Nature Reviews Cancer

RESEARCH HIGHLIGHTSNature Reviews Cancer | AOP, published online 12 April 2013; doi:10.1038/nrc3514

LUNG CANCER

Gastric giveawaysNKX2-1 is a crucial regulator of pulmonary differentiation that is often overexpressed in lung adeno carcinomas, and its loss correlates with a poor prognosis. Published data from Tyler Jacks and colleagues have shown that NKX2-1 can restrain the capacity of oncogenic KRAS-driven lung cancers to evolve as a poorly differentiated, more aggressive disease. Jacks and colleagues now have a potential mechanism to explain this result. The authors generated mice in which an allele of oncogenic Kras (KrasG12D) and both alleles of Nkx2-1 could be activated and deleted, respectively, from the lungs. Tumours in which Nkx2-1 was deleted had a strikingly different morphology (glandular and mucinous) from those expressing NKX2-1 (papillary-like and non mucinous). Tumour burden was also greater in mice in which Nkx2-1 was deleted owing to a greater number of initial lesions rather than to increased rates of proliferation. However, deletion of Nkx2-1 in established tumours (37 months after KrasG12D expression) led to a substantial change in the rate of proliferation; this increased tumour burden, and tumour cells in which Nkx2-1 was deleted progressively adopted a mucinous phenotype. These findings indicate that loss of Nkx2-1 both promotes tumour initiation and progression, and alters lung tumour cell fate. Why does loss of Nkx2-1 result in a mucinous, glandular phenotype? Comparison of mRNA expression levels from the different tumour models described above, along with MetaCore pathway analyses, indicated that Nkx2-1 loss results inNATURE REVIEWS | CANCER 2013 Macmillan Publishers Limited. All rights reserved

cells adopting an mRNA profile that is closest to that of the stomach epithelium. The HNF4 transcription factor that regulates gastrointestinal differentiation was expressed in all Nkx2-1-deleted tumours. Moreover, in 37 human lung adenocarcinoma samples, two stomach epitheliumrestricted proteins (GKN1 and CTSE) were predominately expressed in the NKX2-1-negative tumours. So, does NKX2-1 repress the expression of these stomach epithelium-specific genes? The authors could not find evidence that NKX2-1 functions as a direct repressor of such genes; however, many of the genes that had an increased level of expression as a result of NKX2-1 loss were bound by Forkhead box A1 (FOXA1) and FOXA2 transcription factors. Careful genetic and epigenetic analyses indicated that NKX2-1 recruits FOXA1 and FOXA2 to a pulmonary-specific set of genes and promotes their expression. However, when NKX2-1 expression is absent, FOXA1 and FOXA2 bind to new sites in the genome, many of which are near genes that are normally expressed in the gastrointestinal tract. Further conditional gene deletion studies indicated that HNF4 is partly responsible for the gastric phenotype that is evident in Nkx2-1deleted lung tumours, but that deletion of both Hnf4a and Nkx2-1 did not result in any morphological changes: these tumours remained mucinous and well differentiated, and grew poorly. This is an interesting finding because loss of Hnf4a and Nkx2-1 leads to the partial derepression of Hmg2a, an embryonal transcription factor that is associated with the development of poorly differentiated, poor-prognosis

lung adenocarcinomas that have lost expression of NKX2-1. Thus, although HNF4 can repress Hmg2a, the loss of both Nkx2-1 and Hnf4a does not seem to convey a selective advantage. These results indicate that NKX2-1 drives a pulmonary-specific differentiation programme and that this represses a default gastrointestinal differentiation pathway that involves HNF4. However, even if both of these cell fate transcription factors are lost in lung epithelial cells, this does not automatically result in a more primitive cell type, suggesting that other programmes need to be overcome for lung tumour cells to adopt a truly dedifferentiated and aggressive phenotype.Nicola McCarthyORIGINAL RESEARCH PAPER Snyder, E. L. et al. NKX2-1 represses a latent gastric differentiation program in lung adenocarcinoma. Mol. Cell 21 Mar 2013 (doi:10.1016/j.molcel.2013.02.018)

Nkx2-1 loss results in cells adopting an mRNA profile that is closest to that of the stomach epithelium

VOLUME 13 | MAY 2013

Lara Crow/NPG


Methylation rebootWhether epigenetic alterations can drive or maintain tumours is widely debated. However, apart from agents that cause global loss of epigenetic marks (for example, DNA methylation inhibitors), experimental tools to examine this question are lacking. Stefan Stricker, Steve Pollard and colleagues have used induced pluri potent stem cell (iPSC) reprogramming techniques to reset the epigenome of glioblastoma stem cells (GSCs) in order to investigate the contribution of epigenetic changes to malignant behaviour. Using a panel of 14 GSC lines derived from independent primary human glioblastomas that already expressed high levels of MYC, the authors exogenously expressed the transcription factors Krppellike factor 4 (KLF4) and OCT4 (also known as POU5F1) to induce reprogramming. In two of the cell lines (G7 and G26), this resulted in the generation of iPSC-like cells (iG7 and iG26 cells) with transcriptional profiles more like those of iPSCs than those of normal neural stem cells (NSCs). These cells could form noninfiltrative teratomas (a property of pluripotent cells) when injected either subcutaneously or into the kidney capsule of non-obese diabetic/ severe combined immunodeficient (NOD/SCID) mice, but many of the teratoma cells expressed the neural progenitor marker Nestin and Ki67, suggesting that they were biased towards a neural lineage and remained proliferative. Interestingly, iG7 and iG26 cells also retain the structural chromosomal aberrations that are present in the parental G7 and G26 cell lines. To determine what changes have occurred in the iG7 and iG26 cells, the authors examined DNA methylation patterns. Initial profiling of G7 and G26 GSCs compared with NSCs revealed 691 cancer-specific methy lation variable positions (cMVPs), and the profiles of the GSCs were similar to those of glioblastoma generated by The Cancer Genome Atlas project. Importantly, a large proportion (450 of 691) of cMVPs in G7 and G26 cells was reset in iG7 and iG26 cells, and two known tumour suppressor loci (CDKN1C and TES) that are hypermethylated in GSCs were demethylated in the reprogrammed cells, as was a large proportion of Polycomb repressive complex 2 (PRC2) target genes. Therefore, it seems that abnormal cancer-associated methylation patterns are at least partially erased by the reprogramming procedure. The functional consequences of methylation resetting were assessed using the G7 cells and iG7 derivatives, along with iG7 cells that had been directed to become neural progenitors (N-iG7 cells) or mesodermal progenitors (M-iG7 cells). More comprehensive methylation profiling revealed 60,977 cMVPs between normal NSCs and parental G7 cells; 44% of these were reset genome-wide (55% in regulatory regions) in the iG7 cells. When iG7 cells were differentiated to N-iG7 cells, 83% of the normalized cMVPs persisted. Despite the lack of many cancer-specific methylation marks in N-iG7 cells, these cells were able to form aggressive tumours that were indistinguishable from those formed from G7 cells following injection into the forebrain

resetting of methylation marks is not sufficient to prevent tumorigenesis

of NOD/SCID mice, suggesting that the resetting of methylation marks is not sufficient to prevent tumorigenesis. However, if the cells were directed towards a mesodermal lineage (M-iG7 cells) they were only capable of forming benign compact tumours in the mouse brain, suggesting that sending the iG7 cells down a different developmental pathway can suppress tumorigenesis. Several interesting questions remain, such as how the reprogramming procedure affected other epigenetic changes and why some methylation abnormalities were not fully restored. Furthermore, it will be interesting to see whether this applies to other cancer types.Sarah Seton-RogersORIGINAL RESEARCH PAPER Stricker, S. H. et al. Widespread resetting of DNA methylation in glioblastoma-initiating cells suppresses malignant cellular behavior in a lineage-dependent manner. Genes Dev. 27, 654669 (2013)

NATURE REVIEWS | CANCER 2013 Macmillan Publishers Limited. All rights reserved


BANANASTOCK

EPIGENETICS


M E TA B O L I S M

Glutamine connectionsAs a result of the Warburg effect (aerobic glycolysis) in cancer cells, fewer glucose-derived metabolites feed into the Krebs cycle. Thus, cancer cells typically have an increased reliance on alternative metabolites to replenish Krebs cycle intermediates, and the amino acid glutamine is one such metabolite. Three new studies have characterized molecular links between glutamine metabolism and key cancer signalling pathways. Hyperactivation of the transcription factors hypoxia-inducible factor1 (HIF1) and HIF2 through loss-of-function mutations in the von HippelLindau (VHL) tumour suppressor gene commonly occurs in renal cell carcinoma (RCC). The authors of the first new study had previously shown that VHL-mutant RCC cell lines use glutamine to generate citrate and lipids through the reductive carboxylation of glutaminederived -ketoglutarate, and that HIF activity in hypoxic cells promotes the conversion of glucose to lactate, thus preventing the use of glucose in the Krebs cycle. To determine whether these findings are linked in RCCs with mutant VHL, Gameiro etal. carried out manipulations in RCC cells such as by expressing mutant VHL or VHL-insensitive HIF subunits and used metabolic profiling in vitro and in tumour-bearing mice to show that loss of HIF regulation by VHL is sufficient to switch the Krebs cycle inputs from mostly glucose-derived inputs to mostly glutamine-derived inputs. They also found that replenishing citrate levels could block the switch to glutamine usage, thus intracellular citrate deficiency might promote the switch to glutamine use. Finally, the authors highlighted the therapeutic relevance of this reliance on glutamine metabolism by showing that VHL deficiency sensitizes RCC cells and xenografts to inhibitors of glutaminase, the enzyme that catalyses the first step of glutamine metabolism. In a separate study, Son etal. analysed the contribution of activated KRAS to metabolism in pancreatic ductal adenocarcinoma (PDAC) cells that are dependent on glutamine. Using genetic and pharmacological interventions the authors found that the growth of KRAS-mutant PDAC cells and tumour xenografts does not rely on glutamate dehydrogenase1 (GLUD1; also known as GDH1) in the conversion of glutamine to -ketoglutarate for use in the Krebs cycle, and instead requires a non-canonical route involving the GOT1 aspartate transaminase to convert glutamine-derived aspartate to oxaloacetate, which can be used to generate malate and pyruvate. This series of reactions increases NADPH levels and helps to maintain the cellular redox state through the generation of reduced glutathione. Indeed, glutamine deprivation results in increased reactive oxygen species (ROS) levels in PDAC cells, and inhibitors of glutaminase synergized with H2O2 to kill PDAC cells. Knockdown of KRAS resulted in GOT1 downregulation and reduced metabolic flux through this noncanonical route, thus demonstrating the role of mutant KRAS in specifying the use of this pathway. As glutamine deprivation can occur naturally through the increased use of glutamine reserves by tumours, and is a goal of various therapeutic approaches, it is important to understand how cells respond to glutamine deprivation. Reid etal. tested whether 4 (also known as IGBP1) has a role in the response to glutamine deprivation,

