Neuronal Development: SAD Kinases Make Happy Axons

Abstract

The polarity proteins LKB1 and SAD-A/B are key regulators of axon specification in the developing cerebral cortex. Recent studies now show that this mechanism cannot be generalized to other classes of neurons: instead, SAD-A/B functions downstream of neurotrophin signaling in sensory neurons to mediate a later stage of axon development — arborization in the target field.

A key event in neuronal development is the specification of different functional domains of the neuron. Polarity proteins, including Par (partitioning-defective) proteins, and their effectors have emerged as critical factors controlling the first step in this specification — neuronal polarization [1,2]. LKB1, the mammalian homologue of Caenorhabditis elegans Par-4, plays a pivotal role in the polarization of cortical and hippocampal neurons [3,4]. SAD kinases function directly downstream of LKB1 in regulating cortical neuronal polarization [3,5]. The prototype SAD kinase, SAD-1, was first identified in C. elegans as a regulator of synaptic vesicle clustering [6]. In mammals, SAD kinases consist of two functionally redundant family members, SAD-A and SAD-B (hereafter referred to as SAD-A/B). Although the roles of LKB1 and SAD-A/B in cortical neuronal polarization have been established, considerable uncertainty remains regarding upstream regulators and downstream mediators of LKB1 and SAD kinase functions. Furthermore, whether LKB1 and SAD-A/B regulate polarization in all classes of neurons is unknown.

In a new study recently published in Neuron, Lilley et al. [7] tested whether the requirement for LKB1 and SAD kinases in axon specification can be generalized to other types of neurons. The authors found that, although these proteins are widely expressed in both the central and peripheral nervous systems, LKB1 and SAD kinases are surprisingly dispensable for axon formation in many classes of neurons outside the cortex and hippocampus. For example, genetic deletion of either LKB1 or SAD-A/B in sensory and spinal motor neurons failed to disrupt axon formation and did not interfere with axon projections to skin and muscles. Instead, the authors found that SAD-A/B functions are especially crucial for a late stage of axon development in sensory neurons that are responsive to the neurotrophin NT-3 — i.e. axonal arborization in the spinal cord and brainstem. Surprisingly, this requirement is not shared by LKB1.

Lilley et al. [7] focused on their analysis on proprioceptive sensory neurons, which transmit sensory information regarding muscle length and limb position to the central nervous system. Proprioceptive neurons are pseudounipolar with a peripheral axon contacting muscle spindles and a central axon entering the spinal cord (Figure 1). The central axon gives off collaterals that target spinocerebellar neurons, interneurons and motor neurons. Importantly, signaling mediated by NT-3 and its receptor TrkC has been identified as a critical regulator of the arborization of proprioceptive central axons in the vicinity of motor neuron pools [8,9]. In the new study, Lilley et al. [7] demonstrated that deletion of SAD-A/B specifically disrupted proprioceptive axon terminal arborization in the ventral spinal cord and thus phenocopied the effects of NT-3 deletion. This finding suggested that NT-3/TrkC is a key upstream regulator of SAD kinases. The authors therefore pursued the linkage between NT-3/TrkC signaling and SAD-A/B functions.

Figure 1.

Regulation of SAD kinases downstream of NT-3/TrkC signaling. (A) In the absence of NT-3, SAD kinases are inactivated by phosphorylation of their carboxyterminal inhibitory domain (CTD) via CDK5 and other unknown kinases. Phosphorylation of this domain inhibits the activation of SAD kinases by upstream regulators, such as LKB1. (B) NT-3/TrkC stimulation activates a number of canonical signaling pathways. Over a time frame of minutes, PLCγ activation and intracellular release of Ca²⁺ triggers dephosphorylation of the SAD carboxy-terminal domain. Dephosphorylation of this domain enables phosphorylation of T175/T187 in the activation loop by LKB1 and other kinases and results in the activation of SAD kinases. (C) Long duration NT-3/TrkC signaling through the Raf–MEK–ERK cascade prevents degradation of SAD kinase, leading to higher protein levels. Both short- and long-term mechanisms mediate arborization of proprioceptive axons in the vicinity of motor pools.

