Abstract
Neurons are highly polarized cells, typically with a long axon and relatively short dendrites. A wealth of recent data has identified a number of signaling molecules that are involved in neuronal polarization. Kinesin superfamily proteins (KIFs) contribute to the establishment and maintenance of neuronal polarity by selectively transporting various proteins and vesicles to either the axon or dendrites. Now evidence is emerging that KIFs also play an important role in axonal formation, the initial event of neuronal polarization. In particular, KIF13B transports phosphatidylinositol (3,4,5)-trisphosphate, which, based on current hypotheses, is one of the most upstream molecules in the intracellular signaling cascades involved in axonal formation.
Neurons are highly polarized cells. Many are characterized by a single long axon and multiple shorter dendrites. A number of signaling molecules contribute to neuronal polarization. Kinesin superfamily proteins (KIFs) participate in the establishment and maintenance of neuronal polarity by selectively transporting various proteins and vesicles to either the axon or dendrites (1–3). KIFs are a class of microtubule-based mechanochemical enzymes with a conserved motor domain for adenosine triphosphate (ATP)–dependent motility, a stalk domain, and various tail domains for cargo binding. KIFs are classified into 14 subclasses (kinesin-1 to kinesin-14) based on the primary structure around the motor domain. Each subclass contains several members: For example, kinesin-1 (conventional kinesin) consists of KIF5A, 5B, and 5C in mammals. KIFs may also play an important role in axonal formation, the initial event of neuronal polarization. Specifically, KIF13B transports one of the first polarity signals proposed to contribute to axonal formation, phosphatidylinositol (3,4,5)-trisphosphate (PIP3) (4).
The process of neuronal polarization has been studied extensively in cultured dissociated rodent hippocampal neurons (5). After being plated, neurons extend several undifferentiated neurites about 20 μm long (stages 1 to 2). Then one process begins to grow rapidly and becomes the axon, whereas the others stay short and become dendrites (stage 3). The axon and dendrites gradually mature, acquiring their specific characteristics, such as pre- and postsynaptic structures (stages 4 to 5). There are three major questions regarding the mechanisms of neuronal polarization (Fig. 1). The first question is, what are the signals and the molecular mechanisms that allow one neurite to begin to grow rapidly at stage 3 (axonal formation)? Many of the current studies focus on this question (6, 7). The underlying idea is that the accumulation or activation of signaling molecules at the tip of one neurite initiates a signaling cascade to alter local cytoskeletal dynamics in the growth cone, which is a specialized region of rapid axonal growth (Fig. 1A). The second question is, why does only one axon grow? Or, in other words, how is the growth of the other neurites inhibited so that they stay short? Although the first and second questions together address axonal specification, the second question is currently far less studied. The "stay short" signal may be a retrograde signal from the axon that is then transmitted anterogradely to the future dendrites (Fig. 1B). Thus, the process may potentially involve microtubule-based transport, which is open to future research. The third question is, how does the subsequent maturation of polarity occur? This question encompasses how axonal and dendritic proteins are selectively transported to axons and dendrites (polarized transport) (Fig. 1C).
Roles of KIFs in neuronal polarization. (A) Axonal formation (hippocampal neuron transition from stage 2 to 3). One neurite among several undifferentiated neurites begins to grow rapidly because of the accumulation of signaling molecules, which leads to the local activation of cytoskeletal dynamics for axonal growth at the tip. PIP3, which is proposed to be one of the most upstream intracellular signaling molecules in the cascade, is either produced locally by receptor-mediated PI3K activation or transported to the tip by KIF13B. Other signaling molecules reported to be transported by KIFs are indicated in the figure. (B) Axonal specification (stage 3). Once an axon grows rapidly, other neurites remain short. Some inhibitory signals may be transported retrogradely from the axon and then anterogradely to the tips of the dendrites. KIFs and dynein might be involved in this step. (C) Maturation of polarity (stages 4 to 5). Presynaptic proteins are delivered to the axon and postsynaptic proteins are delivered to the dendrites in a polarized manner by various KIFs. In each panel, the relevant KIF isoform (when known) is listed next to the cargo it transports.
KIFs and dyneins are microtubule-based motor proteins that transport various membrane organelles, protein complexes, and RNA within neurons (1–3). Many proteins are transported selectively to the axon or dendrites by means of KIFs. For example, synaptic vesicle precursors are transported to axons by KIF1A (of the kinesin-3 subfamily) (8), and the glutamate N-methyl-d-aspartate (NMDA) receptor subunit NR2B is transported to dendrites by KIF17 (of the kinesin-2 subfamily) (9). The mechanisms of selective transport may involve interactions between the motor head and microtubule and between the motor tail and cargo. As cargo delivery molecules, KIFs play an important role in the maturation and maintenance of neuronal polarity. Recently, the transport of signaling molecules by KIFs in the initial stages of axonal formation has been reported. Two proteins involved in the establishment of polarity are delivered to axons by KIFs: Par3 is transported to axons by KIF3A (kinesin-2) (10, 11), and CRMP-2 (collapsin response mediator protein 2) is transported to axons by KIF5 (kinesin-1) (12, 13). However, these molecules are relatively downstream of the signaling cascades currently thought to be involved in axonal formation [for review, see (5, 6)]. A complex of Par3, Par6, and atypical protein kinase C (aPKC) was initially identified as crucial for the anterior-posterior polarity of the single-cell embryo (14). In neurons, the Par3/Par6/aPKC complex accumulates at the tip of the axon: This polarized localization is important for axonal specification (15). Although analogy with the worm embryo cell and accumulating cell biological evidence supports the Par3/Par6/aPKC paradigm for neuronal polarization, this model has been challenged by in vivo knockout studies in Drosophila (16).
