(RRP1B) Modulates Metastasis Through Regulation of Histone Methylation

Published OnlineFirst August 4, 2014; DOI: 10.1158/1541-7786.MCR-14-0167

Molecular Cancer Oncogenes and Tumor Suppressors Research

Metastasis-Associated Protein Ribosomal RNA Processing 1 Homolog B (RRP1B) Modulates Metastasis through Regulation of Histone Methylation

Minnkyong Lee1, Amy M. Dworkin1, Jens Lichtenberg1, Shashank J. Patel1, Niraj S. Trivedi2, Derek Gildea2, David M. Bodine1, and Nigel P.S. Crawford1

Abstract Overexpression of ribosomal RNA processing 1 homolog B (RRP1B) induces a transcriptional profile that accurately predicts patient outcome in breast cancer. However, the mechanism by which RRP1B modulates transcription is unclear. Here, the chromatin-binding properties of RRP1B were examined to define how it regulates metastasis-associated transcription. To identify genome-wide RRP1B-binding sites, high-throughput ChIP-seq was performed in the human breast cancer cell line MDA-MB-231 and HeLa cells using antibodies against endogenous RRP1B. Global changes in repressive marks such as histone H3 lysine 9 trimethylation (H3K9me3) were also examined by ChIP-seq. Analysis of these samples identified 339 binding regions in MDA- MB-231 cells and 689 RRP1B-binding regions in HeLa cells. Among these, 136 regions were common to both cell lines. Gene expression analyses of these RRP1B-binding regions revealed that transcriptional repression is the primary result of RRP1B binding to chromatin. ChIP-reChIP assays demonstrated that RRP1B co-occupies loci with decreased gene expression with the heterochromatin-associated proteins, tripartite motif-containing protein 28 (TRIM28/KAP1), and heterochromatin protein 1-a (CBX5/HP1a). RRP1B occupancy at these loci was also associated with higher H3K9me3 levels, indicative of heterochromatinization mediated by the TRIM28/HP1a complex. In addition, RRP1B upregulation, which is associated with metastasis suppression, induced global changes in histone methylation.

Implications: RRP1B, a breast cancer metastasis suppressor, regulates gene expression through heterochromatinization and transcriptional repression, which helps our understanding of mechanisms that drive prognostic gene expression in human breast cancer. Mol Cancer Res; 12(12); 1818–28. 2014 AACR.

Introduction tions (3). The sequential acquisition of random somatic Breast carcinoma is the most commonly diagnosed malig- mutations within the primary tumor is the most widely nancy in women in the United States, with over 235,000 accepted model of metastasis at the molecular level. How- new cases and over 40,000 deaths in 2012 (1). Metastasis is ever, it is becoming increasingly clear that although somatic mutation is the primary driver of metastasis, the propensity the main cause of death in breast cancer and metastatic fl disease is considered incurable (2). Tumorigenesis and of a primary tumor to metastasize is strongly in uenced by metastasis form a spectrum of progression that are induced interindividual germline variation (reviewed in ref. 4). by genetic alterations and disruption of epigenetic modifica- The consideration of metastasis susceptibility as a her- itable trait with a complex inheritance pattern has led to the identification of several germline-encoded metastasis modifier genes, including ribosomal RNA processing 1Genetics and Molecular Biology Branch, National Human Genome fi Research Institute, NIH, Bethesda, Maryland. 2Genome Technology 1 homolog B (RRP1B). RRP1B was identi ed as a germline Branch, National Human Genome Research Institute, NIH, Bethesda, susceptibility gene for breast cancer metastasis using Maryland. expression quantitative trait locus mapping in the FVB/ Note: Supplementary data for this article are available at Molecular Cancer N-Tg(MMTV-PyVT)634Mul/J mouse mammary tumor- Research Online (http://mcr.aacrjournals.org/). igenesis model (5). Specifically, two concurrent and inde- M. Lee and A.M. Dworkin contributed equally to this article. pendent experimental results identified Rrp1b as a novel fi fi Corresponding Author: Nigel P.S. Crawford, National Human Genome germline modi er of metastasis: rst, RRP1B is a binding Research Institute, NIH, 50 South Dr. Rm 5154, Bethesda, MD 20892. partner of the metastasis modifier SIPA1; second, Phone: 301-402-6078; Fax: 301-480-8711; E-mail: [email protected] RRP1B is a germline regulator of extracellular matrix gene doi: 10.1158/1541-7786.MCR-14-0167 expression, which is a class of genes frequently dysregu- 2014 American Association for Cancer Research. lated in tumors prone to metastasizing (5). Following its

