Control of Oncogenic Signaling Through Spatial Organization of Kinases and mRNAs

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

Our research is divided into two main subprojects. The first focuses on how 3'UTRs of mRNAs encoding proteins that are targets of oncogenic kinases such as ERK and CK2 regulate mRNA localization in tumor cells. This subcellular mRNA compartmentalization can either inhibit or stimulate phosphorylation of the encoded protein. Secondly, we are studying how kinases partition to the perinuclear cytoplasm in tumor cells to form perinuclear signaling centers, or PSCs. Our work indicates that PSCs serve as critical signaling engines that drive malignant transformation and cancer. 1.3'UTR regulation of protein activity (UPA) and its underlying mechanisms. 3'UTR regulation of protein activity was discovered through our earlier observation that the CEBPB 3'UTR inhibits RAS-induced post-translational activation of C/EBPb in tumor cells, where the 3'UTR effectively uncouples C/EBPb from RAS signaling. UPA thus constrains the pro-senescence activity of C/EBPb and its ability to induce pro-inflammatory senescence-associated secretory phenotype (SASP) genes. UPA requires a long G/U-rich element (GRE) motif in the 3'UTR and its cognate binding protein, HuR/ELAV1. These components exclude CEBPB mRNAs from a perinuclear cytoplasmic compartment containing endosome-associated ERK1/2 and CK2, thus preventing newly translated C/EBPb from accessing its activating kinases. More recently, we found that the RNA decay proteins UPF1 and Staufen (STAU1/2) are essential UPA factors enriched within the perinuclear cytoplasm. Together with HuR and the nonsense-mediated decay factors SMG6 and 7, these proteins promote perinuclear mRNA decay (PMD) of CEBPB transcripts. Depletion of UPF1 or STAU in tumor cells increased the nuclear-proximal population of CEBPB transcripts, leading to C/EBPb activation and senescence. High resolution imaging showed that these perinuclear CEBPB mRNAs frequently colocalize with CK2 foci. These observations suggest that when UPA is disabled, C/EBPb undergoes co-translational phosphorylation which is contingent on close proximity of CEBPB transcripts with activating kinases. We identified a STAU binding site (SBS) adjacent to the GRE which, when deleted, activated the pro-senescence functions of C/EBPb but not its ability to induce SASP genes. This observation and other data imply that distinct 3'UTR sequences are involved in repressing the two C/EBPb functions, most likely by inhibiting different PTMs. We propose that various mRNA decay factors (e.g., STAU) which recognize discrete 3'UTR sequences are tethered to different types of signaling endosomes (e.g., those carrying ERK, CK2 or other kinases). Hence, individual 3'UTR elements may promote mRNA decay in the vicinity of particular kinases and thereby inhibit protein phosphorylation on specific sites. Future work will expand upon this novel relationship between modular 3'UTR motifs, mRNA decay near perinuclear kinases and inhibition of specific PTMs on the encoded protein. To examine the in vivo relevance of UPA, we generated mice with a deletion that removes the Cebpb GRE and part of the adjacent SBS. This strain was tested in a Kras model of lung tumorigenesis. Although overall lung tumor burdens in deltaGRE mice were similar to WT animals, regions of malignant adenocarcinoma were significantly reduced. These findings provide the first in vivo evidence that UPA constrains C/EBPb activity and thus facilitates tumor progression to carcinomatous lesions. We are currently performing senescent cell analysis and RNA-seq studies on tumors of different stages in the two genotypes to assess whether the delGRE mutation increases senescence and how this allele alters the C/EBPb transcriptome. A key goal of our work is to determine whether UPA is a general mechanism that regulates many other proteins. p53 is one such candidate, as it undergoes activation by oncogenic RAS in OIS cells and its 3'UTR harbors a U-rich element (URE) that binds HuR. We found that the TP53 3'UTR suppresses the cytostatic activity of a p53 transgene expressed in p53-deficient RAS tumor cells, without affecting p53 protein levels. Like C/EBPb, the 3'UTR excludes TP53 mRNA from the kinase-rich perinuclear region and appears to inhibit p53 phosphorylation on an activating CK2 site, Ser392. RNA FISH showed that endogenous TP53 transcripts also partition away from the nuclear-proximal area in tumor cells but are perinuclear in senescent fibroblasts. Similarly, TP53 mRNA undergoes perinuclear translocation in cancer cells treated with chemotherapeutic DNA damaging agents (which induce p53-dependent senescence or apoptosis), correlating with increased Ser392 phosphorylation. Depletion of the p53 ubiquitin ligase, MDM2, also caused p53 transcripts to become perinuclear, indicating that MDM2 restrains p53 activity both directly, through p53 degradation, and indirectly by enforcing UPA. Our findings demonstrate that 3'UTR-dependent changes in mRNA localization control p53 function. 2. Mechanisms and function of RAS pathway perinuclear signaling centers (PSCs) as key signaling engines in cancer cells. Oncogenic RAS induces perinuclear translocation of the effector kinases p-ERK and CK2 and the signaling scaffold KSR1. These proteins form signaling hubs on endosomes tethered to the ER network, termed "perinuclear signaling centers". PSCs are critical to the UPA mechanism and are widely present in cancer cell lines and tissues. We propose that PSCs are also key signaling engines that drive cancer, allowing oncogenic kinases to access targets that are important for neoplastic transformation. We found that the endosomal adaptor TOLLIP is required for perinuclear localization of RAB11A+ endosomes harboring CK2 and KSR1, but not ERK, which resides on a different class of signaling endosome. TOLLIP is perinuclear in human cancer cells and KRasG12D-driven mouse tumors but is pan-cytoplasmic in non-transformed cells, coinciding with the presence of PSCs. TOLLIP associates with KSR1 through a conserved region that binds to the KSR1 pseudo-kinase domain. This interaction recruits CK2 signaling complexes to endosomes. We performed a series of phosphoproteomics experiments which show that perinuclear CK2 phosphorylates selective substrates, including proteins involved in ribosome biogenesis and translation. One such target is the atypical kinase RIOK1, which regulates 18S rRNA processing and 40S subunit maturation. Mutant analysis suggests that phosphorylation on Ser22 by perinuclear CK2 is essential for RIOK1 activity in tumor cells. We observed that KRasG12D-driven lung tumors in Tollip-/- mice progress less efficiently to the malignant adenocarcinoma stage. Furthermore, tumor cell lines carrying KRAS mutations but not HRAS or BRAF lesions require TOLLIP for proliferation/survival. TOLLIP is therefore a key signaling adaptor in KRAS tumor cells whose inhibition is a potential vulnerability of these cancers. TOLLIP contains a phospholipid binding domain (C2) that likely determines its association with particular classes of endosomes. in collaboration with Dr. Nadya Tarasova (CIL) to identify small molecules that are projected to dock in predicted TOLLIP pockets. Compounds verified to bind TOLLIP and that disrupt PSCs in tumor cells may be considered for further development as potential antitumor agents. As cancer cells driven by mutant HRAS, BRAF and other oncogenes display reduced dependence on TOLLIP but nevertheless exhibit perinuclear CK2, we believe that an additional adaptor protein(s) provides a redundant perinuclear tetheri *TRUNCATED*