The role of prostate tumor overexpressed 1 in cancer progression

Prostate-Tumor-Overexpressed-1 (PTOV1) is a conserved adaptor protein discovered as overexpressed in prostate cancer. Since its discovery, the number of binding partners and associated cellular functions has increased and helped to identify PTOV1 as regulator of gene expression at transcription and translation levels. Its overexpression is associated with increased tumor grade and proliferation in prostate cancer and other neoplasms, including breast, ovarian, nasopharyngeal, squamous laryngeal, hepatocellular and urothelial carcinomas. An important contribution to higher levels of PTOV1 in prostate tumors is given by the frequent rate of gene amplifications, also found in other tumor types. The recent resolution of the structure by NMR of the PTOV domain in PTOV2, also identified as Arc92/ACID1/MED25, has helped to shed light on the functions of PTOV1 as a transcription factor. In parallel, by studying its interaction with RACK1, we have discovered PTOV1 action in promoting mRNAs translation. Here, we will focus on the role of PTOV1 in cancer, re-examine its pro-oncogenic effects and re-evaluate the most relevant interactions and evidences of its cellular functions. The data are used to formulate a model for the mechanisms of action of PTOV1 in line with its recently described activities and cellular pathways modulated in cancer.

The gene PTOV1 spans 9.51 Kb and includes 12 exons of which exons 3 to 6 code for the first PTOV homology block, or A domain, and exons 7 to 12 for the second block, or B domain [1] (Figure 1A). At least two transcripts resulting from differential splicing have been described: the first one includes 1,875 bps and translates for a protein of 416 residues, the second one includes 1,539 bps and translate for protein of 374 amino acids [6]. Two alternative splicing events at the 5' and 3' ends in the latter transcript result in a shorter protein with different N-terminal and C-terminal residues. These differences likely affect the function of the final product.
A related gene containing one PTOV-domain, identified as PTOV2, was discovered on the same chromosome, 14 Kb upstream from the putative first exon of PTOV1 [1]. PTOV2 was later found to correspond to the subunit 25 of the Mediator complex, MED25, composed of up to thirty proteins, that is an essential component of the eukaryotic RNA polymerase II complex regulating Review eukaryotic transcription [7][8][9]. The Mediator functions as a transcription coactivator in all eukaryotes. It is also referred as the vitamin D receptor interacting protein (DRIP) coactivator complex and the thyroid hormone receptor-associated proteins (TRAP) [10]. It is required for the transcription of most class II genes in yeasts and mammals and is essential for activator-dependent transcription. The Mediator acts as a bridge between the RNA polymerase II and the activator transcription factors [10]. MED25 was also identified as p78/Arc92/ACID1 [8,11]. In this review, we will refer to PTOV2 as MED25.
The orthologous PTOV1 gene in Drosophila melanogaster (acc. AC013074) shows a similar modular arrangement, but a lower degree of similarity between the two domains A and B (33% identity; 57% similarity) [1].
The orthologous MED25 gene, found on chromosome 3R at 3Mb from PTOV1, was also predicted to encode a protein with a single PTOV block embedded within an 863-amino acid protein.
The putative promoter region of PTOV1 reveals the presence of consensus sequences for transcription factors SP1 and AP2 and a putative Androgen Responder Element (ARE), similar to those observed in the PSA gene. In agreement with these observations, PTOV1 expression is androgen-responsive [1]. This has been confirmed in vascular smooth muscle cells (VSMCs) where the gene was suggested to play a critical role in androgen related atherogenesis in the human aorta through the regulation of proliferation of neointimal VSMCs [12].  [13].

