UNRelenting Translation UNRestrains Melanoma Migration

Abstract

The cytoplasmic RNA-binding protein UNR influences key developmental processes by controlling mRNA turnover and translation initiation. In this issue of Cancer Cell, Wurth et al. report that UNR is highly expressed in melanoma and enhances invasion and metastasis at least partly by inducing translation elongation of VIM and RAC1 mRNAs.

RNA-binding proteins (RBPs) regulate every post-transcriptional facet of gene expression, including pre-mRNA splicing and maturation and mRNA transport, storage, turnover, modification, and translation (Glisovic et al., 2008). Specialized RBPs with affinity for distinct RNA sequences bind subsets of mRNAs and selectively influence their post-transcriptional fate, thus coordinating the expression of the proteins they encode. In this manner, RBPs are central regulators of cellular processes (e.g., proliferation, senescence, apoptosis, immune activation, and the stress response), as well as tissue homeostasis and diseases like neurodegeneration, diabetes, inflammation, and cancer (Yeo, 2014; Neelamraju et al., 2015). Well-studied RBPs with sequence preferences include HuR, HuD, AUF1, CUGBP1, PTB, TTP, LIN28, CPEB, FMRP, and PUM1.

Wurth and colleagues (2016) found that one such sequence-specific RBP, UNR (upstream of N-RAS, also named CSDE1), was highly expressed in melanoma. Interventions to alter UNR levels in human melanoma cells and mice revealed that UNR promoted melanoma migration and metastasis. Five-year survival rates for patients with metastatic melanoma are 15%–20%, despite advances in melanoma therapy, and there is therefore a critical need to understand the molecular mechanisms that underlie metastasis.

Like most RBPs, UNR was previously found to influence the expression of target mRNAs at several post-transcriptional levels (Ray et al., 2015). For example, it promoted decay of some targets (e.g., FOS and GATA3 mRNAs) but stabilized others (e.g., PTH mRNA), while it enhanced cap-independent translation initiation of some targets (MYC, CDK11A, and APAF1 mRNAs) but suppressed it for another target (UNR mRNA ifself). UNR elicited these actions through binding different segments of target mRNAs. In some cases it bound the 5′ untranslated region (UTR, often at internal ribosome entry sites [IRESs]), other times the coding region, and others the 3′ UTR.

Given the broad spectrum of UNR actions and binding patterns, Wurth and coworkers set out to identify comprehensively the molecular mediators of UNR-elicited melanomagenesis by adopting a systematic three-pronged approach: (1) iCLIP (individual nucleotide-resolution crosslinking immunoprecipitation) analysis to identify all of UNR target RNAs and sites of binding, (2) RNA-seq analysis to detect changes in mRNA steady-state levels after silencing UNR, and (3) ribosome profiling to identify changes in relative association of ribosomes and hence translational regulation.

These analyses revealed several surprising findings. First, UNR bound preferentially single-stranded RNAs in the coding region and 3′ UTR, suggesting functions beyond UNR interaction with 5′ UTR IRESs; these sites were found preferentially around start and stop codons. Second, while UNR promoted the accumulation of many target mRNAs, it selectively elevated mRNAs encoding tumor-promoting factors and lowered mRNAs encoding tumor-suppressing factors, globally implementing a pro-tumor gene expression program. Third, ribosome profiling analysis corroborated earlier findings that UNR influenced translation initiation. Importantly, however, the authors discovered that reducing UNR levels did not change the overall association of many mRNAs with ribosomes; instead, the relative distribution of ribosomes shifted upstream of the mRNA, consistent with a role for UNR in promoting translation elongation and/or termination.

