Search

The non-receptor tyrosine kinase effector of Cdc42, ACK, has been identified as an oncogene. Aberrant expression and mutations of ACK have been found in many types of cancer, such as prostate, breast and pancreatic. Clinically, ACK mutants or overexpression correlates with poor prognosis. Recent studies have uncovered substrates for ACK and shown their downstream signalling pathways can contribute to cancer progression. However, whether ACK can be used as a therapeutic target in cancer remains a debatable issue. It is likely that the ACK signalling network is yet to be completely elucidated. To understand whether ACK is an effective, as well as a safe target for cancer therapy, the Owen/Mott group performed a yeast-2-hybrid screen to identify new binding partners of ACK. 14 novel proteins that interact with ACK were identified, one of which was a small G protein regulator Vav3, the least-studied member of Vav-family proteins. The first part of this project characterised the interactions between ACK and Vav-family proteins in human cells. ACK was shown not only to bind to Vav3, but also to Vav1 and Vav2. From mutagenesis studies, it was shown that ACK binds Vav2 differently to Vav1 and Vav3. Further experiments indicated that the interaction with Vav1 and Vav3 stabilized ACK protein levels, an outcome that could underpin the oncogenic properties of ACK and the situation found in many cancers. For Vav2, it was shown that ACK phosphorylates Tyr142 and directs Vav2 GEF activity towards the small G protein Cdc42, resulting in an increase of HEK293T cell proliferation. Knockdown of Vav2 significantly reduced the level of activated Cdc42 and decreased the proliferative rate in ACK-transformed cells. Finally, the ACK-Vav2-Cdc42 pathway was shown to promote cell migration in an androgen-independent prostate cancer cell line. Taken together, this project has identified a new potential player in ACK-driven oncogenesis.

Activated Cdc42-associated kinase (ACK) is a Rho family effector that is widely implicated in cancer, yet not currently pharmacologically inhibited in the clinic. Through relationships with the androgen receptor in prostate cancer and the oestrogen receptor in breast cancer, ACK regulates transcriptional networks in disease. In this work, new roles for ACK in transcription, mediated by its relationship with the signal transducer and activators of transcription (STAT) family, are uncovered. The STAT family are a 7-member family of transcription factors that are hyperactivated in disease, yet no STAT inhibitors have reached clinical approval and thus, understanding of ACK-driven STAT activation is of huge therapeutic interest. In this work, ACK is shown to interact with STAT family members and augment their phosphorylation at the conserved activation tyrosine residue. This activation event is shown to drive STAT3, STAT5A and STAT5B nuclear translocation and transcriptional activation. Thus confirming ACK as a STAT activator. The ACK-STAT relationship is also demonstrated at an endogenous level, with a specificity to haematological disease cell lines. In the K562 chronic myeloid leukaemia cell line, ACK is identified to interact with STAT5 and by cell line engineering was confirmed to contribute to the nuclear pool of transcriptionally active STAT5. By thorough bioinformatic analysis, the true contribution of ACK as a mediator of STAT5 transcription was uncovered. Thirty one ACK- STAT5 transcriptional targets were identified in K562s, the first known ACK-driven transcriptional signature in a disease type. ACK loss in K562s was also phenotypically challenging, with proliferation, epithelial-mesenchymal transition, DNA damage response, glycosylation, adhesion, F-actin polymerisation and drug sensitivity all being altered upon ACK loss. Therefore, this work confirms ACK as a contributor to hyperactivated STAT5 signalling in a CML cell line and reveals a new route for therapeutic intervention.

This thesis describes the structural and functional analysis of a series of cyclic peptides, engineered as inhibitors of the Rho-family GTPase, Cdc42. The research presented builds on previous work which selected Cdc42 binding peptides using CIS-display. The structure of the highest affinity matured peptide (P7) in solution was solved by NMR. A model of the peptide (P7) in complex with Cdc42 was derived by the application of HADDOCK using the unbound peptide structure, a previously solved structure of Cdc42 and chemical shift perturbation data. The model of the complex indicated peptide binding interactions with the effector binding region of Cdc42. Binding assays demonstrated the ability of the peptides to disrupt downstream Cdc42-effector interactions, including PAK1, ACK and WASP. Phenotypic effects for three of these first-generation peptides conjugated to a nona-arginine cell penetrating peptide (CPP) sequence were established, where inhibition of downstream Cdc42-related signalling was observed in a KRas mutant cell line. These phenotypic effects included inhibition of cellular migration and proliferation. The model of the peptide-protein complex informed the rational design of enhanced affinity peptides. Canonical and non-canonical amino acid substitutions were explored to modulate the peptides affinity for their target, Cdc42. A systematic N-terminal truncation library and alanine scan panel of peptides were synthesized, and their affinity tested. Three alanine variants were identified with a slightly enhanced affinity over the wildtype sequence; the structures of which were then determined by NMR. Additionally, installation of a homocysteine residue at one of the disulphide-bridging positions was shown to enhance affinity. Non-canonical proline analogues were used to probe a prolyl bond between residue positions 11-12 of the peptide sequence, where cis-trans isomerization was occurring. Replacement of proline-12 with a 2-methylproline residue was found to retain affinity and facilitate a cleaner synthesis of the peptide by elimination of a conformational isomer. An experimentally derived structure of one of the enhanced affinity peptides, W14A, in complex with Cdc42 was subsequently solved using NMR. Despite similarities with the HADDOCK model, the orientation of the W14A peptide relative to Cdc42 is markedly different. The structure demonstrates that W14A, at least, is a truer mimic of the classical CRIB effectors than previously known. Overall, a panel of peptides have been thoroughly interrogated both structurally and biophysically and a series of rational improvements have been applied to increase binding affinity. This work presents a high affinity peptide with the potential to act as an early-stage pre-clinical Cdc42 inhibitor for eventual application in Ras-driven cancers.

