sign in sign up
<<
Home , CENPA, Chromosome 5

Oncogene (2012) 31, 3244–3253 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE Cancer-associated alteration of pericentromeric may contribute to instability

RB Slee1,2, CM Steiner1, B-S Herbert1,2, GH Vance1,2, RJ Hickey2,3,5, T Schwarz4,6, S Christan4,7, M Radovich1, BP Schneider1,2,3, D Schindelhauer4,8 and BR Grimes1,2

1Department of Medical and Molecular , Indiana University School of Medicine (IUSM), Indianapolis, IN, USA; 2Indiana University Melvin and Bren Simon Cancer Center, Indianapolis, IN, USA; 3Department of Medicine, IUSM, Indianapolis, IN, USA and 4Department of Medical Genetics, Ludwig Maximilians University, Munich, Germany

Many tumors exhibit elevated chromosome mis-segrega- Oncogene (2012) 31, 3244–3253; doi:10.1038/onc.2011.502; tion termed chromosome instability (CIN), which is likely published online 28 November 2011 to be a potent driver of tumor progression and drug resistance. Causes of CIN are poorly understood but Keywords: pericentromeric heterochromatin; chromo- probably include prior tetraploidization, centro- some instability; JMJD2 demethylase some amplification and mitotic checkpoint defects. This study identifies epigenetic alteration of the as a potential contributor to the CIN phenotype. The centromere controls and consists Introduction of higher-order repeat (HOR) alpha-satellite DNA packaged into two domains: the , Cancer cells accumulate somatic and other harboring the centromere-specific H3 variant centromere genomic rearrangements, which impair the normal A (CENP-A), and the pericentromeric hetero- mechanisms controlling proliferation and differen- chromatin, considered important for cohesion. Perturba- tiation, culminating in malignancy. One mechanism to tion of centromeric chromatin in model systems causes accelerate the loss or gain of tumor suppressor CIN. As cancer cells exhibit widespread chromatin and oncogenes, which often accompanies tumorigenesis, changes, we hypothesized that pericentromeric chromatin is chromosome mis-segregation resulting in the gain or structure could also be affected, contributing to CIN. loss of entire (). The majority Cytological and chromatin immunoprecipitation and PCR of solid tumors exhibit numerical chromosomal (ChIP–PCR)-based analyses of HT1080 cancer cells abnormality (Mitelman et al., 1994). Although clonal showed that only one of the two HORs on chromosomes selection of rare aneuploid cells arising from basal 5 and 7 incorporate CENP-A, an organization conserved mis-segregation rates may partly account for tumor in all normal and cancer-derived cells examined. Con- aneuploidy (Cahill et al., 1999), many tumors are trastingly, the heterochromatin marker H3K9me3 (tri- thought to acquire an elevated mis-segregation rate methylation of H3 lysine 9) mapped to all four HORs and referred to as chromosomal instability (CIN), which has ChIP–PCR showed an altered pattern of H3K9me3 been proposed as a potent driver of tumor progression in cancer cell lines and breast tumors, consistent with and drug resistance (Kops et al., 2005). Likely causes of a reduction on the kinetochore-forming HORs. The CIN include multipolar mitoses following genome JMJD2B demethylase is overexpressed in breast tumors tetraploidization and centrosome amplification, chro- with a CIN phenotype, and overexpression of exogenous matid cohesion defects and impairment of the mitotic JMJD2B in cultured breast epithelial cells caused loss checkpoint pathways that ensure faithful chromosome of centromere-associated H3K9me3 and increased CIN. segregation (Thompson et al., 2010). However, the full These findings suggest that impaired maintenance of spectrum of changes underlying the CIN phenotype pericentromeric heterochromatin may contribute to CIN remains to be uncovered. This study presents new in cancer and be a novel therapeutic target. evidence implicating epigenetic impairment of the centromere as a contributing factor. Correspondence: Dr BR Grimes, Department of Medical and The centromere is the nucleoprotein complex at the Molecular Genetics, Indiana University School of Medicine, Indiana- core of the segregation machinery. Its functions include polis, IN 46202, USA. E-mail: [email protected] sister cohesion, attachment to the mitotic 5Current address: City of Hope, Duarte, CA, USA. spindle and management of the mitotic checkpoint. 6Current address: University of Wurzburg, Germany. assemble on the megabase-sized blocks 7Current address: Technical University of Munich, Germany. of alpha-satellite DNA located at the primary constric- 8Current address: Chromosome Medicine Procurement, Therese- tion of all chromosomes. Alpha-satellite DNA Studer-Str. 47, 80797 Munich, Germany. Received 15 August 2011; revised 25 September 2011; accepted 26 comprises tandem 171 bp monomers. Groups of mono- September 2011; published online 28 November 2011 mers are themselves reiterated, giving rise to a higher-order Pericentromeric heterochromatin changes in cancer RB Slee et al 3245 repeat (HOR) structure, which is unique to each required for proper chromosome segregation in mice centromere (Warburton and Willard, 1996). Despite (Peters et al., 2001), and the recently discovered JMJD2 this sequence heterogeneity, the basic chromatin archi- family of demethylases, which reverse trimethy- tecture of the centromere is well conserved and critical lation on H3K9 and K36 and have been implicated to its function (Verdaasdonk and Bloom, 2011). At the in tumorigenesis (Pedersen and Helin, 2010). JMJD2 attachment region (kinetochore) the core antagonize formation of pericentromeric histone protein H3 is partially substituted by a H3K9me3 in mouse cells (Fodor et al., 2006), whereas centromere-specific H3 variant, centromere protein A overexpression of JMJD2B or JMJD2C has been (CENP-A) (Palmer et al., 1987; Sullivan and Karpen, reported in a range of cancers including breast cancer 2004). Alpha-satellite DNA flanking the kinetochore (Liu et al., 2009; Pedersen and Helin, 2010; Yang et al., assembles the pericentromeric heterochromatin com- 2010). Here, we find that overexpression of exogenous partment, which folds between sister on JMJD2B in human breast epithelial cells can drive loss the metaphase chromosome, and is characterized by of centromere-associated H3K9me3 and increased CIN. abundant methylation of H3 at lysine 9 (H3K9) The possibility that disruption of pericentromeric (Sullivan and Karpen, 2004; Alonso et al., 2010). heterochromatin promotes CIN and cancer progression, CENP-A deposition provides a foundation for the by impairing centromere function, opens new research kinetochore (Vafa and Sullivan, 1997; Warburton avenues that may lead to therapeutic interventions et al., 1997), whereas methylated H3K9 serves as a targeting the affected pathways. docking site for the heterochromatin protein HP1 (Bannister et al., 2001), and has been implicated in centromeric cohesion at (Bernard et al., 2001; Results Guenatri et al., 2004; Alonso et al., 2010). The function of the centromere is therefore vulnerable to disruption Mapping the centromere 5 kinetochore in normal and of the pathways that maintain this unique chromatin cancer cells architecture. CENP-A overexpression is seen in breast A generic scheme of centromere organization showing and colon cancer and has been linked to a CIN assembly of CENP-A at the kinetochore and hetero- phenotype (Perou et al., 2000; Tomonaga et al., 2003; chromatin on flanking HOR alpha-satellite is shown in Amato et al., 2009), and engineered disruption of Figure 1a. To investigate how this scheme applies to an pericentromeric heterochromatin is associated with individual , we examined the centromere of chromosome mis-segregation and tumorigenesis (Ekwall , which harbors two HORs, D5Z1 and et al., 1997; Bernard et al., 2001; Peters et al., 2001; D5Z2 (Finelli et al., 1996). In silico analysis revealed David et al., 2003; Shin et al., 2003; Gonzalo et al., 2005; that only D5Z2 possesses intact CENP-B boxes David et al., 2006; Bourgo et al., 2009). Furthermore, (Rosandic et al., 2006) (not shown), a 17 bp motif targeting of chromatin modifiers to human artificial necessary for de novo centromere formation on human chromosomes disrupts centromere function (Nakano artificial chromosomes (Ohzeki et al., 2002). Consistent et al., 2008; Cardinale et al.,2009;Bergmannet al., 2011). with these observations, immunofluorescence (IF) Epigenetic mis-regulation in cancer is well showed that CENP-A colocalized almost exclusively documented, reflecting aberrant patterns of both histone with D5Z2 in chromosome spreads from the HT1080 and DNA methylation as well as histone acetylation. lung cancer cell line (in 22/27 chromosome 5 examples), Some of the enzymatic pathways involved have been suggesting that a portion of D5Z2 forms the kinetochore targeted therapeutically (Rodriguez-Paredes and (Figure 1b). The remainder of D5Z2 spans the inner Esteller, 2011). In contrast, the impact of epigenetic centromere region (Figure 1b). The infrequent overlap perturbation on centromere function in human cancer of D5Z1 with CENP-A (in 4/27 chromosome 5 remains largely unexplored. The available data support examples) is likely to reflect resolution limitations due the generic model of centromere structure outlined to chromosome compaction. Chromatin immunopreci- above, in which HOR alpha-satellite DNA is partitioned pitation and PCR (ChIP–PCR) analysis confirmed that into two adjacent chromatin domains. However, auto- CENP-A is restricted to D5Z2 and excluded from D5Z1 somal centromeres harbor multiple divergent HOR in HT1080 cells (Figure 1b). HT1080 cells have a near alpha-satellite blocks (Lee et al., 1997) whose contribu- tetraploid and exhibit chromosome instability tions to centromeric chromatin assembly are not well upon passage in culture (Grimes et al., 2004), which defined. The present study combines molecular and could reflect impaired centromere function. To compare cytological approaches to map the kinetochore and the organization of the chromosome 5 centromere in flanking heterochromatin domains to the underlying normal cells, the ChIP–PCR analysis was repeated HOR blocks on chromosomes 5 and 7, revealing a well on primary human umbilical vein endothelial cell conserved centromeric organization in normal cells and (HUVEC) preparations from three individuals. Again, tissues. In comparison, tumor-derived cells with a CIN CENP-A was restricted to D5Z2 (Supplementary Figure phenotype exhibited a profile of epigenetic change at S1). A similar CENP-A distribution was observed in both centromeres, consistent with a loss of kinetochore- both chromosomally unstable HeLa cervical cancer cells proximal trimethylated H3K9 (H3K9me3). H3K9me3 and HME5cdk4 normal diploid human breast epithelial maintenance depends on the opposing activities of cells (Supplementary Figure S1), suggesting that the site the SUV39H1/2 histone methyltransferases, which are of CENP-A/kinetochore assembly is conserved. The

