Oncogene (2010) 29, 2457–2466 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Role of multi-hnRNP nuclear complex in regulation of tumor suppressor ANXA7 in prostate cancer cells

Y Torosyan1, A Dobi2, M Glasman1, K Mezhevaya1, S Naga1, W Huang1, C Paweletz1, X Leighton1, HB Pollard1 and M Srivastava1

1Department of Anatomy, Physiology and Genetics, Institute for Molecular Medicine, Uniformed Services University School of Medicine (USUHS), Bethesda, MD, USA and 2Section for Regulation and Bioinformatics, Center for Prostate Disease Research (CPDR), Rockville, MD, USA

Annexin-A7 (ANXA7) tumor suppressor role has been Introduction shown in various tumors, and ANXA7 expression has been particularly lost in androgen-resistant prostate cancers. Annexin-A7 (ANXA7 or synexin) is a member of the In this study, we studied ANXA7 regulation in normal multifunctional calcium/phospholipid-binding annexin prostate versus androgen-sensitive and -resistant prostate family. Human ANXA7 has shown a tumor suppressor cancer cells. Deletion mapping analysis showed lowest role in multiple studies that involved prostate cancer ANXA7-promoter activities in androgen-sensitive LNCaP (Srivastava et al., 2003, 2007; Torosyan et al., 2009). In prostate cancer cells. Genomatix analysis of ANXA7 addition, Anxa7(±) mice have a cancer-prone pheno- promoter identified a cluster of steroid nuclear hormone type, leading to a spectrum of tumors including prostate receptor elements, including V$GREF (V$GRE.02/ cancer (Srivastava et al., 2003). ARE.02). Gelshift analysis clearly indicated distinct Localized to secretory vesicles (in addition to plasma nuclear occupancy at this ANXA7-promoter site membranes and the nucleus), ANXA7 can mediate (À1086/À890) in prostate cancer (LNCaP, DU145, and Ca/GTP-regulated exocytosis (Pollard et al., 1998). PC3) versus normal prostate (PrEC) cells. In matrix- Showing a typical tumor suppressor gene pattern in assisted laser desorption time-of-flight mass spectrometry- our multitumor tissue microarray study, ANXA7 based search for ANXA7 nuclear regulators, we iden- protein expression is consistently decreased in tumors tified several heterogeneous nuclear ribonucleoproteins (Srivastava et al., 2007). However, although reduced (hnRNPs) (A1, A2/B1 and K) attached to the steroid- in androgen-resistant prostate cancers, ANXA7 is associated ANXA7-promoter site in the androgen-resis- abundant in the adrenal gland, suggesting that its tant PC3 prostate cancer cells with high ANXA7 gene hormone-related expression could be coordinated with copy number, but not in PrEC. The hnPNP role in steroidogenesis and hormone sensitivity. ANXA7 regulation (that was validated by hnRNPA2/B1 The ANXA7 promoter contains steroid-responsive antibody interference) resulted in multiple ANXA7 cDNA elements and can respond to dexamethasone (Srivastava and protein products in PC3, but not in PrEC. Ingenuity and Pollard, 2000). A genuinely low of pathways analysis showed plausible molecular paths ubiquitous ANXA7 has been attributed to inhibitory between ANXA7 and the hnRNP-associated network in promoter sequences, which control ANXA7 depending prostate cancer progression. Thus, a multi-hnRNP com- on the cellular environment. Unlike the C-terminal plex can be responsible for aberrant ANXA7 transcription with conserved annexin repeats, the unique ANXA7 and splicing, thereby affecting ANXA7 expression pattern N-terminal is flanked by the upstream sequences and tumor suppressor function in prostate cancer. harboring an entirely different set of cis-acting and Oncogene (2010) 29, 2457–2466; doi:10.1038/onc.2010.2; enhancer elements compared with other annexins published online 1 March 2010 (Shirvan et al., 1994). In our study on ANXA7 versus p53 effects on prostate cancer cells (Torosyan et al., Keywords: Annexin A7, ANXA7 (or synexin); hetero- 2006), we suggested that ANXA7 can be co-regulated by geneous nuclear ribonucleoprotein; hnRNP; prostate ribonucleoprotein (heterogeneous nuclear ribonucleo- cancer; DNA affinity enrichment/identification of protein (hnRNP)) K. nuclear ; mass spectrometry Alterations in hnRNP-mediated mRNA splicing can have dramatic consequences resulting in mRNA instability and aberrant protein products in cancer. ANXA7 is known to be transcribed into different isoforms with a tissue-specific distribution: while the Correspondence: Dr M Srivastava, Department of Anatomy, Physio- short isoform (NP001147.1, NCBI) is more abundant, logy and Genetics, USU School of Medicine, 4301 Jones Bridge Rd., other ANXA7 isoforms can be limited to certain tissues B2-034, Bethesda, MD 20814, USA. E-mail: [email protected] such as the brain (Magendzo et al., 1991). Hence, the Received 16 July 2009; revised 19 October 2009; accepted 19 December regulatory mechanisms in ANXA7 transcription and 2009; published online 1 March 2010 splicing in different cell environments, including hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2458 neoplastic transformation, need to be elucidated. The which lack functional signaling through two major androgen-resistant PC3 prostate cancer cells, in parti- nuclear regulators, androgen (AR) and glucocorticoid cular, have been shown to possess an intriguingly high (GR) receptors (Yemelyanov et al., 2007). ANXA7 gene copy number (Wolf et al., 2004) that ANXA7 is specifically reduced in androgen-resistant questions the tumor suppressor role of endogenous prostate tumors (Srivastava et al., 2007). In the mean- ANXA7 in these cells. Hence, we undertook a study on time, wild-type ANXA7 decreases and dominant-nega- ANXA7 transcription in normal and cancerous prostate tive DN-ANXA7 increases low-molecular weight AR cells, including PC3, and analyzed ANXA7-promoter in androgen-sensitive LNCaP (Torosyan et al., 2009). activity in juxtaposition to the Genomatix-derived These data link ANXA7 and AR, while dexa- mapping of putative regulatory matrixes. Using a mass methasone response from ANXA7-promoter activity spectrometry (MS)-based approach to identify nuclear (Srivastava and Pollard, 2000) link ANXA7 and GR. factors that can be involved in the regulation of ANXA7 Hence, the specifically low ANXA7-promoter activity in expression in PC3 in particular, we identified a multi- LNCaP could reflect the lack of AR- and GR-associated hnRNP-complex attached to the steroid-associated transcriptional regulation. ANXA7-promoter site. This result suggests that hnRNPs can control ANXA7 splicing, thereby mod- ulating ANXA7 cell survival regulation in cancer. Genomatix-based analysis of ANXA7-promoter segments Using Genomatix, we analyzed ANXA7 promoter (from À1200 to þ 150) and identified a cluster of cis- acting, steroid/lipid-associated elements in the segment Results (F), which constitutes the difference between F1 and F2 (Figure 2). This cluster includes double glucocorticoid- Deletion mapping and enhanced activities of ANXA7- responsive and related elements V$GREF: GRE.02 and promoter segments in normal prostate (PrEC) and ARE.02 (for GR and AR, respectively). As a result, GR prostate cancer (LNCaP, DU145, PC3) cells and AR can compete for binding to the same site, and Analysis of promoter activities of the LUC-reporter their competition can be further modulated by adjacent fused ANXA7-promoter deletion constructs (F1–F5, steroid/lipid-associated regulatory elements. from À240/ þ 34 to À1086/ þ 34 bp) in prostate cancer A relatively abundant active GR can function as a versus normal prostate cells (Figure 1) shows cell- tumor suppressor (Yemelyanov et al., 2007) by inhibit- specific profiles of activity and suggests potential ing multiple pathways in cell transformation. As stress enhancer or repressor roles for multiple cis-elements in response mediators, glucocorticoids are linked to PPAR the ANXA7 promoter. For example, in normal PrEC, signaling that affects lipid pathways (including arachi- the core promoter F5-segment shows the highest activity donate cascade) in energy homeostasis (Rizzo and that is supposedly reduced by repressors in F3 and F4. Fiorucci, 2006). The GREF-overlapping V$PPARG.01 Remarkably, this pattern remains specific for PrEC, suggests a regulatory role for PPAR-g, which can form indicating aberrant regulation of the ANXA7-promoter heterodimers with the retinoid X receptor (RXR). The activity in cancer cells. Most remarkably, the lowest adjacent V$RXRF/VDR_RXR.04 indicates a possible activity for all ANXA7-promoter segments (Po0.05 in indirect co-regulation of ANXA7 by VitD, which can most comparisons including PrEC) is found in LNCaP, inhibit prostate cancer growth (Chen et al., 2000). The other GREF-overlapping site suggests binding for the ROR-g receptor, which controls metabolic homeostasis, 25 including lipids (Kang et al., 2007). -1086/+34 -889/+34 -658/+34 -417/+34 -240/+34 Affected by nutrient availability, the serum response 20 element transcription factor site V$SRFF/SRF.02 can confer growth inhibition by glucocorticoids (Karagianni 15 and Tsawdaroglou, 1994). In the meantime, the key lipogenic transcription factor site V$SREB/SREBP/.02 (fold) 10 can confer indirect androgen activation (Heemers et al., * 2006). The SOX/SRY-sex/testis determining SOX2.01 * 5 and SOX5.01 can also contribute to ANXA7 regulation * * * that is consistent with ANXA7 and SOX5 synexpression Normalized promoter activity Normalized promoter in prostate cancer arrays (Srivastava et al., 2007). 0 F1 F2 F3 F4 F5 Remarkably, the F-segment also contains the transfer PrEC DU145 PC3 LNCaP RNA-activating V$STAF/ZNF76_143.01, which is usually Figure 1 Activity of ANXA7-promoter segments in normal located far upstream of the transcription start site. STAF prostate (PrEC) and prostate cancer (LNCaP, DU145 and PC3) binding can be stimulated by hydrocortisone (Adachi et al., cells. As described in Material and methods, the enhanced 1999), and many STAF target are important for cell promoter activity of five ANXA7-promoter segments (F1–F5) growth. Usually sufficient for direct activation without was quantified (mean±s.d.) using Dual-Luciferase Reporter Assay associated basal promoter elements (Myslinski et al., 2006), and ANXA7-promoter constructs. Statistically significant values (Po0.05) for ANXA7-promoter activities in LNCaP vs other cells STAF can recruit TFIID/TBP, whose adjacent regulatory are marked by asterisks (*). elements were also found in the F-segment.

