The purpose of this study was to investigate ROCK2 activation in squamous cell carcinogenesis and assess its co-operation with rasHa and fos oncogene activation together with loss of PTEN mediated AKT regulation. The analysis of ROCK deregulation with these genes in the MAP Kinase and PI3K pathways, two of the most commonly deregulated signalling systems, employed a well-characterised, transgenic mouse skin model of multi-stage carcinogenesis. A major goal was to study co-operation between these genes in the conversion of benign tumours to malignancy and investigate subsequent progression to aggressive carcinomas, given these are the most significant clinical stages of carcinogenesis from a patient’s viewpoint; and also investigated effects of ROCK2 deregulation on the processes of normal epidermal differentiation. ROCK2 is an effector protein of RhoA, which is a member of the ras superfamily and ROCK2 activation has been associated with tumour progression via an increase in tissue stiffness mediated by changes in actomyosin cytoskeleton leading to increased cellular motility. Thus, ROCK2 expression is commonly associated with the later events in cancer. Furthermore, there are many studies investigating ROCK2 in cancer given that it may be a useful therapeutic target in being downstream of oncogenic ras signalling. Yet, relatively few studies have explored the possibilities of a definite link that confirms the co-operation status of ROCK2 activation with ras/MAPK/fos and /or PTEN/PI3K/AKT pathways in SCC aetiology. Thus, questions exist as to exactly when does ROCK2 activation become causal; and what are the collaborating genes involved in the mechanism that drive the early or late stage events in carcinogenesis. To begin to answer these questions, inducible ROCK2 activation has been introduced into a well-characterised transgenic mouse skin carcinogenesis model that expressed a combination of ras and fos activation, driven by a modified human keratin 1 vector (HK1). Thus, exclusive epidermal expression of activated rasHa and fos oncogenes, in proliferative basal layers, gave hyperplasia and papillomatogenesis; but with no evidence of spontaneous malignant conversion. This stability of phenotype is thus ideal to assess roles for multiple transgene co-operations in the development of benign tumours and their conversion to malignancy. Hence, ROCK transgenic mice that expressed a conditionally active, 4-hydroxytamoxifen (4-HT)-regulated of human ROCK2 transgene were crossed with mice expressing activated rasHa and /or fos exclusively in epidermal transit amplifying keratinocytes (HK1.ras, HK1.fos). Inducible PTEN tumour suppressor gene mutation via exon 5 ablation (K14.Cre/D5PTENflx) and thus loss of AKT regulation was also incorporated into this model. This was achieved via deletion of exon 5 employing the RU486-mediated cre/loxP system, driven by keratin K14 promoter expression in basal layer keratinocytes. Therefore, to facilitate this investigation, a new and unpublished inducible ROCK2 system was employed in order to target the identical keratinocytes as PTEN loss. This new transgenic line of lsl.ROCKer transgenic mice employed the same 4-HT inducible ROCKer transgene but now driven by a generic CAG promoter following cre mediated ablation of the stop cassette once treated with RU486. In bi-genic K14.ROCKer/HK1.ras1205 mice, synergism between ROCK2 activation and (wound-promotion) sensitive HK1.ras1205 line showed direct co-operation and achieved malignant conversion of benign papillomas to well-differentiated squamous cell carcinoma (wdSCCs) histotypes (12 weeks of 4-HT treatment). This placed ROCK2 activity as the causal event driving malignant conversion, but in the absence of a wound promotion stimulus (loss of ear tag), papillomas did not convert. The correct papilloma context was required for ROCK to become causal proved to be the case on employing the wound insensitive HK1.ras1276 line. Here, K14.ROCKer/HK1.ras1276 mice failed to exhibit any papillomas and required the constitutive promotion stimulus from additional fos activation. Interestingly, following cessation of 4-HT, two intriguing observations were recorded. Firstly, that once bi-genic ROCK/ras1205 achieved malignancy, exogenous ROCKer expression appeared to show no involvement once squamous cell carcinomas (SCCs) progressed to poorly differentiated squamous cell carcinomas (pdSCCs), given the elevated expression of endogenous ROCK upon malignant progression. Secondly, the rapid growth of papilloma appeared upon cessation of 4-HT with highly intense p21 expression indicated the requirement of exogenous ROCK2 for malignant conversion and the possibility of a papilloma inhibition by 4-HT anti-cancer activity. Another major novel finding demonstrated ROCK2 activation could act as an initiator in co-operation with fos activation. Direct co-operation between ROCK2 and fos activation produced benign squamous papillomas yet, whereas ROCK2 activation alone induced hyperplasia as did fos activation alone at this time point, given papilloma formation in HK1.fos mice occur over 4-5 months. However, unlike deregulation of MAP Kinase signalling in bi-genic ROCK/ras1205 mice, in bi-genic ROCK/fos mice, no malignant conversion was observed due to high levels of compensatory p53/p21 expression. Thus, this bi-genic K14.cre/lsl.ROCKer/HK1.fos model suggests the requirement of additional mutation event for malignant conversion. An unexpected result appeared in bi-genic ROCK/Δ5PTENflx co-operation experiments where K14.cre/lsl.ROCKer/Δ5PTENflx cohorts exhibited epidermal hyperplasia with folded papillomatous appearance to the epidermis, but without papillomatogenesis even after seven month of period. This either indicates a weak co-operation between ROCK2 and Δ5PTENflx which may be due to the unexpected low levels of p-AKT from a compensatory increased p21 expression; and additional events needed to fill in the oncogenic gap in this bi-genic ROCK/Δ5PTENflx model, or may possibly highlight redundancy in the oncogenic hits provided by ROCK and PTEN. This latter suggest similar links exist between their normal roles in the epidermis which may be accountable for the alterations observed in keratinocyte differentiation. In both in vitro and in vivo experiments, K14.cre/lsl.ROCKer/Δ5PTENflx cohorts showed alterations in epidermal differentiation via anomalous K1 and low levels of K6 expression. Interestingly, activated ROCK2 appeared to induce or accelerate differentiation activity in K14.cre/lsl.ROCKer keratinocytes via increased K1 (early differentiation marker) and reduced K6 (proliferation marker) expression profiles. These results were consistent with in vivo data where K6 was expressed in low levels in K14.cre/lsl.ROCKer hyperplasia histotypes. In contrast, in K14.cre/lsl.ROCKer/Δ5PTENflx keratinocytes, inactivation of PTEN-mediated AKT activity may be accountable for restored keratin K6 and anomalous keratin K1 expression; as keratin K1 was expressed in a similar fashion of normal keratinocytes in K14.cre/lsl.ROCKer/fos keratinocytes. Interestingly, all tri-genic cohorts: K14.cre/lsl.ROCKer/ras1276/fos, K14.cre/lsl.ROCKer/fos/Δ5PTENflx and K14.cre/lsl.ROCKer/ras1276/Δ5PTENflx synergisms exhibited malignant conversion and /or malignant progression in all animals highlighting a novel role of ROCK2 activation. Further, the stage-specific consequences in each model in this study were shown to be influenced by p53/p21 status, where typically p53 expression disappeared in late stage papillomas yet, p21 expression persisted. This demonstrated the importance of compensatory p53/p21 expression in modulating tumour pathogenesis in all these models. Given that this study incorporated PTEN mutation, the influence of AKT activity was investigated in the SCC progression of tri-genic ROCK/ras1276/PTENflx and ROCK/fos/PTENflx cohorts; revealing a crucial antagonism between p21 and AKT. However, this study revealed that the malignancy in tri-genic ROCK/ras1276/fos cohorts was not influenced by p-AKT expression, and as this tri-genic model achieved wdSCCs only. This suggests that as the roles of ROCK in altering cytoskeletal organisation leading to increase in tissue stiffness are overlaid onto both MAP Kinase and AKT deregulation, the outcome is a very aggressive pdSCC. Thus, suggesting ROCK signalling to be a potential therapeutic target for ras/MAPK/fos pathway in carcinogenesis. Overall, this study showed the involvement of ROCK2 activation in the initiation stage for papillomatogenesis with fos oncogene, and demonstrated ROCK2 as a converter again and also in malignant progression with ras/fos/Δ5PTENflx mutations. This indicates the link of ROCK2 signalling with both MAPK and PI3K pathways, thus targeting ROCK2 would aid in development of cancer therapy.

