OPCML Regulates RTK Signaling and Suppresses Tumor Growth

EXAMPLE 5

Summary

OPCML, a GPI anchored tumor suppressor gene is inactivated by somatic methylation in multiple cancers. We previously identified this gene by LOH mapping and demonstrated that it was inactivated by somatic methylation in 80% of ovarian cancers. Restoring OPCML expression by stable transfection suppressed in-vitro growth and in-vivo tumorigenicity. We investigated the role of OPCML in growth signaling pathways. In SKOV-3 and PEO1, ovarian cancer cell lines with no expression of OPCML, we demonstrated that OPCML negatively regulates a specific repertoire of receptor tyrosine kinases (RTKs) EPHA2, FGFR1, FGFR3, HER2 and HER4, and reciprocally, OPCML siRNA knockdown in normal ovarian surface epithelial cells up-regulates these same RTKs. OPCML has no effect on the RTKs EPHA10, FGFR2, FGFR4, EGFR, HER3, VEGFR1 and VEGFR3. Example immunoprecipitation experiments revealed that OPCML binds to EphA2, FGFR1 and HER2 extracellular domains with no such interaction to EGFR, thus OPCML binds directly to RTKS that it negatively regulates. We demonstrate that OPCML is located exclusively in the raft membrane fraction and sequesters RTKs that it binds to the raft fraction, leading to polyubiquitination and proteosomal degradation via a cav-1 endosomal mechanism resulting in systems depletion of this specific RTK repertoire, that does not occur with RTKs that OPCML does not bind. We demonstrate that OPCML abrogates EGF mediated phosphorylation of FGFR1, HER2 and EGFR and the downstream phosphosignaling of pErk and pAKT.

A recombinant modified OPCML-like protein without a GPI anchor, signal peptide or glycosylation was constructed and expressed in E. coli. This rOPCML tumor suppressor protein therapeutic caused growth inhibition by apoptosis in 6/7 ovarian cancer cell lines tested, with no effect on OPCML expressing normal ovarian surface epithelium, by an identical mechanism to the transfected normal protein. rOPCML was then injected intraperitoneally twice weekly in two murine intraperitoneal models of ovarian cancer (nude mouse A2780 and SKOV3) and demonstrated profound inhibition of tumour weight, ascites volume and peritoneal dissemination compared with BSA control.

Mechanism of OPCML TSG Function

OPCML is a non-transmembrane, external lipid leaflet GPI-anchored protein, and is frequently lost from cells by somatic inactivation of the gene. We hypothesised that it may mediate its tumour suppressor properties via interactions with transmembrane signalling proteins, and so we analysed the effect of receptor tyrosine kinase (RTK) growth factor stimulation on OPCML gene expression. Treatment of 4/4 ovarian cancer cell lines with EGF or FGF 1/2 resulted in rapid OPCML RNA and concomitant protein expression (data not shown) suggesting that OPCML may be a putative suppressor-type immediate-early negative feedback regulator.

Stable transfection of OPCML in the basal unstimulated or ligand-stimulated SKOV-3 ovarian cancer cells, resulted in the profound protein down-regulation of a specific repertoire of RTKs: EPHA2; FGFR1; FGFR3; HER2 and HER4 (FIG. 16A) and this RTK down-regulation spectrum is reproducible by transient transfection of a different ovarian cancer cell line, PEO1 (FIG. 16B). These same RTKs were also reciprocally up-regulated when physiological OPCML was knocked down by siRNA in OSE-C2, a normal ovarian surface epithelial cell line (Davies et al, (2003) Experimental Cell Research 288: 390-402) (FIG. 16C). This specific inactivation by OPCML was not seen for other RTKs we have investigated so far including: EPHA10; FGFR2; FGFR4; EGFR; HER3; VEGFR1 and VEGFR3 (FIG. 16). The phenotypic consequences of these signalling effects were confirmed in growth assays in ligand-supplemented media where OPCML-transfectants were significantly growth-inhibited compared with vector control (data not shown).

