Mechanisms driving patchwork membrane formation

The patchwork model of membrane organisation extends far beyond current understanding of membrane organisation. Only few different PM domains have been identified within living cells and only three mutual exclusive domains within budding yeast have been de- scribed (Stradalova et al., 2009). Biological membranes represent the sum of many weak interactions between proteins and lipids, hence the patchwork model of membrane organi- sation cannot be explained by a single theory, but by an interplay between existing models and concepts.

Theories explaining lateral inhomogeneities within biological membranes are the lipid raft concept (see section 2.5.1), the picked fence model (see section 2.5.2), and the hydrophobic matching theory (see section 2.5.3).

While the picked fence model was shown to not occur within budding yeast (Valdez-Taubas and Pelham, 2003), lipid rafts can only explain the emergence of a few domains (L_Ovs. L_D). The hydrophobic matching theory on the other hand has the capacity to explain many PM protein domains. Limitations of the hydrophobic matching theory are the exclusive focus on the TMS length and membrane thickness, therefore neglecting contributions of further protein-lipid interactions. Protein-lipid interactions have been described for a long time on different levels. Annular lipids were already proposed in the second half of the last century (Lee, 1977) and interact only transiently with proteins. In addition selective interaction with proteins by geometric matching the shape or charge of proteins was shown in many cases (Galla and Sackmann, 1975; Lehtonen et al., 1996a; Hite et al., 2010). These tight protein-lipid interactions were reported to be important for correct protein function (Huber et al., 2004; Lee, 2004; Bagatolli et al., 2010a). Raft formation and phase separation was recently reported to be induced by hydrophobic matching (Kaiser et al., 2011; McIntosh et al., 2003; Vidal and McIntosh, 2005). Phase separation is the rational of the lipid raft

concept, explaining lateral segregation of lipid and proteins into mutual exclusive domains (Simons and Ikonen, 1997). Indeed, the hydrophobic matching theory was shown to provide the mechanistic basis for lipid raft formation in artificial membranes (Coskun and Simons, 2011; Lingwood and Simons, 2010).

Lipid raft as a domain formed via hydrophobic matching re-classifies rafts as a membrane compartment among others. Within our patchwork model lipid rafts would represent a PM domain which requires sphingolipids and sterols. Slight variations of sphingolipid requirements for certain proteins would predict different raft like domains, which were indeed recently reported within living cells (Tyteca et al., 2010; Sengupta et al., 2011). The hydrophobic matching theory, in combination with protein-lipid interactions explains the domain plurality within the patchwork model of membrane organisation.

Formation of large PM protein domains, as those which have been described within this study can be explained by the “wetting” model, originally introduced to provide the the- oretical basis for lipid raft formation (Akimov et al., 2008). Small protein-lipid clusters of the same or similar proteins, will recruit matching lipids, either as annular lipids or by tight protein-lipid interactions. This will result in an enrichment of certain lipids around growing protein clusters. Depending on the coherence length, more proteins and lipids will be recruited (see section 2.1.4), eventually resulting in the formation of large microscop- ically resolvable PM protein domains. Proteins which differ in their protein environment will therefore segregate into different domains, while similar proteins are expected to show an overlap with others.

Large network like domains like the ones identified within this study, were never observed in mammalian cells. Several reasons can explain these observations. First, systematic studies of a comprehensive set of PM proteins with high spatial and temporal resolution are currently missing. Second, the diffusion of proteins and lipids is much higher in mammalian cells compared to yeast (Greenberg and Axelrod, 1993; Valdez-Taubas and Pelham, 2003) (Fig. 3.7, Tab. 8.2). The coherence length is larger if protein and lipid diffusion is slow and this will lead to the formation of large protein domains. Additionally proteins can induce a phase separation as a result of surrounding lipids, creating an environment of several nanometers around the whole domain (Akimov et al., 2008). This “wetting” environment, which differs in phase from the bulk lipids will prevent proteins from domain exit. Indeed, large protein and lipid domains were observed in artificial membranes (Kaiser et al., 2011;

de la Serna et al., 2009) and were predicted by molecular dynamic calculations (Wallace et al., 2006).

These examples indicated domain formation purely based on protein-lipid interactions. The idea of a multi domain organisation of PM proteins in yeast was already proposed by a study based on FRAP experiments (Abankwa et al., 2008).

Pure protein-lipid driven PM protein domain formation cannot entirely explain protein domain formation. In addition to protein-lipid based interactions, protein-cytoskeleton, or protein-cell wall interactions are likely to contribute to lateral segregation of PM proteins.

PM Protein Lipid Environment █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ █ Domains Possible Lipid-TMS “Patchwork”

Figure 4.1: Patchwork model of membrane organisation. Biological membranes contain

multiple lipid species (rods) and differ in the TMS of proteins (cylinders). Compati- bility of components is indicated by similar colours. Every TMS interacts with a range of different lipids, resulting in a range of possible overlaps, as indicated by mixing of rods and cylinders. Entirely different TMS and lipids will segregate into different domains, shown with green lipids and TMS. The large number of domains can be explained by domain overlap.

Until now, we have performed the most comprehensive and detailed analysis of PM or- ganisation and domain formation within living cells. The resulting patchwork model of membrane organisation describes the PM as a highly organised and compartmentalised or- ganelle for the first time. These results introduce a new level of complexity to the research field of membrane organisation, which allows to test for protein function depending on PM domain association.

4.1.3 Factors influencing membrane domains in yeast and functional