Phosphatidylinositol 3-kinase (PI3K) activity, which produces PIP3, is essential for the proper localization of Par3; thus, PIP3 is considered to be upstream of Par3/Par6/aPKC signaling. CRMP2 modulates tubulin dynamics and is downstream of glycogen synthase kinase 3β (GSK-3β). GSK-3β, which is important for the polarization of migrating astrocytes, is downstream of PI3K and the kinase Akt in hippocampal neurons. However, hippocampal neurons from GSK-3β knockout mice developed normally, although a possibility remains that GSK-3α may compensate (17). The role of PIP3 in cell polarity was established in neutrophils and the slime mold Dictyostelium: It is upstream of the guanosine triphosphatases (GTPases) Rac and Cdc42. PIP3 accumulates at the tips of future axons and induces axonal specification (15). As summarized here, the study of signaling cascades for axonal formation is still in progress and the hypotheses are not fixed yet, but collectively, these and other studies suggest that PIP3 is upstream of GSK-3β, the Par3/Par6/aPKC complex, CRMP-2, and the Rho and Rac GTPases, which in turn may regulate microtubule and actin dynamics for rapid axonal growth at the growth cone.
Horiguchi et al. (4) reported that KIF13B transports PIP3 vesicles to the axon and regulates axonal formation. KIF13B (also known as GAKIN for guanylate kinase–associated kinesin) belongs to the kinesin-3 subfamily, the members of which are involved in vesicular transport. Previously, KIF13B was found to be a binding partner of human discs large tumor suppressor protein (hDlg); the fly homolog is also known to be important in polarity determination (18). Like other members of the kinesin-3 subfamily, KIF13B contains a forkhead-associated (FHA) domain, which mediates interactions with phopshorylated proteins. Horiguchi and colleagues found by yeast two-hybrid experiments that PIP3BP (PIP3 binding protein, also known as centaurin-α) binds to the FHA domain of KIF13B. Interestingly, PIP3BP does not bind the FHA domain of KIF13A, in spite of 70% amino acid identity between KIF13A and KIF13B. KIF13B transported PIP3-containing liposomes in vitro and this required PIP3BP. PIP3 and PIP3BP accumulated at the axonal tips of hippocampal neurons. Overexpression of wild-type and dominant-negative KIF13B both reduced the polarity of hippocampal neurons. These results do not preclude a role of the local production of PIP3 by PI3K at the growth cone, but they suggest an essential role of the transport of PIP3-containing vesicles by KIF13B for axonal formation.
Accumulation of PIP3 in one neurite initiates the signaling cascades involved in axonal formation. PIP3 is transported to neurite tips by KIF13B. Does KIF13B preferentially transport PIP3 vesicles to the process destined to become the axon? This is a critical question in the study of neuronal polarity, because it addresses the question of how asymmetry is initiated in developing neurons. This issue remains unsolved. One scenario is that PIP3 transport to neurites is unbiased and that local degradation in processes destined to be dendrites or that retention or production of PIP3 at growth cones results in the asymmetric distribution of PIP3. In this case, transport by motor proteins does not play a critical role in axonal specification; it simply provides the materials for signaling equally to each neurite.
Another scenario is that PIP3 vesicles are preferentially transported to the process that will become the axon by KIF13B, and as a consequence, this promotes the specification of the axon. Knocking out KIF13B and testing its role in PIP3 transport and axonal formation may help clarify these two potential mechanisms. Some KIFs do appear to promote axonal specification. When not bound to cargo, the tail domain of full-length KIF5 (of the kinesin-1subfamily) inhibits its motility, and cytosolic KIF5 distributes throughout cells presumably by simple diffusion (19, 20). By contrast, when the tail domain is deleted, KIF5 is constitutively active and accumulates at the tip of the axon but not in dendrites (20). Its accumulation in the axon starts as early as stage 2, when the future axon is not yet any longer than the other neurites (21). These results suggest that some KIFs may identify processes destined to be axons earlier than previously identified signaling molecules and that these KIFs may preferentially transport signaling molecules to those future axons. It will be interesting to test whether KIF13B has such a property and whether PIP3 is preferentially transported to processes destined to become axons.
Exciting new findings seem to represent a convergence of the results of signaling studies and motor protein studies into the mechanisms of neuronal polarity formation. However, some key questions remain. Are motor proteins delivering information that provides the initiating signal that specifies whether a process will become an axon or a dendrite? Or are they delivering materials and signals uniformly throughout the neuron at the earliest stages, and axonal specification occurs when signals are amplified at the process destined to be the axon? What are the signals that keep the dendrites short? Does the axon signal back to the other processes to prevent them from also adopting an axonal fate? Analysis of mice in which specific KIFs are knocked out or their function reduced should help to address these questions.
