E6AP ACTS AS A REPRESSOR OF E STROGEN RECEPTOR SIGNALING IN L UCIFERASE ASSAYS . 90

Although it was possible to show that E6AP regulates estradiol-mediated transcription of an endogenous gene like Arc, the mechanism of repression remains unclear. When Arc was first described as an estrogen-responsive gene, it was assumed that activation of Arc transcription is mediated by non-genomic signaling of estrogen receptors, as inhibitors of PI3K (Phosphoinositide-kinase 3) and MAPK (Mitogen-activated protein kinase) prevent estradiol-mediated Arc transcription (Chamniansawat and Chongthammakun, 2009, 2010). In order to obtain insights into the effect of E6AP on estrogen receptor signaling, luciferase-based estrogen receptor reporter assays were established. While overexpression of ERα and ERβ efficiently activated the reporter, concomitant overexpression of E6AP significantly reduced the reporter activity (Figure 28, Figure 29). Additionally, comparison of reporter activity in wild type cells and stable E6AP knockdown cells, revealed enhanced induction of reporter activity in the absence of E6AP (Figure 30). Hence, in agreement with the data obtained from the endogenous Arc promoter, E6AP acts as repressor in estrogen receptor reporter assays. The used reporter construct is a relatively simple construct, which is composed of three estrogen-responsive elements (EREs), a TATA box and the luciferase gene and therefore is supposed to reflect the classical genomic ER signaling.

However, as already mentioned it has been speculated that estradiol-mediated Arc transcription involves non-genomic ER signaling. Although, at first sight, these observations seem to be conflictive, this must not necessarily be the case. It is widely accepted that both the genomic and the non-genomic pathways of estradiol signaling require the action of estrogen receptors. Thus, one possible explanation would be that E6AP, as an E3 ubiquitin ligase, ubiquitinates estrogen receptors, leading to proteasomal degradation and downregulation of ER signaling irrespective of whether the genomic or non-genomic pathway is used. However, neither E6AP dependent ubiquitination of ERα and ERβ in-vitro nor degradation of receptors upon concomitant overexpression of E6AP could be observed (Figure 28 and data not shown).

Discussion 91

Additionally, a ligase-inactive E6AP variant was still able to display repressor function.

Therefore, it seems unlikely that the mechanism of repression involves ubiquitination and degradation of the estrogen receptor. However, it cannot be excluded that E6AP affects the activity of estrogen receptors, either in a direct or indirect manner.

Pulldown experiments did not reveal binding of E6AP to estrogen receptors, neither in the presence nor in the absence of hormone, which might argue against a direct effect of E6AP on ER function (data not shown). A second potential explanation for an effect of E6AP on both genomic and non-genomic signaling could be provided by the well-known crosstalk of genomic and non-genomic pathways. This crosstalk includes the activation of ER co-activators via the non-genomic pathway. Thus, another potential mechanism of E6AP function would be the regulation of a co-activator of estrogen receptors. However, due to the fact that a ligase inactive E6AP is still able to repress reporter activity it seems unlikely that such a mechanism would involve ubiquitination and proteasomal degradation of a co-activator.

It should be mentioned that it was already published several years ago that E6AP modulates the activity of steroid hormone receptors, including the estrogen receptor.

However, in this context E6AP has not been described as a repressor of estrogen receptor signaling, but quite controversially as a co-activator of estrogen receptor signaling (Nawaz et al., 1999). This finding is based on luciferase reporter assays, but the reporter used in these assays differs strongly from our reporter construct. As already mentioned, the reporter we used in this study is composed of three EREs a TATA box and the luciferase gene, whereas the reporter used in the publication contains a fragment of the Xenopus laevis vitellogenin2a promoter, more precisely the region spanning nucleotides -87 to-303, the TATA box of the E1b promoter and the luciferase gene (Nawaz et al., 1999). Thus, while our reporter construct is likely only activated by estrogen receptors, the other reporter construct may contain binding sites for various co-activators or repressors of estrogen receptors or even other transcription factors. Indeed, analysis of the promoter fragment identified potential binding sites for several different transcription factors (Messeguer et al., 2002). Thus, our data are not necessarily in conflict with the reported co-activator function of E6AP, but instead could indicate that the effect of E6AP on estrogen receptor signaling might be dependent on the actual genomic context. However, comparison of the two different reporter constructs should be performed in the future. Furthermore, detailed analysis of these reporters could probably provide insights into the mechanism of E6AP co-activator and/or repressor function.