because an orthologue of this protein, Tap42, is known to have this function in yeast. Indeed, expression of 4 protected mouse embryo fibroblasts (MEFs) and human fibrosarcoma cells from the cytotoxic effects of glutamine deprivation. Using various genetic, biochemical and proteomic approaches, the authors showed that glutamine deprivation triggers an increase in ROS levels, which leads to the 4-mediated assembly of a B55-subunit-containing protein phosphatase 2A (PP2A) complex. This results in the activation of p53 and the induction of pro-survival p53 target genes, such as Cdkn1a (which encodes p21) and Gadd45a. Consistent with this, the growth of fibrosarcoma xenograft tumours (the cores of which were shown to have low levels of glutamine) was inhibited by B55 knockdown, and p53-deficient HCT116 cells invitro were more sensitive to glutamine deprivation than p53-wild-type controls. Reid etal. concluded that glutamine is required to generate reduced glutathione, again to maintain the cellular redox state. Blocking metabolic functions as a therapeutic approach raises the challenge of avoiding toxicity in normal tissues. However, these molecular links between cancer signalling pathways and altered glutamine metabolism support the applicability of anticancer therapeutic approaches that are based on glutamine deprivation and suggest that their efficacy might be enhanced in p53-deficient tumours or in combination with other ROS-generating stresses.Darren J. BurgessORIGINAL RESEARCH PAPERS Gameiro, P.A. etal. In vivo HIF-mediated reductive carboxylation is regulated by citrate levels and sensitizes VHLdeficient cells to glutamine deprivation. Cell Metab. 17, 372385 (2013) | Son, J. etal. Glutamine supports pancreatic cancer growth through a KRAS-regulated metabolic pathway. Nature 27 Mar 2013 (doi:10.1038/nature12040) | Reid, M.A. etal. The B55 subunit of PP2A drives a p53dependent metabolic adaptation to glutamine deprivation. Mol. Cell 13 Mar 2013 (doi:10.1016/j. molcel.2013.02.008)

Lara Crow/NPG

glutamine is required to generate reduced glutathione



RESEARCH HIGHLIGHTSNature Reviews Cancer | AOP, published online 28 March 2013; doi:10.1038/nrc3508

M E TA S TA S I S

ADORA(2B)tionKnockdown of ADORA2B in LM2 cells limited metastasisDaniel Peeper and colleagues have identified the transcription factor FOS-related antigen 1 (FRA1; encoded by FOSL1) as a mediator of metastasis. They have also found that triple-negative breast cancer cell lines that express FRA1 are more sensitive to antagonists of adenosine A receptors (ADORAs). The authors initially screened two rat epithelial cell lines that were transformed and highly metastatic owing to the engineered expression of the neurotrophic receptor tyrosine kinase TRKB and its ligand brain-derived neurotrophic factor. Microarray gene-expression profiling indicated that FRA1 was the most upregulated transcription factor in these cells. Stable knockdown of FRA1 in the transformed cells to levels seen in the parental cell lines had no effect on the growth of subcutaneous tumours, but blocked the emergence of lung metastases. FRA1 is overexpressed in a number of different types of solid tumour, including breast cancer, and the authors examined the role of FRA1 in LM2 cells (a derivative of the MDA-MB-231 triple-negative breast cancer cell line) that preferentially metastasize to the lungs in mice. Intravenous injection of LM2 cells in which FRA1 was knocked down showed that, despite small numbers of these cells being present in the lungs, they did not grow, unlike parental LM2 cells, unless the cells regained FRA1 expression. Inoculation of LM2 cells into the mammary fat pad showed that the tumours with reduced FRA1 expression grew more slowly than controls but, unlike mice with control LM2 tumours, lung metastases were not detected. In addition, intracardiac injection of MDA-MB-231 cells with and without FRA1 expression indicated that FRA1 is required for metastatic outgrowth in multiple tissues. As a transcription factor, FRA1 is not considered a druggable target; therefore, the authors set up a synthetic lethality screen and identified four compounds that were less toxic to FRA1-depleted MDA-MB-231 cells compared with parental MDA-MB-231 cells. Two of these were adenosine receptor antagonists, and the authors chose to focus on ADORA2B as it is highly expressed in oestrogen receptor (ER)-negative breast cancer cell lines, its mRNA expression level correlates with that of FOSL1 in breast cancer cell lines and FRA1 can

bind to the promoter of ADORA2B. Knockdown of ADORA2B in LM2 cells limited metastasis in mouse models, suppressed cell migration invitro and had only a small effect on cell proliferation. Theophylline, an adenosine receptor antagonist that is clinically approved for the treatment of respiratory diseases, inhibited cell migration in vitro and, although it had no effect as a single agent, when combined with docetaxel it synergistically reduced the growth of LM2 lung metastases in mice. These data, combined with the authors finding that a FRA1associated gene signature could predict the likelihood of breast cancer recurrence, indicate that the FRA1ADORA2B axis should be further investigated.Nicola McCarthyORIGINAL RESEARCH PAPER Desmet, C. J. et al. Identification of a pharmacologically tractable Fra-1/ADORA2B axis in promoting breast cancer metastasis. Proc. Natl. Acad. Sci USA 12 Mar 2013 (doi:10.1073/pnas.1222085110)

PhotoDisc




SENESCENCE

ConnectING endocytosisdisrupting endocytosis could induce senescenceThe inhibitor of growth (ING) proteins are epigenetic regulators that have tumour suppressor roles, particularly as mediators of senescence. Expression of the splice variant ING1A induces senescence invitro, but the mechanisms have been poorly characterized. A new study identifies downstream effectors of ING1A-mediated senescence induction, including a role for defective endocytosis. To characterize the genes that show altered expression in response to ING1A expression, Rajarajacholan, Thalappilly and Riabowol carried out gene expression microarray analyses of Hs68 human fibroblasts over expressing ectopic ING1A compared with control Hs68 cells. They found endocytosis-associated genes among the most differentially regulated genes, particularly the upregulation of the endocytic assembly factor intersectin 2 (ITSN2). Next, the authors tested the functional links between ING1A, endocytosis and senescence. A common marker for endocytosis is the internalization of epidermal growth factor receptor (EGFR) in response to EGF treatment. Overexpressing ING1A in different cell types in vitro inhibited the endocytosis and degradation of EGFR, but this was enhanced in Ing1deficient mouse embryo fibroblasts (MEFs) compared with wild-type MEFs. Therefore, ING1A seems to have a role in regulating endocytosis. Moreover, independently disrupting endocytosis either pharmaco logically by treating with Dynasore or genetically by manipulating the expression of endocytosis proteins could induce senescence. To determine whether the ING1A-regulated endocytosis protein ITSN2 is a key mediator of senescence, the authors overexpressed ITSN2 in fibroblasts, which was sufficient to induce some markers of senescence. These included INK4A expression, senescenceassociated -galactosidase activity and senescence-associated heterochromatic foci, although cells did not adopt the characteristic flattened cellular morphology of senescence. Furthermore, ITSN2 knockdown could diminish various features of senescence in ING1A-overexpressing cells such as cell cycle arrest and E2F target gene silencing thus providing evidence that ITSN2 is required for some senescence outputs. Chromatin immunoprecipitation experiments revealed that ING1A binds upstream of the ITSN2 gene, which is consistent with ITSN2 being a direct transcriptional target of ING1A during senescence induction. Turning to additional senescence settings, the authors found that endogenous ING1A and ITSN2 were upregulated during replicative senescence and in senescence triggered by the oxidative stress agent tert-butyl hydroperoxide, whereas neither protein was upregulated in senescence induced by the DNA-damaging agent doxorubicin. Overall, this implicates the ING1AITSN2 pathway in some, but not all, senescence contexts.

It will be interesting to characterize the mechanistic connections between endocytosis and senescence and to assess the contribution of ING1A, ITSN2 and endocytosis disruption to the tumour-suppressive effects of senescence in vivo.Darren J. BurgessORIGINAL RESEARCH PAPER Rajarajacholan, U.K., Thalappilly, S. & Riabowol, K. The ING1a tumor suppressor regulates endocytosis to induce cellular senescence via the Rb-E2F pathway. PLoSBiol. 11, e1001502 (2013)


Simon Bradbrook/NPG



G L I O B L A S TO M A

The histones have itH3K27M reduced the methyltransferase activity of PRC2A paper published in Science by David Allis and colleagues indicates that one of two common histone mutations found in paediatric glioblastoma is a gain-of-function mutation that inhibits histone trimethylation that is mediated by the Polycomb repressive complex 2 (PRC2). Of cases of diffuse intrinsic pontine glioma (DIPG) and supratentorial glioblastoma multiforme (GBM), 60% have mutually exclusive missense mutations in one of two histone3 (H3)-encoding genes: H3F3A (one of the genes that encodes H3.3) and HIST3H1B (one of the genes that encodes H3.1). These mutations result in K27M, G34R or G34V amino acid changes, with almost 80% of DIPGs having the K27M mutation in either gene. DIPG tumours with H3K27M mutations had reduced levels of H3K27 trimethylation (H3K27me3) genome-wide and a modest increase in H3K27 acetylation levels compared with DIPGs that did not have this mutation. These findings were supported by the transgenic expression of H3.3K27M in a mouse model of glioma. Given that these H3 variants do not account for the majority of H3 expressed in the cell, why do H3K27me3 levels drop across the genome? The authors found that only the mutation of lysine to methionine, and to a lesser extent to isoleucine, had this genome-wide effect. Dimethylation and trimethylation of K27 is mediated by PRC2. In vitro analyses using wild-type and mutated nucleosomes showed that H3K27M reduced the methyltransferase activity of PRC2, specifically by inhibiting the SET domain-containing catalytic subunit EZH2. Moreover, mutation of H3K9 and H3K36 to methionine inhibited other SET domain methyltransferases that specifically methylate these lysines and also reduced their trimethylation levels. These data indicate that lysine to methionine mutations in histones might be more broadly applicable to the development and progression of other tumour types.