The authors first showed that SAD kinases are required for morphological regulation of sensory neurons by NT-3 in vitro. They demonstrated that SAD-A/B deletion drastically reduced NT-3-induced axon outgrowth from dorsal root ganglion explants. The effects were surprisingly specific because SAD-A/B deletions had little effect on axon growth from nerve growth factor (NGF)-responsive neurons. To further understand the underlying mechanisms, the authors showed that SAD-A/B deletion did not disrupt proprioceptive innervation of muscle, TrkC expression, or retrograde signaling triggered by NT-3. They further showed that SAD-A/B function is cell autonomous. These results established that SAD kinases are essential intrinsic mediators downstream of NT-3/TrkC signaling that regulate proprioceptive central terminal arborization.

Lilley et al. [7] next investigated the molecular mechanisms for SAD-kinase regulation downstream of TrkC. The authors found that the function of SAD-A/B can be modulated by NT-3/TrkC signaling over multiple time scales. Long-term (over hours) NT-3 stimulation elevated SAD-A/B protein levels by preventing proteasome-mediated degradation. The authors noted that both SAD-A and SAD-B contain a consensus D-box domain, which is known to target proteins for ubiquitination mediated by the anaphase promoting complex/cyclosome (APC/C) E3 ubiquitin ligase. Mutating the D-box domain in SAD-A resulted in stabilization of SAD-A protein. The authors demonstrated that the regulation of SAD-A/B protein levels downstream of NT-3/TrkC was mediated through the Raf–MEK–ERK kinase pathway (Figure 1). Specific inhibition and activation of the Raf–MEK–ERK signaling cascade was shown to diminish and increase SAD-A/B protein levels, respectively. Presumably ERK signaling mediates the interaction between APC/C and the SAD-A/B D-box domain. SAD kinase regulation by the Raf–MEK–ERK pathway was in line with a prior study showing proprioceptive central arborization was dependent on Raf kinase function [10]. Unexpectedly, mTOR signaling, which had been shown to regulate SAD kinase translation [11], did not appear to be a major regulator of SAD-A/B levels.

In contrast to long-term NT-3 stimulation, transient NT-3 stimulation on a scale of minutes led to SAD-A/B activation via phosphorylation on a threonine residue (SAD-A T175; SAD-B T187) within the activation loop. Surprisingly, two kinases that phosphorylate the activation loop, LKB1 and TAK1, were not essential for the effects of SAD kinases on axonal branching. These results raise the possibility that in dorsal root ganglion neurons multiple kinases are involved in SAD-A/B activation. Further investigation revealed a critical role for phosphorylation of the carboxy-terminal domain for SAD-A/B activation (Figure 1). The authors found that SAD-A is phosphorylated on 18 sites in this domain, among which are 16 proline-directed serine/threonine sites. The authors showed that mutated SAD-A (SAD-A^18A) with all 18 phosphorylation sites converted to alanine was activated to a greater extent in response to NT-3 stimulation and exhibited more enzymatic activity than wild-type SAD-A, indicating that phosphorylation of the SAD-A carboxy-terminal domain inhibited SAD-A activation. In addition, overexpression of SAD-A^18A induces axon terminal branching, recapitulating the effects of NT-3 treatment [7,12]. These data lead to the suggestion that NT-3/TrkC signaling regulates the dephosphorylation of the carboxy-terminal domain of SAD-A. The authors then showed that phosphorylation of this domain of SAD-A is catalyzed by CDK5. This inhibitory phosphorylation is relieved by elevating intracellular Ca²⁺ via phospholipase Cγ (PLCγ) downstream of NT-3/TrkC. Either inhibition of CDK5 or elevation of intracellular Ca²⁺ effectively induced SAD-A/B activation in cultured dorsal root ganglion neurons. Overall the work suggests that in the absence of NT-3, the carboxy-terminal domain of SAD-A is phosphorylated by CDK5 and possibly other kinases. Upon TrkC activation by NT-3, PLCγ/Ca²⁺ signaling rapidly reduced the extent of phosphorylation of this domain of SAD (Figure 1). Presumably the activity of LKB1 and related kinases are sufficient to phosphorylate T175/T187 and activate SAD-A/B. Over longer time periods of NT-3 stimulation, Raf–MEK–ERK activity downstream of TrkC elevates SAD-A/B protein levels by inhibiting degradation (Figure 1).

