GOLPH3 links the Golgi network to mTOR signaling and human cancer

Cancer genomes are now being heavily scrutinized with an increasingly sophisticated battery of technologies that allow rapid determinations of candidate oncogene or tumor suppressor gene identity, expression, and function. Comparisons of the genetic alterations observed in cohorts of patients bearing the same cancer subtype, and cross-cohort comparisons of patients with different tumors, allow sorting of genome copy number variations (CNVs) into two conceptual bins. The first bin contains the innumerable array of genetic alterations that accumulate in tumors over many generations of cell division in the setting of random (and rampant) genetic instability. Nested within all of this genetic ‘noise,’ however, are the variable elements that underlie the significant patient to patient variability observed even histologically similar tumors originating in same tissue type. The second bin of genomic changes has attracted much more attention from cancer cell biologists and pharmaceutical companies alike. Here, reside the genetic alterations that are observed repeatedly in human cancers, either within or across different disease subtypes. The interest lavished on this group of cancer genes is well deserved, because recurrent alterations in specific chromosomal regions indicate that these regions are home to genes whose protein products serve as fundamental, broadly active drivers of carcinogenesis. A logical extension of this line of reasoning is that the pathways regulated by these gene products are strong candidates for lineage dependencies (also termed ‘addictions’) in significant subsets of cancer patients, and hence, contain particularly tantalizing targets for the development of molecularly targeted anticancer agents. A recent publication in Nature highlights the power of integrative genomics, combined with functional screening of individual candidate genes, to identify novel oncogenes and tumor suppressor genes (Scott et al., 2009). Indeed, this report goes one step further, implicating an intracellular organelle that has largely escaped notice as a possible collaborator in the oncogenic process. The authors identified a recurrent amplification site in chromosomal region 5p13, and used a combination of gene overexpression and silencing approaches to identify GOLPH3, a component of the Golgi matrix, as a novel protooncogene. Surprisingly, GOLPH3 overexpression constitutively activates mammalian target of Rapamycin (mTOR) signaling, and, in the in vivo setting, confers lineage dependence on mTOR complex 1 (mTORC1) signaling for progressive tumor growth. Scott et al. (2009) performed array-based comparative genome hybridization analyses (array-CGH) on an initial cohort of melanoma tissues to identify the recurrent 5p13 amplification. These studies were extended to other tumor types, and 5p13 gain was observed in a diverse set of solid tumors, including colorectal cancer and non-small cell lung cancer (NSCLC). The 5p13 amplification occurred with frequencies of 24–56% in the solid tumor cohorts examined by Scott et al. (2009); interestingly, 5p13 gain was scored in only 8% of multiple myeloma samples, suggesting that this chromosomal abnormality might be selected for more heavily in epithelial cancers than in hematopoietic tumors. Additional work is clearly needed to understand the frequency of 5p13 gain in various blood-borne neoplasms; nonetheless, it is striking that 5p13 gain frequency is rather low in multiple myeloma cells, which bear an abnormally active Golgi network. To identify candidate oncogenes in the 5p13 region, the authors examined the expression patterns of the four genes residing in this region, and found that only two of the genes, GOLPH3 and SUB1, were correlatively expressed at the gene copy number and transcript levels. Finally, gene silencing by RNA interference in cancer cell lines bearing the 5p13 amplicon pinpointed the GOLPH3 gene as an overexpression-dependent driver of the transformed phenotype in these cells. In contrast, knockdown of GOLPH3 in a melanoma cell line that neither bears the 5p13 amplicon nor overexpresses the GOLPH3 protein had only a minimal effect on the cancerous behavior of these cells. Collectively, these findings suggest that overexpression of GOLPH3 results in a gain of function that promotes cell transformation. Pertinent to melanoma pathogenesis, Scott et al. (2009) observed that forced expression of GOLPH3 effectively cooperated with mutationally activated B-RAF (BRAFV600E) to transform immortalized primary human melanocytes (Figure 1). Transformation of melanocytes by GOLPH3 overexpression. GOLPH3 dynamically moves between the trans-Golgi network and endosomal structures, graphically depicted as internalizing an activated growth factor receptor dimer. Overexpressed GOLPH3 may deliver its transforming signals through altered growth factor receptor internalization and turnover, and/or through changes in protein modification by glycosyltransferases (GlyTs). GOLPH3 ultimately stimulates oncogenic signaling by constitutively activating both mTOR complex 1 and 2 (mTORC1 and mTORC2). Deregulated mTOR signaling cooperates with mutationally activated BRAFV600E to fully transform human melanocytes. GOLPH3 (also known as GMx33) is a 33 kDa protein of unknown function that shuttles dynamically from the trans-Golgi network (TGN) to endosomal and plasma membranes (Snyder et al., 2006). This protein contains three putative coiled regions and appears to be phosphorylated in a regulated fashion, although the significance of this post-translational modification is unclear. Intriguingly, Scott et al. (2009) discovered that GOLPH3 associates with Vps35, a component of the retromer complex, which is responsible for retrograde transport of certain cell surface receptors and other cargo proteins from endosomes to the TGN (Bonifacino and Hurley, 2008). Furthermore, deletion of VPS35 in budding yeast leads to rapamycin hypersensitivity, consistent with an impairment of TORC1 signaling (Xie et al., 2005). Scott et al. (2009) therefore predicted that, as a Vps35-associated protein in mammalian cells, GOLPH3 might also be linked to the regulation of mTOR signaling. Sure enough, the authors found that GOLPH3 overexpression was correlated with hyperactivation of mTORC2, as well as mTORC1 signaling, in human cells. Furthermore, in xenograft experiments in immunodeficient mice, GOLPH3-overexpressing tumor cells showed increased sensitivity to therapy with the mTORC1 inhibitor, rapamycin. The latter studies suggest that GOLPH3-dependent oncogenesis is associated with lineage dependency on mTOR signaling, and, in turn, sensitivity to mTOR inhibitors in these preclinical models. The report by Scott et al. (2009) underscores the power of integrated genetic, transcriptomic, and functional screening strategies as an engine for the discovery of novel gene and network connections in human cancer biology. Groundbreaking insights such as these nearly always raise new questions to be addressed in future studies. At the top of the list of unresolved issues is a clear perspective on the physiological functions of GOLPH3 in mammalian cells. Recent findings suggest that GOLPH3 might be involved in the regulation of protein glycosylation, which is commonly aberrant in cancer cells (Schmitz et al., 2008; Tu et al., 2008). A logical follow-up to these efforts will be a detailed understanding of the mechanism whereby overexpressed GOLPH3 activates both mTOR complexes in mammalian cells. Enhanced coupling of mTORC1 and mTORC2 to the PI3K-AKT signaling axis is one possibility that should be addressed. Finally, several pharmaceutical companies are developing rapamycin analogs and second generation mTOR kinase inhibitors as anticancer agents, and these clinical trials have generated an intense search for biomarkers that predict tumor sensitivity to mTOR inhibitor-based therapies. If 5p13 gain and/or GOLPH3 overexpression allows patient stratification into drug responsive and non-responsive populations, then the report by Scott et al. (2009) could benefit a large number of patients with melanoma and other life-threatening cancers.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.