ACK is an oncogenic non-receptor tyrosine kinase associated with poor prognosis in human cancers. ACK promotes proliferation, in part, by contributing to the activation of Akt, the major PI3K effector. The work presented here demonstrates that ACK also regulates proliferation via interactions with other key components of PI3K signalling, the PI3K regulatory subunits. ACK directly phosphorylates 4 out of the 5 isoforms (p85α, p55α, p50α, p85β) on a conserved tyrosine located on the iSH2 region (Tyr607 for p85α). Phosphorylation of p85β at this site was found to promote cell proliferation, yet counterintuitively, ACK does not stimulate PI3K catalytic activity. This suggested that ACK mediated phosphorylation of the PI3K regulatory subunits drives proliferation via a PI3K-independent mechanism. The PI3K heterodimer is composed of a regulatory subunit and a catalytic subunit. The regulatory subunits are in excess over the catalytic subunits, and undertake critical functions independent of PI3K. The regulatory subunits are already reported to dimerise via interactions between their N-terminal protein domains and the work presented here identifies a novel mode of dimerisation mediated by C-terminal protein domain interactions. The pTyr607 residue of one monomer binds to the nSH2 domain of another via canonical pTyr-SH2 binding. These regulatory subunit phosphodimers form exclusively in nuclear-enriched cell fractions where pTyr607 p85 resides. It is inferred that ACK targets p110-independent p85 and the regulatory subunit dimers that form undertake pro-proliferative nuclear functions that contribute to Cdc42/ACK driven oncogenesis. These dimers represent a previously undescribed mode of regulation for the Class IA PI3K regulatory subunits and potentially reveal additional avenues for therapeutic intervention. The work presented here demonstrates that ACK interacts with the PI3K pathway at multiple nodes, including Akt, the PI3K regulatory subunits and via PTEN, the terminator phosphatase. With a growing body of evidence implicating ACK in the progression of prostate cancer, CRISPR/Cas9 knock out of ACK in a prostate cancer cell line panel was used to assess the validity of targeting ACK therapeutically. Overall, this work contributes to the growing evidence implicating ACK as a key player in carcinogenesis.

The ras genes are the most commonly mutated oncogenes in human cancers, with mutations occurring in approximately 20% of human tumours. However, more than 30 years of attempts to target Ras proteins therapeutically have yielded no effective therapies in the clinic, leading the proteins to be widely deemed 'undruggable'. In recent years, there has been substantial evidence implicating the Ral GTPases, RalA and RalB, which are activated downstream of Ras, as critical drivers of cell growth and metastasis in numerous Ras-driven cancers. Therefore, targeting this pathway may provide an effective method for inhibition of oncogenic Ras signalling. Prior work identified stapled peptides based on the Ral effector, RLIP76, that can bind to the Ral GTPases and disrupt downstream signalling. To improve the affinity of these peptide sequences, an affinity maturation was performed on the Ral-binding domain of RLIP76 from which potential sequence changes were identified. The work described in this thesis aimed to identify sequences from this selection with improved affinity for Ral proteins to guide the design of second-generation stapled peptides targeting the Ral GTPases. In vitro validation of the selection sequences enabled the identification of several sequence substitutions that together improved binding to Ral proteins by more than 20-fold. The effects of individual residue substitutions on the affinity for Ral proteins were determined using biophysical assays and two 1.5 Å co-crystal structures of the tightest-binding mutants in complex with RalB revealed the key interactions formed. The sequences were successfully translated into stapled peptides based on RLIP76, resulting in peptides with improved affinity compared to the wild-type parent sequence. The peptides have been shown to be selective for the active form of Ral, with undetectable binding to a panel of related small GTPases in in vitro assays. The binding site of the lead peptide on RalB has been determined by NMR and was found to overlap with multiple Ral-effector interactions. The peptides were able compete with multiple Ral-effector interactions in vitro and in cellular lysates. This work demonstrates how manipulation of a native binding partner can assist in the rational design of stapled peptide inhibitors targeting a protein-protein interaction.