Introduction Squamous cell carcinoma of the auricle has an unusually high rate of lymph node metastases when compared to similar tumours at other sites. The lymph nodes affected are close to the base of the skull and in the neck. Development of metastasis carries a poor prognosis and most patients will subsequently die of failure of loco-regional control. Despite the likelihood of a poor outcome nothing can be done for patients prior to development of metastasis, as the risk of spread is not sufficiently high to warrant intervention in all patients. They are therefore treated with a ‘wait and see policy’ and only offered treatment once clinical evidence of metastatic spread is detected. This thesis sets out to examine what can be done, at the time of initial presentation with an auricular squamous cell carcinoma to identify patients who would benefit from treatment to the regional lymph node basins. Materials and Methods The thesis is divided into four separate studies. A systematic review examines the evidence available to date, an anatomical study examines the lymphatic drainage of the auricle in cadavers, a sentinel lymph node biopsy study examines the use of this technique to identify early tumour spread and a retrospective analysis of cases of auricular squamous cell carcinoma in our unit examines histopathological prognostic indictors of metastatic spread. Results The systematic review found that these tumours have a metastatic rate of about 11%. Patients developing metastasis usually die from failure of loco-regional control. Depth of tumour invasion, tumour size and mode of invasion seem to be potential indicators of metastatic risk. There is a strong argument for prophylactic intervention to the regional lymph nodes but there is no consensus of opinion as to when this should be carried out The anatomical study comprised 5 cadaveric dissections. They showed that the first echelon nodes draining the auricle lie in the superficial parotid gland, post-auricular/ mastoid nodal group and level II of the neck. There are anastamotic pathways around the mastoid and post-auricular nodes that could permit embolic tumour cells to bypass them. Five lymphatic pathways draining the auricle are described and some of these lie on the lateral and anterior surfaces of the mastoid bone and traverse the insertion of sternocleidomastoid. 28 cases of auricular squamous cell carcinoma were enrolled for sentinel lymph node biopsy. None of them were found to have any metastatic spread. One case showed non-viable tumour cells in a lymph node. There was a high incidence of complications (14%) directly related to the sentinel node biopsy procedure. The retrospective analysis identified 229 cases of auricular squamous cell carcinoma treated in our unit from 1992 - 2004. 212 of these cases had the primary pathology available for analysis. 24 (of 212) patients developed metastasis. 17 patients died as a result of their disease usually due to failure of control at the regional lymph node basin. Primary tumours with a depth of invasion greater than 8mm have metastatic rate of 56%. Tumours with a depth of invasion between 2-8mm and evidence of cartilage destruction, lymphovascular invasion or a non-cohesive invasive front have 24% metastatic rate. Tumours outwith these high-risk groups did not metastasise. Conclusions Elective lymph node dissections of the superficial parotid gland, post-auricular/mastoid and level II nodes should be considered in patients with primary auricular squamous cell carcinomas with a depth of invasion >8mm or a depth of invasion between 2 - 8 mm and evidence of cartilage destruction, lymphatic invasion or a non-cohesive invasive front. This should ideally be done as part of an observational study to evaluate the cost / benefit ratio for these patients. The neck dissection must clear the mastoid bone to a sub-periosteal level on its anterior and lateral surfaces. This will require the removal of the upper portion of sternocleidomastoid. Sentinel lymph node biopsy requires further study to evaluate it as a method for early detection of metastatic spread in auricular squamous cell carcinoma. This could be done as part of an observational study of elective neck dissections.