STUDY LIMITATIONS 1 Y2H analysis limitations

The Y2H system is one of the most widely used and well established genetic techniques available to identify novel protein-protein interactions in vivo (Berggård et al. 2007). The Y2H system has proven to be relatively inexpensive, do not require special equipment and can easily be performed in any molecular laboratory with high efficacy and over the years, a myriad of interactions have successfully been identified using this technique. However, with the extreme application of this method a number of limitations have come to light (Bruckner et al. 2009). When using a classic cDNA library Y2H screen, as was done in the present study, the aim is to search for pairwise interactions between bait and prey proteins that have been pooled. This means that the pooled prey library do not only contain full length ORFs but small DNA fragments as well leading to false positives being identified due to non-specific interactions (Van Criekinge and Beyaert 1999).

Moreover, since Y2H relies on the reconstitution of a transcription factor and the expression of reporter genes, all interaction in this assay takes place within the nucleus of the yeast cell (Fields

112 and Song 1989; Bruckner et al. 2009). This, of course does not represent the true cellular compartments where interacting proteins reside in vivo. Therefore interactions may be detected in Y2H analysis that cannot take place in vivo because the putative interacting proteins are located in cellular compartments that makes it impossible for them to interact. Furthermore, certain proteins may become toxic to the host cell when expressed or could degrade vital yeast proteins or proteins necessary for the assay (Van Criekinge and Beyaert 1999).

One should also be aware that Y2H is notorious for producing false positive interactions, i.e. showing reporter gene activation where no bait-prey interaction has taken place. This could be due to the bait construct being able to autonomously activate the transcription of reporter genes (auto activation) (Cusick et al. 2005; Lalonde et al. 2008; Bruckner et al. 2009). In the present study, the KCNE2-bait construct was unable to act as an auto-activator of reporter gene transcription.

The detection of false negative interactions is also a possibility when performing an Y2H screen. This happens when the interaction between the bait and prey was facilitated by a fusion or anchor protein; leading to the prevention of interaction due to a change in molecular structure (Van Criekinge and Beyaert 1999). Moreover, false negatives may also be caused by interactions that require specific post-translational modifications and unless the enzymes responsible for these modifications are present in the yeast, no interaction will be detected (Cusick et al. 2005; Bruckner et al. 2009).

Additionally, Y2H screening only identifies defined binary interactions in a complex and not all the components of a larger complex which could lead to certain protein interactions being overlooked (Berggård at al. 2007; Bruckner et al. 2009).

Another limitation of this system can be ascribed to the bait and prey proteins fusing to the DNA binding domain. This fusion of proteins might change the conformation of the bait and/or prey proteins and could influence their functions and binding properties (Fields and Song 1989; Bruckner et al. 2009).

When taking the above mentioned limitations into consideration, it is evident that independent verification of protein-protein interaction is crucial. Thus, in order to address this matter, a co-

113 localization assay was selected to determine whether the interacting proteins are located in the same subcellular compartment at any given time.

However, although co-localization of proteins provides evidence of spatial closeness and the presence of two or more proteins in the same cellular compartment, it does not offer prove of a physical interaction between these proteins (Dunn et al. 2011). Therefore, an additional verification technique is necessary to determine if the interaction is a true specific binary protein interaction. In the present study we used a co-immunoprecipitation (Co-IP) assay to confirm this.

4.6.2 Three-dimensional co-localization limitations

Some of the limitations identified throughout the use of 3D co-localization are as follows: The use of highly specific primary antibodies raised against the proteins of interest and fluorophore- labeled secondary antibodies required a great deal of optimization to determine the adequate concentration for each antibody to be used. This was necessary in order to acquire suitable images for analysis which proved to be somewhat time consuming. Additionally, when investigating endogenous proteins (like in the present study), components of the protein complex may not be expressed sufficiently in the cell line studied (Berggård et al. 2007).

Finally, when using conventional light microscopy the resolution is limited to approximately 200nm by the diffraction of light, causing objects closer than 200nm to each other to appear as a single object (Lalonde et al. 2008). However, the method used in the present study allowed z- stack images to be taken roughly 260nm (0.26µm) apart, suggesting that this distance may separate proteins that appear co-localized by fluorescence microscopy. As a result it is possible to determine whether two proteins share the same subcellular space yet, no conclusive evidence can be given regarding physical association between the proteins of interest.

This 3D co-localization method should therefore be considered as one technique that could be used in conjunction with other techniques - such as Co-IP - in order to verify physical protein- protein interactions.

114 4.6.3 Co-immunoprecipitation limitations

Some Co-IP limitations were encountered during the course of this study. The predicted size of KCNE2 has been estimated between 14kDa and 20kDa however, the size of KCNE2 detected throughout all the Co-IP and WB experiments in the present study were approximately 40kDa - 50kDa. This matter was addressed by conducting experiments with three different KCNE2- specific antibodies from three different manufacturers, yet all yielded similar results of lysates producing an intense protein band at approximately 50kDa.

Another possible explanation for the bigger protein band of KCNE2 could be attributed to complex formation of the protein (Um and McDonald 2007). Um and McDonald also reported a 40kDa protein fragment on western blots for the KCNE2 protein. They analysed the band and found that it represented the correct KCNE2 fragment. Further analysis indicated that the 40kDa fragment resulted from the dimerization of KCNE2 proteins (Um and McDonald 2007). Subsequently, this provided support for the results presented in the current study.