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3246 HT1080 a b 0.04 H3K9me * D5Z1 0.03 D5Z2 D5Z2 0.02 rDNA CENP-A/ D5Z1 kinetochore 0.01 0 CENP-A beads D5Z2 D5Z1

c

normal 5 del 5 ring 5

D5Z2 (green)/ GDNF (red) D5Z1 (green)/ GDNF (red)

Figure 1 Localization of centromere protein A (CENP-A) to D5Z2. (a) Generic scheme of centromere structure. CENP-A (red double dots) is embedded within heterochromatin characterized by H3 lysine 9 methylation (H3K9me). (b) On chromosome 5, CENP-A is restricted to D5Z2 and excluded from the adjacent D5Z1 sequence as shown by immunofluorescence and fluorescence in situ hybridization (IF-FISH) and chromatin immunoprecipitation and PCR (ChIP–PCR) in HT1080 cells. Only D5Z2 (*) is significantly elevated compared with beads control (P ¼ 0.01). Schematic of CENP-A localization at centromere 5 is shown. (c) Analysis of patient sample BF. BF has a normal chromosome 5 (small arrow), a deleted chromosome 5 (del 5; arrowhead) and a 5 (large arrow). Ring 5 retains D5Z2 and the GDNF gene, but lacks detectable D5Z1. Chromosome 5 structure in BF cells is depicted: D5Z1 (black), D5Z2 (green), GDNF (red).