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2459

Figure 2 Genomatix analysis of ANXA7 promoter. According to Genomatix, the core ANXA7 promoter contains a single V$GREF/ GRE.02 site ( þ , matrix and core similarity: 0.87–0.83, respectively) within B100 bp from multiple initiation sites ( Â 5, as indicated by red arrows) including V$STAF/V$ZNF76_143.01 ( þ , matrix and core similarity: 0.81–0.77, respectively). In the distant ANXA7 promoter, the double GREF-elements: V$ARE.02 (–, 50-ACAGGACTTTCTGTCCTAC-30) and V$GRE.02 ( þ ,50-GTAGGACAGA AAGTCCTGT-30) with high core and matrix similarity (0.96–0.99 and 0.87–0.94, respectively) share a binding site (around À1050 bp) that is adjacent to V$STAF/V$ZNF76_143.01 (–, core and matrix similarity: 0.81 and 0.78, respectively). Core matrix sequences are highlighted by capitals and ci-value46C–by bold. In addition, the distant core promoter segment around double-GREF (F, B200 bp) contains a cluster of different steroid/lipid-related matrixes, which are represented by color coding as designated in the figure. More detailed information on the regulatory matrixes identified by Genomatix in extended ANXA7 promoter can be found in Supplementary Table S1.

The dispersed ANXA7 core promoter contains five transcription start sites (marked by red arrows in

Figure 2). Unlike the single transcription start site in Control PrEC DU145 PC3 LNCaP focused promoters, multiple start sites and motifs such as XCPE1 in dispersed promoters can specify different responses to enhancers (Juven-Gershon et al., 2008). Particularly found in TATA-less genes, the XCPE1- driven transcription can be active in the absence of TFIID/TBP (Tokusumi et al., 2007). Accordingly, the GC-reach and TATA-lacking ANXA7 core promoter contains adjacent O$XCPE and V$STAF, surrounded by Sp1-sites. Nuclear hormone receptors can modulate transcrip- tion and in particular. In a steroid receptor-dependent and -selective manner, steroid hor- mones affect the processing of pre-mRNA that is Figure 3 DNA affinity to ANXA7-promoter F-segment in synthesized only from the steroid-sensitive promoters prostate cancer (LNCaP, DU145 and PC3) versus normal prostate (PrEC) cells. The gelshift-based DNA affinity to ANXA7 upstream (Auboeuf et al., 2002). Hence, it is tempting to speculate sequences (À1080/À890) was assessed in nuclear extracts from that the steroid/lipid-related cluster with double-GREF different prostate cell lines (PrEC, DU145, PC3 and LNCaP). next to a putative transcription-initiating STAF-TBP Control contained a free DNA probe and no nuclear proteins. In site (F-segment) could coordinate alternative transcrip- prostate cancer cell samples compared with normal PrEC, filled circles mark the more abundant bands and unfilled circles—the less tional regulation and splicing of the hormone-related abundant bands. and phospholipid-binding ANXA7. bands and only one more abundant band, compared with Distinct binding patterns to ANXA7-promoter (F, À1086/ PrEC (Figure 3). The reduced nuclear protein occupancy À890) in prostate cancer (LNCaP, DU145, PC3) cells at the F-segment in LNCaP suggests a loss of the nuclear compared with normal prostate cells (PrEC) regulators responsible for ANXA7-promoter activity in To examine protein-binding patterns to the aforemen- these androgen-sensitive cells. In the meantime, the tioned elements, we used gelshift assay with nuclear opposite pattern in PC3 cells (three more abundant and extracts from different prostate cells and the F-segment one less abundant band) suggests that PC3 can possess as a probe. The lowest ANXA7-promoter activity in more nuclear regulators attached to the F-segment in androgen-sensitive LNCaP (as shown in Figure 1) is ANXA7 promoter in comparison with their androgen- accompanied by a distinct pattern: four less abundant resistant counterpart DU145 and normal PrEC.