1818 Mol Cancer Res; 12(12) December 2014

RRP1B Regulates Histone H3K9 Trimethylation

identification using modifier locus mapping, the proper- After viral infection, cells were selected in normal growth ties of Rrp1b were investigated by ectopic expression in the medium containing 10 mg/mL puromycin (Sigma-Aldrich), highly metastatic Mvt-1 mouse mammary tumor cell line. transferred to 96-well plates, and individual clones were These experiments demonstrated that Rrp1b dysregulation selected by limiting dilution. Ectopic expression of RRP1B in induced a gene expression signature that predicts survival single clonal isolate colonies was confirmed by qPCR as in multiple human breast cancer datasets with a high previously described (6). degree of reproducibility (5, 6). The relevance of RRP1B was further highlighted by the finding that a coding siRNA transfection germline polymorphism within human RRP1B is consis- Two different sequences of RRP1B Silencer Select siRNA tently associated with clinical outcome in multiple breast (Life Technologies, ID s22978 and s225927) were trans- cancer cohorts representing over 2,000 patients with fected in MDA-MB-231 cells using Lipofectamine RNAi- breast cancer (5, 7). MAX Transfection Reagent (Life Technologies) according Subsequent protein–protein interaction analyses revealed to the manufacturer's reverse transfection protocol. For several probable mechanisms by which Rrp1b dysregulation RNA isolation, cells were plated at 1 105 per 24-well has such profound and clinically relevant effects on global gene and collected 48 hours after transfection. For protein analysis expression (6). Most notably, these initial studies suggested and ChIP assays, cells were plated in 6-well and 150 mm that RRP1B is most likely a facultative heterochromatin plates at 2 105 per mL. protein because it colocalizes with several heterochromatin- associated histone marks. Concomitantly, RRP1B was shown Cell growth and invasion assays to physically interact with a number of heterochromatin- MDA-MB-231 clonal isolates infected with either HA- associated proteins, including tripartite motif-containing pro- RRP1B overexpression or control lentiviral particles were tein 28 (TRIM28; KAP1) and heterochromatin protein 1-a seeded in triplicate on 24 mm plates at a density of 105 cells a – (HP1 ;refs.811), which are potent inducers of gene per plate and incubated in 5% CO2 at 37 C for the indicated silencing. TRIM28 interacts with and recruits SETDB1 (SET periods of time. After incubation, the cells were harvested domain, bifurcated 1; ref. 9), a histone methyltransferase and the number of cells was counted. For soft agar growth which has been shown to colocalize with and mediate tri- assays, cells were plated in triplicate in 0.3% soft agar and methylation of histone H3 at lysine 9 (H3K9me3; ref. 12). allowed to grow for 7 days. Colonies were quantified after This creates high-affinity binding sites for the TRIM28/ staining with 0.005% crystal violet. Cell invasion assays were HP1a complex (13). H3K9me3 is a well-known marker of performed as previously described (18) using the MDA-MB- heterochromatin (14), and a strong association between 231 clonal isolate cell lines described above. TRIM28 and H3K9me3 has been reported (15, 16). In our current study, we aim to elucidate the mechanism Lung colonization assays by which RRP1B regulates metastasis-associated gene One million MDA-MB-231 RRP1B or control cells were expression. We utilized a variety of approaches to define intravenously injected via the tail vein of NU/J female mice the mechanism by which RRP1B suppresses gene expres- (Jackson Laboratory). Twenty-six mice were injected with sion. We have used chromatin immunoprecipitation (ChIP) MDA-MB-231 RRP1B cells and 11 mice were injected with assays to identify chromatin regions bound by RRP1B, and control cells. Lungs were harvested 90 days postinjection and demonstrated that these interactions are often associated pulmonary surface metastases were counted. All animal with binding of the transcriptional repressors HP1a and experiments were performed in compliance with the Nation- TRIM28. Furthermore, we demonstrate that an association al Human Genome Research Institute (Bethesda, MD) of these proteins at discrete genomic loci is concurrent with Animal Care and Use Committee's guidelines. H3K9me3 and downregulation of gene expression. Taken together, these data demonstrate that the clinically relevant Total RNA isolation changes in gene expression induced by RRP1B dysregulation Total RNA was isolated from cells using RNeasy Mini Kit and subsequent effects upon metastasis in breast cancer are, (QIAGEN). All samples were subjected to on-column at least in part, due to regulation of epigenetic mechanisms. DNase I digestion, and RNA quality and quantity were determined using NanoDrop 2000 (Thermo Scientific). Materials and Methods Only samples with A260/A280 ratios between 1.8 and 2.1 Cell culture were used for downstream analysis. MDA-MB-231 and HeLa cells were acquired from ATCC. All cell lines were maintained in DMEM with Microarray analysis 10% FBS and 1% penicillin-streptomycin (Gibco) and Microarray was performed with MDA-MB-231 overex- incubated in 5% CO2 at 37 C. pressing RRP1B and control clonal isolates as previously described (5). Briefly, total RNA was extracted from MDA- Lentiviral transduction MB-231 clonal isolates and processed using Affymetrix Lentiviral particles were generated as previously described GeneChIP Human Gene 1.0 ST array according to the (17) using the pDest-688 vector containing either human manufacturer's protocol. Data were analyzed using Partek hemagglutinin (HA)-tagged RRP1B or no insert as a control. Genomics Suite and heatmaps were generated using R (19)

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1819

Lee et al.

