Role of CEBP transcription factors in cell growth and tumorigenesis

Division Of Basic Sciences - Nci

Investigators

Linked publications, trials & patents

Abstract

Our past and ongoing research has focused on the transcription factor C/EBPbeta as a model for elucidating mechanisms by which RAS signaling activates its downstream targets. Studies by our lab and others demonstrated that C/EBPb has oncogenic functions and is essential for the development of many cancers. However, in primary cells such as mouse embryo fibroblasts (MEFs), C/EBPb participates in oncogene-induced senescence (OIS). OIS is activated by oncogenic stresses and serves as an intrinsic barricade to tumor development. C/EBPb is an auto-inhibited protein whose activity can be stimulated by oncogenic RAS signaling through the MEK-ERK cascade. C/EBPb de-repression involves multiple post-translational modifications (PTMs), including phosphorylation by the RAS effector kinases ERK1/2 and CK2. Activated C/EBPb is essential for cell cycle arrest in senescent cells and also activates transcription of senescence-associated secretory phenotype (SASP) genes, which include pro-inflammatory cytokines/chemokines. An important advance from our lab was the discovery that the 3' untranslated region (3'UTR) of the Cebpb mRNA inhibits C/EBPb activity in tumor cells. This novel mechanism, termed 3'UTR regulation of protein activity (UPA), suppresses the DNA-binding, transcriptional and senescence-inducing activities of C/EBPb. UPA requires a 3'UTR sequence containing several G/U rich elements (GREs) and the ARE/GRE-binding protein, HuR. This system acts by restricting Cebpb mRNAs to the peripheral cytoplasm, excluding them from a perinuclear region where ERK1/2, CK2, and other kinases reside. C/EBPb synthesized in the cell periphery is uncoupled from RAS signaling due to its inaccessibility to activating kinases. By contrast, 3'UTR inhibition and Cebpb mRNA compartmentalization are absent in senescent primary mouse and human cells. Instead, C/EBPb becomes activated in response to RAS signaling and promotes senescence. Thus, UPA acts as a switch to control C/EBPb activation in the appropriate contexts. We have used a proteomics approach to identify additional proteins that interact with the GRE and may contribute to the Cebpb UPA mechanism. One candidate, Upf1, is an RNA helicase that is a key component of nonsense-mediated mRNA decay (NMD). NMD eliminates faulty transcripts that contain premature stop codons. In tumor cells, Upf1 localizes to the perinuclear cytoplasm and is negatively correlated with the presence of Cebpb transcripts. Depletion of Upf1 in human lung cancer cells increased Cebpb mRNA abundance in the perinuclear region and stimulated C/EBPb phosphorylation and DNA binding. The cells also became senescent and induced SASP genes in a C/EBPb-dependent manner. Another GRE-enriched factor, the RNA-binding protein staufen (Stau), is known to regulate mRNA localization and decay and acts together with Upf1 to degrade target mRNAs. Stau interacts with a staufen binding site (SBS) in the Cebpb 3'UTR, adjacent to the GRE. Stau is perinuclear in tumor cells and, accordingly, silencing of Stau1 or Stau2 abrogated perinuclear Cebpb mRNA decay, causing C/EBPb activation and senescence. Thus, Upf1 and Stau, together with HuR, direct perinuclear mRNA decay (PMD) of Cebpb transcripts to suppress C/EBPb activity and senescence in tumor cells. We are currently investigating whether other pro- and anti-oncogenic transcription factors are regulated by 3'UTR-dependent PMD or related RNA localization mechanisms. To address whether UPA constrains the activity of endogenous C/EBPb in vivo, we generated a mouse mutant with a deletion of the Cebpb GRE region. MEFs from delGRE mice that also carry a mutation in the p19Arf tumor suppressor gene were resistant to transformation by oncogenic RAS and showed enhanced senescence, in contrast to Cebpb+/+;p19-/- cells which underwent transformation and senescence bypass. delGRE mice were further interrogated using a model of Kras-driven lung cancer. delGRE animals had lower overall lung tumor burdens and, importantly, displayed impaired development of malignant adenocarcinomas. These findings confirm the key role of GRE-dependent UPA in limiting C/EBPb activity in cancer cells, allowing them to circumvent senescence and progress to advanced tumors. A key finding from our studies of UPA is that oncogenic kinases such as p-ERK and CK2 localize to the perinuclear cytoplasm in tumor cells. These proteins form perinuclear signaling centers (PSCs) in cells expressing oncogenic RAS or BRAF. In fact, our findings suggest that PSCs are present in nearly all cancer cells, regardless of the driving oncogene. We propose that perinuclear kinases access key substrates whose phosphorylation promotes neoplastic transformation and tumorigenesis. PSCs are associated with perinuclear endosomes and require the MAPK signaling scaffold, KSR1, which binds to ERK and CK2. KSR1 similarly undergoes RAS-induced perinuclear translocation and colocalizes with p-ERK and CK2. PSC-containing endosomes are embedded within the perinuclear endoplasmic reticulum (ER) network. To further investigate the mechanisms underlying PSC formation in tumor cells, we have explored the roles of endosomal adaptor proteins known to mediate tethering of endosomes to the ER. One such adapter, Tollip, is essential for the formation of CK2 PSCs but not those involving ERK. Tollip tethers CK2 signaling complexes to endosomes via KSR1, and Tollip and KSR1 interact through conserved domains in each protein. Notably, Tollip is required for transformation by oncogenic KRAS but not HRAS or NRAS. Therefore, HRAS and NRAS may use a distinct endosomal adapter(s) to anchor PSCs to the ER compartment. Mice lacking Tollip do not exhibit overt defects. However, we observed that Kras-induced lung tumors in Tollip-/- mice are nearly completely blocked for progression to late-stage malignant carcinomas. Unlike KRAS tumor cells, growth/survival of normal cells is independent of Tollip. The selective Tollip dependency of KRAS mutant tumors - a particularly lethal class of cancers - underscores the potential appeal of developing anti-cancer drugs directed at Tollip and other PSC components. Such drugs may be better tolerated and have a wider therapeutic window than inhibitors of RAS pathway kinases. PSCs are transiently induced by serum growth factors in normal cells with delayed kinetics (4-6 hr after GF stimulation). We postulate that RAS and other oncogenes persistently activate this late phase of GF signaling in tumor cells, constitutively localizing ERK and CK2 to the perinuclear compartment. To identify selective substrates of PSC kinases, we performed phosphoproteomics over a time course in GF-stimulated cells. One candidate from this study is the atypical kinase, RIOK1, which plays a key role in maturation of the 40S ribosomal subunit. RIOK1 was phosphorylated on a CK2-like site (Ser21/22) with kinetics corresponding to perinuclear translocation of CK2. KRAS tumor cells are strongly dependent on RIOK1 for growth and survival, and a phospho-deficient RIOK1 mutant failed to rescue RIOK1 silencing in a lung tumor cell line. These observations indicate that RIOK1 phosphorylation by perinuclear CK2 controls its oncogenic activity. We are continuing to investigate RIOK1 and other PSC substrate proteins to understand their roles in tumorigenesis and how they are specifically selected for modification by PSC kinases.