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2460 Hence, distinct binding patterns clearly indicate lost Additional MS analysis of the nuclear proteins and acquired nuclear protein–DNA interactions at the recruited to the ANXA7 promoter in PC3 cells identified steroid-associated ANXA7-promoter site in prostate hnRNPU, adenosine triphosphate-dependent DNA cancer versus normal prostate cells. helicase as well as different subunits of the eukaryotic polypeptide elongation factor EF-1 involved in apopto- sis and oncogenesis (data not shown). MALDI-TOF MS-based identification of the nuclear Representing the most abundant nuclear proteins that proteins bound to ANXA7-promoter (F, À1086/À890) are implicated in the spliceosome packaging of excised in prostate cancer (PC3) versus normal prostate cells intron sequences (Reed and Hurt, 2002; Cooper et al., (PrEC) 2009), hnRNPs typically bind to exonic/intronic splicing The more abundant bands in the DNA-affinity pattern silencers, thereby repressing splicing (Graveley, 2009). for the F-segment in PC3 (as shown in Figure 3) However, hnRNPs can hinder communication between correspond to the uniquely high ANXA7 gene copy factors bound to different splice-sites, thereby having a number (Wolf et al., 2004) and complementary DNA positive role in RNA splicing. Most importantly, the (cDNA) expression in PC3 among other NCI-60 cells concentration of splicing factors can alter the kinetic (Stanford Microarray Database, Stanford, CA, USA, equilibrium in splicing, resulting in changes in splice-site as shown in Supplementary Figure S1). Accordingly, selection. PC3 shows greater than threefold increase of ANXA7 Thus, MS-based identification of several hnRNPs protein expression compared with other tested cells, recruited to a putative alternative transcription initia- including PrEC, DU145, and LNCaP (data not shown). tion site with double-GREF and STAF implied a To identify ANXA7 nuclear regulators in PC3, we steroid-sensitive and multi-hnRNP-associated control used matrix-assisted laser desorption/ionization time- over ANXA7 transcription including alternative splice- of-flight mass spectrometry (MALDI-TOF-MS)-based site selection in prostate cancer (PC3) versus normal DNA-affinity enrichment/purification of nuclear pro- prostate (PrEC) cells. teins using the F-segment as bait (Figure 4). Consistent with the gelshift results (as shown in Figure 3), the PC3 nuclear extract shows more abundant bands when Specific binding of hnRNPA2/B1 to ANXA7 promoter compared with PrEC. Identified peptide signatures (F, À1086/À890) in PC3 prostate cancer cells match to hnRNP-K, -A1 and -A2/B1 (MS analysis did Nuclear steroid receptors can regulate the hnRNP- not distinguish between A2 and B1 splice isoforms). The associated transcription and mRNA splicing, while specific DNA probe (designated by þ ) in PC3 results hnRNPs such as hnRNPU can bind GR, directly in the reduced recruitment of hnRNP-K and -A2/B1 affecting glucocorticoid-dependent transcription (Eggert proteins and in the disappearance of hnRNPA1, thereby et al., 2001). implying specific binding of identified hnRNPs to the F- Using an antibody interference assay, we tested segment. As expected, the bands containing GAPDH specific GR and hnRNPA2/B1 co-occupancy at the F- have similar levels in both cell types. segment with double-GREF and specific DNA affinity to hnRNPs (Figure 5). The GR antibody does not interfere and consequently does not confirm a direct Nuclear extracts from: PC3 PrEC GR binding. By contrast, the hnRNPA2/B1 antibody Competitor: -+ - + interference eliminates an abundant band, thereby supporting the role of hnRNPA2/B1 that exemplifies hnRNP involvement in the steroid-associated ANXA7 transcriptional control in PC3. It is important to note,

hnRNP K Nuclear extracts : None PC3 None PC3

hnRNP A1 GAPDH Antibodies None None hnRNPA2/B1 None None GR hnRNP A2/B1

Figure 4 MALDI-TOF-MS-based DNA-affinity enrichment/ purification of nuclear proteins attached to ANXA7-promoter F-segment in prostate cancer (PC3) versus normal prostate (PrEC) cells. Double-strand DNA sequences corresponding to F-segment (À1086/À890) were used as a specific probe (absent Figure 5 Specific hnRNPA2/B1 and GR binding to ANXA7- designated by ‘–’ and present—by ‘ þ ’). Arrows mark the promoter F-segment in prostate cancer cells (PC3). Specific binding bands containing the identified by MALDI-TOF MS nuclear of hnRNPA2/B1 and GR was assessed by antibody interference proteins (hnRNPs A1, A2/B1 and K, matched peptides43) assay using ANXA7-promoter F-segment (À1086/À890) as a with specific DNA affinity to ANXA7-promoter sequences in specific DNA probe. The band that was present in control, but PC3 nuclear extracts. disappeared under hnRNPA2/B1 is marked by a circle.

