Curves are attractive

Steps to STIMulation W hen ER calcium levels plummet, an ER membrane calcium sensor called STIM1 directly activates plasma membrane (PM) channels to top up the cell’s calcium. Jen Liou, Tobias Meyer, and colleagues (Stanford University Medical School, Stanford, CA) have now uncovered the molecular steps of STIM1’s pathway to the PM and show that those steps don’t take STIM1 very far. STIM1 sits in the ER membrane holding a calcium molecule in its luminal domain. When the ER loses calcium, STIM1 translocates along the ER membrane, ultimately aggregating at junctions between the PM and the ER. The details of these steps were unknown. Using live cell imaging of fl uorescent STIM1 fusion proteins, Liou et al. now show that, when ER calcium levels drop, STIM1’s fi rst step is to rapidly form oligomers. These oligomers then translocate and form visible aggregates. Aggregation, but not oligomerization, requires STIM1’s polybasic tail—most likely for direct binding to the PM. Formation of these aggregates was rapid, but STIM1 oligomers’ translocation speed through the ER membrane was very slow. Therefore, the distance traveled to ER–PM junctions must be short. Reports have shown that STIM1 promotes calcium infl ux in localized regions of the PM. This is thought to spatially restrict activation of calcium-sensitive targets. STIM1’s slow pace thus explains how this local action is achieved. Reference: Liou, N., et al. 2007. Proc. Natl. Acad. Sci. USA. doi:10.1073/pnas.0702866104. C u r ve s a re a t t ra c t i ve M embrane-bending proteins can attract each other when the curves they create overlap, according to a computer simulation by Benedict Reynwar, Markus Deserno, and colleagues (Max Planck Institute, Mainz, Germany). The energy required for membrane reshaping— as needed for endocytosis, vesiculation, etc.—is too great for one membranebending protein alone to achieve, so cooperation is essential. How cooperation occurs, however, has been a bit of a mystery. Part of the mystery stems from previous theoretical calculations that predicted that curves induced by a membrane-bending protein repel other proteins of its kind. Experimental systems, on the other hand, suggested that mem brane curving might be enough to aggregate the responsible proteins. Ruling out other specifi c protein–protein interactions is diffi cult, however. Now, in silico simulations by Reynwar et al. show that, as suggested by experimentation, curves can be enough to pull together membrane benders. The team created computerized proteins that induced a membrane curve geometry similar to that of a real membrane-bending protein called BAR domain. In a simulated lipid bilayer in which the proteins freely diffused, the induced deformed regions encircling the proteins soon overlapped. The proteins did not then diffuse away but instead aggregated such that the area of deformation grew, eventually leading to vesiculation. The ease with which the simulated proteins aggregated and promoted vesiculation—within just milliseconds—suggests that real cells must keep a tight leash on their benders. Reference: Reynwar, B.J., et al. 2007. Nature. 447:461–464. M EY ER / N A S