Negative Regulation of Specific RTKs by OPCML is Related to Direct Protein Interaction

We further explored as examples EPHA2, FGFR1 and HER2, RTKs that are strongly inactivated at the protein level upon OPCML expression. We also analysed EGFR as an example of a protein that is unaffected by OPCML. Immunoprecipitation (IP) experiments demonstrated protein/protein interactions with EPHA2, FGFR1 and HER2, but no such binding to EGFR (FIG. 17A). These findings were further confirmed using a recombinant OPCML (GST-OPCML D1-3) pull-down assay (FIG. 17B) which was then used to determine that the extracellular domains (ECDs) of the RTKs FGFR1 and HER2 (as examples) were capable of interacting specifically with OPCML (FIGS. 17C&D), showing that the site of interaction lay within the ECD of the RTKs and domain 1-3 of OPCML, defining the site of OPCML action as extracellular.

Downstream Signalling

Upon acute ligand stimulation, OPCML expression led to profound abrogation of phospho-FGFR1-Y766, phospho-HER2-Y1248 and, also, phospho-EGFR-Y1173. Whilst EGFR total protein down-regulation is NOT observed, presumably due to the absence of an RTK ECD physical interaction with OPCML, the consequence of OPCML mediated loss of the activating dimerisation partners of EGFR, (HER2 and HER4), coupled with the continuing availability of the HER3 family member (that results in an inhibitory dimerisation with EGFR), explain the down-regulation of EGFR signalling even though total EGFR levels are unaffected (FIG. 18A). Analysis of FGFR1 signalling showed a similar pattern of phospho inhibition relating to protein down-regulation (FIG. 18B).

Analysis of downstream signalling demonstrated abrogation of phospho-ERK 1 & 2 (T202 & T204) and phospho-AKT-S473 (FIG. 18C), suggesting that both pro-growth and pro-survival pathways are inhibited by OPCML re-expression, via a systems level abrogation of this specific RTK spectrum.

OPCML-Mediated RTK Degradation Mechanism

Using HER2 as a paradigm molecule of OPCML-RTK regulation, we found that the available HER2 in OPCML expressing cells was sequestered in the detergent resistant membrane (DRM) fraction. In the OPCML non-expressing line, HER2 was found equally distributed between the DRM and the detergent soluble (non-raft) fractions. The total level of EGFR was not affected by the expression of OPCML and its distribution showed a much less pronounced but discernible shift to the DRM fraction (FIG. 19A). These data indicate that OPCML expression leads to redistribution of HER2 to the DRM fraction in the plasma membrane (that broadly correlates with membrane “rafts”). IFM was employed to examine the trafficking of OPCML in cells; EEA-1 (a marker of the early endosome) and Caveolin-1 (a marker of the raft-caveolar pathway) were used to investigate this apparent redistribution. A decrease in HER2 co-localisation with EEA1 shows that the sequestration of HER2 to the DRM fraction decreases its endocytosis via clathrin-mediated pathways. While an increase in co-localisation with caveolin-1 was observed, the immunofluorescence pattern suggests this is a function of the redistribution of HER2 into the DRM fraction (housing lipid-raft domains) where caveolin is also localised, as HER2 did not appear to be exclusively localised to caveolae in the presence of OPCML expression. Furthermore, in the presence of OPCML the staining was organised into specific sub-cellular particles, suggestive of distinct vesicular compartments (FIGS. 19B&C).

This analysis demonstrated that OPCML expression was associated with increased ubiquitination of HER2 (that binds OPCML), which was strongly increased upon EGF stimulation (FIGS. 19D&E). Exposure to MG-132, a potent inhibitor of the proteasomal 26S proteinase, attenuated HER2 degradation with no such effect on EGFR expression (that does not bind OPCML). In contrast, chloroquine, a weak base that alkalinises the lysosome, showed no inhibition of HER2 degradation (FIGS. 19H&I). This suggested that the proteasomal pathway was preferentially utilised for OPCML-mediated HER2 degradation. Furthermore, disruption of cholesterol (a component of DRM fraction/lipid rafts) using methyl-β-cyclodextrin (Mβ-CD) also inhibited the degradation of HER2 and increased HER2 phosphorylation (FIG. 19J) suggesting that cholesterol-rich lipid-raft structures are important for OPCML-specific internalisation and degradation of HER2.

These findings suggest that OPCML-mediated negative regulation of this specific repertoire of RTKs is the result of direct binding of OPCML to the ECD of that RTK. These multiple but specific binding events result in ‘lipid-raft’ sequestration, enhanced ubiquitination, and a switch away from clathrin-mediated endocytosis to proteasomal degradation of those specific RTKs negatively regulating their signaling through reducing their protein level. Our data, in the context of very recent publications (Howes et al (2010) J. Cell Biol. 190(4): 675-91; Howes et al (2010) Curr. Opin. Cell Bio. 22(4): 519-527)), would suggest that CLIC/GEEC bulk internalization route is a strong candidate pathway for OPCML-mediated degradation of HER2 and that this is linked to RTK inactivation and the observable strong tumour suppressor phenotype of OPCML.

Recombinant OPCML (r-OPCML) inhibits tumour growth in vitro and in vivo Purified recombinant human OPCML domain 1-3 protein (r-OPCML) (FIG. 8) was produced from the bacterial expression vector (pHis-Trx) subcloned with domains 1-3 of OPCML, excluding the signal peptide and GPI anchor sequences (FIG. 8A). Addition of r-OPCML protein to growth media demonstrated a specific, dose-dependent inhibition of cell growth in OPCML non-expressing SKOV-3 ovarian cancer cells, without affecting normal ovarian surface epithelial cells, OSE-C2 (FIG. 20A). We have confirmed that r-OPCML profoundly inhibited cell growth in 6 of 7 additional OPCML non-expressing epithelial ovarian cancer cell lines; 2 of 2 breast HER2-positive and negative cells; and 5 of 5 lung cancer cell lines (FIG. 20B). To determine the mechanism of this pharmacological growth inhibition we performed Annexin V FACS apoptosis assay in SKOV-3 and A2780 demonstrating evidence of early apoptosis induced by r-OPCML at 2-6 hours post exposure depending on cell line (FIG. 20C). We then performed caspase-glo apoptosis assays across a concentration range in SKOV-3 and A2780 ovarian cancer cells and demonstrated that r-OPCML induces apoptosis in both these cell lines in a dose dependent fashion, demonstrating the underlying mechanism of the observed growth inhibition (FIGS. 20D&E). Immunoblotting confirmed that addition of r-OPCML protein to media potently downregulated the same spectrum of RTKs as seen by transfecting OPCML into cancer cells, as well as abrogating pERK and pAKT in both SKOV-3 and A2780 (FIG. 21). This suggests that pharmacological use of extracellular unanchored r-OPCML utilises the same mechanism of action as transfection induced intracellular re-expression of the normal GPI-anchored, glycosylated, OPCML protein. These data were confirmed by IFM for HER2 in SKOV-3, closely mirroring stable transfection of the normal protein in the same cell line (see, FIG. 9).

In view of these in-vitro findings, we proceeded to determine whether r-OPCML protein had potential and relevance as an in-vivo tumour suppressor therapy. Mice with either SKOV-3 or A2780 cancer cells injected intraperitoneally (IP), after tumour establishment, received twice-weekly IP injections of either 1 ml (10 μM) bovine serum albumin (BSA) or 1 ml (10 μM) r-OPCML. The experiment was terminated after 3 weeks due to obvious extensive IP tumour growth and deteriorating condition of BSA-treated control animals whereas r-OPCML treated mice remained well (FIG. 22A). r-OPCML significantly and profoundly suppressed both IP tumour growth and ascites formation in-vivo in both IP models (FIG. 22B-D), and in A2780 tumour bearing mice, profoundly inhibited the number of IP peritoneal deposits compared with BSA control (FIG. 22E). Western blotting of SKOV3 IP tumour recovered from BSA treated and r-OPCML treated animals clearly demonstrated the same spectrum of RTKs inhibited as predicted from the in-vitro analysis (FIG. 22F).