THE PROTEIN STRUCTURE
PTOV1 is an adaptor, conserved in vertebrates (mammals and fish) and in arthropods (insects), although not in fungi (yeasts). The protein interacts with a number of factors both in the nucleus and the cytoplasm to regulate gene expression at transcription and posttranscriptional levels and to promote cancer cell proliferation and motility.
The predicted protein, 416 aminoacids long, reflects the structure of the gene and presents two highly homologous domains arranged in tandem, identified as A domain (146 amino acids) and B domain (143 amino acids), that show 66% identity and 79% similarity among each other [1,13] ( Figure 1B). The protein presents two putative nuclear localization signals (NLS), one in the A domain and the second in the B domain. Recent studies using NMR technology revealed the structure of the PTOV domain of MED25 [14][15][16]. This domain (391-543) exhibits high sequence identity with the A and B domains of human PTOV1 (81% and 73%, respectively). The secondary structure of MED25 forms a seven-stranded β-barrel framed by three α-helices of topology A(↑)B(↓) D(↑)G(↓)F(↑)E(↓)C(↑) in which strands C, E, F and G are flanked by α-helices I and III, and helix II connects strand D and E at one end of the barrel [15,16]. MED25 also presents two different positively electrostatic charged regions that appear to be important for its binding to the trans activation domain (TAD) of the RNA Polymerase II.
The PTOV domain in MED25 shares structural similarity with the β-barrel domains of the Ku70/Ku80 heterodimer and the Spen paralog and ortholog C-terminal (SPOC) domain of SMRT/HDAC1-associated repressor protein (SHARP) [14][15][16]. These proteins are involved in transcriptional activation, double-stranded DNA break repair, and transcriptional repression by interaction with histone deacetylase complexes. For instance, SHARP has been identified as a component of the Notch co-repressor complex [17]. The PTOV domain shares the central β-barrel and two flanking helices with the SPOC domain in SHARP, but differs from the SPOC domain by the absence of four helices that flank the barrel, the presence of a C-terminal helix α3 with a unique location and a long loop connecting β1 and β2 [16]. These observations suggest that the PTOV domain in MED25 and PTOV1 might have a crucial function in chromatin remodeling and for connecting the activity of several transcription factors to the RNA Polymerase II complex. However, in contrast to MED25, PTOV1 did not interact with the Mediator complex [11].
Very recently, an extended (e)AT-hook motif comprising the first N-terminal 43 amino acids was identified in PTOV1, suggesting a function as nucleic acid binding protein ( Figure 1B) [18]. Classical AT-hook motifs in proteins can bind nucleic acids at AT-rich sequences in the minor groove of the DNA, and have been described as characteristics of proteins associated with chromatin remodeling, histone modifications, chromatin insulator and proposed to anchor chromatin-modifying proteins [19,20]. However, the (e)AT-hook discovered at the N-terminal of PTOV1 is a new functional AT-hook-like motif that differs in the basic amino acid patches from the canonical G-R-P (glycine-arginine-proline) core. This motif showed higher RNA binding affinity compared to DNA [18], and its deletion resulted in a stronger signal of the mutated protein in the nucleus. These apparently contrasting observations may be explained considering the following findings. We have previously shown that the nuclear localization of a GFP-PTOV1, where the first 56 amino acids comprising the (e)AT-hook were replaced by GFP, did not significantly accumulate in the nucleus, suggesting that other features, such as two conserved NLS signals, might be responsible for the localization of the (e) AT-hook deleted PTOV1 [21,22]. In addition, we have also shown that PTOV1 enters the nucleus at the beginning of the S-phase of the cell cycle [21]. Furthermore, the higher affinity of the (e)AT-hook for RNA may suggest that PTOV1 can be found in ribonucleic-complexes that shuttle their carriers (mRNA) from the nucleus to the cytoplasm, in agreement with its interaction with RACK1 in ribosomes. Thus, the accumulation of the (e)AT-hook deleted PTOV1 in the nucleus observed by Filarsky and collaborators [18], might be the result of the cell type used, the cycle phase at the time of observations, and/or the lack of ribonucleic complexes formation that would shuttle PTOV1 and its cargo to the cytoplasm.

PTOV1 EXPRESSION IN NORMAL AND CANCER TISSUES
PTOV1, was suggested to be one of the genes most discerning between normal and carcinomatous prostate [23]. Most functions of PTOV1 have been surmised in pathological conditions, such as cancerous cells and tissues. However, its expression has also been detected in normal tissues (human brain, heart, skeletal muscle, kidney, liver, ovary, aorta and salivary gland), although little has been described about its function at those sites [1,12,24,25]. The expression of PTOV1 in human abdominal aorta and VSMCs cells is androgen-responsive and may play a role in androgen related atherogenesis [12]. In normal epithelial cells from different tissues, PTOV1 is mostly undetectable or shows a weak staining [21,26,27]. However, few sporadic prostatic luminal cells in isolated glands display intense staining, mostly cytoplasmic and in some cases also nuclear. These PTOV1-positive cells also stain for chromogranin A, suggesting their neuroendocrine origin [21].
In pre-neoplastic lesions of High Grade Prostate Epithelial Neoplasia, HGPIN, associated with prostate carcinomas, PTOV1 expression is increased compared to the normal prostate epithelium [28,29]. This higher expression in HGPIN was found helpful to discriminate those premalignant lesions associated with cancer, suggesting the potential value of PTOV1 detection in the early diagnosis of PC [28]. Of interest, the proportions of Ki-67 positive nuclei and the levels of PTOV1 in HGPIN areas adjacent to cancer lesions are higher than those found in HGPIN areas away from the cancer, supporting the concept of field cancerization or field effect in prostatic carcinogenesis [29,30]. More recently, the expression of PTOV1 in atypical adenomatous hyperplasia (AAH), a proliferative lesion of the transition zone of the prostate that morphologically resembles low grade carcinoma, has been associated with PC [31].
In prostate adenocarcinoma 71% of T2 and T3 stages overexpressed PTOV1 [21]. This overexpression was limited to the cytoplasm in 59% of samples whereas a strong expression was detected both in the nucleus and the cytoplasm in the remaining samples, with a small proportion of tumors showing a strong nuclear staining and weak cytoplasmic staining. More recently, metastatic primary tumors and metastatic lesions were shown to express significantly higher levels of PTOV1 compared to non-metastatic tumors [32,33]. This high expression of PTOV1 significantly associated to the Ki67 index suggesting its participation in an active proliferative status. Remarkably, this relationship was stronger in tumors with nuclear PTOV1 staining. These findings are supported by observations in vitro, where in quiescent PC cells, PTOV1 localized in the cytoplasm but after serum stimulation it partially translocated to the nucleus at the beginning of the S phase [21]. The transfection of PTOV1 forced PC3 cells to enter the S phase with a subsequently increase in the levels of cyclin D1, indicating that the overexpression of PTOV1 can directly increase the proliferation of PC cells. In addition, PTOV1 is required for the growth and full metastatic potential of PC cells in vivo [32,33]. An important contribution to higher levels of PTOV1 in aggressive prostate tumors might be given by the high rate of PTOV1 amplifications, also found in other tumor types, as shown by analyses of publicly available genomic datasets ( Figure 2A) [34,35]. Significantly, the highest frequency of amplifications (above 18%) is found in metastatic lesions of adenocarcinomas and neuroendocrine prostate tumors (Figure 2A) [36]. Because very high expression of PTOV1 in sporadic luminal cells in prostate glands associated to cancer very likely correspond to neuroendocrine cells [21], altogether the above findings suggest that PTOV1 expression might be connected to the aggressive features of neural subtypes of PC defined by expression profiles [37,38].
More recently, numerous reports have described the overexpression of PTOV1 in different types of tumors. These included tumors of the breast, pancreas, liver, colon, kidney, bladder, laryngeal, cerebral gliomas and ovary [25][26][27][39][40][41][42][43]. The association between its overexpression and high grade of malignancy was strong in primary hepato-cellular carcinoma, epithelial ovarian cancer, breast cancer and clear cell renal carcinomas, where PTOV1 expression was closely correlated with the clinicpathological characteristics and tumor aggressiveness. In addition, a significant association between high levels of PTOV1 and unfavorable prognosis and poor survival has been observed in several types of carcinomas. In breast cancer, 99.4% of the cancer samples analyzed expressed PTOV1, of which 49.1% showed high expression. The median survival of patients with high PTOV1 levels was 78 months versus 115 months in patients with low PTOV1 expression [27]. In laryngeal squamous cell carcinoma, PTOV1 expression correlates with advanced clinical stage and it was shown to be an independent predictor of overall survival and progression-free survival [41]. HPV-positive head and neck cancers have better prognosis than HPVnegative cancers [43][44][45]. In this context, the levels of expression of PTOV1 in combination with the infection status with the human papillomavirus (HPV) could predict outcome in early-stage laryngeal squamous cell carcinoma [41]. Association to a better outcome was observed in HPV-positive/PTOV1-negative subgroup. In contrast, the HPV-negative/PTOV1-positive subgroup showed the worst outcome.
Similarly to prostate carcinomas, the subcellular distribution of PTOV1 in breast cancer and nasopharyngeal carcinomas was detected both in the nucleus and cytoplasm of carcinomatous cells [21,27,39]. In aggressive prostate tumors, PTOV1 stained intensely in the nucleus of local and distal (bones) metastatic cells [33]. In urothelial carcinoma (UC) nuclear staining was significantly more frequent compared to benign tissue and a reduced cytoplasmic expression significantly correlated with higher pathological stage and grade, suggesting a functional shift for PTOV1 from the cytoplasm to the nucleus in the progression of these tumors [42]. These observations are in line with previous reports showing that intense nuclear PTOV1 expression was able to distinguish in a significant manner high-grade urothelial carcinoma [40]. Thus, the nuclear presence of PTOV1 might be critical for proliferation and tumor progression, suggesting that its transcription regulating abilities are relevant functions.
As mentioned above, gene amplifications for PTOV1 are frequent in prostate carcinomas, pancreas and present in other tumors ( Figure 2A). In addition, mutations are also frequent in several tumors types. About 57% of the mutations are localized inside the A domain ( Figure  2B). The majority are missense mutations, more frequent in colorectal, stomach, lung squamous cell carcinoma, esophageal, and breast cancer. Most of the missense mutations have a predicted functional impact score neutral or low according to Mutation-Assessor [46]. However, the R117L/Q mutation found in LSCC and prostate cancer respectively, has a predicted functional impact score of medium, suggesting it might be a critical change for the www.impactjournals.com/oncotarget function of PTOV1 in these tumors. A recurrent in frame mutation (K212del) is found in adenocortical carcinoma, colorectal, breast and stomach carcinoma.

PTOV1 TRANSCRIPTIONAL REGULATORY FUNCTIONS
Homology models of MED25 PTOV and human PTOV1 have shown that amino acid residues involved in TAD binding in MED25 are generally conserved in PTOV1, suggesting that both domains may serve as activator-binding modules [16]. However, in contrast to MED25, PTOV1 did not interact with the Mediator complex [11]. The differential biological activities of PTOV in MED25 and PTOV1 were suggested to be due to modulation of protein-protein interactions patterns by some amino acid residues which are differently grouped peripherally around the charged surface region [14,15]. It was described that the retinoic acid (RA) activates the response of its receptor (RAR) using MED25 that, through its VWA domain, interacts with the Mediator and, through the PTOV domain, binds to the activators (histone acetyltransferase CBP) [47]. These observations suggest a role for the PTOV domain of MED25 in chromatin remodeling and pre-initiation complex assembly to recruit activators to the basal transcriptional machinery [11]. Additional evidences confirmed that the domain PTOV of MED25 is responsible for binding the TAD domain of several transcription factors, including ERM/ETV5, a PEA3 member of ETS-related transcription factors [48], the nuclear receptor Hepatocyte Nuclear Factor 4 alpha [49], the transcription factor ATF6 alpha [50], a master regulator of endoplasmic reticulum (ER) stress response genes, the retinoic acid receptor (RAR) [51], STAT6 and chromatin remodelers [52]. Because PTOV1, in contrast to MED25, did not interact with the Mediator complex [11], its action may plausibly modulate, or hamper, the MED25 PTOV module binding the activator in those cells where it is overexpressed, as described below for the Retinoic Acid Receptor (RAR). In agreement with the above observations, PTOV1 was recently identified as a regulator of transcription of several genes, including RAR, HES1, HEY1, and Dickkopf-1.

The retinoic acid receptor promoter
Retinoids are promising chemotherapeutics that inhibit cell growth by inducing apoptosis, senescence and differentiation of cancer cells [53,54]. Unfortunately, intrinsic or acquired resistance to these agents frequently occurs after cancer therapy [55]. The formation of the complex MED25-RAR and RA, induced the stimulation of RAR promoter activity [47]. PTOV1 was shown to suppress the MED25-enhanced RAR activity by binding the activator CREB-binding protein (CBP) ( Table 1 and Figure 3) [51]. Thus, the expression of PTOV1 prevented CBP binding to MED25 inducing a repression of the RAR promoter. CBP belongs to a family of large multifunctional transcriptional coactivators that through their acetyl transferase action modify histones and other proteins, regulating a large number of transcription activators and cellular functions (Table 1) [56]. Both PTOV1 and MED25 proteins interact with the acetyl transferase CBP through the PTOV domain [32,51]. Chromatin IP (ChIP) assays showed that PTOV1 itself is not recruited to the RA-responsive RARβ2 promoter. Instead, the increased PTOV1 expression inhibited CBP chromatin binding by forming a chromatin-free PTOV1-CBP interaction that sequesters away CBP from MED25 [51]. In this context, in response to RA the LIM family member Zyxin was shown to interact and cooperate with PTOV1 in RAR repression, by forming a ternary complex with CBP and PTOV1 that antagonized MED25 for CBP binding [57]. These data suggest a potential molecular mechanism for PTOV1 in RA resistance.

The HES1 and HEY1 promoters
Notch is an evolutionarily conserved signaling pathway that regulates cell fate, tissue homeostasis, cell differentiation, proliferation and growth [58]. Activation of the Notch receptor induces its entrance to the nucleus where it acts as a transcription factor for numerous targets, including HES1 and HEY1 genes, two well known downstream regulators of the pathway [59,60]. In the absence of activated Notch a transcriptional repressor complex, that include SMRT/NCoR and HDAC1, is formed on target promoters [61].
Aberrant Notch signaling has been detected in different types of cancer to suppress or activate cancer progression depending on the cell context and tumor type [62][63][64]. In PC, its role in progression has been studied in vitro and in vivo with contradictory results [65][66][67]. We have shown that PTOV1 in metastatic prostate tumors is significantly overexpressed and represses the transcription of the downstream targets of Notch, the HES1 and HEY1 genes, by interacting with SMRT, RBP-Jκ, NCoR, HDAC1 and HDAC4 (Figure 3) [32]. PTOV1 is bound to the chromatin of these promoters when Notch is inactive. Its repressive action was reverted by trichostatin A (TSA), an HDAC inhibitor, indicating the requirement for the activity of HDACs. Interestingly, the repression by PTOV1 was abolished by the overexpression of CBP, in agreement with previous reports showing that HES1 transcription was activated by CBP [68]. PTOV domains shares structural similarity with the SPOC domain of SHARP, a known component of the Notch repressor complex [17,69], and thus PTOV1 may be a facultative additional Notch co-repressor restricted to cancerous events. Additional in vivo evidence supports the role of PTOV1 as a negative regulator of the Notch pathway [32]. In the Beclin 1. Plays a key role in autophagy as a component of the phosphatidylinositol-3-kinase (PI3K) complex which mediates vesicle-trafficking processes in multiple cellular processes, including tumorigenesis, neurodegeneration and apoptosis [123,124].
CBP /CREBBP Co-IP [32,47] cAMP-response element binding protein (CREB) binding protein. It is involved in the transcriptional coactivation of many different transcription factors. It plays critical roles in embryonic development, control of growth, and homeostasis by coupling chromatin remodeling to transcription factor recognition. The protein encoded by this gene has intrinsic histone acetyltransferase activity and also acts as a scaffold to stabilize additional protein interactions with the transcription complex. Acetylated histones and non-histones proteins may act as a transcriptional activator [56,125,126].

FLOT1
Co-IP Co-localization Co-fractionation [22] Flotillin-1. It associates to lipid rafts and localizes to non-caveolar membranes. This protein plays a role in vesicle trafficking, cell morphology and signaling. Elevated expression of FLOT1 has been reported in different types of tumors. Its nuclear localization is associated to induction of proliferation. FLOT1 regulates the function of Aurora B kinase in metaphase cells [111,113,116] HDAC1 Pull-down [32] Histone deacetylase 1. It is a component of the histone deacetylase complex which plays a key role in regulation of eukaryotic gene expression. Together with metastasis-associated protein-2 (MDM2), it deacetylates p53 and modulates its effect on cell growth and apoptosis. It is an element of the Notch repressor complex [32,132].

HDAC4
Pull-down [32] Histone deacetylase 4. It possesses histone deacetylase activity and represses transcription when tethered to a promoter. It is an element of the Notch repressor complex. Inhibitors of function increase cytotoxicity of docetaxel in gastric cancer cells [133,134] HDAC6 Protein-peptide [100] Histone Deacetylase 6. Deacetylates lysine residues on the N-terminal part of the core histones. Histone deacetylation gives a tag for epigenetic repression alters chromosome structure and affects transcription factor access to DNA and plays an important role in transcriptional regulation, cell cycle progression and developmental events. The protein contains an internal duplication of two catalytic domains which appear to function independently of each other [100,135].

KLHDC2
Affinity capture-MS [136] Kelch domain containing 2. Represses CREB3-mediated transcription by interfering with CREB3-DNA binding [137] NCoR Pull-down assay [32] Nuclear receptor corepressor 1. Transcriptional repressor of thyroid-hormone and retinoic-acid receptors. It is part of a complex which also includes histone deacetylases and other transcriptional regulators [138] PEX19 Affinity capture-MS [136] Peroxisomal biogenesis factor 19. Required for early peroxisomal biogenesis [139] PIN1 Two hybrid [129] Peptidylprolyl cis/trans isomerase, NIMA-interacting 1. It catalyzes the cis/trans isomerization of peptidyl-prolyl peptide bonds. This PPIase specifically binds to phosphorylated ser/ thr-pro motifs to catalytically regulate the post-phosphorylation conformation of its substrates. The conformational regulation catalyzed by this PPIase has a profound impact on key proteins involved in the regulation of cell growth, genotoxic and other stress responses, the immune response, induction and maintenance of pluripotency, germ cell development, neuronal differentiation, and survival [140][141][142].
It is involved in different cellular processes including apoptosis, cell spreading, cell proliferation, regulation of insulin receptor and resistance to drugs [94,97] RASSF6 Affinity capture-MS [143] Ras association (RalGDS/AF-6) domain family (N-terminal) member 9. Involved in the induction of apoptosis. It may act as a Ras effector protein [144].
RBP-Jκ/CBF-1 Co-IP Pull-down assay [32] Recombination signal binding protein for immunoglobulin kappa J region. Is a transcriptional regulator in the Notch signaling pathway. It acts as a repressor when not bound to Notch proteins and an activator when bound to Notch proteins. It is thought to function by recruiting chromatin remodeling complexes containing histone deacetylase or histone acetylase proteins to Notch signaling pathway genes [32,145].

RPS6
Co-IP Co-localization Co-fractionation [33] Ribosomal protein S6. It is a component of the 40S subunit. It is the major substrate of protein kinases in the ribosome. Phosphorylation is induced by a wide range of stimuli, including growth factors, tumor-promoting agents, and mitogens. Dephosphorylation occurs at growth arrest. The protein may contribute to the control of cell growth and proliferation through the selective translation of particular classes of mRNA. Overexpressed in esophageal squamous cell carcinoma, renal cell carcinoma metastases and hyperphosphorilated in lung cancer [146][147][148].
SFN/14-3-3 sigma Affinity capture-MS [149] Stratifin. Adapter protein implicated in the regulation of numerous signaling pathways including cancer metabolic reprogramming. When bound to KRT17, regulates protein synthesis and epithelial cell growth by stimulating Akt/mTOR pathway. It is a p53regulated inhibitor of G2/M progression [150,151].

SPTAN1
Two hybrid [152] Spectrin, alpha, non-erythrocytic 1. Filamentous cytoskeletal protein involves in the stabilization of the plasma membrane and organization of intracellular organelles. Also involved in DNA repair, cell cycle regulation, cell migration and its deletion is related to some encephalopaties [152,153] UBC Reconstituted complex [154] Ubiquitin C. Polyubiquitin precursor. Provides extra ubiquitin during stress and it loss cannot be compensated by induction of the other Ub genes [155].
Regulates kinetochore localization of Plk1 to promote proper chromosome alignment in mitosis [156].
The table summarizes the interactors reported for PTOV1 based on STRING [120] and BIOGRID [121]. It describes the experimental evidences for each interaction and includes a brief explanation on the role reported for the interacting partner. IP: immunoprecipitation; MS: mass spectrometry. Drosophila melanogaster wing model, the expression of the human PTOV1 exacerbated Notch deletion mutant phenotypes and suppressed the effects of constitutively active Notch. In human tissues, the normal prostate epithelium revealed high levels of expression of HES1 and HEY1 proteins, supporting activated Notch signaling, whereas metastatic samples expressed significantly lower levels of these proteins, suggesting a Notch repressed state [32]. In contrast, the expression of PTOV1 in the normal prostate epithelium was mostly absent, but the protein was significantly overexpressed in metastatic samples. In human PC cell lines, the downregulation of PTOV1 induced an upregulation of the endogenous HEY1 and HES1 genes, and reciprocally, the ectopic expression of PTOV1 in PC cells and HaCaT keratinocytes, where Notch acts as tumor suppressor, caused the inhibition of expression of HEY1 and HES1 genes [70,71]. All together, these observations support a pro-oncogenic role for PTOV1 as a negative regulator of Notch signaling in PC progression. They also support a tumor-suppressor function of Notch in PC, similarly to previous reports in skin, myeloid leukemia, and cervical carcinoma cells [62,70,72].

The dickkopf-1 promoter
Aberrant activation of Wnt/β-catenin signaling was reported in breast cancer and strong evidence point to a possible epigenetic silencing of negative regulators of Wnt, although the regulatory mechanisms underlying these epigenetic changes are poorly understood [73]. Very recently, PTOV1 expression was shown to activate Wnt/β-catenin signaling in breast cancer [74]. In the canonical Wnt pathway, binding of Wnt ligand to frizzled receptors and lipoprotein receptorrelated protein-5 or 6 (LRP5/6) co-receptors initiates a cascade, which results in β-catenin activation, its nuclear translocation and transcription of target genes [75]. Dickkopf-1 (DKK1) is negative regulator of Wnt signaling. Silencing its expression was tightly associated with DNA hypermethylation and histone deacetylation [76,77]. DKK1 methylation has been reported in 27% of breast cancer cell lines and 19% of breast cancer patients [78].
In human breast carcinoma, high levels of PTOV1 expression correlated with high levels of nuclear β-catenin and low levels of DKK1 [74]. The overexpression of PTOV1 in breast cancer cell lines induced the nuclear translocation of β-catenin and increased β-catenin/ TCF transcriptional activity. PTOV1 overexpression repressed DKK1 transcription via the recruitment of HDACs to the promoter and a concomitant decrease of histone acetylation [74]. Treatment with TSA reverted the repression of DKK1. These findings suggest a role for PTOV1 as a novel epigenetic regulator of the Wnt/βcatenin pathway in breast tumorigenesis.

ROLE OF PTOV1 IN CANCER STEM CELLS
Extensive research over many years have posited a cancer stem cell (CSC) model for tumorigenesis in which a small proportion of cells, possibly originated from normal stem cells, reside inside tumors and may be subject of additional genetic changes or epigenetic transitions that can drive tumor initiation, progression to metastasis and drug resistance [79,80]. These cells, identified and isolated in a number of solid cancers models, including prostate cancer, have self-renewing and differentiation abilities and are capable of recapitulating the characteristics of the original tumor from which they were isolated [81]. As the normal stem cells, CSCs possess the ability to grow in vitro as self-renewing spheres. However, the mechanisms implicated in CSCs self-renewal and differentiation, critical to understand CSCs biology and their role in tumorigenesis, are still under investigation. In the prostate, recent studies have shown that CSCs can also be isolated from cancer cell lines, and the analysis of their transcriptome shows expression profiles that support CSCs identity [82].
Recently, it was shown that PTOV1 promotes the in vitro formation of spheres in HaCaT transformed keratinocytes and PC3 prostate cancer cells, and promoted tumor growth in vivo [32,33]. In addition, in breast cancer cells PTOV1 was shown to enhance tumor growth, increase the number of mammospheres, the proportion of the side population (SP), and of CD24 -/ CD44 + cells [74]. These findings reveal the ability of the protein PTOV1 to promote CSCs-like properties in cells from different tumor types. Significantly, the associated overexpression of the protein in more aggressive tumors with poor prognosis provides support for a role of PTOV1 in favoring CSCs self-renewal and tumor progression.

ROLE OF PTOV1 IN PC RESISTANCE TO CHEMOTHERAPY
Because of the high resistance of metastasis to conventional androgen-depletion-therapy (ADT) and chemotherapy, metastatic PC is virtually incurable [83]. The numerous evidences that PTOV1 is expressed significantly more in aggressive tumors and metastatic lesions and its implication in the mechanisms leading to cancer progression, suggested a potential action of this protein in recurrence. Indeed, PTOV1 is expressed at higher levels in cells resistant to chemotherapy with docetaxel compared to parental sensitive cells (Cánovas et al., submitted 2016). The ectopic overexpression of PTOV1 in docetaxel sensitive PC cells resulted in a significant increase in the number of prostatospheres and a significantly greater capacity of the cells to survive to www.impactjournals.com/oncotarget docetaxel treatment (Cánovas et al., submitted 2016). These effects are associated to a significant increase in the levels of genes associated with the resistance to docetaxel (ABCB1, CCNG2, TUBB4A, TUBB2B) and with the stemness phenotype (LIN28A, ALDH1A1, MYC, NANOG). In turn, the knockdown of PTOV1 very significantly inhibited self-renewal and proliferation in all cell models. In addition, cells knockdowns for PTOV1 showed a striking cell cycle arrest at the G2/M phase that was associated to a significant increase in apoptosis. Docetaxel resistant cells knockdown for PTOV1 showed more extreme phenotypes. These findings identify PTOV1 as a promoter of docetaxel resistance and a survival factor for CRPC cells and give support to potential anticancer therapies directed to eliminate the action of this protein in future therapeutic interventions.

SEARCH FOR PTOV1 INTERACTING PROTEINS TO UNVEIL ITS FUNCTIONS
Discovering new proteins interactions can be helpful in defining the functions in which a protein of interest is involved. PTOV1 has been detected at different subcellular locations, including submembrane sites, lipid rafts, cytoplasm especially the perinuclear region, and the nucleus. These locations are likely associated with different interacting proteins. A summary of the proteins interacting with PTOV1 is given in Table 1. These proteins are involved in several cellular processes, although it is not clear how each interaction contributes to a role of dysregulated PTOV1 expression in cancer progression. Figure 3 summarizes the interactions of PTOV1 with other proteins in relationship with their described functions. Groups including transcriptional regulation and DNA replication, cell cycle regulation-mitotic functions contain numerous interactors. Similarly, groups including protein synthesis, ubiquitination and membrane trafficking functions contain each three interactors. Together, these observations indicate that PTOV1 participates in different cellular events at different subcellular locations. Interactions described by high-throughput capture affinity assays that lack further functional analyses will not be reviewed here. Those interactions studied more in detail, where some functional aspects are revealed, are reviewed below.

Receptor of activated protein C kinase, RACK1
We have described the interaction of PTOV1 with RACK1, also known as guanine nucleotide binding protein beta polypeptide 2-like 1, GNB2L1 [33]. RACK1 contains seven repeats of a short Trp-Asp [W-D] dipeptide, highly conserved in eukaryotes, and although was first described to interact with Protein Kinase C (PKC) isoforms [84,85], it is now known to interact with numerous proteins and is involved in a diverse array of cellular processes. These include apoptosis regulation of insulin receptor and IGF-1R signaling, cell spreading, cell proliferation, STAT3 activation, and UV radiation [86][87][88]. In 2004, an interesting report described RACK1 as part of the 40S small ribosomal subunit, and localized the protein at the head of the 40S subunit close to the mRNA exit channel [89], suggesting that it may function as a molecular link connecting cell signaling with the protein translation machinery. Later, RACK1 was described to recruit active PKCβII on ribosomes to phosphorylate eukaryotic initiation factor 6 (eIF6) that allows the assembly of the 80S ribosome on the pre-initiation mRNA complex [90,91]. RACK1 has been described as an adaptor required for the PKCmediated phosphorylation of Ser129 of JNK [92]. More recently, the recruitment of JNK by RACK1 has been identified as part of a mechanism underlying the quality control of newly synthesized proteins under stress conditions [93]. After stress induction, RACK1 recruits activated JNK to 40S on actively translating ribosomes to phosphorylates the elongation factor eEF1A2, which in turn promotes the ubiquitination and degradation of damaged newly synthesized polypeptides [93]. RACK1 overexpression has been reported in different tumors and it was found to be a differential diagnostic biomarker and predictor for poor clinical outcome in breast and pulmonary carcinomas [94][95][96]. In hepatocellular carcinoma it promotes chemoresistance and tumor growth by localization on ribosomes and phosphorylation of eukaryotic initiation factor 4E (eIF4E), preferential translation of cyclin D1, Myc, survivin and Bcl-2 [97]. PC3 cells stimulated with IGF-1 or phorbol esthers, show PTOV1 and RACK1 colocalized at membrane ruffles [33]. RACK1 interacted with full-length PTOV1 and with the B domain, but not with the A domain, suggesting un-equivalent functions between the two PTOV1 domains. The PTOV1-RACK interaction was localized on 40S ribosomes by polysome profiling experiments in PC3 cells. Significantly, PTOV1 failed to co-sediment with the 40S ribosomal subunit in RACK1 knockdown cells, indicating that RACK1 is necessary for the association of PTOV1 with ribosomes [33]. In addition, PTOV1 co-immunoprecipitated and colocalized with the ribosomal protein S6 (RPS6) corroborating its interaction with ribosomes. Of notice, PTOV1 was not detected in gradient fractions corresponding to polysomes, suggesting that its action is directed to translation initiation rather than elongation. In addition, the overexpression of PTOV1 caused a significant stimulation of bulk protein synthesis, including of c-Jun. In turn, increased levels of PTOV1 and c-Jun induced SNAI1 transcription, promoted an epithelial-mesenchymal-transition (EMT) and a significant increase of cell invasiveness in vitro and tumor growth and metastasis in vivo [33]. These findings indicate that the increased expression of PTOV1 in cancer and its ability to bind to RACK1 on ribosomes and to increase protein synthesis is one efficient way for this protein to promote cancer progression.
The identification of the N-terminal (e)AT-hook motif in PTOV1 and the observed direct interaction of this motif with RNA chains [18], the structural resemblance of PTOV domains with SPOC domains often present in proteins containing RNA binding motifs [16], and the ability of the protein to bind to 40S ribosomes and regulate the rate of mRNA translation, point to a role for PTOV1 both in the nucleus, regulating gene transcription and in the cytoplasm, conceivably as part of ribonucleoprotein (RNP) complexes that may be critical in regulating translation initiation of at least a set of mRNAs, such as c-Jun (Figure 4).

BUZ/Znf-Ubp domains of the histone deacetylase HDAC6
The BUZ (binder of ubiquitin zinc finger) domain, also known as the Znf-UBP (zinc finger-ubiquitinspecific processing protease domain), is present in a subfamily of ubiquitin-specific processing proteases (USPs), the E3 ubiquitin ligase BRCA1-associated protein 2 (BRAP2) and in the histone deacetylase 6 (HDAC6) [98,99]. The BUZ domain is a sequencespecific protein-binding module that recognizes the free C-termini of proteins. Through its C-terminal sequence (RGMGG), PTOV1 interacts with the BUZ domain of HDAC6 with a low KD value that required the Gly-Gly motif present at the C-terminal of HDAC6 [100]. HDAC6 although it has been detected in the nucleus, it is mostly a cytoplasmic deacetylase that catalyzes the cleavage of the acetyl group of ε-amino groups of lysines and can regulate growth factor-induced chemotaxis by association with the cytoskeleton [101].
The role of HDAC6 in tumor progression is controversial. Some evidences suggest an oncogenic role: its overexpression is associated to mutated K-ras, and correlated with more aggressive tumors and lower survival rate in several tumor types [102][103][104][105][106]. In contrast, several reports suggest a tumor-suppressorlike function, where its reduced nuclear localization was associated with distant metastasis, worse overall survival, and poor prognosis [107,108]. Because PTOV1 is a promoter of motility and invasion, the above observations would be in line with a role for the PTOV1-HDAC6 interaction in the cytoplasm in promoting the pro-oncogenic role of PTOV1. We speculate that, by interacting with HDAC6, PTOV1 might have a more direct access to the cytoskeleton and act in promoting cell motility. Further investigation would be required to ascertain the contribution of HDAC6 to the oncogenic capacities of PTOV1.

Flotillin-1
PTOV1 was shown to interact with the lipid-raftassociated protein Flotillin-1 [22] a protein that belongs to the Reggie/Flotillin family. Lipid rafts play a central role in membrane trafficking and signaling [109]. Flotillin-1, localized to non-caveolar lipid-rafts [110], has been involved in neuronal regeneration [111], and in insulin signaling in adipocytes, where it generates a signal crucial in the regulation of glucose uptake in adipocytes [112].
In PC cells, Flotillin-1 interacts with the B domain of PTOV1, as expected the two proteins colocalized in lipid rafts and, surprisingly, in the nucleus [22]. After a mitogenic stimulus, Flotillin-1 entered the nucleus concomitantly with PTOV1 shortly before the beginning of the S phase. The overexpression of Flotillin-1 caused a significant increase in cell proliferation, and both PTOV1 and Flotillin-1 are required for PC proliferation. However, while the presence of PTOV1 and an intact carboxy terminus of Flotillin-1 are required for its nuclear entry, the depletion of Flotillin-1 did not affect the nuclear localization of PTOV1 [22]. In additional work, we have shown that Flotillin-1 is required for the stability and function of the Aurora B kinase in mitosis [113]. These data suggest that PTOV1 may drive PC progression in part through the regulation of expression and nuclear localization of Flotillin-1 necessary to support Aurora B kinase mitotic function.
Additional recent reports confirmed the prooncogenic effects of increased Flotillin-1 levels in tumors [114][115][116] and its action in promoting invasion and metastasis through EMT, activation of NF-kB, Wnt/β-catenin and TGF-β pathways [114,115]. Flotillin-1 is an important regulator of H-ras activation and invasion in triple-negative breast cancer and its expression inversely correlates with patient diseasefree survival rates [116]. In gastric cancer, Flotillin-1 overexpression was shown to be the result of miR-485-5p downregulation that correlated with poor prognosis [117]. These findings suggest that the interaction of PTOV1 with Flotillin-1 might amplify their action in tumor progression.

DISCUSSION: A MODEL FOR PTOV1 ACTIONS
The hitherto little known oncogenic PTOV1 protein has gained recent attention as a regulator of multiple cellular functions and pathways that tend to enhance cell growth and self-renewal in multiple cell types. Here, we propose a mechanistic model that collects the most recently described actions of PTOV1 ( Figure 4). PTOV1 shows dual functions in the regulation of gene expression at transcriptional and translational levels. For the role as transcription factor, PTOV1 is www.impactjournals.com/oncotarget In the cytoplasm, PTOV1 is found at perinuclear and submembranes regions associated to Flotillin-1 and also RACK1 and ribosomes. The latter, likely occur in RNA-protein complexes (RNP) that modulate mRNA translation, including the synthesis of the oncogene c-Jun. In the nucleus, PTOV1 can regulate the expression of a number of genes related to cell proliferation, survival, EMT, and chemoresistance, by direct or indirect (genes shown in parenthesis) association to specific promoters to activate or repress transcription. In turn, c-Jun/AP1 may be directly or indirectly contributing to the action of PTOV1 as a transcription factor. PTOV1 is a repressor for the regulation of HES1, HEY1 and DKK1. These effects result in the negative regulation of Notch signaling in PC and the activation of Wnt/β-catenin signaling in breast cancer. In both tumors, these effects culminate with increased tumor growth, invasion, metastasis, and chemoresistance. PTOV1 as a transcriptional repressor requires the presence of HDACs, suggesting that it is an epigenetic regulator. An additional mechanism for transcriptional repression was suggested for the RAR promoter, where PTOV1 sequesters the activator CBP from MED25, and results in suppression of transcription of RAR targets. The action of PTOV1 of sequestering activators from MED25 might be associated to inhibition of other MED25 targets. The recent identification of the nucleic acid-binding motif (e)AThook, at the N-terminal region of PTOV1 gives support to its role in regulation of gene expression by direct DNA or RNA binding. found associated with the regulatory regions of at least three genes where it recruits deacetylases and other DNA modifying enzymes to modulate transcription. The association with transcriptional repressor complexes, including several HDACs and NCoR, has been reported for the repression of a few genes (HES1, HEY1, and DKK1). Transcriptional repression by PTOV1 might also occur through association with activators, such as CBP in the RARβ2 gene. In this case, the complex PTOV1-CBP is not bound to chromatin, but the protein competes with MED25 for binding and sequestering the activator away from the Polymerase II complex. Conceivably, the association CBP-PTOV1 could also act promoting transcription in different genes.
Additionally, PTOV1 was discovered to regulate protein synthesis by direct association with RACK1 and 40S ribosomes in translation pre-initiation complexes. In PTOV1 overexpressing cells significantly more c-Jun mRNA levels were loaded on polysomes compared to actin, but PTOV1 is not found in polysomal fractions, suggesting that its action is directed to translation initiation, possibly mediating the recruitment of specific mRNA-protein complexes to ribosomes.
Significant for these dual functions described for PTOV1 is the very recent identification of the N-terminal (e)AT-hook motif that, allowing the protein to bind directly to nucleic acids, gives support to the roles in gene expression regulation by direct DNA or RNA binding. Nucleic acids binding through the (e)AThook would feasibly allow simultaneous interactions of PTOV1 with different factors binding the A and B domains. For example, protein complexes associated to promoters, or ribonucleoprotein complexes charged with specific mRNAs to be translated. Interestingly, the subcellular localization of the specific interactions mediated by the A domain with HDAC1 and RBP-jκ, are coupled to actions in the regulation of transcription [32]. In contrast, the interactions with cytoplasmic RACK1 and Flotillin-1 exclusively engaging the B domain are linked to activities more closely related to protein synthesis. All together these evidences suggest that PTOV1 might be a new moonlighting protein able to perform different activities in the cell [118], and may utilize separate protein surfaces for its multiple actions [119]. The overexpression of a single protein with multiple actions that converge in activation of proliferation, survival and drug-resistance is energetically convenient for a tumor cell, and might contribute to the acquisition of stemness features. Many questions remain to be solved to understand what is the biological role of PTOV1 in normal tissues, or how does the protein promote tumor progression. The evidences so far suggest that the potential therapeutic effects of its specific targeting in aggressive cancer cells is worthy of being studied.