me that this membrane perturbation is not sufficient to explain tubes formation or membrane aggregation. The basic residues in the surface must be in the right positions to induce membrane aggregation. 7 Membrane Effects of Peptides conjugated lipids the authors found that the transition is driven by membrane aggregation, which is consistent with our model in which the ahelix induces the membrane adhesion. A further conformational change could explain membrane breaking apart and tube enlargement for pAntp. Considering the tubes formation by non amphipathic peptides, and in agreement with our data, it was reported that polylysine induces the formation of membrane tubes in GUVs. Tubes formation could be a consequence of the lipid asymmetry induced by the peptides. The starting driving force for tubulation would be based on the phospholipid asymmetry between inner and outer layers which generates invagination. In model membranes, the two leaflets are symmetric and therefore the surface tension is nil and no budding should occur. Our hypothesis is that highly positively charged CPPs such as R9 or pAntp recruit lipids on the outer monolayer. This process generates a local asymmetry so that the outer layer is compressed and the inner is relatively dilated. In agreement with GUVs experiments, the model predicts that the invagination grows in the form of tubes inside the vesicles. This driving force would be stabilized further by the capability of peptides to bridge membranes. This results in the formation of thin tubes observed for R9, pAntp and RW9. As Astragalus polysaccharide proposed by Goncalves et al., membrane adhesion could also be powered by intervesicle peptide bridges. Thus, the membrane bridging property of peptides could stabilize tube growth. It is interesting to note that the R9 and RW9 tubes often remain thin Membrane Effects of Peptides whereas other peptide-induced tubes become larger more frequently. Moreover, it is difficult to explain all these effects without cooperativity between peptides. The deformation of the membrane bilayer and the tubular growth would need the clusterisation of peptides on the 16041400 membrane surface. The clusterisation would be obtained by phospholipids bridging basic residues of adjacent peptides. This 
property is apparently not related to the amphipathicity of the peptide and seems to be more related to the high density of positive charges present in the peptides. LUV’s aggregation is correlated with the GUVs tubulation. Peptides efficient for tubulation such as R9, RW9 and pAntp are also able to aggregate LUVs efficiently at the lower Peptide/Lipid ratios. In conclusion, we have established that a combination of the number of basic residues and their relative position is important for membrane perturbations. The non-amphipathic basic peptides are able to induce membrane aggregation and tube formation but do not form pores and show low cell toxicity. The amphipathic peptides are able to form tubes and bridges and permeabilize membranes. However, the properties of the amphipathic peptides are widely modulated by their length and the distribution of charged versus hydrophobic residues at the surface of the helix gives to the peptides different degree 9726632 of detergent character and then different pore forming efficiency. The model of lipid-peptide interaction illustrated in figure 8 sketches the behaviour of two archetypal peptides. pAntp, which does not adopt a perfect amphipathic helix, has an intermediate behaviour and seems to ass