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2461 however, that the lack of GR interference does not regulate ANXA7 transcription, splicing and expression rule out the possibility of GR-hnRNP interaction and, in cancerous versus normal cells. thereby, the indirect GR inducibility of ANXA7- promoter activity and alternative splicing. ANXA7 downregulation and hnRNP-associated synexpression in prostate cancer progression Baseline and hnRNPB1-induced ANXA7 expression in (Ingenuity Pathways Analysis, IPA) prostate cancer (PC3) versus normal prostate cells To further examine ANXA7-hnRNP associations, we (PrEC) used IPA sample analysis of Prostate Cancer Subtyping Next, we examined hnRNP effects on ANXA7 cDNA in the progression from benign prostatic hyper- and protein expression in PC3 versus PrEC (Figure 6). plasia (BPH) to localized prostate cancer (PCA) and Transfection with the predominantly nuclear B1-iso- its metastatic form (MET). Analysis of concordant form of hnRNPA2/B1 represents a panel of different to ANXA7 expression profiles in hnRNP-associated hnRNPs (data not shown). Compared with PrEC, PC3 networks represents putative molecular paths between possesses essentially higher levels (Po0.05) of baseline ANXA7 and hnRNPs in prostate cancer progression as well as the hnRNPB1-induced ANXA7 cDNA that (Figure 7). is consistent with the aforementioned ANXA7 up- Consistent with its tumor suppressor role, ANXA7 regulation in PC3. The baseline and hnRNPB1-induced expression in BPH gradually declines, reaching a ANXA7 protein levels are also drastically higher in PC3 substantial twofold decrease in MET (1.2 and À1.7, compared with PrEC. The initially elevated ANXA7 respectively). Similarly, hnRNP-A1 and -E2 decrease protein expression in PC3 cells is further increased by from BPH to MET (approximately 2- and 1.6-fold, hnRNPB1, while PrEC does not respond with a respectively), and hnRNPA3 is slightly reduced in both detectable increase. Validating cell-specific character of PCA and MET (hnRNPs A2/B1 and K were not the hnRNP-associated ANXA7 upregulation, ANXA7 available on arrays). ANXA7 GTP-ase activity as well protein levels in PC3 versus PrEC show similar profiles as Ca-binding and co-activation (Pollard et al., 1998) after transfection with other hnRNPs including A1, A2 and A3 (not shown). Most remarkably, the same ANXA7 antibody recog- nizes a single protein product in PrEC versus three products in PC3, especially in the hnRNPB1-transfected cells. ANXA7 protein products with similar MW (B50 kDa) in PC3 could correspond to three ANXA7 isoforms (Ensembl/Havana database, Hinxton/Cam- bridge, UK). Thus, identification of multi-hnRNP-complex bound to the steroid-sensitive ANXA7-promoter site in PC3 coincides with the enhanced expression of multiple ANXA7 products that is further induced by the hnRNPA2/B1 transfection in PC3, but not in PrEC. These data clearly indicate that hnRNPs distinctly

PrEC PC3

Control hnRNP B1 Control hnRNP B1 - ANXA7 (cDNA)

- beta-actin (cDNA)

- ANXA7 (protein)

- beta-actin (protein) Figure 7 ANXA7 and hnRNP-associated synexpression in the prostate cancer progression (IPA). The ANXA7 and hnRNP- Figure 6 The baseline and hnRNPB1-induced ANXA7 cDNA associated network was generated using Ingenuity Pathways and protein expression in prostate cancer (PC3) versus normal Analysis, IPA and Prostate Cancer Subtyping (a sample micro- prostate (PrEC) cells. ANXA7 message and protein expression was array-based analysis from IPA). Bar charts represent all expression assessed in PrEC and PC3 cells after transfection with a panel of values (fold change) in different stages: benign prostatic hyperpla- different hnRNPs (not shown), as described in Material and sia (BPH), prostate cancer without (PCA) and with metastases methods. The presented data (cropped images) show ANXA7 (MET), from left to right. Color coding in bar charts and gene cDNA and protein expression (reverse transcriptase (RT)–PCR symbols represents up- (red) and downregulated (green) genes. and Western, respectively) in response to hnRNPB1 versus control, Gene symbols are colored using the expression overlay for representing typical hnRNP-associated profiles in normal PrEC metastatic prostate cancer (MET); blank shapes represent the versus cancerous PC3. genes or chemicals with no available expression data.