Discussion 92

5.2.5 E6AP represses Estradiol-mediated SFPQ transcription

After identifying the Arc gene as an endogenous target of the estrogen receptor, the transcription of which is negatively regulated by E6AP, we analyzed proteins whose expression levels are increased in the absence of E6AP in proteomics approaches, for potential estradiol responsiveness of the respective genes. One protein that on the one hand was found to be upregulated in the absence of E6AP and on the other hand possessed a potentially estrogen responsive promoter, was the proline- and glutamine-rich splicing factor SFPQ. Quantitative PCR analysis of SFPQ mRNA expression clearly showed that SFPQ transcription could be induced by overexpression of ERα (Figure 32 A). In addition, induction of estradiol-mediated SFPQ transcription was significantly increased in E6AP knockdown cells (Figure 32 B). Furthermore, overexpression of E6AP strongly inhibited SFPQ transcription in ERα overexpressing cells. Thus, it can be assumed that in addition to Arc, SFPQ represents a second endogenous target gene, which is induced by estrogen receptor signaling and negatively regulated by E6AP (Figure 32 A).

The identification of a second estrogen responsive gene, which is negatively regulated by E6AP expression, strongly indicates that the effect observed for Arc expression is not an exception, but instead might rather be a general mechanism for estrogen responsive genes. Estradiol signaling regulates a broad spectrum of physiological effects, including, development, differentiation, reproduction, behavior and processes in the nervous system(Cordoba Montoya and Carrer, 1997; Couse and Korach, 1999;

Mangelsdorf et al., 1995; Yamamoto, 1985). Thus, it can be speculated that loss of E6AP expression leads to a global deregulation of estrogen responsive genes. Such global transcriptional deregulation could probably explain the complex phenotype observed in Angelman Syndrome better than deregulation of one or even several E6AP ubiquitination substrates. Due to the fact that Angelman Syndrome is an imprinting disorder, the symptoms of Angelman patients are very likely restricted to the nervous system, as Angelman Syndrome is caused by loss of maternal E6AP expression in specific brain regions, while other tissues are supposed to be unaffected (Rougeulle et al., 1997). This might explain why Angelman patients do not display reproduction defects. However, impaired fertility is one of the phenotypes that is observed in Ube3a-/- mice. If the proposed model of deregulated estrogen receptor signaling is correct, it could explain why these mice suffer from fertility problems. In contrast to the patients, these mice lack E6AP expression in all tissues, including the reproductive organs and thus, might display additional phenotypes. Thus, identification of additional estrogen responsive genes would probably help to reveal the proteins and pathways, which are responsible for the individual phenotypes. A combination of a microarray approach and proteomics approaches would very likely be a suitable tool to identify genes and pathways that are deregulated in Angelman patients.

Discussion 93

In this context it should also be mentioned that Angelman Syndrome is supposed to be the result of loss of E6AP ubiquitin ligase activity. However, the results obtained in the reporter assays indicate that ligase activity is not required for the repressive function of E6AP. Different Angelman Syndrome-associated E6AP mutations have been described and range from the deletion of the whole chromosomal region to single point mutations. Additionally, many mutations result in premature stop codons and, thus, the expression of truncated E6AP variants (Fang et al., 1999). All of these mutations have been associated with a loss of E3 ligase activity. However, no mouse model is currently available that expresses an inactive E6AP mutant. Thus, it cannot be concluded at present that inactivation of E3 ligase activity alone is indeed sufficient to cause the observed phenotypes. If the model proposed in this thesis is correct (Figure33), E6AP mutants found in Angelman patients should be impaired in their ability to modulate estrogen receptor signaling. However, overexpression of an Angelman Syndrome associated E6AP mutant and C-terminal truncated versions of E6AP in reporter assays still displayed repression of estrogen receptor signaling (data not shown), revealing a weak spot in our model. However, one should consider that these assays were overexpression experiments that might not reflect normal physiological conditions. Thus, careful titration experiments with wild-type E6AP and Angelman Syndrome mutants as well as ligase-inactive E6AP have to be performed to study if these mutants display less efficient repression of estrogen receptor signaling at protein levels that may be comparable to physiological E6AP levels. At the moment, very preliminary data indicate that this might actually be the case (data not shown).

Furthermore, as already mentioned, different mutation types in Angelman Syndrome have been associated with varying severance of symptoms. Deletion of the entire genomic locus is for example associated with a more severe phenotype than E6AP point mutations (Lossie et al., 2001; Moncla et al., 1999). Thus, supportive to our model, it can be speculated that point mutations within E6AP, which cause Angelman Syndrome, might still display repressor function, however with impaired efficiency, while a complete loss of E6AP expression is associated with a more severe phenotype due to an entire loss of repressor function. Nonetheless, additional experiments have to be performed in the future to elucidate the role of E6AP mutants in estrogen receptor signaling.

It should also be mentioned that estrogen receptor signaling is likely not the only transcriptional pathway that is deregulated in the absence of E6AP, as we additionally observe inhibitory effects of E6AP on Androgen and Progesterone receptor signaling in reporter assays (data not shown). Finally, gene expression might be differentially regulated by E6AP, meaning that E6AP might act as a repressor for a number of genes, while it might function as a co-activator for another set of genes.

Discussion 94

Taken together, this study proposes a new model to explain at least parts of the Angelman Syndrome phenotype. However the situation is likely to be much more complicated than proposed at the moment and more work has to be invested in future studies.