Caspase-3, Shears for Synapse Pruning

Abstract

Motor neurons regulate neuromuscular junction formation by using agrin to stimulate acetylcholine receptor clustering and using acetylcholine to disperse unnecessary receptor clusters on muscle fibers. Wang et al. now report in this issue of Developmental Cella critical role for caspase-3 in intracellular mechanisms of acetylcholine-induced dispersal.

A synapse is a communicating contact between two neurons or a neuron and its target cell. It is fundamental to how we sense, think, and respond to the environment. Vast numbers of synapses are formed during embryonic development, presumably as a protective mechanism to ensure that all neurons are connected. Consequently, neonatal animals contain redundant, functionally immature synapses. To establish precise connectivity in the nervous system, during postnatal period, excessive or in appropriate synapses are eliminated. This coincides with maturation of maintained synapses and occasional formation of new synapses. Moreover, the addition and disassembly of synapses is involved in activity-dependent plasticity in adult animals. Unlike synaptogenesis that has been studied extensively, molecular mechanisms of synapse elimination are largely unknown. In this issue of Developmental Cell, Wang and colleagues (2014) unravel an intracellular mechanism of synapse elimination.

Wang et al. (2014) studied the development of the neuromuscular junction (NMJ), a peripheral synapse between motor neurons and skeletal muscle fibers that uses acetylcholine (ACh) as neurotransmitter. The NMJ has structural and functional features characteristic of a chemical synapse. Being large and easily accessible experimentally (compared with central nervous system (CNS) synapses), the NMJ has contributed greatly to the understanding of general principles of synaptogenesis. Unlike a neuron in the brain that receives inputs from many synapses for information integration, a muscle fiber has no need to integrate signals but faithfully executes orders from a single motor neuron to ensure efficient and precise muscle contraction. Hence, in adult mammals, one muscle fiber has only one NMJ.

In mice, prior to the arrival of motor axons, diaphragm muscle fibers form primitive, small AChR clusters (Figure 1A). These aneuralclusters are dotted in the middle region of muscle fibers and outline a poorly-defined central band - a phenomenon called prepatterning. Innervation by motor axons will have to carry out two tasks: first, to disperse the aneural AChR clusters and second, to induce large AChR clusters, some of which may be converted from aneural ones, to form premature NMJs. At birth, the premature NMJs appear like plaques and often are convergently innervated by multiple neurons. In ensuing two weeks, the number of motor axons innervating each muscle fiber is reduced to one and the NMJs appear as characteristic “pretzels” in shape (Figure 1A)(Wu et al., 2010).

Figure 1. Different stages of NMJ development and mechanisms of AChR cluster formation and dispersal.

(A.) Developing NMJ. A neural AChR clusters are dotted in the middle region of muscle fibers prior to the arrival of motor axons (left). Immature NMJs could be innervated by multiple axons (middle) where mature NMJs each receive input from one motor neuron (right). (B.) Proposed signaling pathways activated by motor neurons to disperse a neural AChR clusters and to induce AChR clusters.

Motor neurons induce AChR clusters by releasing agrin, which binds to LRP4, a member of the low-density lipoprotein receptor family, and activates the receptor tyrosine kinase MuSK(Wu et al., 2010). Downstream of the agrin/LRP4/MuSK pathway, Dok7 is believed to promote MuSK activity and rapsyn is believed to bridge the AChR with the cytoskeleton(Wu et al., 2010). Much less is known how extra AChR clusters or synapses are eliminated during development. In mutant mice lacking choline acetyltransferase (ChAT), the ACh biosynthetic enzyme, AChR clusters are increased in number and populate a broader area of muscle (Brandon et al., 2003; Misgeld et al., 2002), suggesting that motor neurons use ACh to disperse aneural AChR clusters and restrict induced clusters within the central region. The effect of AChis thought to be mediated by cyclin-dependent kinase 5 (Cdk5) (Fu et al., 2005; Lin et al., 2005). There are two models explaining how muscle activity stimulates Cdk5 (Figure 1B). In one, rapsynis dissociated from the protease calpain to increase its protease activity, which leads to cleavage of the Cdk5 co-activator p35(Chen et al., 2007). In another, p35 is recruited to Cdk5 that is associated with the intermediate filament protein nestin(Yang et al., 2011).

Wang et al. (2014) identify a pathway that is critical in eliminating aneural AChR clusters. They investigated mechanisms of CCh, a non-hydrolysable AChR agonist that disperses AChR clusters in muscle cells(Chen et al., 2007; Lin et al., 2005) and found that it stimulates caspase-3, a protease implicated in apoptosis. To determine whether caspase-3 can be activated in a physiological setting, the authors expressed channelrhodopsin-2 in motor neurons and co-cultured them with myotubes. Laser stimulation of transfected motor neurons led to activation ofcaspase-3inareas of non-innervated AChR clusters. These experiments elegantly demonstrated that presynaptic activity of motor neurons could activate caspase-3 in muscle. Remarkably, CCh-stimulated elimination of AChR clusters was attenuated by pharmacological inhibition or genetic ablation of caspase-3 incultured muscles. In agreement, caspase-3 mutation increased the number of aneural AChR clusters in embryonic diaphragm. These observation corroborate that caspase-3 is a mediator of muscle activity to eliminate a neural AChR clusters.

How does caspase-3 work? Local activation of caspase-3 by CCh did not cause cell death, but instead regulated AChR clusters in an unexpected manner. Dishevelled (Dvl1) is an adapter protein implicated in Wnt signaling; previously Dvl1 has been shown to bridge a MuSK-Dvl1-PAK1 complex for agrin-induced AChR clustering (Luo et al., 2002). In a nice set of biochemical and cell biology experiments, Wang et al. (2014) demonstrated that CCh stimulation caused cleavage of Dvl1 and inhibition of Dvl1 cleavage prevented cluster dispersal by CCh or by motor neuron activation. Expression of caspase-3-resistant form of Dvl1 by chest viral infection increased the number of a neural AChR clusters in embryonic diaphragm. CCh also reduced the interaction between MuSK and Dvl1, and dispersed agrin-induced clusters of APC, a microtubule-regulator protein implicated in both Wnt signaling and AChR clustering (Wu et al., 2010).

APC seems to act downstream of Dvl1, because CCh-induced dispersal of APC clusters was inhibited by preventing Dvl1 cleavage. Finally, CCh-induced caspase-3 activity or cluster dispersal was inhibited by 17-AAG, an inhibitor of HSP90β, suggesting that HSP90β helps to keep caspase-3 in check, in addition to protecting rapsyn(Luo et al., 2008).

The findings of Wang et al. (2014) are provocative and reveal a role for caspase-3 in synaptogenesis. They also raise many questions. Is this mechanism involved in postnatal synapse elimination or converting multi-innervated NMJs to mature ones with single motor neuron terminal (Figure 1A)? Is it involved in sculpturing of the postsynaptic membrane for “pretzel”-like morphology? These questions could be addressed by studies of caspase-3 mutant mice because they survive to 6 months of age. The rescue effect of caspase-3 ablation in agrin mutant mice appears to be partial, suggesting existence of other effectors. What is the relationship of caspase-3 with these effectors, for example, Cdk5 and its regulators? Dvl1 and APC are mediators of Wnt signaling. Considering recent implications of Wnts in NMJ formation(Wu et al., 2010), it would be interesting to determine if caspase-3-mediated Dvl1 cleavage acts by altering Wnt pathways. Results of this paper also call for future studies to determine whether caspase-3 is involved in CNS synapse pruning.