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2462 present possible direct connections of ANXA7 to the copy number of ANXA7 (Wolf et al., 2004) can explain hnRNP-associated network. the increase in total gene and protein expression of The hnRNPK-interacting (Mikula et al., 2006) and ANXA7 in PC3, multiple ANXA7 products in PC3 Ca-binding calgizzarin (S100A11) that can bind versus PrEC indicate additional hnRNP-associated ANXA1 (Sakaguchi et al., 2007) is decreased similar effects on ANXA7 splicing in PC3. According to to ANXA7. A diverse cell growth regulator, hnRNPK ANXA7 expression profiling in cancer cells (Gene interacts with and is proteolyzed by in response Expression Omnibus and Stanford Microarray Data- to Ca2 þ (Kimura et al., 2003). Calcium signaling also base), only PC3 and PPC-1 cells, that share similar regulates the recruitment and binding of PI3K regula- karyotypes, have ANXA7 elevation, and an ANXA7 tory subunit alpha, PI3KR1, or PI3K p85-alpha. In expression profile in general is not concordant with cancer samples with high Gleason grade, PI3K pathway hnRNPs (Supplementary Figure S2–S3). Hence, the is marked by essential decreases (Hellwinkel et al., 2008) predominant effect of nuclear hnRNPs on ANXA7 including PIK3R1, which is similarly reduced (greater appears to be by alternative splicing rather than the than twofold) in the prostate cancer progression (IPA). change of total expression. The Ca/phospholipid-binding ANXA7, whose defi- Disruption of splicing is a hallmark in cancer, and ciency affects Ca-release by inositol-1,4,5-trisphos- some hnRNPs are oncogenes or cancer biomarkers that phate(IP3)-generating agonists (Srivastava et al., 1999), regulate tumor progression by controlling the expression can be directly linked to the PI3K-cascade including the of proteins involved in proliferation, apoptosis, angio- hnRNP-K and -E2-interacting PIK3R1. genesis and metastasis (Hilbe et al., 2004; Carpenter The downregulated in MET transportin-1 (TNPO1) et al., 2006; Cooper et al., 2009). Owing to pleiotropic can bind the nuclear localization and export signals hnRNP-effects on mRNA translation and stability, the (NLS/NES) to mediate nucleocytoplasmic trafficking, a hnRNP-associated ANXA7 regulation can distinctly process regulated by RanGTPase. The structure of affect ANXA7 splicing and expression in normal versus TNPO1, bound to hnRNPA1 NLS, explains the cancerous tissues. It is therefore not surprising that mechanism of substrate displacement by RanGTPase, ANXA7 splice variants have cancer-specific distribution predicting that the ANXA7-homologous ANXA11 can (see Alternative Splicing Annotation Project, http:// be a TNPO1 substrate (Lee et al., 2006). Although some bioinfo.mbi.ucla.edu/ASAP). The ANXA7 variant_1 hnRNPs are strictly nuclear, the cargo hnRNP-A1, -A2 (134178) is detected more frequently in tumors, and and -K, which are bound to the ANXA7 promoter, variant_2 (134182 and 134183) is mostly detected in contain bidirectional signals conferring both import normal tissues. Accordingly, identification of ANXA7 and export (Nakielny and Dreyfuss, 1999). With the along with GR and hnRNP-SNRPB2 in the invasion- RanGTPase cycle having a key role in transport promoting MT1-MMP network was rightfully attribu- directionality, the ANXA7 GTPase activity (Pollard ted to cancer-specific splicing (Rozanov et al., 2008). et al., 1998) may link ANXA7 to the hnRNP-mediated Consistent with numerous ASAP-predicted ANXA7- nucleocytoplasmic shuttling. transcripts, Genomatix identifies regulatory elements Cell survival regulating 14-3-3-proteins are implicated that indicate alternative regulators acting through in RNA machinery involving pre-mRNA splicing (Shi multiple initiation sites in ANXA7 promoter. The and Manley, 2007) and sorting of RNA granules XCPE1, identified in ANXA7 core promoter, usually (Courchet et al., 2008). The anti-apoptotic 14-3-3-zeta shows weak transcriptional activity alone that could (YWAZ) as well as angiotensinogen (AGT), which is explain genuinely low ANXA7 expression. However, also involved in cell survival, have a profile opposite to XCPE1 exerts significant activity when accompanied by ANXA7 and similar to FYN and EGFR oncogenes. specific activator-binding sites (Tokusumi et al., 2007). The ubiquitin-antagonist sentrin (SUMO1), which is The recruitment of transcription factors to XCPE1 may involved in hnRNP-A1 and -K sumoylation (Li et al., require bridging interactions such as PPAR-g coactiva- 2004) is increased in BPH (1.6). Being particularly tor-1, suggesting a specific role for the lipid-associated abundant in the prostate, SUMO1 is controlled by PPAR-g in ANXA7 activation. Meanwhile, TBP- 5alpha-dihydrotestosterone, DHT (Caron et al., 2008) associated factor 1 (TFIID) has been shown to bind to and can inhibit the AR transcriptional activity (Zheng the TATA-lacking ANXA7 promoter (Kim et al., 2005). et al., 2006). Thus, utilization of the TFIID-STAF-site in a distant Thus, IPA shows multiple links between ANXA7 and promoter (F-segment) versus the XCPE1-site in the core the hnRNP-associated network, presumably involved in promoter can result in distinct transcription initiation ANXA7 control in the prostate cancer progression. and RNA splicing of ANXA7. Indicating differential cytotoxicity, the transfected short ANXA7-isoform reduces cell survival in prostate cancer cells, including PC3, but not in normal PrEC Discussion (Torosyan et al., 2009). However, wild-type and dominant-negative forms of the short ANXA7 have The ANXA7 promoter appears to be a sophisticated more similar cytotoxicity in PC3 compared with other transcription gateway modulated by diverse signals cancer cells, in which predominant cytotoxicity is shown leading to alternative transcription initiation and spli- by the wild-type-ANXA7. To reconcile these counter- cing. In particular, although the specifically high gene intuitive results, we suggest that the hnRNP-controlled

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2463 ANXA7 transcription and splicing in parental PC3 regulation of ANXA7 and lipoxygenases (Torosyan results in some ANXA7-isoforms with reduced cyto- et al., 2006). toxicity. As a result, multiple endogenous ANXA7 Thus, this study provides mechanistic insights into the products in PC3 appear to intervene with the effects of hnRNP-associated ANXA7 control that can hinder exogenous ANXA7. ANXA7 cell survival regulation. As cancer-related Representing RNA link in cancer (Cooper et al., RNA changes can represent potential biomarkers and 2009), aberrant splicing and inappropriate expression entry points for therapeutic intervention, the hnRNP co- can be based on a disrupted balance between hnRNP regulation of tumor suppressor ANXA7 uncovers an and SR-proteins in the spliceosome assembly that attractive target in prostate carcinogenesis. Further regulates alternative splicing producing appropriate elucidation of the steroid/hnRNP-associated control ratios of pro- and antiapoptotic isoforms. Hence, the over ANXA7 transcription, RNA processing, and notion of tumor-type-specific regulation of individual translation can shed light on ANXA7 tissue-specific annexins, including ANXA7 (Duncan et al., 2008), can distribution and distinct cell survival effects in normal be updated by the introduction of splicing in ANXA7 versus cancer cells. control. Consequently, the annexin’s pattern in cancer can be considered pathogenic only after the identifica- tion of corresponding isoforms and their functional Materials and methods consequences. Similar to ANXA7 cell-type specific and isoform- Cell culturing associated effects, small interference RNA-silencing of Normal prostate (PrEC) and prostate cancer (LNCaP, DU145 hnRNP-A1 and -A2/B1-induces apoptosis in cancerous, and PC3) cells (ATCC, Manassas, VA, USA) were routinely but not normal, cells (Patry et al., 2003). Suppression grown, as described previously (Torosyan et al., 2009). of hnRNPA2 increases p21 (He et al., 2005), while downregulation of hnRNPK and p21 switches between ANXA7-promoter segment enhanced activity apoptosis and growth arrest induced by p53 (Enge et al., ANXA7-promoter segments (F1–F5) were generated, as 2009). Hence, the multi-hnRNP-regulated ANXA7 described previously (Srivastava and Pollard, 2000). Using could interfere with p53 cell survival pathways in Dual-Luciferase reporter assay (Promega, Madison, WI, p53/PTEN-null PC3. USA), activities of co-transfected ANXA7-promoter segments The lack of proper AR signaling in androgen-resistant were normalized to an internal control and quantified in PC3 can affect the recruitment of hnRNPs to the target triplicate experiments (mean±s.d.). promoters, intervening with the AR-associated tran- script production and splicing. The overexpression of an ANXA7-promoter analysis by Genomatix hnRNP and an estrogen response (ERE)-binding ANXA7-promoter sequences (À2500/ þ 500 bp) were analyzed protein, both of which compete with ER for binding using MatInspector Release professional 8.0. from Genomatix to the ERE-element, has been shown to result in software package (GmbH, Munich, Germany). Identified estrogen resistance in breast cancer (Chen et al., 2005). regulatory elements were mapped to corresponding ANXA7- promoter segments (F1–F5). The overlapping double Similarly, the overrepresented multi-hnRNP-complex at glucocorticoid-responsive and related elements V$GREF: ANXA7 promoter in PC3 could affect GR and AR V$GRE.02 and V$ARE.02 for AR and GR, respectively, binding to the GREF/ARE-elements. As GR and AR were found in the segment (F) that comprised the difference can mediate opposite effects on cell metabolism and between F1 and F2. proliferation, their signaling pathways can be mutually inhibitory (Chen et al., 1997). Hence, the multi-hnRNP Preparation of specific probe complex at the GREF-site can influence the ANXA7 The F-segment (À1086/À890) was selected as a specific probe tumor suppressor role by affecting the formation of AR- and synthesized using following primers: forward—50-CTCAG or GR-homodimers versus AR-GR-heterodimers, and GGAAGGCTTAAATCCTG-30 and reverse—50-CAAATAG thereby causing diverse cell survival effects. The role of GTGTGTGGAGTGAAAAG-30 (Gene Probe Technologies, AR-signaling in ANXA7 transcription is also empha- Gaithersburg, MD, USA) for subsequent use in gelshift sized by the lowest ANXA7-promoter activity in assay and MS-based DNA-affinity enrichment/purification of androgen-sensitive LNCaP. nuclear proteins. hnRNPK has been identified as a novel AR transla- tion regulator within the EGFR/PI3K/PTEN/Akt/ Radioactive probe labeling and electrophoretic mobility shift mTOR pathway (Mukhopadhyay et al., 2009) that can (gelshift) assay serve a platform role in stabilizing the AR-promoter The F-segment of ANXA7 promoter was synthesized as a activating complex (Shi et al., 2008). The hnRNPK level specific radioactive probe using the aforementioned primers within the nuclear matrix in particular correlates with and g-32P-adenosine triphosphate with 1X Polynucleotide the Gleason score in prostate cancer (Barboro et al., Kinase buffer (Amersham, GE Healthcare, Piskataway, NJ, USA), as suggested by the manufacturer. The radioactively 2009). Hence, the nuclear hnRNPK at the ANXA7 labeled primers were added into the PCR mixture with Pfu- promoter further connected the phospholipid-binding polymerase (Stratagene, Agilent Technologies, Santa Clara, and hormone-related ANXA7 to the major lipid- CA, USA). The PCR was performed by GeneAmp 2400 relevant cell survival cascade and AR signaling in (PerkinElmer, Waltham, MA, USA) and subjected to phenol- prostate carcinogenesis, consistent with suggested co- chloroform extraction and ethanol precipitation. The radio-

