Anderson DJ, Hersh LB. Protein kinase A 2009;285:5541–54. regulates cholinergic gene expression in PC12 cells: REST4 silences 7. Windhorst S, Kalinina T, Schmid K, Blechner C, Kriebitzsch N, Hinsch the silencing activity of neuron-restrictive silencer factor/REST. Mol R, et al. Functional role of inositol-1,4,5-trisphosphate-3-kinase-A for Cell Biol 1999;19:6788–95. motility of malignant transformed cells. Int J Cancer 2010 Nov 17. 15. Zhu WG, Srinivasan K, Dai Z, Duan W, Druhan LJ, Ding H, et al. [Epub ahead of print]. Methylation of adjacent CpG sites affects Sp1/Sp3 binding and 8. Windhorst S, Blechner C, Lin HY, Elling C, Nalaskowski M, Kirchber- activity in the p21(Cip1) promoter. Mol Cell Biol 2003;23:4056–65. ger T, et al. Ins(1,4,5)p3 3-kinase-A overexpression induces cytoske- 16. Coulson JM. Transcriptional regulation: cancer, neurons and the rest. letal reorganization via a kinase-independent mechanism. Biochem J Curr Biol 2005;15:R665–8. 2008;414:407–17. 17. Westbrook TF, Martin ES, Schlabach MR, Leng Y, Liang AC, Feng B, 9. Shames DS, Minna JD, Gazdar AF. DNA methylation in health, dis- et al. A genetic screen for candidate tumor suppressors identifies rest. ease, and cancer. Curr Mol Med 2007;7:85–102. Cell 2005;121:837–48. 10. Chekhun VF, Lukyanova NY, Kovalchuk O, Tryndyak VP, Pogribny IP. 18. Reddy BY, Greco SJ, Patel PS, Trzaska KA, Rameshwar P. RE-1- Epigenetic profiling of multidrug-resistant human MCF-7 breast ade- silencing transcription factor shows tumor-suppressor functions and nocarcinoma cells reveals novel hyper- and hypomethylated targets. negatively regulates the oncogenic TAC1 in breast cancer cells. Proc Mol Cancer Ther 2007;6:1089–98. Natl Acad Sci U S A 2009;106:4408–13. 11. Wierstra I. Sp1: Emerging roles—beyond constitutive activation of 19. Fukuoka J, Fujii T, Shih JH, Dracheva T, Meerzaman D, Player A, et al. Chro- TATA-less housekeeping genes. Biochem Biophys Res Commun matin remodeling factors and BRM/BRG1 expression as prognostic indi- 2008;372:1–13. cators in non-small cell lung cancer. Clin Cancer Res 2004;10:4314–24.

506 Mol Cancer Res; 9(4) April 2011 Molecular Cancer Research

Lydia Chang, Heidi Schwarzenbach, Sönke Meyer-Staeckling, et al.

Mol Cancer Res 2011;9:497-506.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/9/4/497

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2011/04/18/1541-7786.MCR-10-0556.DC1

Cited articles This article cites 18 articles, 12 of which you can access for free at: http://mcr.aacrjournals.org/content/9/4/497.full#ref-list-1

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://mcr.aacrjournals.org/content/9/4/497. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.