organization of CENP-A, D5Z1 and D5Z2 is summar- S3). This result supports our earlier study, showing that ized in Figure 1b. the heterochromatin protein HP1a, which binds to FISH analysis showed an absence of detectable D5Z1 H3K9me3, is often undetectable between the CENP-A sequence on a mitotically stable chromosome 5-derived signals in HT1080 cells (Grimes et al., 2004). Consistent ring chromosome in a patient sample (Figure 1c). The with the IF findings, ChIP–PCR with an H3K9me3 ring 5 chromosome, formed from a within the antibody showed reduction of H3K9me3 on the D5Z2 cen-5p13 region (Schuffenhauer et al., 1996), retained kinetochore sequence relative to both the flanking D5Z1 D5Z2 (Figure 1c) and has a kinetochore (Supplementary (P ¼ 3 Â 10À5) and GAPDH (P ¼ 0.04) sequences, when Figure S2). These findings are consistent with D5Z1 HT1080 cells were compared with HUVECs (Figures 2a representing a chromosome arm sequence outside of the and b). We next investigated whether the H3K9me3 functional centromere. distribution at the centromere is altered in two additional chromosomally unstable cancer cell lines (HeLa and HCC1937 (a BRCA1 deficient breast cancer Altered pericentromeric H3K9me3 distribution on cell line)) relative to four normal diploid controls (three chromosome 5 in cancer cell lines and invasive HUVEC samples and HME5cdk4 cells) using ChIP– breast tumors PCR. In comparison to the D5Z1 reference sequence, Pericentromeric heterochromatin is considered to be the tumor-derived cell lines exhibited consistent loss important for sister chromatid cohesion and is char- of H3K9me3 from the kinetochore-forming D5Z2 acterized by a high density of methylated H3K9, which sequence (Pp0.002; Figure 2b). is absent from the adjacent kinetochore/CENP-A The preceding analysis identifies an association domain (Sullivan and Karpen, 2004). Using IF in between reduced kinetochore proximal heterochromatin normal HUVECs, H3K9me3 was readily detectable on and a CIN phenotype in genetically unmatched cell metaphase chromosomes between the CENP-A/kineto- lines. To exclude population variation as the source for chore signals, and often extended into the surrounding epigenetic differences in centromere structure, the ChIP– chromatin (Figure 2a, Supplementary Figure S3). PCR comparison was repeated in a non-tumorigenic ChIP–PCR with an H3K9me3 antibody showed that cell line and its tumorigenic derivative. Breast epithelial both D5Z1 and D5Z2 precipitated with similar effi- cells were isolated from healthy tissue of a patient ciency and were enriched 15- and 17-fold relative to the harboring an inherited heterozygous p53 , GAPDH reference (Figure 2a). To test the possibility and human telomerase reverse transcriptase (hTERT) that pericentromeric heterochromatin is altered in was introduced to create the immortal, diploid and non- cancer cells, the dual CENP-A/H3K9me3 IF was tumorigenic cell line, HME50hTERT (Supplementary repeated in HT1080 cells. In contrast with HUVECs, Table S1; Herbert et al. in preparation). Addition of H3K9me3 was not detectable between the majority of H-RasV12 to HME50hTERT resulted in the HME50TR sister kinetochores (Figure 2a, Supplementary Figure cell line (Hochreiter et al., 2006). HME50TR cells

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3247

Figure 2 Altered H3K9me3 distribution on centromere 5 in cancer cell lines and invasive ductal carcinoma (IDC) breast tumors. (a) Dual centromere protein A (CENP-A) (green)/H3K9me3 (red) immunofluorescence (IF) and chromatin immunoprecipitation and PCR (CHIP–PCR) analysis of human umbilical vein endothelial cells (HUVECs) (left) and HT1080 cells (right) (see also Supplementary Figure S3). When compared with HUVECs, HT1080 cells show a significant reduction of H3K9me3 on D5Z2 relative to the flanking D5Z1 sequence (P ¼ 3 Â 10À5) and GAPDH (P ¼ 0.04) using ChIP–PCR with an H3K9me3 antibody. (b–d) ChIP–PCR analysis of H3K9me3 at centromere 5 in cell lines and tissues. Enrichment is shown relative to the D5Z1 arm reference that is set at 1. (b) Normal controls are HUVECs (N1) and HME5cdk4 (N2), whereas tumor-derived cell lines are HT1080 (T1), HeLa (T2) and HCC1937 (T3). All tumor-derived cell lines exhibited reduced H3K9me3 on D5Z2 relative to D5Z1 when compared with normal controls (Pp0.002). (c) and tumorigenicity in a breast cancer progression series (the HME50 series) are indicated. Cytogenetic characterization of HME50hTERT and HME50-T cell lines is provided in Supplementary Table S1. In the ChIP–PCR panel, N is the HME50hTERT cell line and T is HME50-T cell line. (d) Seven matched IDC tumor (T) /normal (N) breast tissue pairs were examined. The range of H3K9me3 enrichment on D5Z2 relative to D5Z1 is shown in Supplementary Figure S4. formed tumors in mice after passage through soft agar, chromosomes, we characterized a second centromere from which the HME50-T line was derived (Hochreiter on , where two HOR alpha-satellite et al., 2006) (Figure 2c). HME50-T exhibited a CIN sequences, D7Z1 and D7Z2, map to the primary phenotype with chromosome numbers ranging from constriction (Wevrick and Willard, 1991; Finelli et al., 74–87/cell (Supplementary Table S1). In addition, 1996). HOR units of both sequences were found to ChIP–PCR showed an altered chromatin profile at the contain intact CENP-B box sequences (not shown). chromosome 5 centromere in the tumor-derived cell line Interestingly, and consistent with a previous study (Haaf when compared to its diploid parent, with a reduction and Ward, 1994), IF-FISH on chromosome spreads in H3K9me3 on D5Z2 relative to D5Z1 (P ¼ 0.001; from HT1080 cells showed CENP-A localizing predo- Figure 2c). Thus, disruption of the pathways regulating minantly to D7Z1, while largely excluded from D7Z2 centromeric chromatin structure occurred concomi- (Figure 3a). D7Z1 overlapped with the CENP-A double tantly with transformation and acquisition of a CIN dot signals and also spanned the inner centromere phenotype. region in the majority (33/35) of chromosome 7 To explore the in vivo relevance of these findings, examples (Figure 3a). These data suggest that factors pericentromeric heterochromatin structure was analyzed additional to the CENP-B box influence the site of in a panel of invasive ductal carcinoma (IDC) breast CENP-A/kinetochore formation. Confirming the tumors and adjacent healthy breast tissue. Consistent IF-FISH results, ChIP–PCR showed that only D7Z1 with the cell line data, a more modest but significant was detectably enriched by precipitation with an anti- reduction of kinetochore proximal H3K9me3 on D5Z2 body to CENP-A, thus identifying D7Z1 as the location was detected at centromere 5 in IDC tumors (P ¼ 0.014; of the kinetochore (Figure 3a). This organization Figure 2d; Supplementary Figure S4). appeared conserved in all normal and tumor-derived In contrast to H3K9me3, the ChIP–PCR showed a samples (Supplementary Figure S1; not shown). The distribution of CENP-A at the chromosome 5 centro- relationship between D7Z1, the flanking D7Z2 sequence mere that was conserved among all the normal and and CENP-A is illustrated in Figure 3a. tumor-derived samples examined (Figure 1b; Supple- ChIP–PCR with an H3K9me3 antibody showed mentary Figure S1; not shown). similar enrichment for D7Z1 and D7Z2 relative to GAPDH in HUVECs (Figure 3b). Following the pattern at the chromosome 5 centromere, H3K9me3 Analysis of centromere 7 chromatin structure in normal was reduced on the kinetochore-forming D7Z1 sequence cells, cancer cells and IDC tumors in HT1080 cells relative to the flanking D7Z2 sequence To determine whether the altered chromatin structure at (P ¼ 0.005) and GAPDH (Po0.05) when compared with the centromere of chromosome 5 extends to other HUVECs (Figures 3b and c). Similar to chromosome 5,

