To address point 1/, we reasoned that cells should have particular mechanisms to distinguish actin filaments belonging to different structures. Because actin is a highly conserved protein during evolution, we worked on two possible strategies to assign actin filaments an identity within actin networks. Based on a comparative study of the eukaryotic genome, we hypothesized that some organisms might use an additional family of proteins, tropomyosins, that can wrap around actin filaments to give actin filaments specific biochemical properties (a mechanism possible in yeast and animals that have few actin isoforms but many tropomyosins), while other organisms could use different, though similar, actin isoforms to assemble different types of filaments (a possible mechanism in amoebae and plants that have multiple actin isoforms but no tropomyosin).
A first work therefore addressed the question of how tropomyosins could drive actin-binding protein segregation in cells. Our results indicate that their ability to bind cooperatively and with a high stoichiometry to actin filaments leads them favor the longest actin filaments, such as those found at the level of the linear networks, and avoid the short actin filaments present in branched networks. Consequently, all actin-binding proteins binding cooperatively with tropomyosin will follow to the same population of actin filaments while proteins binding competitively with tropomyosin will be relocalized to other networks (Antkowiak, Le Goff, et al., to be submitted).
A second work focused on the question of how different actin isoforms, despite being so similar (most actins in eucaryotes are 85%+ identical), can be nevertheless assemble into different types of actin networks. We found that minor changes in the actin molecule were sufficient to inhibit their interaction with one or several actin regulators. Because most regulators are implicated in the assembly of specific types of actin networks, an actin isoform which does not interact with one of these factors will automatically be redistributed to other actin networks. Eventually, we were also able to generate yeast cells which, instead of using their unique wild-type actin to assemble branched and linear actin networks, could use two actins, one specialized in branched-network assembly and the other specialized in linear actin assembly. Comparison of this strain with a wild-type strain showed that a major difference of the two systems is for one to be able to redistribute a single actin between multiple networks according to the needs of the cell, while the other assembles actin networks more independent from each other.
To address point 2/, we developed two different experimental setups.
To characterize the relationship between ABP composition of an actin network and its response to mechanical deformations, we combined the use of paramagnetic cylinders with actin assembly from yeast extracts. Paramagnetic cylinders can be used in a magnetic field as high throughput force sensors, while yeast extract allow for testing actin networks assembled from a variety of yeast strains. Together, these tools enabled us to define precisely the mechanical properties of these networks, the importance of various families of crosslinkers, as well as the importance of these properties for force generation in the cell.
Finally, this project also allowed the development of new tools to understand the molecular mechanisms of actin dynamics. We discovered that a family of fluorescent nucleotides can bind actin without altering its polymerization properties and its binding to its regulatory proteins. This discovery opens the door to many future projects interested not only in actin dynamics, but also in how actin networks consume energy within the cell.