based on the most significant gene expression data. Data on the remainder of the library. Typically 12 to 14 cycles were submitted to Gene Expression Omnibus (accession were required. The PCR reaction was cleaned up using number GSE53979). Agencourt Ampure Xp beads (Beckman Coulter). Libraries were quantitated by qPCR (KAPA Biosystems) and fl Chromatin immunoprecipitation sequenced on an Illumina GAiix on single read ow cells. ChIP assays were performed using a modified protocol Early data were generated using version 4 chemistry and later (20). Briefly, cells were fixed with 1% formaldehyde and data were generated using version 5 chemistry. Read lengths lysed with lysis buffer (50 mmol/L HEPES-KOH, pH 7.5, were 35 bases. 140 mmol/L NaCl, 1 mmol/L EDTA, 0.1% Triton X-100, 0.1% sodium deoxycholate, and protease inhibitors). Cell ChIP-sequencing analysis lysates were sonicated with a Daigger Ultrasonic Processor Read mapping. ChIP DNAs were sequenced in dupli- and Sonicator Rod for a total of 3 minutes with 30 seconds cate. ChIP-seq reads were mapped to the hg18 human pulse and 20 seconds off cycles to shear the DNA to a final genome build and analyzed using the samtool suite. Using size of 200 to 500 base pairs. After preclearing with Protein G a database of repetitive elements (21) and the UCSC Sepharose beads (GE Healthcare), antibodies targeting Genome Browser track characterizing ribosomal RNAs either RRP1B (sc-83327, Santa Cruz Biotechnology), (22), a bowtie library (23) was generated for the character- H3K9me3 (ab8898, Abcam), or IgG (12-370, Millipore) ization of unmapped reads. The set of unmapped reads was were added to the lysate and incubated at 4C for 3 hours. then aligned to the repetitive regions of the human genome Protein G Sepharose beads were added and incubated using the bowtie tool. H3K9me3 ChIP-seq reads were overnight. The complex was washed twice with lysis buffer, mapped to the hg19 human genome build. once with high salt buffer (50 mmol/L HEPES-KOH, pH Peak calling. Using the Input reads as background for 7.5, 500 mmol/L NaCl, 1 mmol/L EDTA, 0.1% Triton HeLa and MDA-MB-231 cells, respectively, MACS (mod- X-100, 0.1% sodium deoxycholate), twice with LiCl buffer el-based analysis of ChIP-seq, version 1.4.1) was used to (10 mmol/L Tris-HCl, pH 8.0, 0.25 mol/L LiCl, 0.5% determine peaks under a statistical significance range of P ¼ NP-40, 0.5% sodium deoxycholate, 1 mmol/L EDTA), and 10 3 to 10 7. For H3K9me3 ChIP-seq data, Sole-Search once with TE buffer, followed by elution in TE buffer (Version v2) was used. Visual inspection of the different peak containing 1% SDS. Cross-links were then reversed and profiles allowed for the identification of relevant replicates the DNA was purified with QIAquick PCR purification kit and P value settings. (QIAGEN). The DNA obtained was analyzed either by Peak partitioning. Peaks were partitioned using the high-throughput sequencing or qPCR. RefSeq gene annotations available for the UCSC Genome For ChIP-reChIP, MDA-MB-231 lysates were collected Browser. The resulting genomic partitions include proximal as described above and incubated overnight with the first promoters (1 kb upstream to 50 bp downstream of TSS), antibody, either against TRIM28 (ab22553, Abcam) or distal promoters (10 kb upstream to 1 kb upstream of TSS), HP1a (05-689, Millipore), and Protein G Sepharose beads. downstream regions (10 kb directly downstream of the Immunocomplexes were washed twice with lysis buffer, once transcription end site) as well as coding regions, exons, with high salt buffer and once with TE buffer, followed by introns, untranslated, and intergenic regions. In addition elution. The eluted samples were diluted with lysis buffer to using genomic partitions, peaks were categorized using and incubated with the second antibody, anti-RRP1B, for 3 UCSC Genome Browser track annotations for noncoding hours at 4C. Protein G Sepharose beads were then added and ribosomal RNAs, alternative events, the top scoring 1% and incubated overnight with rotation at 4C. The following and 5% conserved elements and DNaseI hypersensitive sites, day, samples were processed according to the ChIP protocol. as well as track data derived for the location of CpG Islands All immunoprecipitations were performed in triplicate. (24) and repetitive elements (21). Venn diagrams were generated using R (19). High-throughput sequencing Approximately 100 ng of ChIP DNA with a fragment size qRT-PCR gene expression analysis of 200 bp was used to produce a library compatible with cDNA was synthesized from RNA isolated from MDA- Illumina sequencing on a GAiix. Single-end adapters and MB-231 cells using the iScript cDNA Synthesis Kit (Bio- sample DNA were input into a Beckman SPRI-TE robot Rad) following the manufacturer's protocol. Real-time along with SPRI-TE reagents for automated library con- qPCR was performed to detect cDNA levels of RRP1B struction. Postconstruction size selection was performed and a variety of genes (Supplementary Table S1) using an using a 2% agarose gel stained with SYBR-gold and imaged ABI 7900 Sequence Detection System (Life Technolo- on a Dark Reader. A band representing the range 250 to 350 gies). Reactions were performed using ABI SYBR Green bp was cut out and gel extracted using QIAquick Gel Master Mix as per the manufacturer's protocol. The Extraction Kit (QIAGEN). To ensure minimal amplifica- cDNA level of each gene was normalized to GAPDH tion, test PCR amplifications were set up for each library. cDNA levels using custom-designed primers for SYBR Aliquots were removed at every two cycles from six through green-amplified target genes. All ChIP-qPCR samples 16 cycles and analyzed for yield on a 2% agarose gel. On the were amplified in duplicate. Statistical significance was basis of these results, large-scale amplification was performed calculated by Student t test.

1820 Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research

RRP1B Regulates Histone H3K9 Trimethylation

Results RRP1B is due to changes in cellular growth rate, invasiveness, and migratory capacity. Ectopic expression of RRP1B in a human breast cancer fi cell line significantly affects cell growth, invasiveness, Microarray analyses of these cells depicted a signi cant motility, and lung colonization change in gene expression due to the overexpression of Previous studies have shown that knockdown of Rrp1b RRP1B with 219 downregulated and 73 upregulated genes increases the metastatic capacity of Mvt-1 and 4T1 mouse compared with that of the control (Fig. 2). These data mammary tumor cell lines (25). To gain a more compre- support the conclusion that the decrease in metastasis and hensive understanding of the role of this gene in metastasis, changes in cell behavior were associated with changes in gene we used lentiviral-mediated ectopic expression of RRP1B in expression elicited by RRP1B. Analysis of the associated the human breast cancer cell line MDA-MB-231. Over- functions of these genes using Ingenuity Pathway Analysis expression of RRP1B was confirmed through Western blot- (IPA, www.ingenuity.com) indicated that cellular growth ting and qRT-PCR (Supplementary Fig. S1). Ectopic and proliferation were among the top associated networks expression of RRP1B significantly decreased cell growth rates (Supplementary Table S2). compared with control cell lines (Fig. 1A). In soft agar assays, fewer colonies were observed with the cells overexpressing RRP1B binds to various genic regions within the genome RRP1B compared with the control cell lines (Fig. 1B). When RRP1B was previously shown to interact with multiple seeded in Transwells, cells overexpressing RRP1B displayed nucleosome-binding proteins, implying that RRP1B may be decreased invasiveness compared with the control cell lines involved in chromatin modification, which in turn could (Fig. 1C). In addition, decreased pulmonary metastases were account for a proportion of the transcriptional changes seen observed with the overexpression of RRP1B when intrave- with dysregulation of RRP1B (6). However, the consequence nously injected into the tail vein of NU/J female mice (Fig. of these interactions and the precise role of RRP1B binding is 1D). These data, which complement our earlier studies (25), unclear. The aim of these experiments was to use ChIP- confirm that RRP1B is a metastasis suppressor, and that the sequencing (ChIP-seq) to define regions of chromatin less metastatic phenotype induced by overexpression of bound by endogenous RRP1B. We hypothesized that