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3248

Figure 3 Altered distribution of H3K9me3 on centromere 7 in cancer cell lines and invasive ductal carcinoma (IDC) breast tumors. (a) Immunofluorescence and fluorescence in situ hybridization (IF-FISH) in HT1080 cells mapped centromere protein A (CENP-A) to D7Z1. Overlap with D7Z2 was rarely observed (in 7/35 chromosome 7 examples). Chromatin immunoprecipitation and PCR (ChIP– PCR) in HT1080 cells confirmed that CENP-A is restricted to D7Z1. Only D7Z1 (*) is significantly elevated compared with beads control (P ¼ 0.002). Organization of D7Z1, D7Z2 and CENP-A on chromosome 7 is shown. (b) ChIP–PCR indicated that H3K9me3 is reduced on D7Z1 relative to D7Z2 (P ¼ 0.005) and GAPDH (Po0.05) in HT1080 cells compared with human umbilical vein endothelial cells (HUVECs). (c, d) ChIP–PCR analysis of H3K9me3 at centromere 7 in cell lines and tissues. Enrichment is shown relative to D7Z2 which is set to 1. (c) Normal cell lines are: HUVECs (N1), HME50hTERT (N2) and HME5cdk4 (N3). Tumor-derived cell lines are HT1080 (T1), HeLa (T2), HME50-T (T3) and HCC1937 (T4). All tumor-derived cell lines exhibited reduced H3K9me3 on D7Z1 relative to D7Z2 when compared with normal controls (Pp0.01). (d) Six matched IDC tumor/normal breast tissue pairs were examined. The range of H3K9me3 enrichment on D7Z1 relative to D7Z2 in normal breast tissue (N) and IDC tumors (T) is shown in Supplementary Figure S4.

a wider comparison of tumor and normal cells revealed 17 revealed aneuploidy and instability for both chromo- an altered pericentromeric heterochromatin profile with somes based on deviation from the modal chromosome consistent loss of H3K9me3 from D7Z1 relative to number (Figure 4c). D7Z2 (Po0.01; Figure 3c). Again, the findings mirror In order to test the effect of JMJD2B overexpression changes in vivo, as H3K9me3 was also reduced on D7Z1 on chromosome segregation, diploid HME50TR cells relative to D7Z2 in IDC breast tumors (P ¼ 0.001; (Figure 2c) were transiently transfected with a myc- Figure 3d; Supplementary Figure S4). tagged JMJD2B expression construct. As shown in Figure 5a, cells expressing JMJD2B-myc exhibited a profound loss of nuclear H3K9me3. Significantly, the JMJD2 demethylase-driven loss of pericentromeric pronounced H3K9me3 foci, which colocalize with H3K9me3 may contribute to CIN CENP-A and likely include pericentromeric H3K9me3 Transcriptome analysis of a panel of 10 estrogen (Figure 5c) were largely abolished. No effect on the receptor-, progesterone receptor-, human epidermal H3K9me3 level was observed in control cells transfected growth factor receptor-2-negative (triple-negative) with an enzymatically inactive H188A JMJD2B mutant breast tumors (TNBTs) revealed robust upregulation construct (Figure 5b) (Whetstine et al., 2006). JMJD2B- of JMJD2A and JMJD2B histone lysine demethylases transfected cells were accumulated in mitosis using relative to laser-captured normal breast epithelium from nocodazole treatment, and then blocked in anaphase, 10 healthy volunteers (Figure 4a; Radovich et al., permitting a direct evaluation of segregation error rates submitted), while JMJD2B expression was also signifi- by FISH. A six-fold increase in the mis- cantly elevated in HME50-T when compared with segregation rate was detected in JMJD2B-transfected HME50hTERT (not shown). In comparison to normal cells compared with cells transfected with the JMJD2B breast, ChIP–PCR analysis of centromeres 5 and 7 in mutant control (P ¼ 0.002; Figure 5d). TNBTs revealed a significant loss of H3K9me3 from the kinetochore-forming sequences relative to both a ribosomal DNA arm reference and the flanking alpha- Discussion satellite sequence (Figure 4b; Supplementary Figure S5). In the three TNBTs examined (included in the ChIP– Tumors frequently exhibit CIN, leading to large scale PCR analysis), FISH with probes for centromeres 7 and genetic changes that can promote tumor progression

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3249

Figure 4 JMJD2 histone demethylase overexpression, pericentromeric H3K9me3 loss and chromosome instability (CIN) in TNBTs. (a) JMJD2A and JMJD2B are elevated in TNBTs (n ¼ 10) compared with laser-captured normal breast epithelium (n ¼ 10). Relative expression levels (reads per kb-exon per million mapped reads) are plotted for each sample. Expression of JMJD2C and JMJ2D did not differ significantly (not shown). (b) Chromatin immunoprecipitation and PCR (ChIP–PCR) analysis of TNBTs and normal breast tissue (n ¼ 4–6/group). Enrichment is shown relative to the rDNA arm reference, which is set to 1. H3K9me3 is significantly reduced on the kinetochore-forming sequences of chromosomes 5 (D5Z2) and 7 (D7Z1) in TNBTs compared with normal breast tissue (P values A and C, respectively), but did not significantly differ on flanking D5Z1 and D7Z2 sequences (P-values B and D). The range of H3K9me3 enrichment levels on D5Z1, D5Z2, D7Z1 and D7Z2 in individual TNBT tumors and normal tissue samples is shown in Supplementary Figure S5. H3K9me3 loss from the kinetochore-forming sequences at centromeres 5 and 7 was also significant relative to the flanking alpha-satellite sequences when TNBTs were compared with normal breast tissues. P-values are 0.027 and 0.014, respectively. (c) FISH image of an aneuploid TNBT nucleus using centromere 7 (red; D7Z1) and 17 (green) probes. 20–22 nuclei were scored for each of three TNBTs. Modal copy number and range for chromosomes 7 and 17 in the three TNBTs is shown. CIN was measured as % nuclei exhibiting deviation from modal copy number for chromosomes 7 and 17 in three TNBTs. In comparison % deviation for chromosomes 17 and X in four uncultured stromal vascular fraction control (C) samples (n ¼ 1197 nuclei) was 1.17% (Grimes et al., 2009).

