In this study, integrative analyses of genome-wide copy number and expression data coupled with reinforcing knockdown and over-expression assays in vitro
and in vivo
led to the identification of GOLPH3 as a bona fide
oncoprotein frequently targeted for copy number gain/amplification in diverse human cancers. A suspect role for the Golgi apparatus in regulating cancer-relevant signaling has been speculated based on observation that some cytoplasmic membrane oncoproteins, such as RAS, can functionally signal when temporally present at the Golgi apparatus24
. However, proteins such as GOLPH3 that are predominantly localized to the TGN have not been directly linked on a genetic level to cancer; therefore, GOLPH3 represents a first-in-class Golgi oncoprotein. Mechanistically, we show that enhanced activation of mTOR signaling represents a molecular basis for GOLPH3’s oncogenic activity. In this light, enhanced and sustained mTOR activation in vivo
would be expected to confer a significant growth advantage to cancer cells, a likely basis for increased GOLPH3 gene copy number or expression in a large fraction of human cancers.
Although we show through genetic and pharmacological studies that activation of mTOR is essential for GOLPH3’s tumorigenicity, the precise structural and biochemical basis for GOLPH3’s activity remains to be fully elucidated. Our molecular data on the physical interaction between GOLPH3 and the retromer complex, which is responsible for protein trafficking between endosomes and the TGN16
, for the first time genetically implicates this biological process in cancer. This is consistent with recent reports on the essential role of the retromer and retrograde transport in regulation of the Wntless receptor and proper secretion of the WNT morphogen25
, which is important in both normal and neoplastic development. Along the same line, depletion of VPS35 in Drosophila inhibited endocytosis of RTKs with concomitant alterations in downstream signaling26
. Taken together, these data raise the possibility that GOLPH3 might function with VPS35 and the retromer to regulate receptor recycling of key molecules thereby influencing downstream signaling through mTOR.
It has recently been discovered that Vps74, the yeast homolog of GOLPH3, is required for proper docking and localization of glycosyltransferases to the Golgi apparatus27;28
. Protein glycosylation is one of the most prevalent forms of post-translational modification, and altered glycosylation is a hallmark feature of cancers29
. It is noteworthy that glycosylation is known to be important for growth factor-activation of transmembrane receptors, since glycosylation mediates receptor sorting, ligand binding and endocytosis29;30
. Thus, it is plausible that human GOLPH3 might serve a similar function in glycosyltransferase docking as in S. cerevisiae
and therefore might influence the downstream mTOR signaling response through its effect on membrane RTKs.
The PI3K-AKT-mTOR signaling cascade is activated in nearly all cancers and hence represents an intense focus for cancer drug development. However, the clinical response to rapamycin and its analogs has been feeble11
. Our demonstration of GOLPH3’s role in activating mTOR signaling and conferring increased sensitivity to rapamycin in preclinical setting raises the possibility that GOLPH3 expression level or copy number status may predict sensitivity to mTOR inhibitors. Indeed, endpoint analysis of our preclinical treatment studies showed that rapamycin was significantly more effective against xenograft tumors expressing high level of GOLPH3 (p = 0.0268; 1205LU-GOLPH3 vs. 1205LU-EV tumor volumes at endpoint; Supp Fig. S5b
), thereby suggesting that GOLPH3 levels may be a positive predictor of rapamycin sensitivity. Although the predictive value of GOLPH3 as a biomarker remains to be demonstrated, the formulation of this hypothesis highlights the importance of mechanistic insights beyond functional and clinicopathological validation in the translation of cancer genomics.