PERSPECT IVES Muscle memory

In the movie ‘Burn after reading’, George Clooney sings the praises of ‘muscle memory’ – the concept that muscles have excitable circuitry capable of directing complex behaviour on their own. Sure enough, when the moment comes his muscles remember to rapidly draw a gun and shoot a man who inadvertently surprises him. Normally, Coen Brothers movies are not a reliable source for information about neuroscience, but in this case they touched upon an interesting controversy. Can muscles really drive behaviour independent of the nervous system? Alternating muscle contractions in locomotion are thought to be generated by a central pattern generator – a neuronal circuit (Grillner, 2006). But muscles can fire action potentials and produce rhythmic contractions without neuronal input – no one doubts that the heart keeps its own counsel. Is it possible that body muscles could direct locomotory behaviour on their own? In a recent issue of The Journal of Physiology, Zhao-Wen Wang’s group (Liu et al. 2011) suggests that it is possible. Nematodes move by propagating a rhythmic sinusoidal wave consisting of alternating dorsal and ventral contractions along their body. These rhythmic waves propel them along surfaces or in liquid. Early behavioural studies on the large parasitic nematode Ascaris lumbricoides demonstrated that the initiation and propagation of contractile waves continue after the head and the tail are excised (Crofton, 1971). This result indicates that the head and tail ganglia are not required for coordinated locomotion – worms do not need a brain to move. Do they need a nervous system at all? Weisblat and Russell observed thatAscaris bodymuscles generate spontaneous voltage spikes independent of the nervous system (Weisblat & Russell, 1976). Moreover, Walrond and Stretton demonstrated that voltage spikes in the muscles propagate along the body via gap junctions between adjacent muscles cells (Walrond & Stretton, 1985). These early studies suggest that muscles themselves are capable of generating and propagating oscillatory contractions down the length of the worm. C. elegans is one hundred times smaller than Ascaris but the number of motoneurons is identical. Similarly, C. elegans body-wall muscles are coupled by gap junctions and genetic disruption of these gap junctions results in an extremely uncoordinated worm (Liu et al. 2006). Mutants with severe defects in neurotransmission or lacking acetylcholine receptors are still capable of moving, although very slowly (Richmond et al. 2001; Francis et al. 2005). Thus, there is reason to believe that C. elegans muscles may function in a similar fashion to Ascaris. On the other hand, evidence that the motoneurons provide instructional input into muscle contractions is overwhelming: mutants lacking innervations from the motoneurons, as opposed to defects in neurotransmission, are essentially paralysed. Moreover, the inputs from the motoneurons and the response in the muscles are thought to be graded rather than coded by action potentials for several reasons. First, synaptic output from motoneurons is graded; the rate of tonic miniature currents is proportional to neuronal membrane potential (Liu et al. 2009). Second, nematode muscles simultaneously receive inputs from excitatory and inhibitory motoneurons; muscle contractions appear to be a graded integration of these inputs. Third, nematodes lack voltage-gated sodium channels that would amplify synaptic inputs into all-or-nothing depolarizations in the muscle. These conclusions have been experimentally supported: physiological recordings demonstrate that body muscles are non-spiking and non-oscillating cells (Jospin et al. 2002). A modelling study further indicated that the conductance of gap junctions between muscles is too small to play significant roles in rhythm generation (Boyle & Cohen, 2008). These authors concluded that C. elegans body muscles function simply to execute the orders imposed by the neurons. However, would these conclusions change if it could be shown that body muscles are capable of spiking and oscillating? The article by Liu et al. (2011) reports that there are all-or-none action potentials in C. elegans muscles. This study from Zhao-WenWang’s laboratory demonstrates that these action potentials are calcium dependent and occur in spontaneous trains. By recording from mutant animals, Liu and colleagues identified the ion channels that contribute to these action potentials: calcium influx through L-type and T-type calcium channels, and calcium release from internal stores via the ryanodine receptor are responsible for the upstroke and peak of the muscle action potential; shaker and BK potassium channels shape the repolarization. Interestingly, Liu et al. found thatbodywall muscles continue to fire action potentials even when both acetylcholine and GABA neuronal inputs were blocked. These results suggest that C. elegans muscles may have intrinsic oscillations independent of the nervous system. Do these spike trains cause muscle contraction? The authors simultaneously monitored intracellular calcium fluctuations, muscle action potentials and muscle contractions. They found that in dissected preparations calcium transients were correlated with clusters of action potentials, and that in intact animals, calcium transients were correlated with muscle contractions. These observations indicate a one-to-one correlation between calcium spike trains andmuscle contraction in vivo. Furthermore, the authors noticed that intact animals with greatly reduced synaptic inputs still exhibit calcium transients in muscles. These data suggest how animals without functional neuronal inputs are still capable of moving. In the end this study is consistent with the model that muscles may be able to drive a contractile wave on their own. So is the issue settled? One caveat is that the calcium transients detected from the animals with severe synaptic defects have reduced amplitudes and somewhat irregular shape compared to those from wild-type animals. It is not clear whether these smaller calcium transients are caused by all-or-none action potentials or subthresholdmembrane potential fluctuations. Thus, these results do not preclude the