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2464 actively labeled PCR product was purified on 4% 1XTAE primers for ANXA7 (forward—50-ATGTCATACCCAGGC polyacrylamide gel by electrophoretic separation, excised from TATCCCCCA-30 and reverse—50-AGGCATGAAGAGGGC the gel, eluted in 1X Binding Buffer and processed as described CAGGATCAGT-30) and beta-actin (forward—50-AAGAGAG by Dobi et al., 2002. The radioactively labeled DNA probe and GCATCCTCACCCT-30 and reverse—50-TACATGGCTGG poly(d(I-C)) were mixed with PrEC and PC3 nuclear extracts. GGTGTTGAA-30). After incubation, the samples were analyzed on 4% non- denaturing polyacrylamide gel by electrophoresis in 1XTBE Western blotting buffer. The gel was fixed in 10% acetic acid, transferred to Western blotting was performed by standard procedure using Whatman paper, and analyzed by autoradiography. PrEC and PC3 cell lysates and antibodies for ANXA7 and beta-actin (BD Transduction Laboratories). DNA affinity/purification of nuclear proteins followed by mass spectrometry Statistical analysis The DNA affinity/enrichment/purification of proteins from Statistical analysis was performed using a Student’s two-sided PrEC and PC3 nuclear extracts was performed according to t-test for independent samples; P-values o0.05 were consid- the protocols modified by A Dobi, as previously described ered significant. Intensity of bands (arbitrary units) in ANXA7 (Torosyan et al., 2006). The specific probe corresponding to expression analysis was assessed using ImageJ (http://rsb. the F-segment (À1086/À890) of ANXA7 promoter was onfo.nih.gov/ij). synthesized as double-stranded DNA with biotinylation (Gene IPA (www.ingenuity.com, Ingenuity Systems, Redwood Probe Technologies, Gaithersburg, MD, USA). The bands City, CA, USA) was used to examine potential common that were visualized in nonspecific competition in PC3 (À) and pathways for ANXA7 and hnRNPs. Using the Prostate corresponded to the bands competed-out by the specific DNA Cancer Subtyping analysis data (IPA), we first identified probe ( þ ) were excised, digested, and analyzed by MALDI- ANXA7 expression profile, which represented a gradual TOF MS using ThermoFinnegan LCQ (San Jose, CA, USA) downregulation from benign prostate hyperplasia to the and Voyager DE STR MALDI-TOF Workstation (Applied localized and metastatic prostate cancers. Next, we generated Biosystems, Foster City, CA, USA). hnRNP-associated networks with genes that have been substantially altered during the prostate cancer progression (RXtwofold). Using the generated networks with immediate Antibody interference assay binding or interacting relations with different hnRNPs (A1, Antibody interference assay was performed as previously A2/B1, A3, E2 and K), we selected the genes with concordant described (Dobi et al., 2002) using PC3 nuclear extracts and to ANXA7 expression. Using GTP and Ca2 connections, we antibodies for GR (BD Transduction Laboratories, San Jose, þ integrated the Ca-binding ANXA7-GTPase with hnRNP- CA, USA) and hnRNPA2/B1 (a gift from Dr G Dreyfuss, associated genes (nodes) and relationships (edges), creating a University of Pennsylvania School of Medicine). The anti- network with plausible molecular paths between ANXA7 and bodies were mixed with PC3 nuclear extracts in Binding hnRNPs in the prostate cancer progression. Buffer, and the reactions were incubated at 0 1C. The synthesized specific probe (F-segment) and poly(d(I-C)) were mixed in 1 Â Binding Buffer and added to the antibody- containing mixtures. After incubation, the samples were Abbreviations aliquoted and electrophorezed on 4% non-denaturing poly- acrylamide gel in 1XTBE buffer. The gel was fixed in 10% ANXA7, Annexin A7; AR, androgen receptor; GR, gluco- acetic acid, transferred to Whatman paper and analyzed by corticoid receptor; hnRNP, heterogeneous nuclear ribonucleo- autoradiography. protein; IPA, ingenuity pathways analysis; MALDI-TOF MS, matrix-assisted laser desorption time-of-flight mass spectro- Cell transfection metry. PrEC and PC3 cells were grown in six-well plates (24 h), transfected with different hnRNP constructs: A1, A2, B1 and A3 (Burd et al., 1989) using Lipofectamine reagent (Invitro- Conflict of interest gen, Carlsbad, CA, USA), and harvested (24 h) for cell lysates. The authors declare no conflict of interest. RNA extraction cDNA synthesis, and reverse transcriptase–PCR After hnRNP transfection, RNA was extracted from PrEC and PC3 cells using RNAqueous-4PCR Kit (Ambion, Austin, Acknowledgements TX, USA) and transcribed into cDNA using Reverse Transcription System (Promega). Reverse transcriptase– We thank Dr Gideon Dreyfuss (University of Pennsylvania PCR was performed using GeneAmp PCR System 2400 School of Medicine) for providing hnRNPA2/B1 antibody. (PerkinElmer) and Expand Long Template PCR System This work was supported by grants funded by the US (Roche Diagnostics, Indianapolis, IN, USA) with following Department of Defense (DoD).