One surprising finding is the specificity of SAD kinase functions for NT-3/TrkC. The Raf–MEK–ERK and PLCγ/Ca²⁺ signaling pathways regulating SAD-A/B functions are core signaling cascades downstream of all three major neurotrophin receptors — TrkA, TrkB and TrkC. However, disruption of SAD-A/B did not noticeably affect the development of central axons that are responsive to the TrkA ligand, NGF. As there are differences in the cytoplasmic sequences of TrkA and TrkC, presumably recruitment of distinct signaling molecules differentially affects SAD-A/B regulation. A possibly related phenomenon is that, in vitro, NT-3 stimulates predominantly branching of TrkC-expressing neurons, whereas NGF stimulates predominantly elongation of TrkA-expressing neurons [12].

Another unresolved issue is exactly how the reported results relate to the responses of proprioceptive axons in vivo. Regulation of central arbors of proprioceptive neurons by peripheral sources of NT-3 has been established [8,13]. The results reported by Lilley et al. [7] suggest that long-term regulation of SAD-A/B protein levels primarily by peripheral NT-3 may mediate these established effects. The more rapid responses would presumably be related to the secretion of NT-3 by motor neurons in the spinal cord that might attract axons or mediate branching in the vicinity of motor neuron pools. Interestingly, a recent study deleting NT-3 specifically in motor neurons showed abnormalities of central projections, consistent with a role for centrally derived NT-3 in shaping the projection [9].

The discovery of an important role of SAD kinases in mediating axon branching, the demonstration of a specific link between SAD-A/B and NT-3/TrkC signaling, and the identification of novel and intricate mechanisms underlying SAD activation represent a major advance in understanding how neurons generate morphological responses to extracellular cues. In another major advance, Courchet et al. [14] recently demonstrated that terminal arborization of callosally projecting cortical neurons involves LKB1 signaling mainly through the kinase NUAK1, but not SAD-A/B. NUAK1 is required for mitochondria immobilization, which is essential for distal axonal branch formation. In contrast, sensory axon arborization requires SAD-A/B and sensory axon development is apparently independent of LKB1 at earlier developmental stages [7]. It is probably not surprising that varying neuronal classes responding to distinct upstream cues and with very different molecular characteristics of target fields would employ specific molecular mechanisms to mediate target field branching. As molecular mechanisms that underlie additional examples of target field arborization are uncovered, presumably a set of general principles will emerge.

Finally, where are things headed in the future for SAD kinases? Interestingly, the prototype SAD was discovered as a mediator of synaptic vesicle clustering in C. elegans [6]. Further, work in mammals has already demonstrated that at least one SAD kinase isoform, SAD-B, is localized to presynaptic terminals, where it associates with synaptic vesicles and regulates neurotransmitter release [15]. It is important to emphasize that during and after axon arborization in target fields, synaptic vesicle clustering in distal axons and synapse formation are the next steps in axonal development. In a sort of ‘preview of coming attractions’, Lilley et al. [7] state that they have found in unpublished work that SAD-A/B deletions affect maturation of synapses in many classes of neurons. Thus, we can look forward to an elegant dissection of the regulation and functions of SAD-A/B in synapse formation along the lines we have seen in the work described here.