(Kops et al., 2005). This study finds evidence that an Here, we employed cytological and ChIP–PCR-based altered centromere structure may contribute to the CIN methods to compare the chromatin architecture of the phenotype. Centromere function depends on incorpora- centromere in normal and tumor-derived cell lines, and tion of CENP-A into at the kinetochore tissues. The analysis focused on the centromeres of domain and the formation of flanking pericentromeric chromosomes 5 and 7, each of which harbors two heterochromatin. Overexpression of CENP-A has been adjacent blocks of HOR alpha-satellite DNA with linked to CIN in colon tumors and causes chromosome divergent sequence composition. At both centromeres, mis-segregation in colon cancer cells (Tomonaga et al., IF-FISH and ChIP–PCR analysis mapped CENP-A 2003; Amato et al., 2009). In model organisms, and the kinetochore domain to one of the two HORs. disruption of the pathways which maintain pericentro- This organization was conserved in all samples exam- meric heterochromatin has been linked to chromosome ined, representing a total of 14 individuals. Significantly, mis-segregation and cancer (Bernard et al., 2001; Peters the results show that CENP-A is restricted to a subset of et al., 2001; David et al., 2003; David et al., 2006; the available HOR alpha-satellite blocks on autosomal Bourgo et al., 2009). centromeres by a mechanism that remains to be Aberrant DNA methylation and histone acetylation determined. In contrast to the CENP-A distribution, are known to be associated with tumorigenesis and can ChIP–PCR revealed that pericentromeric H3K9me3 result in epigenetic gene mis-regulation such as silencing extended over both HORs on chromosomes 5 and 7, of tumor suppressor genes (Esteller, 2008). Successful although at least in the case of chromosome 5, the cancer treatment strategies have been developed, which kinetochore flanking HOR was dispensable for centro- target the DNA methyltransferases or histone deacety- mere function. When compared to primary HUVEC lases involved (Rodriguez-Paredes and Esteller, 2011). samples and normal breast epithelial cells, cancer cell Cancer-associated changes in histone methylation pat- lines with a CIN phenotype exhibited a marked terns have also been reported at various genomic alteration of the pericentromeric H3K9me3 distribution locations, including the X centromere (Esteller, 2008; that was consistent with a loss of kinetochore proximal Mravinac et al., 2009). However, the association heterochromatin. Furthermore, IF analysis of the between aberrant pericentromeric histone modification chromosomally unstable fibrosarcoma-derived HT1080 and CIN in cancer cells has not been examined. cell line pointed to a more genome-wide loss of

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3250

Figure 5 JMJD2B overexpression reverses H3K9me3 and increases chromosome instability (CIN) in HME50TR cells. (a, b) Split images showing H3K9me3 (red) in the presence or absence of JMJD2B-myc (green). Blue indicates DAPI staining. (a) H3K9me3 is reduced in the presence of wild-type JMJD2B-myc (arrow). (b) H3K9me3 is unaffected in cells expressing a mutant JMJD2B-myc control protein (arrow). (c) Centromere protein A (CENP-A) (red) colocalizes with heterochromatic foci (green) in HME50TR cells. (d) Representative FISH images of daughter nuclei showing normal X chromosome (red) segregation (2:2) and aberrant segregation (4:0) in anaphase blocked cells. X chromosome mis-segregation is elevated in HME50TR cells transfected with JMJD2B-myc (1) compared with controls cells expressing mutant JMJD2B-myc (2) (P ¼ 0.002 using a w2 test). Data represent a total of X497 scored segregation events and three independent transfections per condition. Western analysis confirmed similar levels of JMJD2B-myc in transfected cells (Supplementary Figure S6).

H3K9me3 from the vicinity of the kinetochore. The an effect on centromere function and chromosome possibility of a mechanistic link between the altered segregation would also be predicted but had not been pericentromeric H3K9me3 distribution and a CIN tested. Our analysis of a panel of TNBTs revealed phenotype was further strengthened by examining an robust overexpression of JMJD2A and JMJD2B relative ex vivo breast cancer model, which comprises serially to normal breast epithelium from healthy donors, as derived human breast epithelial cell lines recapitulating well as a pattern of pericentromeric H3K9me3 alteration tumor progression and the acquisition of CIN. In that mirrored the above findings. Consistent with a comparison to its untransformed diploid parent, the recent report detailing high levels of aneuploidy in tumorigenic and chromosomally unstable derivative cell TNBTs (Malorni et al., 2010), the three TNBTs line, HME50-T, exhibited a similar pattern of pericen- examined in this study also exhibited aneuploidy and a tromeric H3K9me3 alteration. A more modest but CIN phenotype. Endogenous JMJD2B was also over- significant change in pericentromeric H3K9me3 was expressed in the ex vivo breast cancer progression model also seen in a comparison of invasive ductal carcinoma (not shown) and in diploid HME50TR cells over- tumors with healthy genetically matched breast tissue. expression of exogenous JMJD2B largely abolished The reduced magnitude of change detected in tumors H3K9me3 staining, including the pericentromeric may reflect the presence of normal cells within tumor H3K9me3 foci, and resulted in a significant increase in samples, and/or the observation that CIN is a feature of the chromosome mis-segregation rate. only approximately 45% of invasive breast tumors In light of these findings, elevated JMJD2 demethy- (Lingle et al., 2002). Given the documented importance lase activity in tumors has the potential to promote CIN of pericentromeric heterochromatin for centromere via pathways including pericentromeric H3K9me3 loss. function, the association found in the present study However, other mechanisms must be considered. In supports a model in which failure to maintain adequate particular, reversal of H3K36 methylation, which is also kinetochore proximal heterochromatin contributes to enriched at centromeric regions and may be important CIN in human cancer. for centromere function (Bergmann et al., 2011), could As the H3K9me3 modification has a role in epigenetic be involved. Furthermore, the effect of JMJD2 over- gene regulation, the oncogenic consequences of JMJD2 expression on downstream gene regulation means that demethylase overexpression are likely to include tran- an impact on CIN independent of pericentromeric scriptional effects on downstream genes (Liu et al., 2009; H3K9me3 loss is also possible. kinetics Pedersen and Helin, 2010; Yang et al., 2010). However, preclude detection of an effect on centromere function