AB 10 MDA-MB-231 cell growth 120 MDA-MB-231 colony formation 9 8 , P < 0.050 100 P < 0.001 7 80 cells/mL)

6 6 5 4 60 3 RRP1B Control 2 40 Average colony count colony Average

Cell count (10 1 0 20 123456 0 Time (days) RRP1B Control

CDMDA-MB-231 cell invasiveness Pulmonary surface metastasis count 5 80 70 4 P = 0.020 P = 0.008 60 3 50 2 40

30 1 Metastasis count 20 0 10 Average percentage invasion percentage Average 0 –1 RRP1B Control RRP1B Control

Figure 1. RRP1B suppresses metastasis in vivo and decreases cell invasiveness in vitro. A, cell proliferation assay. B, soft agar assay. C, Transwell invasion assays. D, pulmonary surface metastases from intravenous injection of RRP1B overexpressing cells and control cells in NU/J mice. All data represent the average of three biologic replicates.

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1821

Lee et al.

Color key

681012 Value

Figure 2. Overexpression of RRP1B in MDA-MB-231 cells causes a significant change in gene expression. Global gene expression was quantified in five clonal isolates of MDA-MB-231 cells ectopically expressing RRP1B and control. Heatmap represents relative gene expression levels in the five clonal isolates of MDA-MB-231 cells ectopically expressing RRP1B and control detected by microarray analysis (WT, RRP1B; ctrl, control). WT9 WT5 WT8 WT7 ctrl3 ctrl4 ctrl5 ctrl6 ctrl1 WT10

RRP1B Control

RRP1B induces prognostic changes in gene expression and 231 and HeLa cells are shown in Table 1. These data suppresses metastasis by interacting with these chromatin- demonstrate that RRP1B occupancy peaks were compar- associated factors at target gene sites. We performed genome- atively evenly distributed relative to genic position in both wide ChIP-seq in the MDA-MB-231 breast cancer cell line cell lines. and confirmed in HeLa cells using an antibody targeting Of the peaks identified by ChIP-seq, eight common endogenous RRP1B. Although not a breast cancer cell line, RRP1B-binding regions were randomly selected for valida- we deemed ChIP-seq analysis in HeLa cells to be of impor- tion by ChIP-qPCR. A genomic region on chromosome 8 tance since earlier protein–protein interaction studies in with no evidence of RRP1B binding was selected as a this cell line defined RRP1B as an interactor of chromatin- negative control. In both MDA-MB-231 and HeLa cells, associated factors (6). A total of 339 peaks binding RRP1B we confirmed significant enrichment of RRP1B at binding were identified in MDA-MB-231 cells and 680 peaks were regions compared with the IgG control, and no significant identified in HeLa cells. Among these, 203 and 544 peak enrichment of RRP1B was detected at the negative control regions were unique to MDA-MB-231 and HeLa, respec- region (Fig. 3C and D). To confirm the specificity of the tively, and 136 peak regions were common to both cell antibody used for ChIP-seq, we performed ChIP-qPCR lines (Fig. 3A). To define RRP1B occupancy in relation to assays using an alternative RRP1B antibody, which was gene position, the locations of RRP1B-binding peaks were generated using a different epitope of this protein. RRP1B partitioned into four bins relative to RefSeq gene coordi- binding was confirmed at the same eight peak regions using nates; distal promoter, proximal promoter, downstream, this alternative RRP1B antibody, thus demonstrating the and genic regions (Fig. 3B). The locations of RRP1B binding specificity of the antibody used to generate global occupancy peaks relative to gene position in MDA-MB- ChIP-seq data (Supplementary Fig. S2).

1822 Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research

RRP1B Regulates Histone H3K9 Trimethylation

Figure 3. RRP1B interacts with multiple chromatin regions. A, Venn diagram of unique and common binding regions detected by ChIP-seq in MDA-MB-231 and HeLa cells. B, division of chromatin regions based on relative genic locations. C, veriﬁcation of common RRP1B binding regions via ChIP-qPCR in HeLa cells. D, MDA-MB-231 cells. , P 0.05, compared with IgG controls.

RRP1B associates with potent inducers of downofRRP1B.Thismaybeduetoachangeinan heterochromatinization to suppress gene expression upstream regulator of these genes as a result of silencing Next, we examined how RRP1B binding to chromatin RRP1B. influences gene expression at these loci. In MDA-MB-231 Given that RRP1B binding is frequently associated with clonal isolates ectopically expressing RRP1B,theexpres- reduced gene expression (Fig. 2), we chose to focus on the sion of 40 genes with RRP1B occupancy peaks either mechanism underlying these observations. We hypothesized within 5 kb up- or downstream was quantified by qPCR that the decreased gene expression observed with RRP1B (Supplementary Table S3-1). Transcriptional repression overexpression is due to increased association of RRP1B with was observed for the majority of genes that were dysre- the heterochromatin-associated proteins, TRIM28 and gulated in RRP1B overexpressing cells compared with that HP1a, at the chromatin region of these genes. We per- of the control. Genes that displayed a significant decrease formed ChIP-reChIP-qPCR assays to examine whether in comparison with the control were selected for further RRP1B shares common occupancy with either TRIM28 or investigation (Fig. 4 and Supplementary Table S3-2). To HP1a at binding peaks. Two consecutive immunoprecipi- examine how changes in RRP1B expression levels influ- tations were performed; first using either a TRIM28 or enced the transcription of genes with nearby RRP1B HP1a antibody, followed by a second immunoprecipitation occupancy peaks, MDA-MB-231 cells were transfected using an RRP1B antibody. The majority of transcriptionally with one of two different siRNA sequences targeting repressed genes with a nearby RRP1B-binding peak had an human RRP1B. Knockdown of RRP1B was confirmed associated region of TRIM28 and/or HP1a binding. For by qPCR and Western blotting (Supplementary Fig. S1). TRIM28, co-occupancy with RRP1B was detected at sig- qRT-PCR analysis demonstrated that the expression of the nificant levels at six of the twelve binding regions within 5 kb majority of these genes, with the exception of SYNJ2 and of genes that displayed a decreased expression in RRP1B- c-MYC, was rescued and increased significantly compared overexpressing cells; c-MYC, DNM2, EGFR, NCOR2, with that of the RRP1B overexpressing cells (Fig. 4). RAB5B, and SP1 (Fig. 5A). For HP1a, co-occupancy with Expression of SYNJ2 and c-MYC decreased in comparison RRP1B was determined to be significant at all binding with the control with both the overexpression and knock- regions that were examined, other than SYNJ2 (Fig. 5B).

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1823

Lee et al.

Table 1. Analysis of RRP1B-binding regions hypothesized that the association of these proteins to the common chromatin regions would be accompanied by a detected through ChIP-seq concomitant increase in this repressive, heterochromatin- associated histone mark. To examine this, we performed Peak position relative to gene HeLa MDA-MB-231 ChIP-qPCR with an antibody against H3K9me3 with Distal promoter 150 64 MDA-MB-231 cells stably overexpressing RRP1B and con- Proximal promoter 156 54 trol cells. At RRP1B-binding regions where reduced gene Downstream 263 102 expression was observed in RRP1B-overexpressing cells, nine fi Genic 205 90 of the twelve genes examined displayed signi cantly Genic increased levels of H3K9me3 in the cells overexpressing 50 UTR 61 21 RRP1B compared with the control cells (Fig. 5C). At these Exons 60 13 same regions, with the loss of RRP1B expression via siRNA, fi Introns 129 74 there was a signi cant decrease in H3K9me3 levels com- 30 UTR 33 19 pared with that of the control (Fig. 5D). These data indicate that RRP1B–chromatin interactions are not only associated NOTE: Regions bound by endogenous RRP1B in MDA-MB- with decreased gene expression, but are also accompanied by 231 and HeLa cells were defined with MACS. Genomic transcriptionally repressive changes in histone methylation sequences were partitioned into four bins relative to RefSeq typically associated with binding of the TRIM28/HP1a genes; distal promoter, proximal promoter, downstream, heterochromatinization complex. and genic. The genic bin was further divided by 50 UTR, 0 On the basis of these observations, we chose to examine exons, introns, and 3 UTR. The number of peaks for each bin the global effects of RRP1B dysregulation on levels of is shown here. H3K9me3. We performed ChIP-seq with H3K9me3- immunoprecipitated samples in MDA-MB-231 cells stably overexpressing RRP1B and control cell lines. A global Both TRIM28 and HP1 have been shown to colocalize with increase in the number of H3K9me3-enriched regions was markers of gene silencing and recruit transcriptional core- observed in cells overexpressing RRP1B (Table 2). Analysis pressors (8–11, 16). These data demonstrate that a prom- using Sole-Search identified 78 unique peaks for control inent mechanism by which RRP1B reduces gene expression cells, 364 unique peaks for RRP1B-overexpressing cells and is through an association with the TRIM28/HP1a hetero- 190 common regions between the two groups (Fig. 5E and chromatin complex. Supplementary Table S4). It is expected that some regions, such as the 78 regions identified here, will have decreased RRP1B binding is associated with increased H3K9 levels of H3K9me3 with RRP1B overexpression. This is trimethylation supported by our microarray data and gene expression data, Given that both TRIM28 and HP1a bind to regions of which demonstrate that although the predominant effect of RRP1B occupancy, and that TRIM28/HP1a binding is RRP1B overexpression is transcriptional repression, there is a associated with an increase in H3K9me3 levels (15, 26), we subset of genes that exhibit increased expression.

Figure 4. RRP1B-binding regions are frequently associated with decreased gene expression. The consequences of RRP1B binding were examined in RRP1B overexpressing cells and compared with that of RRP1B knockdown cells. Fold change was calculated using the control for each treatment. RRP1B siRNA represents the average fold change of cells transfected with either one of the two different RRP1B siRNA sequences. , P 0.05; , P 0.005, compared with RRP1B overexpressed.

1824 Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research

RRP1B Regulates Histone H3K9 Trimethylation

Figure 5. RRP1B interacts with TRIM28 and HP1a at regions associated with H3K9me3 and decreased gene expression. A, ChIP-reChIP with TRIM28 and RRP1B antibodies in MDA-MB-231 cells. , P 0.05, compared with both IgG/TRIM28 and IgG/RRP1B controls. B, ChIP-reChIP with HP1a and RRP1B antibodies in MDA-MB-231 cells. , P 0.05, compared with both IgG/HP1a and IgG/RRP1B controls. C, H3K9me3 ChIP-qPCR in MDA-MB-231 clonal isolates overexpressing RRP1B and control cell lines. D, H3K9me3 ChIP-qPCR in MDA-MB-231 cells transfected with either RRP1B siRNA or negative control siRNA. E, Venn diagram of unique and common H3K9me3-enriched regions identiﬁed by ChIP-seq in MDA-MB-231 clonal isolates overexpressing RRP1B and control cell lines.

