Fuentes y Detectores 28

development

Injection of virus particles gave good infections for both c-Frizzled-1 and c-Frizzled-7 RCAS viruses, and allowed me to look for possible effects of misexpressing these Frizzled receptors. Disappointingly, however, I either failed to generate a phenotype, or failed to spot one. In chapter 6 , 1 will deal in more detail with the implications of the negative results from these experiments, as well as the experiments one might consider to resolve the issue of whether or not Frizzled signalling has the roles in inner ear development that we

hypothesise. However, there are a number of possibilities which relate specifically to the use of Frizzled receptor overexpression which I will cover here.

4.3.3.1 Overexpression o f Frizzled receptors in other systems

In the introduction to this chapter, I described successful uses of Frizzled receptor overexpression to perturb PCP signalling (Medina et al. 2000), (Usui et al. 1999). Do they,

however, represent an exception to the rule? Might it be that Frizzled receptor overexpression will only work in certain developmental circumstances?

In the cases where Frizzled overexpression has been successful, it has already been shown that the mutant case, loss of that particular Frizzled receptor, gives a phenotype. For example, loss of Drosophila frizzled-1 was shown to perturb PCP signalling in the fly wing (Vinson and Adler 1987) long before overexpression offrizzled-1 was shown to also have a PCP phenotype (Krasnow and Adler 1994). In essence, the overexpression experiments are conducted in a system where genetic redundancy is not thought to be a concern. In our case, we do not know the mutant phenotype for loss o f either c-frizzled-1 or c-frizzled-7.

However, we do know that their expression patterns in the inner ear are extremely similar (at least at E4), raising the spectre of genetic redundancy at the receptor level for Frizzled signalling. This must therefore stand as a potential reason behind the failure of

overexpression of c-frizzled-1 and c-frizzled-7.

In addition to a role in PCP determination, Frizzled-1 in Drosophila is required to orient the asymmetric cell division of the sensory organ precursor cell (SOP) (Lu et al. 1999). In this case, however, there are no reported uses of overexpression of a Frizzled receptor to disrupt this control mechanism. Rather, the reports present loss-of-function data. Again, with regard to a potential role for c-Frizzled-1 and c-Frizzled-7 in cell fate

specification via control of asymmetric cell division in the chick inner ear, it is not clear that overexpression of a single Frizzled receptor will disrupt the process.

4.3.3.2 Practical and theoretical reasons governed our choice o f c-Frizzled-1 and c- Frizzled-7for misexpression experiments

The in situ and antibody data in chapter three suggested that four frizzled genes and one flamingo gene would all be interesting candidates for roles in either cell fate and/or PCP determination. I have only analysed two of these, c-frizzled-1 and c-frizzled-7 by RCAS-mediated misexpression.

The availability of full-length clones initially dictated this choice: both c-frizzled-1 and

c-frizzled-7 were available, already cloned into RCAS vectors. This was not the case for c-

frizzled-5 and c-frizzled-10 (the probes used in chapter 3 were generated from partial

cDNA’s). Another reason for selecting c-frizzled-1 and c-frizzled-7 was the sequence analysis, which revealed that both were closely related to other frizzled-7 genes, which have been implicated in convergent extension control in frog and fish (Medina et al. 2000). Therefore, my initial experiments concentrated on the use of c-frizzled-1 and c-frizzled-7

misexpression.

Although I did embark on the construction of a full-length c-frizzled-10 construct, I discontinued this in the light of the negative results with c-frizzled-1 and c-frizzled-7.

Rather than continue with efforts to clone full-length c-frizzled-10 (and c-frizzled-5), I instead chose to pursue the option of making Dishevelled constructs (see chapter five) which, I hoped, would circumvent the need to misexpress the correct frizzled gene, and would be less subject to problems of redundancy. In vertebrates, as in Drosophila, there are many frizzled genes, but only a few dishevelled genes (one in Drosophila), on which Frizzled signalling depends. Moreover, the modular nature of the Dishevelled protein

makes it possible to target specific intracellular signalling pathways downstream from Frizzled receptors in a dominant-negative fashion. These experiments are described in chapter 5.

5 Chapter 5: Disruption of Frizzled signalling by modulating

Dishevelled function

5.1 Introduction

In chapter 4 , 1 described misexpression of c-Frizzled-1 and c-Frizzled-7 using RCAS viruses. The results did not reveal a function for these receptors in the chick inner ear. Yet these genes are expressed there in a highly suggestive pattern, and there are strong reasons to expect that the Frizzled signalling pathway should be important in ear development. The failure to detect a phenotype following overexpression of c-Frizzled-1 and c-Frizzled-7 does not rule out the existence of a function for Frizzled signalling in the ear. I therefore sought to assess Frizzled pathway function in the inner ear by a different means. In this chapter, I describe the use of Dishevelled-1 deletion constructs. These were designed to impinge specifically on canonical or PGP pathway functions of Frizzled signalling, in a dominant negative fashion. The results from these experiments, as from those in chapter 4, proved negative. Although Dishevelled deletion constructs were successfully tagged and shown to be expressed in the inner ear from E5 through E l 6 , 1 was unable to detect any disruption of either cell fate or PCP determination in the inner. Though still not conclusive, when taken in conjunction with the data from chapter 4 this suggests that our original hypothesis about the role of Frizzled signalling is incorrect, or at least simplistic.

5.1.1 Misexpression o f full-length Frizzled receptors is a limited approach to disrupting wild type Frizzled signalling

While exploring the consequences of misexpressing full-length frizzled genes in the chick inner ear, it became clear that a more incisive set of molecular tools would be useful. Even before I came to the conclusion that c-Frizzled-1 and c-Frizzled-7 overexpression gave no phenotype in the inner ear, there were theoretical reasons for wanting alternative viral constructs. Whatever results were obtained using the viruses driving overexpression of full-length receptors, it would be desirable to do an inverse experiment to test the effect of blocking the signalling pathway. With the full-length receptors as our only molecular tool, we had no way of doing this. One might contemplate using a dominant-negative Frizzled construct, but there is no well-established way to generate a dominant-negative allele. Another approach would be through gene knock-out in the mouse, but we do not currently know enough about frizzled expression patterns in the mouse inner ear to know the best candidates. In any case, it is always possible that genetic redundancy would mask effects at the receptor level, and could only be overcome by impinging on downstream signal

transduction.

Another problem arises from the branched nature of the Frizzled signalling pathway, which is thought to trifurcate downstream from the receptor into Canonical, PGP and Ca^"^ intracellular branches. Any phenotype observed through misexpression of a Frizzled receptor could be due to effects on any of these three downstream pathways. How to distinguish them?

To address these problems, I decided to make use of published data on the downstream effector, Dishevelled, to design a series of deletion constructs. These would serve as tools to block Frizzled signalling, so as to do the reciprocal experiment to overexpression of full- length Frizzled receptors and would also make it possible to interfere specifically with either the canonical or the PCP signalling pathway.