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3251 rapidly enough to exclude involvement of downstream HeLa and HT1080 cell lines were obtained from ATCC gene activity. In future, centromeric targeting of a (Manassas, VA, USA). HCC1937 were a kind gift from Dr JMJD2 fusion construct, as recently reported with Shay (Hammond Cancer Center, UT Southwestern, Dallas, another histone lysine demethylase, LSD1 (Bergmann TX, USA). Cell lines were cultured for less than 6 months at a et al., 2011), could show if altered centromeric time and verified by morphology. chromatin structure in isolation is a driver of CIN. It is notable that our ChIP–PCR analysis finds a Normal and tumor breast tissues selective reduction of H3K9me3 on the kinetochore- Breast tumors and normal breast tissues were purchased from forming sequence relative to the adjacent HOR alpha- the IU Simon Cancer Center (IUSCC) tissue bank (Indiana- satellite DNA and other arm sequences. While the polis, IN, USA) and Origene (Rockville, MD, USA). Normal mechanism is unclear, it may be speculated that the breast tissues from healthy volunteers for the transcriptome analysis were obtained from Susan G. Komen for the Cure relatively open chromatin conformation at the kineto- Tissue Bank at the IUSCC. chore (Sullivan and Karpen, 2004) and alpha-satellite transcription (Wong et al., 2007; Ting et al., 2011) favor access of the relevant histone modifiers, or that some Chromatin immunoprecipitation and PCR other targeting mechanism is involved. Clearly, causes Chromatin was extracted from a minimum of 100 mg per tissue sample and 6  106 cells per cell line. Nuclear isolation and of pericentromeric H3K9me3 perturbation other than chromatin digestion by micrococcal nuclease were performed demethylase overexpression can be envisaged and as described (Yoda et al., 2004). Chromatin fractions were warrant investigation. Potential candidates include incubated with anti-CENP-A (Abcam, Cambridge, MA, mis-expression of the SUV39H1/2 methyltransferases USA), anti-H3K9me3 (Abcam) or zero antibody (beads whose inactivation in mice eliminates pericentromeric control; Millipore, Bedford, MA, USA). Samples were H3K9me3 and drives CIN, and tumorigenesis (Peters processed according to the manufacturer’s protocol (Milli- et al., 2001). In addition, pRb, one of the most pore). Relative quantitation of target sequences in the input frequently mutated genes in human cancer, as well as chromatin, the immunoprecipitated fractions and controls was the chromatin remodeling complexes SWI/SNF and performed by real-time PCR using SYBR green chemistry and mSds3, are all required for faithful chromosome an Applied Biosystems 7500 cycler (Foster City, CA, USA). Each PCR reaction was performed in duplicate. Primers are segregation and tumor suppression and have been listed in Supplementary Table S2. For cell lines, a minimum of implicated in the maintenance of pericentromeric three replicate cultures was examined, with the exception of heterochromatin and H3K9 methylation (David et al., HCC1937, which was based on two replicates. HUVEC 2003; Gonzalo et al., 2005; David et al., 2006; Bourgo replicates represented three individual donors. The number et al., 2009; Sullivan et al., 2011). of tumor and normal breast tissue samples used in ChIP–PCR Considering the multitude of factors influencing are indicated in Figures 2–4. The Student’s t-test was used to chromosome segregation it would be unsurprising if measure significance of differences in ChIP–PCR in cell lines cancer cells evolve many different routes to a CIN and TNBT/normal tissue comparisons. A paired t-test was phenotype and evidence suggests this to be the case used for comparison of matched IDC tumor/normal tissues. (Thompson et al., 2010). Disruption of pericentromeric Error bars represent s.e.m. heterochromatin may represent one such pathway, either sufficient in itself or requiring additional co- Immunofluorescence and fluorescence in situ hybridization operative changes. In either case, this epigenetic CENP-A (Abcam), H3K9me3 (Abcam), CENP-E or c-myc component to the CIN phenotype could present new (Santa Cruz Biotechnology, Santa Cruz, CA, USA) antibodies therapeutic targets in the form of reversible chromatin were used in IF (see supplementary protocols). FISH probes were pG-A16 (D5Z1) (Hulsebos et al., 1988), pZ5.1 (D5Z2) changes. Perhaps agents such as inhibitors of JMJD2 (Archidiacono et al., 1995), D7Z1 (p7t1) (Waye et al., 1987), demethylases, under current development (Rose et al., D7Z2 (pMGB7) (Waye et al., 1987), GDNF (Schindelhauer 2010; Lohse et al., 2011), could have dual therapeutic et al., 1995) or commercial CEP 17 and CEP X probes (Abbot potential, targeting both epigenetic gene mis-regulation Molecular, Des Plaines, IL, USA). DNA was counterstained and CIN. with DAPI. Prior to micrococcal nuclease digestion, a fraction of TNBT nuclei harvested for ChIP were pelleted and resuspended in fixative (3:1 v/v methanol:acetic acid), and Materials and methods processed for FISH (Figure 4). Images were captured at  100 magnification using a Spot RTKE camera (Diagnostic Instruments, Sterling Heights, MI, USA), mounted on a Leica Cell culture DN5000B microscope (Leica Microsystems, Buffalo Grove, HeLa, HCC1937 and HME50-T cells were grown in DMEM/ IL, USA) and modified to increase brightness and/or contrast 10% FBS, HT1080 cells in alpha-MEM/10% FBS and using Photoshop. HUVECs in EGM2 medium (Lonza, Walkersville, MD, USA)/10% FBS. HME5cdk4 are primary breast epithelial cells expressing exogenous CDK4 to extend lifespan (Herbert JMJD2B constructs, transfection and anaphase assays et al., 2002; Ramirez et al., 2003). HME5cdk4, HME50hTERT A myc-tagged JMJD2B expression construct was purchased and HME50TR cells were grown in supplemented MCDB 171 from Origene. An inactivating H188A mutation (Whetstine medium (Invitrogen, Carslbad, CA, USA) (Ramirez et al., et al., 2006) was introduced using site-directed mutagenesis 2003). BF lymphoblastoid cells were grown in RPMI1640 (Quickchange; Stratagene, La Jolla, CA, USA). JMJD2B medium/15% FBS and have the designation 47,XY, constructs were transfected into HME50TR cells using del(5)(p10p13), þ r(5)(p10p13) (Schuffenhauer et al., 1996). Lipofectamine 2000 (Invitrogen). At approximately 48 h

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3252 following transfection, cells were accumulated in mitosis using Mariano Rocci and Theo Hulsebos for centromere nocodazole, blocked in anaphase and processed for FISH probes; Tim Yen for CENP-E antibodies; Mervin Yoder for (Sullivan and Warburton, 1999). Transfection efficiency was HUVEC samples; Virginia Thurston for advice on the patient estimated at B60% using a GFP reporter (not shown). BF sample and Andrew Ross, Nikki Collins, Janak Bhavsar, Tony Brown, Waylan Bessler and Laura Mead for technical assistance. This research was supported in part by INGEN Conflict of interest (Indiana Genomics Initiative), the IUSM Divi- sion and grants from the US DOD BCRP (W81XWH-09-1- The authors declare no conflict of interest. 0424) and American Cancer Society (IRG-84-002-24) awarded to BRG, IUSM Biomedical Research Grants awarded to RBS and an NCI grant (N01 CN-43300) to BSH. DS was supported Acknowledgements by Deutsche Forschungsgemeinschaft and the Friedrich Baur Institute. INGEN is supported in part by Lilly Endowment. We thank Kinya Yoda for advice with ChIP assays; Simone MR was supported by an NIH pre-doctoral fellowship (NRSA Schuffenhauer and Jan Murken for BF cells; Hunt Willard, 1 T32 CA 111198 Cancer Biology Training Program).