NPG

It is not yet clear why aberrant epigenetic silencing as a result of reduced PRC2 activity contributes to the development of paediatric gliomas, and further research is needed to gain a better understanding of these difficult-to-treat childhood cancers.Nicola McCarthyORIGINAL RESEARCH PAPER Lewis, P. W. et al. Inhibition of PRC2 activity by a gain-of-function H3 mutation found in pediatric glioblastoma. Science 28 Mar 2013 (doi:10.1026/ science.1232245)



RESEARCH HIGHLIGHTS

In the newsIMMUNOTHERAPY SQUAREDPhase I clinical trial data presented at the 2013 American Association for Cancer Research (AACR) meeting support the use of a new combination immunotherapy strategy in patients with advanced ovarian cancer. LanaE. Kandalaft, from the University of Pennsylvania, USA, who presented the study, said Its not a slam dunk but the more we do, the more we learn (The Philadelphia Inquirer, 7 Apr 2013). Patients initially received a personalized antitumour dendritic cell (DC) vaccine, in which tumour antigens present in tumour tissue collected during each patients surgery were used to activate the patients own DCs. They also received bevacizumab, which targets vascular endothelial growth factor (VEGF). A clinical benefit was seen in 19 of 31 patients, and eight of these had no measurable disease at the end of the study. One of the eight has remained disease-free for 42 months following vaccination, although it is difficult to say conclusively whether the vaccination was responsible for her response. The 11 patients who responded to the vaccine but who still had residual disease then underwent adoptive Tcell therapy, which seemed to amplify the immune response, as the DC vaccine had already educated the T cells to respond to the tumour antigens. This allowed seven patients to achieve stable disease, and one had a complete response. Although the trial is still ongoing, Louis Weiner of Georgetown University, USA, who was not involved with the study, noted that it shows that its now possible to devise very efficient and complex but feasible combination strategies (starting with) a vaccination that will basically point the immune system in the direction of the tumour and that this may help overcome some of the innate resistance mechanisms that cancers use (MedPage Today, 7 Apr 2013).Sarah Seton-Rogers



RESEARCH HIGHLIGHTS

IN BRIEFTHERAPEUTICS

Picking and choosingDesigning drugs against the fibroblast growth factor (FGF) pathway is complicated by the fact that there are several FGF family members with different biological activities. Blocking the hormonal FGFs causes toxicities; so, Harding et al. developed a soluble decoy receptor, FP-1039, that binds only the mitogenic FGFs. FP-1039 blocked FGF- and vascular endothelial growth factor (VEGF)-induced angiogenesis in vivo, and inhibited tumour growth in several xenograft mouse models with minimal toxicity. Xenografts of cancer cells with FGF pathway alterations were particularly sensitive to FP-1039.ORIGINAL RESEARCH PAPER Harding, T. C. et al. Blockade of nonhormonal fibroblast growth factors by FP-1039 inhibits growth of multiple types of cancer. Sci. Transl. Med. 5, 178ra39 (2013)

THERAPEUTICS

An alternative explanationThe heat shock protein 90 (HSP90) chaperone recruits protein kinase clients via cell division cycle 37 (CDC37). Polier et al. have found that CDC37 can directly prevent binding of ATP to kinases and so may affect kinase activity. Interestingly, the ATP-competitive kinase inhibitors vemurafenib (a BRAF inhibitor) and lapatinib (an ERBB2 and epidermal growth factor receptor inhibitor) block the binding of CDC37 to the oncogenic kinases BRAF and ERBB2, thus preventing them from accessing HSP90. As this leads to kinase degradation, it could account for some of the therapeutic effects of these inhibitors.ORIGINAL RESEARCH PAPER Polier, S. et al. ATP-competitive inhibitors block protein kinase recruitment to the Hsp90-Cdc37 system. Nature Chem. Biol. 17 Mar 2013 (doi:10.1038/nchembio.1212)

BREAST CANCER

Improving mouse modelsKnight et al. have developed a mouse model that resembles the claudin-low subtype of triple-negative breast cancer (TNBC) by expressing a weakly oncogenic MET receptor tyrosine kinase under the control of the mouse mammary tumour virus promoter concomitant with conditionally deleting Trp53 in the mammary gland. Tumours in these mice have a similar molecular signature to human claudin-low TNBC, and require MET for proliferation and for maintaining the claudin-low morphological phenotype. Therefore, MET inhibitors may be effective against TNBC.ORIGINAL RESEARCH PAPER Knight, J. F. et al. Met synergizes with p53 loss to induce mammary tumors that possess features of claudin-low breast cancer. Proc. Natl Acad. Sci. USA 18 Mar 2013 (doi:10.1073/pnas.1210353110)

I M M U N OT H E R A P Y

Modified CARAutologous chimeric antigen receptor (CAR)-modified Tcells that target the B cell antigen CD19 and that express the CD137 (a costimulatory receptor) signalling domain have been used in adults with chronic lymphocytic leukaemia (CLL). Gruppet al. tested these cells in two children with relapsed and refractory pre-B cell acute lymphoblastic leukaemia (ALL). Although several adverse events occurred, both patients had a complete remission, and this is still ongoing in one patient after 11months. The other patient relapsed after 2 months, with cells that no longer expressed CD19, indicating that other molecules may need to be targeted in some patients.ORIGINAL RESEARCH PAPER Grupp, S. A. et al. Chimeric antigen receptor-modified T cells for acute lymphoid leukemia. New Engl. J. Med. 25 Mar 2013 (doi:10.1056/ NEJMoa1215134)




O VA R I A N C A N C E R

At the starting linehilum cells are susceptible to transformationSeveral sites have been proposed to give rise to the cell of origin for epithelial ovarian cancer (EOC), including the ovarian surface epithelium (OSE) and the uterine (Fallopian) tubes. Furthermore, the stem cell niche in the OSE has not yet been defined. It has been hypothesized that a transitional zone (termed the hilum), which is located amid the OSE, the mesothelium and the tubal epithelium, and at which nerves and vessels enter the ovary, may contain cells with stem cell character istics and may also be a candidate location for the EOC cell of origin. Andrea Flesken-Nikitin, Alexander Nikitin and colleagues have investigated this possibility. Aldehyde dehydrogenase 1 (ALDH1) is a marker of stem and progenitor cells in several tissues, so the authors separated primary OSE cells into those expressing or those not expressing members of the ALDH family using fluorescenceactivated cell sorting. ALDH+ cells represented a small proportion of the total cells (5.3%), and single ALDH+ cells formed spheres at a higher frequency in vitro than ALDH cells. Furthermore, in ovarian tissues from several strains of mice, cells specifically positive for ALDH1 were primarily found in the hilum, and cells micro dissected from the hilum formed spheres that could be propagated for at least seven generations. Bromodeoxyuridine (BrdU) pulse-chase experiments indicated that the ALDH1+ cells in the hilum proliferate slowly relative to other ovarian tissues and, interestingly, the number of BrdU+ cells in the hilum increased just after ovulation, indicating that these cells are activated to repair the OSE following ovulation. The authors showed that leucinerich repeat-containing Gproteincoupled receptor 5 (LGR5), another stem cell marker, is also highly expressed in ALDH1+ cells, and the authors conducted lineage tracing using a mouse model in which LGR5-directed expression of a tamoxifen-inducible Cre induced the expression of a red fluorescent protein variant following tamoxifen treatment. At early time points after tamoxifen treatment, hilum cells were labelled exclusively, but these went on to contribute to regeneration throughout the OSE. Are ALDH1+ and LGR5+ cells in the hilum susceptible to carcino genesis? In mice, conditional inactivation of Trp53 and Rb1 (the loss of which is common in human high-grade serous ovarian cancers) resulted in cells from the hilum that

proliferated at a higher rate in vitro than other OSE cells with the same mutations. Furthermore, hilum cells deficient in both Trp53 and Rb1 formed metastatic high-grade serous carcinomas at a high frequency when transplanted intraperitoneally, whereas cells from elsewhere in the OSE did not. The hilum also contained the earliest detectable abnormal cells following inactivation of Trp53 and Rb1. Together, these data indicate that the hilum cells are susceptible to transformation and may be a cell of origin for EOC. This research raises the interesting questions of whether this zone is also susceptible to transformation in humans and whether similar transitional zones in other organs may be cancer-prone. Importantly, if human EOC also arises from the hilum, specific examination of this site may assist early diagnosis of these difficult-to-detect and, therefore, deadly tumours.Sarah Seton-RogersORIGINAL RESEARCH PAPER Flesken-Nikitin, A. et al. Ovarian surface epithelium at the junction area contains a cancer-prone stem cell niche. Nature 495, 241245 (2013)

PHOTODISC




MELANOMA

More horsesforced expression of PGC1 protected cells against PLX4720In the April issue of Nature Reviews Cancer we highlighted a paper on peroxisome proliferator-activated receptor- coactivator 1 (PGC1; encoded by PPARGC1A) and increased oxidative metabolism in a subset of melanoma cells. Now, Rizwan Haq and colleagues have extended these findings by linking the expression of PGC1 to that of oncogenic BRAF. The authors used published gene expression profiles of melanomas with mutant BRAF that had been treated with vemurafenib, an inhibitor of BRAF-V600E, to look for genes that had altered expression levels after treatment. The expression of genes involved in the citric acid (also known as Krebs) cycle and oxidative phosphorylation and ATP generation was increased, and these findings were supported by quantitative PCR analyses in three melanoma cell lines. Moreover, treatment of these cell lines with PLX4720 (the preclinical analogue of vermurafenib) showed that loss of BRAF activity induced an increase in the numbers of mitochondria and their activity,

and decreased lactate production (indicative of increased oxidative phosphorylation), but these effects were not evident in a BRAF wild-type cell line. Further experiments showed that BRAF-V600E suppressed the expression of PPARGC1A, a regulator of mitochondrial metabolism, and treatment of BRAF-mutant melanoma cell lines with PLX4720 increased the expression of PGC1. The effect of BRAF-V600E on PGC1 seems to occur as a result of increased MEKERK signalling, as an inhibitor of MEK, PD0325901, also increased PPARGC1A mRNA levels. Interestingly, the link between oncogenic BRAF and PGC1 was not evident in colon cancer cell lines with mutant BRAF, nor was it apparent in microarray data from breast, lung and colon cancers treated with PD0325901, indicating a lineage-specific effect. Consistent with this, the authors found that the expression of PGC1 correlated with that of microphthalmia-associated transcription factor (MITF), the expression of which is limited to the melanocytic lineage. In silico

analyses and chromatin immunoprecipitation and luciferase reporter assays showed that MITF binds to the PPARGC1A promoter and that increased expression of PCG1 is absent in cells in which BRAF-V600E is inhibited by PLX4720 and in which MITF expression is knocked down. MITF is amplified in 30% of melanomas, and melanomas that had high expression levels of genes involved in oxidative phosphorylation also had increased expression levels of MITF. Moreover, MITF expression correlated with PGC1-regulated gene expression. A comparison of immortalized human melanocytic isogenic cell lines that expressed BRAF-V600E and that only differed in their expression of MITF showed that MITF protects against reduced ATP generation when BRAF-V600E is inhibited. Moreover, the isogenic cells expressing MITF were more sensitive to a mitochondrial uncoupling agent. Using eight patient-derived melanoma cell lines the authors also found that PGC1 was induced after treatment with vemurafenib and ATP levels responded accordingly; however, the magnitude of these responses varied considerably between the cell lines. High levels of PGC1 are associated with a poor prognosis, and the authors found that forced expression of PGC1 protected cells against PLX4720, and that sensitivity to PLX4720 increased when this drug was combined with a mitochondrial uncoupler. These findings indicate that the increased expression of MITF and PGC1 represents an adaptive metabolic response that limits sensitivity to BRAF inhibitors.Nicola McCarthyORIGINAL RESEARCH PAPER Haq, R. et al. Oncogenic BRAF regulates oxidative metabolism via PGC1 and MITF. Cancer Cell 23, 302315 (2013) FURTHER READING Vazquez, F. et al. PGC1 expression defines a subset of human melanoma tumors with increased mitochondrial capacity and resistance to oxidative stress. Cancer Cell 23, 287301 (2013) | McCarthy, N. Horses for courses. Nature Rev. Cancer 13, 222 (2013)


PHOTODISC



LEUKAEMIA

Gimme shelter

Mutation of POT1 results in increased levels of telomerebased genomic rearrangements

Maria Blasco, Elias Campo, Carlos Lpez-Otn and colleagues have found that protection of telomeres 1 (POT1), a component of the shelterin complex that caps the ends of telo meres, is mutated in a subset of patients with chronic lymphocytic leukaemia (CLL). Mutation of POT1 results in increased levels of telomere-based genomic rearrangements. Initially, the authors examined exome sequence data from 127 matched tumour and normal samples from patients who were diagnosed with CLL but who had not started treatment. In addition to the known driver mutations, such as SF3B1, NOTCH1 and MYD88, they found that approximately 5% of the samples had a mutation in POT1 and that these mutations were point mutations. Sanger sequencing of POT1 in an additional 214 samples from patients with CLL supported the initial findings and, overall, 12 somatic

point mutations in POT1 were found in 341 samples. Interestingly, POT1 mutations occurred exclusively in patients with a wild-type immunoglobulin heavy chain variable gene cluster (IGHV@), a biological status that correlates with a poorer prognosis. Indeed, the authors found that POT1 mutations also correlate with a more advanced disease at diagnosis. Three of the 12 somatic mutations resulted in a truncated protein, and the remaining nonsynonymous mutations were all predicted to have deleterious effects on the function of POT1. The majority of the mutations reside in the two oligonucleotideoligosaccharide binding (OB) folds present in the amino terminus of POT1. These folds enable POT1 to bind to the telomeric single-stranded (ss) DNA sequence TTAGGG, suggesting that this crucial function is compromised in POT1 mutants. Indeed, POT1

mutations found in other types of tumours also predominately occur in the OB folds. Immunofluorescence, chromatin immunoprecipitation and electrophoretic mobility shift assays indicated that mutations that disrupt the OB folds do not inhibit the localization of POT1 to telomere ends, which is perhaps consistent with the fact that POT1 binds to TPP1, another component of shelterin, through its carboxyl terminus, but that mutant POT1 is unable to bind telomeric ssDNA. The authors also found that only one allele of POT1 is mutated in CLL, suggesting that the mutations in POT1 result in a dominantnegative protein. In agreement with this hypothesis, a wild-type POT1 human fibrosarcoma cell line (HT1080) expressing exogenous mutant POT1 proteins had longer telomeres, but there was no change in the activity or level of telomerase. Moreover, these cells also had an increased incidence of chromosomal aberrations, including the fusion of sister chromatids at metaphase, increased telomere fragility and significantly increased levels of chromosomal breaks and fusions. Similar results were evident in CLL cells from patients with POT1 mutations, although there were no obvious differences in telomere lengths compared with patient samples with wild-type POT1. This probably reflects the fact that telomere lengths are highly variable in humans. Collectively, these data indicate that POT1 mutations result in telomere uncapping, leading to an increased incidence of telomererelated chromosomal fusions. It will be interesting to determine why POT1 mutations only seem to occur in patients with wild-type [email protected] McCarthyORIGINAL RESEARCH PAPER Ramsay, A. J. et al. POT1 mutations cause telomere dysfunction in chronic lymphocytic leukemia. Nature Genet. 17 Mar 2013 (doi:10.1038/ng.2584)


Lara Crow/NPG


REVIEWSDysregulation of the basal RNA polymerase transcription apparatus in cancerMegan J.Bywater1,2, Richard B.Pearson14, Grant A.McArthur1,3,5,6 and Ross D.Hannan1,2,3,4,7

Abstract | Mutations that directly affect transcription by RNA polymerases rank among the most central mediators of malignant transformation, but the frequency of new anticancer drugs that selectively target defective transcription apparatus entering the clinic has been limited. This is because targeting the large proteinprotein and proteinDNA interfaces that control both generic and selective aspects of RNA polymerase transcription has proved extremely difficult. However, recent technological advances have led to a quantum leap in our comprehension of the structure and function of the core RNA polymerase components, how they are dysregulated in a broad range of cancers and how they may be targeted for transcription therapy.Division of Cancer Research, Peter MacCallum Cancer Centre, Locked Bag 1, Melbourne 8006, Victoria, Australia. 2 Sir Peter MacCallum Department of Oncology, The University of Melbourne, Parkville 3010, Victoria, Australia. 3 Department of Biochemistry and Molecular Biology, University of Melbourne, Parkville 3010, Victoria, Australia. 4 Department of Biochemistry and Molecular Biology, Monash University, Clayton 3800, Victoria, Australia. 5 Department of Medicine, St Vincents Hospital, University of Melbourne, Fitzroy 3065, Victoria, Australia. 6 Division of Cancer Medicine, Peter MacCallum Cancer Centre, Locked Bag 1, Melbourne 8006, Victoria, Australia. 7 School of Biomedical Sciences, University of Queensland, Brisbane 4072, Queensland, Australia. Correspondence to R.D.H. e-mail: ross.hannan@ petermac.org doi:10.1038/nrc34961

The human genome is immensely complex, containing more than 25,000 genetic loci organized in compact chromatin structures. The intricate, tightly regulated mechanisms by which the genes contained in these domains are transcribed and how these processes are dysregulated during cancer have been the focus of considerable attention for the past 30years. Central to the expression of the genome are three functionally distinct classes of nuclear DNA-dependent RNA polymerases (RNA Pols)1 that were shown to transcribe genes that encode distinct classes of RNAs. RNA Pol I transcribes ribosomal RNA; RNA Pol II transcribes protein-encoding mRNA; and RNA Pol III transcribes 5S rRNA and tRNA2,3. However, subsequent studies and more contemporary genome-wide approaches have identified additional targets for these RNA Pols, in particular for RNA Pol II48 and RNA Pol III912. The specificity of transcription reflects the structurally distinct subunit compositions of the three classes of RNA Pol, which contain both common and unique subunits13 (FIG.1). Indeed, recent detailed structural studies demonstrate that, although the active centre region and core enzymes are similar 14,15, each RNA Pol exhibits strong structural differences on its surface, which is consistent with the gene class-specific functions of each RNA Pol1518. The altered patterns of gene expression associated with malignant transformation have generally been attributed as a downstream consequence of mutations

in genes that encode cellular signalling molecules such as growth factor receptors, kinases, small GTPases and other intracellular adaptor molecules. However, it became evident that genomic rearrangements or mutations in sequence-specific DNA-binding RNA Pol transcriptional activators and repressors were also either crucial drivers or contained secondary mutations that were required for the malignant transformation of a broad range of cell types. Perhaps the most prototypical examples of such sequence-specific transcription factors are MYC and p53, the oncogenic and tumour suppressive activity of which, respectively, is nearly ubiquitously deregulated during malignant transformation. More recently, cancer genome-sequencing projects have revealed that mutations occur in a wider range of RNA Pol transcriptional apparatus components than previously thought, and these include corepressors, coactivators, components of the mediator complex, chromatin modifiers and even factors involved in RNA Pol transcription elongation (TABLE1). For the most part, the precise roles of these mutations and the contributions that they make to the aetiology of cancer are unknown, simply because the new technologies used to identify novel cancer-specific mutations have surpassed our ability to test for the causative role of these mutations in malignant transformation. Although traditionally the major focus has been on RNA Pol II in malignancy, the odd RNA Pols, RNA Pol I and RNA Pol III, are not idle bystanders in cancer aetiology.VOLUME 13 | MAY 2013 | 299


REVIEWSAt a glanceThe core RNA polymerase (Pol) subunits and general transcription factors (GTFs) are rarely mutated in cancer, although some GTFs are consistently overexpressed in tumours and this is thought to contribute to malignancy in somecases. Subunits of the mediator complex are increasingly being found to be mutated or amplified in tumours, where they have oncogenic or tumour suppressive activities and functions, depending on the genetic background and cellular context. Regulators of post-initiation stages of transcription, particularly components of RNA Pol II super elongation complexes (SECs), are recurrently mutated in cancer, particularly haematological malignancies through translocation with the mixed lineage leukaemia (MLL) family of transcription factors. The resultant fusion proteins facilitate the enhanced transcription elongation of homeobox (HOX) transcription factors that are involved in embryonic development and haematopoietic cell differentiation, which drives malignancy. RNA Pol I and RNA Pol III are consistently dysregulated in cancer, which is mostly mediated through upstream oncogenic and tumour suppressive signalling pathways rather than through mutations. The most potent and pervasive oncogenic and tumour suppressive components of the transcription apparatus seem to be those that are capable of modulating all three RNAPols. RNA Pol I transcriptional overactivity has been shown to be necessary for the survival of haematological tumour cells, and the RNA Pol I GTF SL1 has been successfully targeted using a small-molecule inhibitor to therapeutically treat transgenic mouse models of cancer invivo. RNA Pol I transcription therapy is currently entering PhaseI trials in humans for the treatment of lymphoma and leukaemia. Components of the core transcription apparatus, including the mediator complex and the SEC, represent bone fide therapeutic targets for cancer treatment not only as advanced broad-spectrum cytotoxics but also potentially as part of the new paradigm of personalized medicine.

Mediator complexA multiprotein complex that functions as an RNA Pol II transcriptional co-activator in all eukaryotes, although it is unable to bind specific DNA sequences, it functions as an adaptor between sequencespecific transcription factors bound at regulatory elements, and RNA Pol II and GTFs.

General transcription factors(GTFs). Also known as basal transcriptional factors. A class of protein transcription factors that bind to specific sites on DNA to activate transcription and are essential for basal (as opposed to activated) RNA Pol I, RNA Pol II and RNA Pol III transcription.

TATA boxA DNA sequence (cis-regulatory element) that is found in the core promoter region of a subset of RNA Pol II and RNA Pol III genes. For RNA Pol II the TATA box is involved in recruiting the RNA Pol II general transcription factor (GTF) transcription initiation factor IID (TFIID) that contains TATA-box-binding protein (TBP).

Although the range of target genes these RNA Pols transcribe is considerably smaller compared with RNA Pol II, their non-coding RNA transcripts account for over 50% of all transcription in actively growing cells, and their dysregulated expression is a consistent feature of tumour cells. In the case of RNA Pol I, its overactivity has been shown to be necessary for the survival of haematological tumour cells and can be therapeutically targeted invivo19. Moreover, the demonstration that classical RNA PolIIdependent sequence-specific transcription factors can directly interact with, and profoundly modulate, the core RNA Pol I and RNA Pol III transcription apparatus in most cancer settings, suggests that hijacking the control of these RNA Pols might be essential for malignant trans formation. Indeed, the most potent and pervasive oncogenic and tumour suppressive components of the transcription apparatus seem to be those that are capable of modulating all three RNAPols. In this Review, we illustrate the common and distinct mechanisms by which all three RNA Pols are subject to dysregulation in cancer. Specifically, we focus on the core RNA Pol transcription apparatus, the mediator complex, its associated coactivators, corepressors and components that are involved in transcription elongation. We also highlight the current progress and utility of drugs that aim to directly target the transcriptional apparatus for cancer therapy.

RNA Pol general transcription factors Despite the multisubunit nature of the core RNA Pols, accurate transcription initiation on the respective core

promoter elements requires additional RNA Pol-specific general transcription factors (GTFs) that interact, directly or indirectly, with the polymerase20 (FIG.1). GTFs are required to position the RNA Pol at the gene promoter to help melt (open) the two DNA strands, and confer to the enzyme the full competence for initiating and elongating transcription. Each RNA Pol uses a distinct set of GTFs, and only one GTF, TATA-box-binding protein (TBP), is common to all three RNA Pols. TBP is a component of the essential RNA Pol I transcription initiation factor SL1 (also known as the TIF-1B), the RNA PolII transcription initiation factor TFIID and the RNA Pol III transcription initiation factor TFIIIB (FIG.1). However, recent studies suggest that some additional RNA Pol II GTFs might also function in RNA PolI transcription (FIG.1). For example, the TFIID subunit 12 (TAF12), a component of RNA Pol II TFIID that interacts with TBP, also associates with the TBP-containing SL1 initiation factor of RNA Pol I21. SL1, which contains at least five TATA-box-associated factors (TAFs) in addition to TBP, is essential for recruiting RNA Pol I to the ribo somal RNA gene (rDNA) promoter and is particularly relevant as it is the only GTF against which small-inhibitory molecules have been developed for use in cancer therapy 19,22. Similarly, the RNA Pol II complex TFIIH also associates with RNA Pol I23, and probably functions in RNA Pol I transcription elongation2426. Owing to their highly conserved, non-redundant function, mutations in genes encoding core RNA Pol II components and GTFs are extremely rare, and have not been routinely associated with cancer aetiology. Perhaps the exception is TFIIH, in which loss-of-function mutations of the TFIIH subunit ERCC2 (also known as XPD and RAD3), which is involved in nucleotide excision repair27, are associated with an increased susceptibility to certain cancers in addition to the well-known role of ERCC2 in human genetic disorders such as the cancerprone syndrome xeroderma pigmentosum2830. By contrast, several core GTFs have been shown to be subject to changes in expression that are associated with and even causative of malignant transformation19,3134. One of the best examples is TBP. In certain cell types TBP levels have been shown to be limiting for the transcription of RNA Pol I and RNA Pol III target genes35. The situation for RNA Pol II genes is more complex. Although TBP is required for both TATA-box-containing and TATA-box-less promoters, only the RNA Pol II promoters that contain TATA boxes can be activated by ectopic TBP expression36,37. Thus, overexpression of TBP results in selective changes in the cellular transcriptional profiles of genes with TATA box-containing promoters rather than in a generic effect on transcription perse. The overexpression of TBP has been demonstrated in colon31 and colorectal32 cancer as the result of activated HRAS signalling. Moreover, TBP overexpression alone is sufficient to increase anchorage-independent growth in Rat1A cells, in addition to facilitating their ability to grow as xenografts31, and this was dependent on enhanced RNA Pol III transcription38. However, the enforced expression of mutant TBPs that cannot participate in RNA PolIIdependent transcription, or bind towww.nature.com/reviews/cancer

300 | MAY 2013 | VOLUME 13 2013 Macmillan Publishers Limited. All rights reserved

REVIEWSaRNA Pol I TAFI12 RRN3 TBP TAFIB TAFI41 TAFIA SL-1 TFIIIB TAFIC TFIIIB

c

RNA P1 Pol III D B F1 TBP TFIIIC BR A B

tRNA

UCE UBF dimer

Core

RNA 1 Pol III P BD F1 TBP TFIIIC BR A C TFIIIA TFIIIB

5S rRNA

bTFIID TBP TATA RNA Pol II TFIIB TFIIF TFIIE TFIIH

BDSNAPC PSE

P1

RNA Pol III RNU6-1

TFIIA

F2 BR TBP TATA

Figure 1 | Basal transcription machineries and promoter structures of the eukaryotic DNA-dependent RNA polymerases I, II and III. a | Assembly of the RNA polymerase (Pol) I pre-initiation complex (PIC) at ribosomal RNA |gene Nature Reviews Cancer (rDNA) promoters begins with the binding of upstream binding factor (UBF) to the upstream control elements (UCEs) and core element of the rDNA promoter, leading to the recruitment of the SL1 initiation factor, which contains TATA-box-binding protein (TBP) and at least five TATA-box-associated factors (TAFs). The resultant stable UBFSL1 complex recruits an initiation-competent form of RNA Pol I, which contains RRN3 that mediates interactions between RNA Pol I and SL1 (REF.189). b | For RNA Pol II transcription, TBP initiates PIC assembly by binding to the TATA box at the promoter. TFIIA and TFIIB interact with TBP and reinforce its binding to DNA. In turn, TFIIB recruits RNA Pol II and TFIIF, thus positioning RNA Pol II over the transcription start site (TSS). TFIIH mediates melting of the TSS to form the open complex that is stabilized by TFIIE190. The dashed outline of TBP indicates that it is part of the TFIID complex. c | RNA Pol III PICs differ in composition depending on the class of genes transcribed. Most RNA Pol III-transcribed genes (for example, those that encode tRNAs) have internal promoters that comprise two sequence blocks (A and B) that are located in the transcribed region. The A and B blocks are recognized by TFIIIC that recruits TFIIIB, which is composed of the subunits Brelated factor 1 (BRF1), BDP1 and TBP. Finally, TFIIIB recruits RNA Pol III. For 5S rDNA promoters the B block is replaced by a sequence, termed block C, to which TFIIIA binds and recruits and orientates TFIIIB, following which transcription initiation proceeds as for tRNA genes. For a small number of RNA Pol III-transcribed genes (for example, U6 snRNA (RNU61)) the promoters are located upstream of the gene and contain TATA boxes bound by TBP, and proximal sequence elements (PSEs) bound by a complex called small nuclear RNA-activating protein complex (SNAPC). These upstream promoters are bound by a different form of TFIIIB from tRNA genes, which is composed of BRF2, BDP1 and TBP9.

SL1An essential transcription initiation factor complex unique to RNA Pol I consisting of TATA-box-binding protein (TBP) and at least five TBP-associated factors (TAFs) that functions to recruit RNA Pol I to the ribosomal RNA gene promoters.

Nucleotide excision repairA DNA repair mechanism that removes mutations resulting from ultraviolet-induced DNA damage.

Xeroderma pigmentosumAn autosomal recessive genetic disorder of nucleotide excision repair caused by mutations in at least eight separate genes including XPC, ERCC2 and POLH, in which the ability to repair DNA damage caused by ultraviolet light is deficient.

consensus TATA sequences, failed to transform cells31. Thus, it seems that the ability of TBP to bind to consensus TATA-containing promoters and changes in RNA PolIIdependent transcription are also necessary for TBPmediated transformation. The specific contribution of RNA Pol I transcription to TBP-mediated transformation remains to be determined. These studies are important as they provided the first real evidence that the regulation of a GTF could affect the malignant potential of cells. However, the mechanism behind this process is less clear. Increased cellular TBP levels have been shown to induce both RNA PolIdependent and RNA Pol III-dependent transcription39,40. Thus, it has been proposed that, as their products, tRNAs and rRNAs, are limiting for cellular growth rates, increases in their synthesis would promote the biosynthetic capacity of cells that is needed for oncogenic transformation31. Consistent with this, as only a subset of RNA Pol II promoters are responsive to changes in cellular TBP concentration36,37,41, increased TBP expression could lead to qualitative and quantitative changes in cellular proteins that regulate growth control31.

Similar to the core RNA Pol II components, mutations in RNA Pol I and RNA Pol III subunits are extremely rare42 and have not yet been associated with malignant transformation. By contrast, several RNA PolI and RNA Pol III GTFs are overexpressed or amplified in cancer, and this seems to contribute to malignancy. With respect to RNA Pol Ispecific GTFs, an RNA Pol I regulon consisting of genes encoding >90% of the essential factors and subunits involved in RNA Pol I transcription initiation, which contain Eboxes in their promoters, is selectively upregulated by MYC19,43 (BOX1). Moreover, the increased expression of RNA Pol Ispecific GTFs is required to maintain MYC-driven malignancy, as normalization of their increased levels leads to rapid apoptosis of tumour cells19. Importantly, these studies suggest that the transforming capability of the potent oncogene and transcription factor MYC partly depends on its ability to modulate RNA Pol I transcription19. With regard to RNA Pol III GTFs, Brelated factor 1 (BRF1; also known as TFIIIB90), the structurally related BRF2 (also known as TFIIIB50) and TFIIIC have been found to be overexpressed in multiple types of cancer 33,34,4448 (FIG.1). In higher eukaryotes, two formsVOLUME 13 | MAY 2013 | 301


REVIEWSTable 1 | DNA-dependent RNA polymerase I, II and III in cancer*RNA Pol I Polymerase coreA190

RNA Pol IIRPB1

RNA Pol IIIC160

FunctionActive centre Active centre

Dysregulation in cancerRNA Pol I subunits are overexpressed in MYC-driven tumours

Refs19,42

RPB2 AC40 AC19

C128 AC40

RPB3 RPB11 RPB9 RPB5 RPB8

AC19 C11 RPB5 RPB6 RPB8 RPB10 RPB12 C17 C25

A12.2 RPB5 (ABC27)

RNA cleavage

RPB6 (ABC23) RPB8 (ABC14.5)

RPB6

APB10 (ABC10a) RPB12 (ABC10b)

RPB10 RPB12 RPB4 RPB7

Polymerase stalk GTFs

A14 A43

Initiation complex formation Initiation complex formation Initiation complex formation and start site selection Initiation complex formation and start site selection Open complex stabilization Open complex stabilization Open complex stabilization DNA binding (TATA-boxcontaining promoters only) TBP-associated factor, TBP and polymerase binding DNA opening and start site selection Overexpressed in colon and colorectal cancer Overexpressed or amplified in lung, breast and bladder cancer SL1 components are overexpressed in MYC-driven tumours Mutations associated with increased cancer susceptibility in xeroderma pigmentosum 31,32 33,34, 4448 19,43

A49 (N-terminal domain) A34.5

TFG1 (TFIIF) TFG2 (TFIIF) TFA1 (TFIIE)

C37 C53 C82 C34 C31

A49 (C-terminal domain) TBP TAFs (SL-1)

TFA2 (TFIIE)

TBP TAFs (TFIID and TFIIB)

TBP TAFs (TFIIIB and BRF)

TFIIH

TFIIH

Transition of the PIC to the open complex, elongation and nucleotide excision repair

2830

of TFIIIB have so far been identified4951. The form of TFIIIB that is required for proper initiation from geneinternal RNA Pol III promoters is comprised of TBP, BDP1 and BRF1. Proper initiation from gene-external RNA Pol III promoters requires TBP, BDP1 and BRF2 (FIG.1). Of particular note is the fact that the gene encoding BRF2 has been shown to be amplified in breast cancer, bladder cancer and in a panel of lung squamous cell carcinoma specimens, and this amplification correlated with increased BRF2 transcript levels. Importantly, this is the only genetic lesion in a GTF that has been identified to date in cancer 34,4547. Moreover, ectopic expression of BRF2 in human bronchial epithelial cells induced a transformed phenotype, whereas RNA interference (RNAi)-mediated knockdown of BRF2 expression in lung squamous cell carcinoma lines with BRF2 amplification decreased proliferation and anchorageindependent growth only in cell lines with BRF2 amplification34. Strikingly, these data implicate BRF2 as a novel lineage-specific oncogene in lung squamous cell carcinoma. The mechanism by which overexpression of BRF1 or BRF2 contribute to malignant transformation is not yet clear. It may be due to effects on302 | MAY 2013 | VOLUME 13

translation, as preliminary evidence from experiments with transgenic mice suggests that overexpression of the BRF1dependent transcript TRNAM1 (which encodes tRNAiMET) can result in the increased proliferation of several cell types, as well as in a predisposition to lymphoma in the context of a RAS mutation (R. White, personal communication).

Sequence-specific transcription factors Although the basal transcription machinery described above is sufficient to support accurate initiation and efficient transcription of DNA templates in vitro 52, selective RNA Pol recruitment and transcription of target genes invivo requires additional sequence-specific DNA-binding transcriptional activators, cofactors or mediator complexes to overcome repressive chromatin and/or to activate transcription above basal levels (FIG.2). The number of sequence-specific DNA-binding transcription factors that modulate RNA Pol activity is impressively large (~2,500). The vast majority are involved in regulating the transcription of protein-coding genes and non-coding RNAs, which are transcribed by RNA Pol II20. In general, these sequence-specificwww.nature.com/reviews/cancer

2013 Macmillan Publishers Limited. All rights reserved

REVIEWSTable 1 (cont.) | DNA-dependent RNA polymerase I, II and III in cancer*RNA Pol I Cofactors and mediatorsUBF

RNA Pol II

RNA Pol IIIPIC formation, elongation and chromatin remodelling

Dysregulation in cancerA target of oncogenic signalling and a MYC target gene, overexpressed in hepatocellular carcinoma Target of oncogenic signalling and MYC Amplified and mutated: CDK8 (colorectal cancer); MED1 (prostate, breast and ovarian cancer); MED12 (prostate cancer); MED19 (bladder cancer, breast cancer, lung tumours, hepatocellular carcinoma and osteosarcoma); and MED29 (pancreatic cancer)

Refs19,43, 137, 138 19,43, 9699 69, 7991

RRN3 (also known as TIF-1A) Mediator

Mediates interactions between GTF and RNA Pol I Recruits RNA Pol II PIC, also post-initiation roles

SAGA TFIIIC SNAPC

Transcription coactivator, interacts with TBP and TFs Recruitment of RNA Pol III (gene internal binding) Recruitment of RNA Pol III (gene external binding) Involved in promoter-proximal pausing, binds in a stable complex with DSIF Involved in promoter-proximal pausing TFIIS Prevents promoter-proximal Overexpressed or amplified in a pausing and facilitates high rates of variety of human tumours elongation Prevents pausing and facilitates high rates of elongation Communicates with transcriptional activators; recruitment and activation of histone modification factors; facilitation of elongation on chromatin templates; and the recruitment of 3 end-processing factors necessary for accurate termination of transcription Transcripton elongation complex comprised of transcription elongation factors such as ELL and P-TEFb, recruited to rapidlytranscribed genes in response to differentiation signals to release paused RNA Pol II FACT Fascilitates transcription through histones ELL component of SECs are translocation partners of MLL gene family in AML 118, 119 Mutations associated with increased cancer susceptibility in xeroderma pigmentosum 123, 124 2830

Elongation componentsDSIF

NELF

DSIF TFIIS

TFIIH

TFIIH

PAF1

PAF1

SEC

FACT

FACT

CDK, cyclin-dependent kinase; DSIF, DRB sensitivity-inducing factor; FACT, facilitates chromatin transcription; GTF, general transcription factor; NELF, negative elongation factor; PAF1, polymerase-associated factor 1; PIC, preinitiation complex; PTEFb, positive transcription elongation factor; SAGA, Spt-AdaGcn5acetyltransferase; SEC, super elongation complex; SNAPC, small nuclear RNA-activating protein complex; TAF, TBP-associated factor; TBP, TATA-box-binding protein; TF, transcription factor; TIF-1A, transcription initiation factor 1A; UBF, upstream binding factor. *Alernative names are indicated in brackets. The core RNA polymerase (Pol) I, II and III subunits are indicated.

transcriptional activators, such as MYC and p53, bind to specific DNA sequences that are located near the core promoter or distant promoter regions (called enhancers) and they interact with GTFs and the core RNA Pol II complex to enhance transcription (FIG.2). The sequencespecific transcriptional activators also recruit cofactors, mediator complexes and chromatin-modifying enzymes that lead to distinct post-translational modifications ofNATURE REVIEWS | CANCER

histones at the promoters. In contrast to GTFs, sequencespecific transcriptional activators are not usually essential for transcription: instead they facilitate the efficient transcription (or repression) of what may be a single gene or a cohort of genes, meaning that their alteration is less likely to be lethal. Consistently, sequence-specific transcriptional activators are frequently dysregulated in tumours through altered upstream signalling, mutation,VOLUME 13 | MAY 2013 | 303

2013 Macmillan Publishers Limited. All rights reserved

REVIEWSamplification and deletion. As their contribution to malignant transformation has been extensively documented elsewhere, they are not the focus of this Review. However, of particular importance to this discussion, is that although most sequence-specific transcriptional activators and repressors were originally identified in the context of RNA PolII more recent studies demonstrate that some can also directly interact with and regulate the RNA Pol I and RNA Pol III transcriptional machinery, thus providing a mechanism to coordinate transcription from all three RNA Pols (BOX 1) and also their dysregulation in cancer 9,53.REFS5760).

Cofactors and the mediator complex The multisubunit complex associated with RNA Pol II, which is termed the mediator complex, was discovered in Saccharomyces cerevisiae 54,55 and was subsequently shown to be conserved in mammals56 (reviewed in

This mediator complex, although unable to bind specific DNA sequences, serves as an adaptor between sequence-specific transcriptional activators bound at regulatory elements in the DNA, and RNA PolII and GTFs52. The human mediator complex contains ~30 subunits, of which at least 22 are homologues of the 25 S.cerevisiae mediator subunits6163. The mediator complex seems to be universally required at all RNA Pol II genes, and thus can be seen as a major conduit of regulatory information from enhancers to promoters, connecting regulatory proteins (activators and repressors) with the RNA Pol II transcription machinery 5759 (FIG.2). Specific mediator subunits modulate distinct gene expression programmes via interactions with appropriate sequence-specific transcription factors. Such interactions, combined with the effects of the core complex on RNA Pol II and GTFs, allows for the selective activation of transcription at specific target genes.

Box 1 | Transcription factors that drive all three RNA polymerases are powerful oncogenes and tumour suppressorsIn contrast to RNA polymerase II (RNA Pol II)-transcribed genes, for a which p53 predominantly functions as a sequence-specific transcription factor, p53 represses RNA Pol I transcription indirectly via an association with the TATA-box-binding protein (TBP) and TBP-associated factor 1C (TAF1C), which are components of the SL1 initiation factor. This interaction subsequently prevents the interaction of SL1 with the RNA Pol I Ac cytoarchitectural transcription factor upstream binding factor (UBF)160 (see the figure, part a). Similarly, p53 disrupts the interaction of TFIIIB TRRAP with TFIIIC and RNA Pol III161 by binding the TBP component of TFIIIB and thus decreasing RNA Pol III recruitment161 (see the figure, part c). MYC p53mediated repression of RNA Pol I and RNA Pol III transcription is E-box abolished by a p53 mutation (p53R175H) that frequently occurs in cancer160,162,163. Owing to the pivotal roles that RNA Pol I and RNA Pol III have in driving ribosome biogenesis and protein synthetic capacity that is essential for the accelerated growth of malignant cells, it is thought that the ability of p53 to repress RNA Pol I and RNA Pol III is essential for its tumour suppressive activity. Transcriptional activation of RNA Pol I by the transcription factor MYC has been shown to operate in both an indirect and a direct manner, mediated through the binding of MYC with its heterodimerization partner MYC-associated factor X (MAX) to canonical Eboxes (which are the consensus motifs for MYC target genes). b Indirectly, MYC facilitates the RNA Pol IImediated transcriptional activation of a large cohort of RNA Pol I-associated transcription factors, Ac including RRN3, UBF, general transcription factors (GTFs) and subunits HAT of the RNA Pol itself43 (see the figure, part b). This is necessary for MYC TRRAP to upregulate RNA Pol I transcription and is conserved from flies to

RNA Pol I TAFI12 p53 TAFIC RRN3 TBP TA FIB TAFI41 TAFIA SL-1 UCE UBF dimer Core 45S rRNA

RNA Pol I RRN3 UBF

TFIID RNA Pol II MAX humans164. In a more direct mechanism, increased MYC binding to the TBP MYC TFIIH TFIIA TATA TFIIB TFIIF TFIIE ribosomal RNA gene (rDNA) repeats in mouse cells has been reported in E-box UBF, RRN3 and POLR1 response to growth factor stimulation, which correlated with increased binding of RNA Pol I and transformation/transcription domain-associated protein (TRRAP), and a subsequent increase in total RNA synthesis165,166 (see the figure, part a). The exact molecular mechanism by which MYC directly transactivates the rDNA repeats is unclear, but may involve looping of the rDNA167. Similarly, MYC c is found at the promoters of the RNA Pol III-transcribed genes encoding TFIIIB tRNAs for leucine and tyrosine and 5S rRNA in response to serum. Rather TBP than binding to Eboxes, MYC binds the TFIIIB complex to facilitate an p53 increase in RNA Pol III transcription through the recruitment of TRRAP RNA Pol III Ac and its associated histone acetyltransferase (HAT) and GCN5 TFIIIB GCN5 cofactors168 (see the figure, part c). Most importantly, in addition to the well-recognized role of MYC in driving malignant transformation TRRAP TBP TFIIIC through RNA Pol II transcription, recent studies have demonstrated MYC that the ability of MYC to modulate ribosome biogenesis, RNA Pol I and 5S rRNA A C RNA Pol III transcription is also necessary for its oncogenic activity19. TFIIIAThe dashed outline of TBP indicates that it is part of the TFIID complex.

Nature Reviews | Cancer304 | MAY 2013 | VOLUME 13 2013 Macmillan Publishers Limited. All rights reserved www.nature.com/reviews/cancer

REVIEWSPre-initiation complex(PIC). The complex of proteins that is necessary for the initiation of RNA Pol I, RNA Pol II and RNA Pol III transcription in eukaryotes where it helps position RNA Pol over gene transcription start sites, denatures the DNA, and positions the DNA in the RNA Pol active site for transcription.

Together, these cofactors are thought to provide the basis for the complexity of RNA Pol II transcriptional regulation in eukaryotes that is required for intricate processes such as differentiation and development 64. Recent studies have shown that, in addition to modulating pre-initiation complex (PIC) formation at specific genes, the mediator complex also operates at postinitiationsteps, including the efficient elongation (potentially regulated by mediator of RNA Pol II transcription subunit 26 (MED26)) and capping of transcripts6567.4 TF TF

As the mediator complex is essential for the transcription of RNA Pol II target genes it has generally been considered that the mediator complex is unlikely to be dysregulated in cancer. However, some mediator complex subunits interact with specific transcription factors and thus can regulate the expression of distinct subsets of genes that are involved in development and differentiation, and these subunits increasingly seem to be subject to mutation or amplification in cancer 68. For example, the gene encoding cyclin-dependent kinase 8 (CDK8), which functions

CappingThe process in which a guanine nucleotide is connected to the 5 end of newly RNA Pol II transcribed mRNA via an unusual 5 to 5 triphosphate linkage that facilitates nuclear export of the mRNA, prevents its degradation by exonucleases, and promotes translation and 5 proximal intron excision.

1

TF MP

2 TF

Ligand TF

3 PTM

Histone modication (Me or Ac)

8 5 AD 6 CF RE TF 7 SECs AFF1 or AFF4 P-TEFb ENL or ELL2 AF9

Nucleosome

RE

9 CDK8TF

Mediator 11

NELF

12

PAF1 PAFC

10 IID TBP IIB IIA TATA RNA Pol II IIF IIE IIH

Transcription initiation NELF RNA Pol II DSIF

Paused transcription

CTD

Productive elongation RNA Pol II

DSIF

Figure 2 | Activation of RNA polymerase II transcription and its druggable interfaces. RNA polymerase (Pol) II transcriptional activation is initiated by transcription factors (TFs) that bind to their response elements (REs) in the Reviews | Cancer regulatory region of their target genes. These TFs recruit chromatin coactivator factors (CFs) thatNature covalently modify nucleosomes at specific histone residues (for example, methylation (Me) of arginine and lysine residues or acetylation (Ac) of lysine residues in the histone tails) and thus reorganize nucleosomes through the function of recruited ATP-dependent remodellers. The activators then recruit the mediator complex that generically consists of the core mediator (MED) subunits and the kinase module (cyclin-dependent kinase 8 (CDK8)). In turn, the mediator complex recruits the pre-initiation complex (PIC), which consists of the basal transcription factors TFIIA, TFIIB, TFIID, TFIIE, TFIIF, TFIIH and RNA Pol II (this form of RNA Pol II is known as RNA Pol IIA). The subsequent initiation of transcription is associated with the restructuring of the mediator complex, including the loss of the kinase module. Following RNA Pol II promoter clearance transcription can proceed to the elongation phase. This is associated with the binding of elongation factors such as DRB sensitivity-inducing factor (DSIF), negative elongation factor (NELF) and positive transcription elongation factor (PTEFb). At this time, the RPB1 (a component of RNA Pol II) carboxy-terminal domain (CTD) is phosphorylated at Ser2 and Ser5 by TFIIH and PTEFb (this form of RNA Pol II is known as RNA Pol II0). Alternatively, RNA Pol II may be subject to promoter-proximal pausing (at approximately +50 nucleotides from the transcription start site). Paused RNA Pol II is associated with DSIF but phosphorylated only at Ser5. In response to stimuli such as growth factor signalling or developmental cues paused RNA Pol II can be converted into a productive elongation complex, and this is assisted in some cases by the binding of the mediator complex64. There are many druggable interfaces in the RNA Pol II complex. The binding of TFs to REs proximal to gene promoters is facilitated by proteinprotein interactions (1,4), ligand binding (2) or post-translational modifications (PTMs) (3) that alter the affinity of a TF for its respective RE. Thus, TFDNA binding can be prevented by inhibiting these affinity-altering events or can be directly prevented at the TFDNA interface (5). Activation domain (AD) replacement (6) represents a strategy to exogenously activate transcription. Transcription can also be targeted at the level of CF recruitment (7) and CF activity (8). Recruitment of the mediator complex (9), which facilitates transcriptional activation by its associated TFs (10), could also be targeted. Mediator recruitment of RNA Pol II (11) also represents a druggable interface. The activity of PTEFb (that is, CDK9) (12) and the ability of interacting chromatin readers (for example, bromodomain-containing protein 4 (BRD4)) to interact with specific histone or DNA marks (not shown) can also be targeted. MP, masking protein.


VOLUME 13 | MAY 2013 | 305

REVIEWSin the kinase module of the mediator complex (FIG.2), is recurrently amplified in colorectal cancer69. Furthermore, overexpression of CDK8 alone transforms mouse 3T3 fibroblasts, and additional findings suggest that this putative oncogene may exert its transformative effects partly by facilitating enhanced catenin transcriptional activation. However, a dominant-negative inhibitor of catenin failed to completely inhibit CDK8driven transformation, indicating a role for CDK8 in oncogenic transformation beyond catenin regulation70. The mediator complex was originally identified as a transcriptional co-activator bound to nuclear hormone receptors and it is essential for their function as transcriptional activators71. It was subsequently shown that the binding of nuclear hormone receptors to the mediator complex was mediated via an interaction with MED1 (REFS7278). Interestingly, MED1 has been increasingly implicated in both nuclear hormone receptor-responsive and nuclear hormone receptor-resistant cancers. For example, MED1 is overexpressed in both androgen receptor (AR)-positive and ARnegative prostate cancer cell lines, and in a high proportion of clinically localized human prostate cancers79. Moreover, reduced expression of MED1 via RNAi not only decreased transcription of androgen-responsivegenes, such as kallikrein-related peptidase 3 (KLK3; which encodes prostate-specific antigen (PSA)), but also had both androgen-dependent and androgen-independent effects on proliferation and apoptosis in prostate cancer cell lines79. Similarly, MED1 is also recurrently amplified in oestrogen receptor (ER)-positive breast and ovarian cancer cells68,80, where it functions as a node of crosstalk for the human epidermal growth factor receptor 2 (HER2; also known as ERBB2) and ER pathways in the development of HER2mediated tamoxifen resistance. Specifically, MED1 is activated by phosphorylation in a HER2dependent manner. Activated MED1 is recruited to the ER target gene promoters by tamoxifen in HER2overexpressing cells and is required for their expression. RNAi-mediated attenuation of MED1 expression sensitizes HER2overexpressing cells to tamoxifen treatment. Finally, increased levels of MED1 in tumours significantly correlate with tamoxifen resistance in patients with ERpositive breast cancer treated with adjuvant tamoxifen monotherapy 81,82. Conversely, loss of MED1 promotes the invasion and metastasis of human non-small-cell lung cancer cells by inducing the expression of metastasis-associated genes, through as yet undefined mechanisms83. Several other mediator complex components are overexpressed or amplified in cancer 8488 (TABLE1). In particular, MED29 seems to have paradoxical roles in cancer, as it is recurrently amplied and overexpressed in pancreatic cancer 89,90, whereas its overexpression has been shown to reduce the growth of pancreatic cancer cells in xenografts, and this correlated with the differential expression of genes involved in cell cycle control91. The exact molecular mechanisms involved are not yet clear but MED29 possibly serves as a cofactor for specific transcriptional activators of a cell cycle gene expression programme91. Thus, MED29 is likely to be an important regulator of key cellular functions in pancreaticcancer,306 | MAY 2013 | VOLUME 13 2013 Macmillan Publishers Limited. All rights reserved

and, like some other mediator sub units (such as MED1), it has dual oncogenic and tumour suppressive characteristics. These characteristics possibly depend on the genetic background of the cancer cells and their surrounding environment or on the transcription factors that are important to the cell or tissue type in which the tumour occurs91. At least in some cases, it seems that the overexpression of mediator components can interfere with the transcriptional activators with which they normally interact. This probably occurs because the excess free mediator subunits interact with the activation domain of transcription factors, which blocks further interaction with the complete mediator complex 92. Intriguingly, the mediator complex does not seem to interact with RNA Pol I or RNA Pol III, probably because the non-coding transcripts produced by these RNA Pols generally have ubiquitous cellular functions and thus a more limited set of transcriptional modulators is sufficient to integrate cellular signalling to control their activity. However, both RNA Pol I and RNA Pol III are still subject to modulation by a variety of corepressors and coactivators, some shared with RNA Pol II, others unique. Forexample, the RNA Pol I transcription initiation factor RRN3 (also known as TIF-1A) is essential for recruiting RNA Pol I to the PIC (FIG.1). RRN3 binds RNA Pol I in a position resembling the location of mediator complex binding on RNA Pol II9395, suggesting a topological conservation between the RNA Pols of regulatorbinding sites15. RRN3 is thought to be one of the major factors that integrates nutrient, energy and growth factor signalling and cellular stress with RNA Pol I transcription initiation and is highly conserved from yeast to humans. Although no tumour-specific mutations, amplifications or rearrangements of RRN3 have been reported, its activity is modulated by oncogenic signalling pathways including PI3KmTOR and RASRAFERK9699. RRN3 is also a direct transcriptional target of MYC43 and it is over expressed in MYC-driven tumours, which correlates with the hyperactivation of RNA Pol I transcription19 (BOX1). Moreover, even a modest knockdown of RRN3 expression leads to the rapid apoptosis of MYC-driven lymphoma cells19. This suggests that MYC-driven tumour cells are more dependent on optimal RNA Pol I transcription rates compared with non-transformed cells19. However, whether the increased activity of RRN3 and RNA PolI transcription is sufficient to drive malignancy in the absence of other oncogenic lesions in RNA Pol II or RNA Pol III has not been demonstrated. An example of a mediator-like protein that regulates RNA Pol III transcription is the transcriptional corepressor MAF1, which has a central role in directly repressing RNA Pol III that is conserved from yeast to humans100. Specifically, in response to nutrient deprivation and various forms of stress, yeast MAF1 is dephosphorylated, which leads to nuclear accumulation, increased association with RNA Pol III-transcribed genes and direct physical interactions with RNA Pol III. MAF1 association results in the inhibition of both denovo assembly of TFIIIB by promoter-bound TFIIIC and RNA PolIII-dependent transcription from pre-assembled TFIIIBDNA complexes101. MAF1 binding at RNA Pol IIIwww.nature.com/reviews/cancer

REVIEWSpromoters is thought to be repressed by phosphorylation by mTOR complex 1 (TORC1), suggesting that the repressive effects of MAF1 on RNA Pol III would be frequently derepressed incancers, as mTORC1 and upstream oncogenic signalling pathways such as PI3K are frequently hyperactivated in tumour cells. Mammalian MAF1 also represses a limited subset of RNA Pol II genes, including TBP. Thus, in mammals, MAF1 also indirectly suppresses RNA Pol I. However, so far, there are no reports of genetic lesions leading to loss of function of MAF1 in cancer 102. Nonetheless, overexpression of MAF1 reduced anchorage-independent growth in U87 cells, indicating a possible role of this protein in suppressing cellular transformation100 although whether this is mediated by RNA Pol I, RNA Pol II or RNA Pol III is not clear. the accurate termination of transcription elongation. The SECs, which contain PTEFb, the elongation factors ELL, ELL2 and ELL3 and a dynamic cohort of additional sub units, are recruited to many rapidly transcribed genes in response to differentiation signals to release paused RNA Pol II for the dynamic induction of transcription 66. RNA Pol II elongation through chromatin is also facilitated by histone chaperones, such as the facilitates chromatin transcription (FACT) complex 115, which aid RNA Pol II passage by destabilizing nucleosomes through the transient release of histone H2AH2B dimers116,117. Components of the SECs were originally identified through investigation into the translocation partners of the gene encoding mixed lineage leukaemia (MLL)118, and MLL translocations frequently occur in human acute leukaemias of myeloid and lymphoid lineages119. The first of these translocation partners, ELL, was found to stimulate elongation by RNA Pol II invitro. Following this, AFF1, AFF4, AF9 and ENL were all identified as members of the SEC, in addition to being common translocation partners of MLL in leukaemia119. The recruitment of SECs to the elongating complex is mostly mediated via the interaction of SECs with PAF1, although recent proteomic studies have identified MED26 as a docking site for SECs. MED26 may potentially function as a molecular switch by interacting first with TFIID during transcription initiation and then with SECs during the transition into early elongation65. The fusion of these SEC components with the DNA-binding domain of MLL negates the need for the interaction of MLL with PAF1 and the subsequent recruitment of the SEC to MLL target genes120. In this way, the expression of MLL chimaeras facilitates the enhanced transcriptional elongation of MLL target genes (for example, classI homeobox (HOX) genes, which are transcription factors that are involved in embryonic development and haematopoietic cell differentiation120) and thus the runaway transcription of genes that are normally under tight control during lymphocyte development. Consistent with this, the developmentally regulated HOX gene cluster has been found to be most frequently upregulated in the presence of MLL translocations, and the altered expression of the HOX genes was shown to be required for leukaemogenesis118. There is also emerging evidence for the contribution of non-SEC components of the RNA Pol II elongation apparatus, specifically FACT and TFIIS, to malignant transformation. In mammals, three different forms of TFIIS have been identified: transcription elongation factor A1 (TCEA1), TCEA2 and TCEA3 (REFS121,122). TCEA1 is consistently overexpressed in a variety of human tumours 123; and TCEA2 resides in a region of chromosome 20 that is frequently amplified in cervical cancer, with TCEA2 mRNA overexpressed in these tumours124. Moreover, knockdown of TFIIS has been shown to inhibit cancer cell prolif

Nature Reviews Cancer

Documents

Transcript of Nature Reviews Cancer