ACK is a non-receptor tyrosine kinase and an effector protein for Cdc42. ACK has been implicated in carcinogenesis events especially in promoting cell migration and invasion, indicating a need for a fuller understanding of the cellular roles of ACK both in normal and pathological conditions. A previous yeast-2-hybrid screen in the lab identified RhoGDI-3 as an interacting partner for ACK. RhoGDI-3 is a member of RhoGDI family of proteins, which are negative regulators that maintains Rho-family GTPases in the inactive GDP-bound state and sequesters them in the cytosol. RhoGDIs have also been found to become deregulated in cancer and it is possible therefore that they play role in ACK-driven cancer progression. In this work, it is shown that ACK binds but does not phosphorylate the RhoGDIs in cells or in vitro. ACK was shown to shuttle between the cytoplasm and the nucleus. RhoGDI-1 and -2 have been found only in the cytoplasm, while RhoGDI-3 is localized to both cytoplasm and the nucleus, under the conditions tested. The interaction between RhoGDI-3 and ACK occurs predominantly in the nucleus and RhoGDI-3 levels were shown to decrease following co-expression with ACK, especially in the nucleus. Data generated here shows that ACK-mediates RhoGDI-3 proteasomal degradation potentially by regulating RhoGDI-3 ubiquitination. In order to determine the cellular effects of the interaction between ACK and the RhoGDIs, all possible Rho-family GTPases that interact with the RhoGDIs were determined in a systematic study. RhoGDI-1 and 2 were found to have relatively restricted activity, mainly binding members of the Rho and Rac subfamilies. RhoGDI-3 displayed wider specificity interacting with several novel interacting partners within the Rho, Rac and Cdc42 subfamilies. Unexpectedly RhoGDI-3 was found to form complexes with the atypical Rho GTPases such as RhoD, Wrch2, Rnd2, Miro2 and RhoH which are not regulated by standard GDP/GTP cycling. The GTP levels of these target proteins were found to decrease following co-expression with RhoGDI-3, confirming its role as a negative regulator of these Rho GTPases. ACK was shown to regulate the activation of these target proteins, including RhoA, RhoB, Rac1 and RhoH, which are known to be involved in cell proliferation and migration. Aberrant activation of these target protein is frequently observed in cancer, suggesting a role for RhoGDI to drive ACK oncogenicity.

Prior work has indicated that preventing Cdc42 interacting with its downstream effector proteins can reverse Ras-driven oncogenesis. The experiments demonstrating this used the G protein binding region (GBD) of the Cdc42-specific effector ACK, which is 42 amino acids long and binds with 40 nM affinity. The aim of this work was to find a peptide which exhibits similar effects in reversing oncogenic characteristics in cells but with more promising therapeutic properties: for example, a smaller size, improved binding and improved protease resistance. Firstly, the Cdc42-ACK binding interface was characterised thermodynamically, producing a comprehensive dataset of the contribution of individual ACK residues to complex affinity. This information revealed regions of the ACK GBD with higher concentrations of vital interactions but also residues that were unimportant for maintaining affinity. An \emph{in vitro} display technology (CIS-display) was employed to select for shorter peptides with improved binding to Cdc42. By selecting the best binders from libraries based on the ACK GBD, peptides have been generated that are half the length of the full binding domain but bind with affinities approaching the wild-type ($\approx$100 nM). More interestingly, a serendipitous discovery in a na\"{\i}ve cyclic library revealed a 16-residue sequence that binds with 350 nM affinity, subsequently called C1. The biophysical properties of selected peptides were characterised and demonstrate that C1 is reliant on an intramolecular disulphide bond for tight binding. An N-terminal nona-arginine sequence was added to C1 to facilitate entry into cells: internalisation was confirmed by confocal microscopy. Subsequently it was found that C1 reduced signalling through the ACK and PAK effector pathways, negatively impinging on MAPK signalling. Levels of signalling returned to baseline within 24 hours, however, and the transiency of the effect was thought to be linked to peptide degradation: an effect likely in a construct dependent on a disulphide bond for cyclisation and binding affinity. Subsequent maturation of C1 using a focussed CIS-display library has selected second generation peptides. These have been characterized and display binding affinities to Cdc42 of $\approx$20 nM: a 17-fold increase in binding over C1. This has enhanced the affinity to a level even higher than wild-type ACK and should improve the cellular potency of the peptides identified in this selection. %any more?! Having successfully selected peptides both shorter in length and with higher affinity than the wild-type sequence, efforts were channelled into developing a method to cyclize the peptide by a chemical linkage that would be stable in a cellular environment. The most successful approach involved desulphurisation of the disulphide bond to a thioether, making the cyclic linkage non-labile in the reducing environment of the cytosol, while retaining original binding affinity. The peptides developed in this work bind to Cdc42 with nanomolar affinity \emph{in vitro} and can enter cells and impact upon signalling processes connected with oncogenesis. As such, they provide a potential therapeutic lead for future development.