References

Alonso A, Hasson D, Cheung F, Warburton PE. (2010). A paucity of Fodor BD, Kubicek S, Yonezawa M, O’Sullivan RJ, Sengupta R, heterochromatin at functional human . Perez-Burgos L et al. (2006). Jmjd2b antagonizes H3K9 trimethyla- Chromatin 3:6. tion at pericentric heterochromatin in mammalian cells. Genes Dev Amato A, Schillaci T, Lentini L, Di Leonardo A. (2009). CENPA 20: 1557–1562. overexpression promotes genome instability in pRb-depleted human Gonzalo S, Garcia-Cao M, Fraga MF, Schotta G, Peters AH, Cotter SE cells. Mol Cancer 8: 119. et al. (2005). Role of the RB1 family in stabilizing histone methylation Archidiacono N, Antonacci R, Marzella R, Finelli P, Lonoce A, at constitutive heterochromatin. Nat Cell Biol 7: 420–428. Rocchi M. (1995). Comparative mapping of human alphoid Grimes BR, Babcock J, Rudd MK, Chadwick B, Willard HF. (2004). sequences in great apes using fluorescence in situ hybridization. Assembly and characterization of heterochromatin and euchroma- Genomics 25: 477–484. tin on human artificial chromosomes. Genome Biol 5: R89. Bannister AJ, Zegerman P, Partridge JF, Miska EA, Thomas JO, Grimes BR, Steiner CM, Merfeld-Clauss S, Traktuev DO, Smith D, Allshire RC et al. (2001). Selective recognition of methylated lysine Reese A et al. (2009). Interphase FISH demonstrates that human 9 on by the HP1 chromo domain. Nature 410: 120–124. adipose stromal cells maintain a high level of genomic stability in Bergmann JH, Rodriguez MG, Martins NM, Kimura H, Kelly DA, long-term culture. Stem Cells Dev 18: 717–724. Masumoto H et al. (2011). Epigenetic engineering shows H3K4me2 Guenatri M, Bailly D, Maison C, Almouzni G. (2004). Mouse centric is required for HJURP targeting and CENP-A assembly on a and pericentric satellite repeats form distinct functional hetero- synthetic human kinetochore. EMBO J 30: 328–340. chromatin. J Cell Biol 166: 493–505. Bernard P, Maure JF, Partridge JF, Genier S, Javerzat JP, Allshire Haaf T, Ward DC. (1994). Structural analysis of alpha satellite DNA RC. (2001). Requirement of heterochromatin for cohesion at and centromere proteins using extended chromatin and chromo- centromeres. Science 294: 2539–2542. somes. Hum Mol Genet 3: 697–709. Bourgo RJ, Siddiqui H, Fox S, Solomon D, Sansam CG, Yaniv M Herbert BS, Wright WE, Shay JW. (2002). p16(INK4a) inactivation is et al. (2009). SWI/SNF deficiency results in aberrant chromatin not required to immortalize human mammary epithelial cells. organization, mitotic failure, and diminished proliferative capacity. Oncogene 21: 7897–7900. Mol Biol Cell 20: 3192–3199. Hochreiter AE, Xiao H, Goldblatt EM, Gryaznov SM, Miller KD, Cahill DP, Kinzler KW, Vogelstein B, Lengauer C. (1999). Genetic Badve S et al. (2006). Telomerase template antagonist GRN163L instability and darwinian selection in tumours. Trends Cell Biol 9: disrupts maintenance, tumor growth, and metastasis of M57–M60. breast cancer. Clin Cancer Res 12: 3184–3192. Cardinale S, Bergmann JH, Kelly D, Nakano M, Valdivia MM, Hulsebos T, Schonk D, van Dalen I, Coerwinkel-Driessen M, Kimura H et al. (2009). Hierarchical inactivation of a synthetic Schepens J, Ropers HH et al. (1988). Isolation and characterization human kinetochore by a chromatin modifier. Mol Biol Cell 20: of alphoid DNA sequences specific for the pericentric regions of 4194–4204. chromosomes 4, 5, 9, and 19. Cytogenet Cell Genet 47: 144–148. David G, Dannenberg JH, Simpson N, Finnerty PM, Miao L, Turner Kops GJ, Weaver BA, Cleveland DW. (2005). On the road to cancer: GM et al. (2006). Haploinsufficiency of the mSds3 chromatin aneuploidy and the mitotic checkpoint. Nat Rev Cancer 5: 773–785. regulator promotes chromosomal instability and cancer only upon Lee C, Wevrick R, Fisher RB, Ferguson-Smith MA, Lin CC. (1997). complete neutralization of p53. Oncogene 25: 7354–7360. Human centromeric . Hum Genet 100: 291–304. David G, Turner GM, Yao Y, Protopopov A, DePinho RA. (2003). Lingle WL, Barrett SL, Negron VC, D’Assoro AB, Boeneman K, Liu mSin3-associated protein, mSds3, is essential for pericentric W et al. (2002). Centrosome amplification drives chromosomal heterochromatin formation and chromosome segregation in instability in breast tumor development. Proc Natl Acad Sci USA mammalian cells. Genes Dev 17: 2396–2405. 99: 1978–1983. Ekwall K, Olsson T, Turner BM, Cranston G, Allshire RC. (1997). Liu G, Bollig-Fischer A, Kreike B, van de Vijver MJ, Abrams J, Ethier Transient inhibition of histone deacetylation alters the structural SP et al. (2009). Genomic amplification and oncogenic properties of and functional imprint at fission yeast centromeres. Cell 91: the GASC1 histone demethylase gene in breast cancer. Oncogene 28: 1021–1032. 4491–4500. Esteller M. (2008). Epigenetics in cancer. N Engl J Med 358: Lohse B, Kristensen JL, Kristensen LH, Agger K, Helin K, Gajhede M 1148–1159. et al. (2011). Inhibitors of histone demethylases. Bioorg Med Chem Finelli P, Antonacci R, Marzella R, Lonoce A, Archidiacono N, 19: 3625–3636. Rocchi M. (1996). Structural organization of multiple alphoid Malorni L, Shetty PB, Hilsenbeck SG, Rimawi MF, Elledge RM, De subsets coexisting on human chromosomes 1, 4, 5, 7, 9, 15, 18, and Angelis C et al. (2010). Triple-negative breast cancers: biomarkers 19. Genomics 38: 325–330. and outcomes. J Clin Oncol 28(15 Suppl): 10621.