Discussion in ref. 30). However, the origins of these signatures and the Over the last decade, various multigene "poor-outcome" specific mechanisms regulating gene expression during breast cancer signatures have been identified (27–29). The tumorigenesis are not fully understood. prognostic power of such signatures has proven so great It is becoming increasingly clear that germline variation that they have been deployed in the clinic for risk stratifi- has a significant influence upon gene expression within the cation in newly diagnosed patients with breast cancer. primary tumor (31). It is also apparent that germline genetic Examples of such signature-based clinical tests include the variation strongly modulates the propensity of primary 21-gene Oncotype Dx and 70-gene MammaPrint (reviewed breast carcinomas to metastasize (reviewed in ref. 4), which

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1825

Lee et al.

Table 2. H3K9me3-enriched regions detected the H3K9me3-enriched regions unique to RRP1B overexpressing MDA-MB-231 cells were ZNF genes (Supple- through ChIP-seq mentary Table S4). On the basis of the data presented here, we conclude that RRP1B serves not only to localize TRIM28 Peak position relative to gene Control RRP1B and HP1a to specific genomic loci, but also increases Distal promoter 74 167 H3K9me3 at these same loci, which leads to transcriptional Proximal promoter 17 70 repression. These conclusions are supported by our RRP1B Downstream 159 355 knockdown experiments, which demonstrated that there Genic 336 663 was a decrease in H3K9me3 levels (Fig. 5D) coupled with no significant changes in TRIM28 or HP1a levels at these loci NOTE: ChIP-seq analysis of H3K9me3-enriched regions in compared with the control (Supplementary Fig. S3). More MDA-MB-231 cells stably overexpressing RRP1B and con- specifically, a lower level of RRP1B expression lessened its trol cell lines were identified using Sole-Search. ability to guide the TRIM28/HP1a heterochromatinization complex to target loci. RRP1B may be influencing both gene expression and is of significance given that more than 90% of deaths in metastasis through additional mechanisms as well. For breast cancer are a direct consequence of metastasis (2). example, it was apparent that activation of RRP1B sup- RRP1B was previously identified as a germline metastasis presses the expression of oncogenes (e.g., c-MYC) and suppressor that regulates gene expression in a manner that transcription factors with wide-ranging effects (e.g., SP1). accurately predicts survival in human breast cancer (5, 6). In Also, our work demonstrates that three genes exhibiting this report, we used a combination of ChIP analyses to define transcriptional repression (DNM2, PHLDA, and PTCH2) interactions between endogenous RRP1B and chromatin, do not exhibit a significant increase in H3K9me3 levels, as coupled with expression analyses. This approach has defined shown in Fig. 5C. This suggests that additional, and as yet the manner in which RRP1B interacts with chromatin, and uncharacterized, H3K9me3-independent mechanisms are has revealed one possible mechanism by which RRP1B in part responsible for RRP1B-mediated transcriptional regulates metastasis. repression. Furthermore, although the majority of genes ChIP-seq experiments in MDA-MB-231 and HeLa cells adjacent to RRP1B occupancy peaks did display reduced revealed that genomic regions occupied by RRP1B are highly expression upon RRP1B overexpression, some genes did diverse with RRP1B binding to a wide variety of elements in display increased expression (Fig. 2) and a reduction in relation to genic position. Given that these cell lines were H3K9me3 levels (Fig. 5E). These complex effects upon derived from two different tissues, confirmation of our gene expression are a reflection of the highly promiscuous ChIP-seq findings in additional breast cancer cell lines may manner in which RRP1B interacts with other proteins (34). be of importance. However, given the degree of RRP1B For example, it has been previously demonstrated that occupancy peak overlap between these two different cell RRP1B interacts with PARP1 (6), which is a known positive lines, as shown in Fig. 3A, we predict that a similar degree of regulator of transcription (35). In addition, RRP1B also peak overlap would be observed with other breast cancer cell colocalizes with markers of euchromatin (6). This suggests lines. Yet, such experiments are beyond the scope of our that RRP1B likely regulates transcription by interacting with current work. multiple transcriptional complexes, in addition to regulating Our interest in the TRIM28/HP1a heterochromatin- other cellular processes, such as ribosome biogenesis through associated complex stems from an earlier observation show- its interaction with protein phosphatase 1 (34). Supporting ing that both of these proteins physically interact with its role in rRNA processing, several nucleolar proteins (e.g., RRP1B, and colocalize with H3K9me3 (6). Both HP1a nucleophosmin, nucleolin) have been shown to interact with and H3K9me3 have been purified with several transcrip- RRP1B. Interestingly, previous studies reported that these tionally repressive protein complexes (13, 26, 32, 33). nucleolar proteins also regulate transcription by directly HP1a, a component of the HP1 tetramer, interacts with binding to chromatin (36, 37). RRP1B, which is primarily TRIM28 (8, 16), which was also identified as a RRP1B- located in nucleoli, most likely interacts indirectly with binding partner, primarily in the perinucleolar region (6). chromatin, as it does not possess any known DNA-binding TRIM28 assembles a macromolecular complex that contains motifs, and recruits chromatin-binding proteins as a means proteins such as Mi2a, SETDB1 (SET domain, bifurcated of transcriptional regulation. Given that heterochromatin is 1), and HP1, which acts to create a stable heterochromatic concentrated in the nucleus in perinucleolar clusters, it is not microenvironment (8–10, 13, 16). We found that RRP1B is surprising that the primarily nucleolar protein RRP1B acts as located near genes with decreased expression that also bound a regulator of heterochromatinization. TRIM28 and/or HP1a. In addition, higher levels of Tumor expression profiling has become a useful tool H3K9me3 were seen at most RRP1B-binding regions with for assessing prognosis in breast cancer; however, the origin reduced expression in adjacent genes. Previous studies have of these prognostic signatures remains somewhat unclear. In shown that TRIM28 and H3K9me3 are specifically 0 this study, we demonstrate that RRP1B regulates metastasis- enriched at the 3 end regions of zinc finger (ZNF) genes associated gene expression by interacting with the transcrip- (15, 26). Consistent with these findings, we found that 38 of tional corepressors TRIM28 and HP1a, which act by