The hypothesis that UNR controls translational elongation was tested on several direct UNR targets, many of which (PTEN, SDC4, TRIO, TNC, FN1, RAC1, VIM, and CTTN mRNAs) encode proteins with roles in melanoma progression. In this manner, UNR can regulate their coordinated production, according to the ribonomic model proposed by Keene and Tenenbaum (2002). The products of these mRNAs in turn ensure that melanoma cells proliferate, escape senescence, avoid apoptosis and anoikis, and migrate to and colonize other tissues. The authors analyzed RAC1 and VIM mRNAs in detail. RAC1 mRNA encodes a GTPase in the RAS superfamily with functions in cell signaling, proliferation, and cytoskeletal architecture. RAC1 promotes melanoma metastasis by promoting invadopodia formation (Revach et al., 2016). VIM mRNA encodes VIM (vimentin), a component of the intermediate filaments that protect melanoma cells as they change morphology, and key marker of epithelial-mesenchymal transition (EMT) (Vuoriluoto et al., 2011). Melanoma cells are not classical epithelial cells but undergo a mesenchymal-like transition and upregulate VIM. UNR silencing did not change VIM or RAC1 mRNA levels, the numbers of ribosomes assembled on these mRNAs, or the initiation of VIM or RAC1 translation. Instead, the ribosomes bound to VIM and RAC1 mRNAs shifted their distribution upstream of the coding region, consistent with diminished elongation of translating ribosomes. Accordingly, VIM and RAC1 protein levels declined markedly after UNR silencing. Together, these results indicated that UNR promoted VIM and RAC1 translation elongation and/or termination, ultimately contributing to EMT and increased invasion. In light of recent data implicating the EMT in drug resistance (Fischer et al., 2015), these results also raise the intriguing possibility that UNR might contribute to drug resistance.

According to an emerging pattern, certain RBPs are beneficial in stressful environments such as cancer. Their ability to control gene expression post-transcriptionally might be advantageous to cancer cells because stresses in the tumor milieu, such as hypoxia, hyperglycemia, oxidative damage, and therapy, can shut off gene transcription. Thus, if the cell can deploy mechanisms that operate with already-transcribed mRNAs, protein expression patterns can still be regulated in the absence of de novo transcription. An RBP such as UNR that can affect mRNA turnover, modulate translation initiation, and/or promote translation elongation can help implement an adaptive gene expression program (Figure 1). Thus, when transcription is halted, post-transcriptional control by RBPs could still enable cancer cell survival, proliferation, metastasis, and therapy resistance. This novel function of UNR adds to only a handful of reported examples of sequence-specific RBPs affecting translation elongation and/or termination: FMRP, HNRNPE1, and the PUF-AGO complex. PUF-AGO and HNRNPE1 elicit this repression by impairing the activity of eukaryotic elongation factor eEF1A, but how other RBPs, including UNR, modulate translation elongation is unclear. Elucidating the molecular details of this function will require further investigation.

Figure 1. Schematic of the Function of RNA-Binding Protein UNR in Melanoma Progression.

Melanoma arises from melanocytes within the epidermis. After transformation due to the activation of oncogenic pathways largely following ultraviolet irradiation, melanocytes become melanoma cells that grow rapidly, acquire traits of mesenchymal cells, and invade and metastasize other tissues. Wurth et al. have found that elevated expression of the RNA-binding protein UNR in melanoma contributes to this process. UNR was previously shown to affect the initiation of translation of different target mRNAs (often via an internal ribosome entry site, IRES), as well as the turnover of other target mRNAs. In melanoma cells, Wurth et al. found that UNR additionally promoted translation elongation of several target mRNAs, including VIM and RAC1 mRNAs, in turn raising the production of VIM and RAC1, two proteins that induce a mesenchymal transition and hence drive invasion and metastasis of melanoma cells.

In closing, these studies underscore the potential value of developing interventions that control UNR function. As reviewed (Anderson and Catnaigh, 2015), UNR abundance is transcriptionally repressed by MYC and MAX. At the post-transcriptional level, alternative splicing of UNR pre-mRNA yields multiple isoforms that include different 5′ UTRs, in turn affecting translation via the 5′ UTR IRES. Two RBPs bind competitively to the UNR 5′ UTR: PTB suppressing translation, HNRNPC promoting translation. The use of three alternative polyadenylation signals further diversify UNR mRNAs at the 3′ UTR, thereby dictating whether instability elements are included or excluded. Post-translationally, UNR is modified by phosphorylation and acetylation, although the impact of these modifications is unknown. Although RBPs generally make poorly druggable targets, efforts to suppress UNR production or activity by intervening at these steps may prevent melanoma progression. Because resistance to metabolic stress and to DNA damage are major problems in the treatment of melanoma, and because UNR promotes EMT, it will be important to see whether targeting UNR ameliorates resistance to chemotherapy. The evidence supporting a “master regulator” function for UNR in melanoma cells adapting to the tumor microenvironment and to metastasis paves the way for promising future work on the diagnostic, prognostic, and therapeutic potential of UNR in the challenging management of metastatic melanoma.