Computational simulations reveal that Abl activity controls cohesiveness of actin networks in growth cones

Abstract

How do single-molecule dynamics produce multimicron-scale changes in actin organization in an extending axon? Comparison of computational simulations to in vivo data suggests that Abl kinase and Arp2/3 expand actomyosin networks by fragmenting them into multiple domains, thus toggling the axon between states of local versus global internal connectivity.

A new view of axon growth and guidance grounded in the stochastic dynamics of actin networks

“…The length of the actin peak in the growth cone fluctuates around a median value of about 12–15 µm long, and it is positioned near the front of the axon, associated with the filopodial-rich domain, but is generally not at the very tip of the axon. Note that while, for simplicity, we refer to this concentration of actin as an ‘actin peak’, we presume that it is in fact a contractile structure rich in myosin II as well as actin, in addition to containing a host of other typical actin-associated proteins [ 52 , 53 ]. Between any two timepoints, the actin peak may move forwards, backwards or not move at all.…”

“…The ‘growth cone’ was then allowed to undergo dynamics for 2000 s (long enough to achieve steady state), incorporating both chemical events, such as polymerization, branching, bundling and contractility, as well as relaxation of physical strain every 25 ms. Kinetic constants for the simulations were taken from published values for the relevant dynamic processes, and concentration ranges of components were selected based on a combination of theoretical considerations and empirical tests [ 52 , 94 ]. While extraordinarily rich in detail, the resulting findings were remarkably simple in terms of the principles that they revealed.…”

“…Like many stochastic networks, actin self-organizes into a ‘small-worlds’ structure. That is, while most individual actin filaments in the network have cross-links to at least one other filament (provided in the form of Arp2/3-linked branches, or by simultaneous binding to the same two-headed myosin motor or the same molecule of α-actinin cross-linker), the network tends to develop such that there are local regions (subnetworks) within which those cross-links are quite dense, but that separate subnetworks are connected to one another by only a much sparser set of cross-links ( figure 5 ) [ 52 ]. The spatial scale of these actin networks and subnetworks, however, depends critically on the distribution of lengths of the individual actin filaments (figures 5 and 6 ).…”

“…The community structure of the actin network, and its striking dependence on filament length distribution, has another consequence with potentially far-reaching functional implications. The larger, more coherent actin networks that form in the presence of a modest proportion of long filaments are also capable of transmitting force over long distances across those networks, whereas the fragmented, mechanically disconnected networks generated with short filaments are not capable of force transmission across a multi-micrometre scale [ 52 ]. This provides a physical basis for the observation that increase of Abl activity expands the actin network beyond the range at which mechanical information (i.e.…”

“…Different from both of these, the probabilistic actin reorganization model described here consigns force to an altogether subsidiary role and instead ascribes mechanism almost entirely to the spatial distribution of chemical signalling. Indeed, periodic dissipation of accumulated forces is an explicit step in the computational simulations of actin dynamics in TSM1 [ 52 , 93 , 94 ]. Here again, however, it may be that the distinction between these growth cone modes is one of degree rather than being categorical, with these modes defining a continuum of mechanisms.…”