Origin and evolution of the self-organizing cytoskeleton in the network of eukaryotic organelles

Abstract

The eukaryotic cytoskeleton evolved from prokaryotic cytomotive filaments. Prokaryotic filament systems show bewildering structural and dynamic complexity and, in many aspects, prefigure the self-organizing properties of the eukaryotic cytoskeleton. Here, the dynamic properties of the prokaryotic and eukaryotic cytoskeleton are compared, and how these relate to function and evolution of organellar networks is discussed. The evolution of new aspects of filament dynamics in eukaryotes, including severing and branching, and the advent of molecular motors converted the eukaryotic cytoskeleton into a self-organizing "active gel," the dynamics of which can only be described with computational models. Advances in modeling and comparative genomics hold promise of a better understanding of the evolution of the self-organizing cytoskeleton in early eukaryotes, and its role in the evolution of novel eukaryotic functions, such as amoeboid motility, mitosis, and ciliary swimming.

Figure 1.

Prokaryotic and yeast cytoskeletal-organellar network. Cytoskeletal-organellar network of (A) prokaryotes, and (B) yeast. The nodes correspond to gene names or cytological terms.

Figure 2.

Human cytoskeletal-organellar network. The nodes correspond to gene names or cytological terms.

Figure 3.

Cluster analysis of actin- and tubulin-like proteins. Sequence-similarity-based clustering was performed on (A-C) prokaryotic and eukaryotic actin-like proteins, and (D-F) prokaryotic and eukaryotic tubulin-like proteins. In both cases, an exhaustive, 90% nonredundant set of Uniprot is shown. The clusters were colored to reflect domain-wide (A, D), or eukaryote-wide (C, F) phyletic distribution. The BLASTP connections of (B) crenactins, and (E) artubulins were shown, with hits of different P-value cutoffs shown in different hues of red.

Figure 4.

Dynamic properties of the cytoskeleton. Dynamic properties and self-organized patterns of the prokaryotic (A-F), and eukaryotic (G-K) cytoskeleton. (A) Filament nucleation by a dimeric nucleus, (B) dynamic instability, (C) filament capping, (D) treadmilling, (E) bipolar growth of antiparallel filaments, and (F) higher-order structures, such as filament pairs, asters, meshes, sheets. Eukaryotes, in addition, display (G) filament branching, (H) dynamic overlap of antiparallel filaments, (I) spindle and asters, (J) filament severing, and (K) actin networks, axoneme, and basal bodies.

Figure 5.

Evolutionary scenario for the origin of processive kinesin and myosin motors. Kinesin and myosin have a common origin and evolved from a GTPase switch. In the first stage, the NTPase is bound to the filament in a nucleotide-dependent manner via a short motif connected to the NTPase domain. The NTPase was engaged in other interactions (e.g., membrane binding), and recruited the filament to an organelle. In the next step, the mechanical elements evolve that can perform one mechanical cycle following the nucleotide cycle. Motion and dissociation are both coupled to the nucleotide cycle and are transduced via a relay helix that is conserved between kinesin and myosin. This nonprocessive motor can now exert force on the bound organelle (e.g., a vesicle). The clustering of several of these motors can move organelles. Monomeric motors may have been nonprocessive (Berliner et al. 1995), or may have used biased one-dimensional diffusion for processivity (Okada and Hirokawa 1999). Myosin and kinesin probably diverged at such a stage by the acquisition of a novel filament-binding site and engagement with the second filament type (the direction is unclear). It is unlikely that the common ancestor of kinesin and myosin had a binding surface for both actin and tubulin filaments. Motor dimerization may have evolved to increase the probability of repeated engagement with the filaments. For processivity, the dimensions of the linker had to match the spacing of the accessible binding sites on the filament (80 Å for microtubules, 360 Å for actin filaments). This allowed the filament-dependent coupling of the nucleotide cycles on the two motor heads (the “mechanically controlled access” model) (Vale and Milligan 2000).