References

Adachi K, Tanaka T, Saito H, Oka T. (1999). Hormonal induction of Auboeuf D, Honig A, Berget SM, O0Malley BW. (2002). Coordinate mouse selenocysteine transfer ribonucleic acid (tRNA) gene regulation of transcription and splicing by steroid receptor transcription-activating factor and its functional importance in the coregulators. Science 298: 416–419. selenocysteine tRNA gene transcription in mouse mammary gland. Barboro P, Repaci E, Rubagotti A, Salvi S, Boccardo S, Spina B et al. Endocrinology 140: 618–623. (2009). Heterogeneous nuclear ribonucleoprotein K: altered pattern

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2465 of expression associated with diagnosis and prognosis of prostate Karagianni N, Tsawdaroglou N. (1994). The c-fos serum response cancer. Br J Cancer 100: 1608–1616. element (SRE) confers negative response to glucocorticoids. Burd CG, Swanson MS, Go¨rlach M, Dreyfuss G. (1989). Primary Oncogene 9: 2327–2334. structures of the heterogeneous nuclear ribonucleoprotein A2, B1, Kim TH, Barrera LO, Zheng M, Qu C, Singer MA, Richmond TA and C2 proteins: a diversity of RNA binding proteins is generated et al. (2005). High-resolution map of active promoters in the human by small peptide inserts. Proc Natl Acad Sci USA 86: 9788–9792. genome. Nature 436: 876–880. Caron D, Winstall E, Inaguma Y, Michaud S, Lettre F, Bourassa S Kimura E, Abe K, Suzuki K, Sorimachi H. (2003). Heterogeneous et al. (2008). Proteomic characterization of mouse cytosolic and nuclear ribonucleoprotein K interacts with and is proteolyzed by membrane prostate fractions: high levels of free SUMO peptides are calpain in vivo. Biosci Biotechnol Biochem 67: 1786–1796. androgen-regulated. J Proteome Res 7: 4492–4499. Lee BJ, Cansizoglu AE, Su¨el KE, Louis TH, Zhang Z, Chook YM. Carpenter B, MacKay C, Alnabulsi A, MacKay M, Telfer C, Melvin (2006). Rules for nuclear localization sequence recognition by WT et al. (2006). The roles of heterogeneous nuclear ribonucleo- karyopherin beta 2. Cell 126: 543–558. proteins in tumour development and progression. Biochim Biophys Li T, Evdokimov E, Shen RF, Chao CC, Tekle E, Wang T et al. Acta 1765: 85–100. (2004). Sumoylation of heterogeneous nuclear ribonucleoproteins, Chen H, Stuart W, Hu B, Nguyen L, Huang G, Clemens TL et al. zinc finger proteins, and nuclear pore complex proteins: a proteomic (2005). Creation of estrogen resistance in vivo by transgenic analysis. Proc Natl Acad Sci USA 101: 8551–8556. overexpression of the heterogeneous nuclear ribonucleoprotein- Magendzo K, Shirvan A, Cultraro C, Srivastava M, Pollard HB, related estrogen response element binding protein. Endocrinology Burns AL. (1991). Alternative splicing of human synexin mRNA in 146: 4266–4273. brain, cardiac, and skeletal muscle alters the unique N-terminal Chen S, Wang J, Yu G, Liu W, Pearce D. (1997). Androgen and domain. J Biol Chem 266: 3228–3232. glucocorticoid receptor heterodimer formation. A possible mechan- Mikula M, Dzwonek A, Karczmarski J, Rubel T, Dadlez M, Wyrwicz ism for mutual inhibition of transcriptional activity. J Biol Chem LS et al. (2006). Landscape of the hnRNP K protein-protein 272: 14087–14092. interactome. Proteomics 6: 2395–2406. Chen TC, Schwartz GG, Burnstein KL, Lokeshwar BL, Holick MF. Mukhopadhyay NK, Kim J, Cinar B, Ramachandran A, Hager MH, (2000). The in vitro evaluation of 25-hydroxyvitamin D3 and Di Vizio D et al. (2009). Heterogeneous nuclear ribonucleoprotein 19-nor-1alpha,25-dihydroxyvitamin D2 as therapeutic agents for K is a novel regulator of androgen receptor translation. Cancer Res prostate cancer. Clin Cancer Res 6: 901–908. 69: 2210–2218. Cooper TA, Wan L, Dreyfuss G. (2009). RNA and disease. Cell 136: Myslinski E, Ge´rard MA, Krol A, Carbon P. (2006). A genome scale 777–793. location analysis of human Staf/ZNF143-binding sites suggests a Courchet J, Buchet-Poyau K, Potemski A, Bre` s A, Jariel-Encontre I, widespread role for human Staf/ZNF143 in mammalian promoters. Billaud M. (2008). Interaction with 14-3-3 adaptors regulates the J Biol Chem 281: 39953–39962. sorting of hMex-3B RNA-binding protein to distinct classes of Nakielny S, Dreyfuss G. (1999). Transport of proteins and RNAs in RNA granules. J Biol Chem 283: 32131–32142. and out of the nucleus. Cell 99: 677–690. Dobi A, Debnam W, Dalgard C, Owusu A, von Agoston D. (2002). Patry C, Bouchard L, Labrecque P, Gendron D, Lemieux B, Toutant J Mammalian expression cloning of nucleic acid binding proteins et al. (2003). Small interfering RNA-mediated reduction in by agarose thin-layer gelshift clone selection. Biotechniques 33: heterogeneous nuclear ribonucleoparticule A1/A2 proteins induces 868–872. apoptosis in human cancer cells but not in normal mortal cell lines. Duncan R, Carpenter B, Main LC, Telfer C, Murray GI. (2008). Cancer Res 63: 7679–7688. Characterisation and protein expression profiling of annexins in Pollard HB, Caohuy H, Minton AP, Srivastava M. (1998). Synexin colorectal cancer. Br J Cancer 98: 426–433. (annexin VII) hypothesis for Ca2+/GTP-regulated exocytosis. Adv Eggert H, Schulz M, Fackelmayer FO, Renkawitz R, Eggert M. (2001). Effects of the heterogeneous nuclear ribonucleoprotein U Pharmacol 42: 81–87. (hnRNP U/SAF-A) on glucocorticoid-dependent transcription Reed R, Hurt E. (2002). A conserved mRNA export machinery in vivo. J Steroid Biochem Mol Biol 78: 59–65. coupled to pre-mRNA splicing. Cell 108: 523–531. Enge M, Bao W, Hedstro¨m E, Jackson SP, Moumen A, Selivanova G. Rizzo G, Fiorucci S. (2006). PPARs and other nuclear receptors in (2009). MDM2-dependent downregulation of p21 and hnRNP K inflammation. Curr Opin Pharmacol 6: 421–427. provides a switch between apoptosis and growth arrest induced by Rozanov DV, Savinov AY, Williams R, Liu K, Golubkov VS, pharmacologically activated p53. Cancer Cell 15: 171–183. Krajewski S et al. (2008). Molecular signature of MT1-MMP: Graveley BR. (2009). Alternative splicing: regulation without regula- transactivation of the downstream universal gene network in cancer. tors. Nat Struct Mol Biol 16: 13–15. Cancer Res 68: 4086–4096. He Y, Brown MA, Rothnagel JA, Saunders NA, Smith R. (2005). Sakaguchi M, Murata H, Sonegawa H, Sakaguchi Y, Futami J, Roles of heterogeneous nuclear ribonucleoproteins A and B in cell Kitazoe M et al. (2007). Truncation of is a regulatory proliferation. J Cell Sci 118: 3173–3183. lever for linking epidermal growth factor signaling with cytosolic Heemers HV, Verhoeven G, Swinnen JV. (2006). Androgen activation phospholipase A2 in normal and malignant squamous epithelial of the sterol regulatory element-binding protein pathway: current cells. J Biol Chem 282: 35679–35686. insights. Mol Endocrinol 20: 2265–2277. Shi L, Ko S, Kim S, Echchgadda I, Oh TS, Song CS et al. (2008). Loss Hellwinkel OJ, Rogmann JP, Asong LE, Luebke AM, Eichelberg C, of androgen receptor in aging and oxidative stress through Myb Ahyai S et al. (2008). A comprehensive analysis of transcript protooncoprotein-regulated reciprocal chromatin dynamics of p53 signatures of the phosphatidylinositol-3 kinase/protein kinase B and poly(ADP-ribose) polymerase PARP-1. J Biol Chem 283: signal-transduction pathway in prostate cancer. BJU Int 101: 36474–36485. 1454–1460. Shi Y, Manley JL. (2007). A complex signaling pathway regulates Hilbe W, Dirnhofer S, Greil R, Woll E. (2004). Biomarkers in non- SRp38 phosphorylation and pre-mRNA splicing in response to heat small cell lung cancer prevention. Eur J Cancer Prev 13: 425–436. shock. Mol Cell 28: 79–90. Juven-Gershon T, Hsu JY, Theisen JW, Kadonaga JT. (2008). The Shirvan A, Srivastava M, Wang MG, Cultraro C, Magendzo K, RNA polymerase II core promoter—the gateway to transcription. McBride OW et al. (1994). Divergent structure of the human Curr Opin Cell Biol 20: 253–259. synexin (annexin VII) gene and assignment to 10. Kang HS, Angers M, Beak JY, Wu X, Gimble JM, Wada T et al. Biochemistry 33: 6888–6901. (2007). Gene expression profiling reveals a regulatory role for ROR Srivastava M, Atwater I, Glasman M, Leighton X, Goping G, alpha and ROR gamma in phase I and phase II metabolism. Physiol Caohuy H et al. (1999). Defects in inositol 1,4,5-trisphosphate Genomics 31: 281–294. receptor expression, Ca(2+) signaling, and insulin secretion in

