RPPA analysis

In order to consider the broader signaling effects following SGK3 overexpression, reverse-phase protein arrays were carried out, looking for differences between SGK3 overexpression compared to the empty vector control. The proteins studied focused on PI3K signalling and some apoptosis/EMT mediators. Although most of the candidate proteins modulated were not statistically significant, this was more due to variations in the up-regulation fold change between each experiment despite each experiment showing at least a 1.2-fold change, therefore due to consistent high expression compared to the control in each experiment, these were included. At 0h baseline, prior to serum stimulation, several proteins are upregulated in the SGK3 overexpressed condition. Of interest are the top 3 overexpressed proteins phospho-Myosin-IIa-S1943, HES1 and connexin-43.

Myosin-IIa has been implicated in increasing invasiveness and motility in poor prognosis tumours. Myosin II has different isoforms, where switching occurs (not too dissimilar from the cadherin switching during EMT). For example, in breast cells, Myosin IIc tends to be expressed in luminal cells, whereas Myosin IIb tends to be up regulated in myoepithelial cells, which tend to have more mesenchymal characteristics. It is thought that TGFb induced EMT results in an isoform switch from Myosin IIc to Myosin IIb and increased phosphorylation of myosin heavy chain IIa on sites known to regulate filament dynamics (S1916, S1943). This transition is thought to be a critical stage in EMT and contributes to increased invasive behavior (Beach et al, 2011). It was proposed that phosphorylation of Myosin IIa results in redistribution of myosin II from posterior to anterior regions, where it can facilitate adhesion and alter cell invasion capability. It is also thought that Myosin II represents a point of convergence of signal transduction pathways that drive glioma invasion so it is possible that the same applies to ovarian and pancreatic cancer. Many studies discuss the role of myosin II isoform switching during TGFb induced EMT; the PI3K pathway is a known non-smad pathway contributing to TGFb induced EMT and is also thought to regulate TGFb/activin signalling (Zhang et al, 2009), so perhaps SGK3 serves as a

mediator in regulating these crosstalks between the two pathways, in a manner similar to that proposed for mTORC2 by Yu et al, 2015.

HES1 (hairy enhancer of split-1) is a transcriptional target of the Notch pathway and its expression correlates with activated Notch. However, its expression is not restricted to the Notch pathway as it can be elevated by other signals such as Hedgehog, Ras and MAPK pathways (Stockhausen et al, 2005; Ingram et al, 2008) Various roles have been proposed for HES1 in different tumour types, from promoting proliferation in rhabdomyosarcoma (Ramskold et al, 2012) to promoting invasion by suppressing the metastasis inhibitor Deltex1 in osteosarcoma (Zhang et al, 2008) and through the STAT3- MMP14 pathway in colorectal cancer (Weng et al, 2015). Additionally, HES1 has been reported to be a bad prognostic marker in medulloblastoma and ovarian cancer.

Connexin-43 also is associated with promoting invasion and migration in several tumour types such as prostate cancer (Zhang et al, 2015), gliomas (Lin et al, 2002) and breast cancer (Stoletov et al, 2013) and have been suggested as a therapeutic target to benefit patients with metastatic disease. High expression of connexin 43 has also been associated with poor progression free survival in invasive urothelial bladder cancer (Poyet et al, 2015).

When analyzing the RPPA dataset at 30 minutes after serum stimulation, interestingly, Myosin-IIa-pS1943 was the most differentially upregulated protein along with epithelial membrane antigen (EMA) also known as MUC1. It is aberrantly expressed in majority of breast carcinomas and is known to interact with b-catenin, HER2 receptors and GSK-3b in a complex that promotes disassembly of adherens junctions and the invasion of cells (Schroeder et al, 2003). Consequently the interactions between MUC1 and b- catenin are higher in metastatic tumours. Other reports have also suggested that MUC1 promotes migration and invasion of hepatocellular carcinoma (HCC) cells via JNK-mediated phosphorylation of SMAD2 (Wang et al, 2015).

Other proteins upregulated included Syk, which is an interesting protein because it has pleiotropic effects dependent on the isoform activated. The two isoforms in question are Syk (S) and Syk (L). Syk (S) promotes EMT and enhances invasiveness, whereas Syk (L) inhibits metastasis in HCC (Hong et al, 2014).

Some reports have also implicated Syk in cell motility and progression of squamous cell carcinomas of the head and neck, where high Syk expression significantly correlated with poor survival in these patients (Luangdilok et al, 2007).

