membrane, whereas the side chain of positively charged (Fig. 2A, blue residues) amino acids interacts with the polar head of negatively charged lipids (Fig. 5B). It is not completely understood how an amphipathic helix would bind negatively curved membranes (Drin and Antonny, 2010). As cardiolipin has been shown to be enriched in membranes interacting with MurG (van den Brink-van der Laan et al., 2003), it is possible that cardiolipin may attract MurG to the forespore through direct interactions with its amphipathic helix. But cardiolipin (diphosphatidylglycerol) is not chemically different than phosphatidylglycerol, one of the main components of bacterial membranes, so it is unlikely that MurG mutant interacts specifically with cardiolipin. We favor a second possibility where the insertion of cardiolipin into the zones of negative curvature in the forespore membrane (Fig. 5A) may induce loose lipid packing in the bilayer which could facilitate the insertion and folding of MurG amphipathic helix into such zones. The increase in fluidity of the bilayer is generated by the separation of lipid polar heads, a phenomenon favored by the presence of conical lipids such as cardiolipin (Antonny, 2006; Drin and Antonny, 2010). Once inserted and its amphipathic helix properly folded, MurG could diffuse around the forespore membrane, explaining the observa- tion that MurG is found throughout the forespore membrane. Point mutations in MurG disrupting its amphipathic helix may lead to the disruption of the folding process of its N-terminal peptide into an amphipathic helix and loss of MurG forespore localization.