Oncogene hnRNPs in ANXA7 regulation in prostate cancer Y Torosyan et al 2466 the anx7(+/À) knockout mouse. Proc Natl Acad Sci USA 96: Torosyan Y, Dobi A, Naga S, Mezhevaya K, Glasman M, Norris C 13783–13788. et al. (2006). Distinct effects of annexin A7 and p53 on arachidonate Srivastava M, Montagna C, Leighton X, Glasman M, Naga S, lipoxygenation in prostate cancer cells involve 5-lipoxygenase Eidelman O et al. (2003). Haploinsufficiency of Anx7 tumor transcription. Cancer Res 66: 9609–9616. suppressor gene and consequent genomic instability promotes Torosyan Y, Simakova O, Naga S, Mezhevaya K, Leighton X, Diaz J tumorigenesis in the Anx7(+/À) mouse. Proc Natl Acad Sci USA et al. (2009). Annexin-A7 protects normal prostate cells and induces 100: 14287–14292. distinct patterns of RB-associated cytotoxicity in androgen-sensitive Srivastava M, Pollard HB. (2000). Low in vivo levels of human anx7 and -resistant prostate cancer cells. Int J Cancer 125: 2528–2539. Wolf M, Mousses S, Hautaniemi S, Karhu R, Huusko P, Allinen M (annexin VII) gene expression are due to endogenous inhibitory et al. (2004). High-resolution analysis of gene copy number promoter sequences. Cell Biol Int 24: 475–481. alterations in human prostate cancer using CGH on cDNA Srivastava M, Torosyan Y, Raffeld M, Eidelman O, Pollard HB, microarrays: impact of copy number on gene expression. Neoplasia Bubendorf L. (2007). ANXA7 expression represents hormone- 6: 240–247. relevant tumor suppression in different cancers. Int J Cancer 121: Yemelyanov A, Czwornog J, Chebotaev D, Karseladze A, Kulevitch 2628–2636. E, Yang X et al. (2007). Tumor suppressor activity of glucocorticoid Tokusumi Y, Ma Y, Song X, Jacobson RH, Takada S. (2007). The receptor in the prostate. Oncogene 26: 1885–1896. new core promoter element XCPE1 (X Core Promoter Element 1) Zheng Z, Cai C, Omwancha J, Chen SY, Baslan T, Shemshedini L. directs activator-, mediator-, and TATA-binding protein-dependent (2006). SUMO-3 enhances androgen receptor transcriptional but TFIID-independent RNA polymerase II transcription from activity through a sumoylation-independent mechanism in prostate TATA-less promoters. Mol Cell Biol 27: 1844–1858. cancer cells. J Biol Chem 281: 4002–4012.

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

Oncogene