1826 Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research

RRP1B Regulates Histone H3K9 Trimethylation

recruiting chromatin-modifying enzymes such as SETDB1. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- tational analysis): M. Lee, A.M. Dworkin, J. Lichtenberg, S.J. Patel, N.S. Trivedi, On the basis of previous studies and our data presented here, D. Gildea, D.M. Bodine, N.P.S. Crawford the manner in which RRP1B regulates both transcription Writing, review, and/or revision of the manuscript: M. Lee, A.M. Dworkin, S.J. Patel, D. Gildea, D.M. Bodine, N.P.S. Crawford and metastasis proves to be highly complex, and its impact Study supervision: N.P.S. Crawford upon both of these processes are not the result of a simplistic, linear effect on one signaling pathway. Rather, RRP1B Acknowledgments dysregulation influences both metastasis and prognostic The authors thank Dr. Kent Hunter for his critical review of this article and Dr. Abdel Elkahloun at the NHGRI Microarray Core Facility and Alice Young at the gene expression through a net effect on multiple pathways NIH Intramural Sequencing Center for their technical assistance. and biologic processes. Grant Support Disclosure of Potential Conflicts of Interest This work was supported by the Intramural Research Program of the National No potential conflicts of interest were disclosed. Human Genome Research Institute, NIH. The costs of publication of this article were defrayed in part by the payment of page Authors' Contributions charges. This article must therefore be hereby marked advertisement in accordance with Conception and design: M. Lee, A.M. Dworkin, N.P.S. Crawford 18 U.S.C. Section 1734 solely to indicate this fact. Development of methodology: A.M. Dworkin, D. Gildea, N.P.S. Crawford Acquisition of data (provided animals, acquired and managed patients, provided Received March 28, 2014; revised July 21, 2014; accepted July 21, 2014; facilities, etc.): M. Lee, A.M. Dworkin, S.J. Patel, D.M. Bodine, N.P.S. Crawford published OnlineFirst August 4, 2014.