Oncogene Pericentromeric heterochromatin changes in cancer RB Slee et al 3253 Mitelman F, Johansson B, Mertens F. Catalog of Chromosome Sullivan BA, Warburton PE. (1999). Studying progression of Aberrations in Cancer, 5th edn. Wiley-Liss: New York, USA, 1994. chromosomes through mitosis by immunofluorescence Mravinac B, Sullivan LL, Reeves JW, Yan CM, Kopf KS, Farr CJ and FISH. In: Bickmore WA (ed). Chromosome Structural et al. (2009). Histone modifications within the human X centromere Analysis-A Practical Approach. Oxford University Press: Oxford, region. PLoS One 4: e6602. pp 81–101. Nakano M, Cardinale S, Noskov VN, Gassmann R, Vagnarelli P, Sullivan LL, Boivin CD, Mravinac B, Song IY, Sullivan BA. (2011). Kandels-Lewis S et al. (2008). Inactivation of a human kinetochore Genomic size of CENP-A domain is proportional to total alpha by specific targeting of chromatin modifiers. Dev Cell 14: 507–522. satellite array size at human centromeres and expands in cancer Ohzeki J, Nakano M, Okada T, Masumoto H. (2002). CENP-B box is cells. Chromosome Res 19: 457–470. required for de novo centromere chromatin assembly on human Thompson SL, Bakhoum SF, Compton DA. (2010). Mechanisms of alphoid DNA. J Cell Biol 159: 765–775. chromosomal instability. Curr Biol 20: R285–R295. Palmer DK, O’Day K, Wener MH, Andrews BS, Margolis RL. (1987). Ting DT, Lipson D, Paul S, Brannigan BW, Akhavanfard S, Coffman A 17-kD centromere protein (CENP-A) copurifies with EJ et al. (2011). Aberrant overexpression of satellite repeats in core particles and with . J Cell Biol 104: 805–815. pancreatic and other epithelial cancers. Science 331: 593–596. Pedersen MT, Helin K. (2010). Histone demethylases in development Tomonaga T, Matsushita K, Yamaguchi S, Oohashi T, Shimada H, and disease. Trends Cell Biol 20: 662–671. Ochiai T et al. (2003). Overexpression and mistargeting of Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA centromere protein-A in human primary colorectal cancer. Cancer et al. (2000). Molecular portraits of human breast tumours. Nature Res 63: 3511–3516. 406: 747–752. Vafa O, Sullivan KF. (1997). Chromatin containing CENP-A and Peters AH, O’Carroll D, Scherthan H, Mechtler K, Sauer S, Schofer C alpha-satellite DNA is a major component of the inner kinetochore et al. (2001). Loss of the Suv39h histone methyltransferases impairs plate. Curr Biol 7: 897–900. mammalian heterochromatin and genome stability. Cell 107: 323–337. Verdaasdonk JS, Bloom K. (2011). Centromeres: unique chromatin Radovich M, Clare SE, Sledge GW, Pardo I, Mathieson T, Kassem N structures that drive chromosome segregation. Nat Rev Mol Cell et al. The transcriptional landscape of triple-negative breast cancer Biol 12: 320–332. compared to normal breast tissues using next-generation RNA Warburton PE, Cooke CA, Bourassa S, Vafa O, Sullivan BA, Stetten sequencing. (submitted). G et al. (1997). Immunolocalization of CENP-A suggests a distinct Ramirez RD, Herbert BS, Vaughan MB, Zou Y, Gandia K, Morales nucleosome structure at the inner kinetochore plate of active CP et al. (2003). Bypass of telomere-dependent replicative senes- centromeres. Curr Biol 7: 901–904. cence (M1) upon overexpression of Cdk4 in normal human Warburton PE, Willard HF. (1996). Evolution of centromeric alpha epithelial cells. Oncogene 22: 433–444. satellite DNA: molecular organization within and between human Rodriguez-Paredes M, Esteller M. (2011). Cancer epigenetics reaches and primate chromosomes. In: Jackson M, Strachan T, Dover G mainstream oncology. Nat Med 17: 330–339. (eds). Evolution. BIOS Scientific Publishers: Oxford, Rosandic M, Paar V, Basar I, Gluncic M, Pavin N, Pilas I. (2006). pp 121–145. CENP-B box and pJalpha sequence distribution in human alpha Waye JS, England SB, Willard HF. (1987). Genomic organization of satellite higher-order repeats (HOR). Chromosome Res 14: 735–753. alpha satellite DNA on human chromosome 7: evidence for two Rose NR, Woon EC, Kingham GL, King ON, Mecinovic J, Clifton IJ distinct alphoid domains on a single chromosome. Mol Cell Biol 7: et al. (2010). Selective inhibitors of the JMJD2 histone demethy- 349–356. lases: combined nondenaturing mass spectrometric screening and Wevrick R, Willard HF. (1991). Physical map of the centromeric crystallographic approaches. J Med Chem 53: 1810–1818. region of human chromosome 7: relationship between two distinct Schindelhauer D, Schuffenhauer S, Gasser T, Steinkasserer A, alpha satellite arrays. Nucleic Acids Res 19: 2295–2301. Meitinger T. (1995). The gene coding for glial cell line derived Whetstine JR, Nottke A, Lan F, Huarte M, Smolikov S, Chen Z et al. neurotrophic factor (GDNF) maps to chromosome 5p12-p13.1. (2006). Reversal of histone lysine trimethylation by the JMJD2 Genomics 28: 605–607. family of histone demethylases. Cell 125: 467–481. Schuffenhauer S, Kobelt A, Daumer-Haas C, Loffler C, Muller G, Wong LH, Brettingham-Moore KH, Chan L, Quach JM, Anderson Murken J et al. (1996). Interstitial deletion 5p accompanied by MA, Northrop EL et al. (2007). Centromere RNA is a key dicentric ring formation of the deleted segment resulting in component for the assembly of nucleoproteins at the nucleolus and 5p13-cen. Am J Med Genet 65: 56–59. centromere. Genome Res 17: 1146–1160. Shin HJ, Baek KH, Jeon AH, Kim SJ, Jang KL, Sung YC et al. (2003). Yang J, Jubb AM, Pike L, Buffa FM, Turley H, Baban D et al. (2010). Inhibition of histone deacetylase activity increases chromosomal The histone demethylase JMJD2B is regulated by estrogen receptor instability by the aberrant regulation of mitotic checkpoint alpha and hypoxia, and is a key mediator of estrogen induced activation. Oncogene 22: 3853–3858. growth. Cancer Res 70: 6456–6466. Sullivan BA, Karpen GH. (2004). Centromeric chromatin exhibits a Yoda K, Morishita S, Hashimoto K. (2004). Histone variant histone modification pattern that is distinct from both CENP-A purification, nucleosome reconstitution. Methods Enzymol and heterochromatin. Nat Struct Mol Biol 11: 1076–1083. 375: 253–269.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene