The 2013 EMBO meeting in Amsterdam opened on Saturday night with a thought-provoking plenary session on cellular organization. Several of the talks focused on the emerging idea that both cellular membranes and the cytoplasm can be compartmentalized by the physical process of phase separation.
One of the original examples of this phenomenon is, of course, the idea of lipid rafts in the plasma membrane. Rafts form when sphingolipids and cholesterol (which form a liquid-ordered bilayer) separate out from phospholipids in the liquid-disordered phase. By recruiting specific membrane proteins, this compartmentalization of the membrane can promote numerous processes such as cell signaling pathways.
The pioneer of this idea, Kai Simons, gave the opening lecture of the session and discussed his recent efforts (including this 2012 JCB paper) to prove the existence of lipid rafts in vivo, a difficult proposition give the tiny size and transient nature of rafts. He also described his recent finding that bacteria – which lack sterols – may instead use a class of molecules called hopanoids to form regions of liquid-ordered membrane and phase separation. In Simons’ view, the plasma membrane is “poised for phase separation” and compartmentalization, but emerging data suggests that the cytoplasm may be similarly poised.
Tony Hyman (one of the conference chairs) pointed out in his talk that many cytoplasmic reactions are compartmentalized without the aid of membranes. (For example, the nucleolus contains over 100 proteins involved in making ribosomes). Hyman began is presentation by describing the behavior of cytoplasmic RNA and protein particles called P granules. In 2009, Hyman and colleagues showed that, in C. elegans embryos, P granules behave like liquid droplets within the cytoplasm. (Picture a separated vinaigrette in which drops of vinegar have separated from the oil). Specifically, P granules seem to be colloidal liquids formed by multivalent, weak interactions between the P granule components. P granules preferentially form at the posterior of C. elegans embryos because the transition to the liquid phase is favored at this end of the embryo. And, although the P granule components can freely diffuse within the granule itself, their diffusion across the phase boundary into the aqueous solution phase of the cytoplasm is limited. This helps the P granule components stay together. More recently, Hyman and colleagues have shown that nucleoli behave like liquids as well.
In the following talk, Michael Rosen extended the idea of phase separation to two dimensional structures that form on membranes, such as the network of nephrin, NCK, and N-WASP that induces actin polymerization at the plasma membrane. Again, these components form a large number of multivalent interactions (boosted by phosphorylation of the nephrin receptor), resulting in a transition to the liquid phase and separation from the surrounding aqueous phase. This separation helps activate the Arp2/3 complex to initiate actin polymerization.
Rosen and Hyman think that similar principles could apply to many cellular structures. It’s certainly an interesting and important way to think about cellular function but, as the speakers themselves pointed out, it isn’t necessarily a new way to look at things. In the first half of the twentieth century – before molecular biology allowed researchers to identify and manipulate individual genes and proteins – scientists were more concerned with the physicochemical processes that drive cell function. It seems that one of those processes, particularly with respect to cellular organization, may be phase transitions and phase separation.