References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer 14. Hublitz P, Albert M, Peters AH. Mechanisms of transcriptional repres- J Clin 2013;63:11–30. sion by histone lysine methylation. Int J Dev Biol 2009;53:335–54. 2. Scully OJ, Bay BH, Yip G, Yu Y. Breast cancer metastasis. Cancer 15. O'Geen H, Squazzo SL, Iyengar S, Blahnik K, Rinn JL, Chang HY , et al. Genomics Proteomics 2012;9:311–20. Genome-wide analysis of KAP1 binding suggests autoregulation of 3. Yoo CB, Jones PA. Epigenetic therapy of cancer: past, present and KRAB-ZNFs. PLoS Genet 2007;3:e89. future. Nat Rev Drug Discov 2006;5:37–50. 16. Sripathy SP, Stevens J, Schultz DC. The KAP1 corepressor functions 4. Hunter KW, Crawford NP. The future of mouse QTL mapping to to coordinate the assembly of de novo HP1-demarcated microenvir- diagnose disease in mice in the age of whole-genome association onments of heterochromatin required for KRAB zinc finger protein- studies. Annu Rev Genet 2008;42:131–41. mediated transcriptional repression. Mol Cell Biol 2006;26:8623–38. 5. Crawford NP, Qian X, Ziogas A, Papageorge AG, Boersma BJ, Walker 17. Alsarraj J, Walker RC, Webster JD, Geiger TR, Crawford NP, Simpson RC , et al. Rrp1b, a new candidate susceptibility gene for breast cancer RM , et al. Deletion of the proline-rich region of the murine metastasis progression and metastasis. PLoS genetics 2007;3:e214. susceptibility gene Brd4 promotes epithelial-to-mesenchymal transi- 6. Crawford NP, Yang H, Mattaini KR, Hunter KW. The metastasis tion- and stem cell-like conversion. Cancer Res 2011;71:3121–31. efficiency modifier ribosomal RNA processing 1 homolog B 18. Crawford NP, Alsarraj J, Lukes L, Walker RC, Officewala JS, Yang HH , (RRP1B) is a chromatin-associated factor. J Biol Chem 2009;284: et al. Bromodomain 4 activation predicts breast cancer survival. Proc 28660–73. Natl Acad Sci U S A 2008;105:6380–5. 7. Hsieh SM, Look MP, Sieuwerts AM, Foekens JA, Hunter KW. Distinct 19. Warnes GR, Bolker B, Bonebakker L, Gentleman R, Huber W, Liaw A , inherited metastasis susceptibility exists for different breast cancer et al. gplots: Various R programming tools for plotting data. R package subtypes: a prognosis study. Breast Cancer Res 2009;11:R75. version 2.11.3 2013 8. Nielsen AL, Ortiz JA, You J, Oulad-Abdelghani M, Khechumian R, 20. Huang J, Sengupta R, Espejo AB, Lee MG, Dorsey JA, Richter M , et al. Gansmuller A , et al. Interaction with members of the heterochromatin p53 is regulated by the lysine demethylase LSD1. Nature 2007; protein 1 (HP1) family and histone deacetylation are differentially 449:105–8. involved in transcriptional silencing by members of the TIF1 family. 21. Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, Walichie- EMBO J 1999;18:6385–95. wicz J. Repbase Update, a database of eukaryotic repetitive elements. 9. Schultz DC, Ayyanathan K, Negorev D, Maul GG, Rauscher FJ III. Cytogenet Genome Res 2005;110:462–7. SETDB1: a novel KAP-1-associated histone H3, lysine 9-specific 22. Fujita PA, Rhead B, Zweig AS, Hinrichs AS, Karolchik D, Cline MS , et al. methyltransferase that contributes to HP1-mediated silencing of The UCSC genome browser database: update 2011. Nucleic Acids euchromatic genes by KRAB zinc-finger proteins. Genes Dev Res 2011;39:D876–82. 2002;16:919–32. 23. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory- 10. Schultz DC, Friedman JR, Rauscher FJ III. Targeting histone dea- efficient alignment of short DNA sequences to the human genome. cetylase complexes via KRAB-zinc finger proteins: the PHD and Genome Biol 2009;10:R25. bromodomains of KAP-1 form a cooperative unit that recruits a 24. Illingworth RS, Gruenewald-Schneider U, Webb S, Kerr AR, James KD, novel isoform of the Mi-2alpha subunit of NuRD. Genes Dev Turner DJ , et al. Orphan CpG islands identify numerous conserved 2001;15:428–43. promoters in the mammalian genome. PLoS Genet 2010;6:e1001134. 11. Ryan RF, Schultz DC, Ayyanathan K, Singh PB, Friedman JR, 25. Lee M, Dworkin AM, Gildea D, Trivedi NS, Moorhead GB, Crawford NP. Fredericks WJ , et al. KAP-1 corepressor protein interacts and RRP1B is a metastasis modifier that regulates the expression of colocalizes with heterochromatic and euchromatic HP1 proteins: alternative mRNA isoforms through interactions with SRSF1. Onco- a potential role for Kruppel-associated box-zinc finger proteins in gene 2014;33:1818–27. heterochromatin-mediated gene silencing. Mol Cell Biol 1999; 26. Iyengar S, Ivanov AV, Jin VX, Rauscher FJ III, Farnham PJ. Functional 19:4366–78. analysis of KAP1 genomic recruitment. Mol Cell Biol 2011;31:1833–47. 12. Frietze S, O'Geen H, Blahnik KR, Jin VX, Farnham PJ. ZNF274 recruits 27. Ramaswamy S, Ross KN, Lander ES, Golub TR. A molecular signature the histone methyltransferase SETDB1 to the 30 ends of ZNF genes. of metastasis in primary solid tumors. Nat Genet 2003;33:49–54. PLoS ONE 2010;5:e15082. 28. van 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M , et al. 13. Iyengar S, Farnham PJ. KAP1 protein: an enigmatic master regulator of Gene expression profiling predicts clinical outcome of breast cancer. the genome. J Biol Chem 2011;286:26267–76. Nature 2002;415:530–6.

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1827

Lee et al.

29. van de Vijver MJ, He YD, van't Veer LJ, Dai H, Hart AA, Voskuil DW , 34. Chamousset D, De Wever V, Moorhead GB, Chen Y, Boisvert FM, et al. A gene-expression signature as a predictor of survival in breast Lamond AI , et al. RRP1B targets PP1 to mammalian cell nucleoli and is cancer. N Engl J Med 2002;347:1999–2009. associated with Pre-60S ribosomal subunits. Mol Biol Cell 2010;21: 30. Paik S. Is gene array testing to be considered routine now? Breast 4212–26. 2011;20:S87–91. 35. Krishnakumar R, Gamble MJ, Frizzell KM, Berrocal JG, Kininis M, 31. Li Q, Seo JH, Stranger B, McKenna A, Pe'er I, Laframboise T , et al. Kraus WL. Reciprocal binding of PARP-1 and histone H1 at pro- Integrative eQTL-based analyses reveal the biology of breast cancer moters specifies transcriptional outcomes. Science 2008;319: risk loci. Cell 2013;152:633–41. 819–21. 32. Shi Y, Sawada J, Sui G, Affar el B, Whetstine JR, Lan F , et al. 36. Gonzalez V, Guo K, Hurley L, Sun D. Identification and characterization Coordinated histone modifications mediated by a CtBP co-repressor of nucleolin as a c-myc G-quadruplex-binding protein. J Biol Chem complex. Nature 2003;422:735–8. 2009;284:23622–35. 33. Ogawa H, Ishiguro K, Gaubatz S, Livingston DM, Nakatani Y. A 37. Li Z, Boone D, Hann SR. Nucleophosmin interacts directly with c-Myc complex with chromatin modifiers that occupies E2F- and Myc- and controls c-Myc-induced hyperproliferation and transformation. responsive genes in G0 cells. Science 2002;296:1132–6. Proc Natl Acad Sci U S A 2008;105:18794–9.

1828 Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research

Metastasis-Associated Protein Ribosomal RNA Processing 1 Homolog B (RRP1B) Modulates Metastasis through Regulation of Histone Methylation

Minnkyong Lee, Amy M. Dworkin, Jens Lichtenberg, et al.

Mol Cancer Res 2014;12:1818-1828. Published OnlineFirst August 4, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-14-0167

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2021/03/13/1541-7786.MCR-14-0167.DC1

Cited articles This article cites 36 articles, 16 of which you can access for free at: http://mcr.aacrjournals.org/content/12/12/1818.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/12/12/1818.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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/12/12/1818. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.