Next, the fold changes between the time points for each condition (i.e. SGK3 overexpressed and empty vector control) were examined to account for serum-independent changes, however the primary goal of this experiment was to assess signalling changes with SGK3 overexpression irrespective of serum dependent/independent changes. Interestingly, it was observed that the although no proteins were upregulated with a significant fold change difference, the majority of the top targets were those implicated in either motility, invasion or collectively EMT (such as vimentin, rheb, fibronectin). Also, Myosin IIa- pS1943 was again the top candidate on the list, suggesting that although further validation is needed, there is a high possibility that this protein and its specific phosphorylation site is important in facilitating the role SGK3 has in poor prognosis. Another interesting protein that was highlighted in this analysis was pAKT-T308 which was upregulated with SGK3 overexpression, suggesting that the AKT pathway and SGK pathway may have complementary roles as effectors of PI3K.

Collectively, the RPPA has highlighted several upregulated proteins with SGK3, most of which have roles in migration, invasion and promoting metastasis and EMT in some tumours. This reinforces the theory for the potential role of SGK3 in mediating these processes in ovarian and pancreatic cancer. Although these analyses comprised of three independent biological experiments, further functional validations are required to assess the roles these proteins have with respect to SGK3 and the signalling cascade that follows. However, as a perspective, it endorses SGK3 involvement in metastatic processes. In particular, it would be interesting to investigate myosin-IIa-pS1943 and assess its precise role with SGK3; as mentioned before, perhaps there is a feedback loop given the existing known crosstalk between TGFb signalling and PI3K, where SGK3 phosphorylates phosphor-myosin-IIa-S1943 as its upregulation at the protein level correlates with SGK3 overexpression in any context of analysis.

4.16.6 Summary

Overall, the data presented here indicates that SGK3 plays a role in regulating invasive and migratory phenotypes, promoting metastasis and EMT. This is likely occurring in part through the transcriptional regulation of EMT markers by SGK3 and partly through the CXCR4-CXCL12 signalling axis, through SGK3 positively regulating its protein levels and potentially mRNA too. Processes that govern metastasis and EMT require a convergence and crosstalk between several key pathways such as the PI3K/b- catenin/MAPK pathways so it is unsurprising that we do not yet have a full picture of all the interactions between SGK3 and other mediators but some are becoming apparent. For example, from the knowledge that SGK and AKT family have similar functions and substrate specificities, coupled with the observation of upregulation of pAKT with SGK3 overexpression in the RPPA indicates that they work complementarily. There are possible links between complementary functions between HER2 and SGK3 too. As mentioned earlier, SGK3 is able to prevent loss of CXCR4 (presumably through inhibition of ubiquitination and subsequent proteosomal degradation) thus allowing the CXCR4 signalling cascade to continue. Similar mechanisms have been proposed for HER2, where it has been suggested that CXCR4 overexpression is mediated by HER2 overexpression and HER2 activates the PI3K pathway causing enhanced translation of CXCR4 mRNA via the activation of PDK1 and AKT/PKB (Li et al, 2004; Slagsvold et al, 2006). Given that we see upregulation of CXCR4 mRNA with SGK3 overexpression too, it is possible that perhaps HER2 is involved in mediating this process. Additionally, HER2 is also known to inhibit ubiquitin-dependent lysosomal degradation of CXCR4 by a mechanism that remains to be characterized (Li et al, 2004). This might again suggest a role for HER2 and SGK3 to work complementarily.

Although the data generated here clearly shows a role for SGK3 in metastasis, there are some limitations within the study that need to be considered. For example, bulk of the work in elucidating the role of SGK3 was carried out in SKOV3 cells. SKOV3 cell lines have come under scrutiny in the last few years with several studies suggesting that they may be a poor choice to model HGSOC because they do not resemble HGS tumours at a molecular level (Domcke et al, 2013; Cunnea and Stronach, 2014). However, they are still accepted as a useful cell line to study ovarian cancer and furthermore harbor an activating PIK3CA mutation and so would be relevant to this studies. It would be useful to carry out these studies

in other more appropriate cell line models that reflect the biology of HGS tumours more accurately. Additionally, with respect to the experiments conducted, it would be beneficial in the future to use CRISPR genome editing to knock out SGK3 and subsequently knock in the mutant forms of SGK3 to decipher the role of SGK3 without any endogenous SGK3 affecting the results. This may possibly enhance the clear effects already being observed with SGK3 modulation.

Overall, there is compelling evidence to suggest that inhibiting SGK3 may have potential to inhibit the metastatic process; this is very important and may prove especially beneficial for tumours such as ovarian cancer, which is typified by peritoneal dissemination, which is the primary cause of poor prognosis.

CHAPTER 5: DNA-PKCS INHIBITION RESTORES SENSITIVITY TO