Flashes of Insight: Applying New Techniques to a Classic Model

THE study of dynamical systems has developed to explain complicated emergent behaviors exhibited by coupled systems, and there are few systems with couplings more complicated than nervous systems. To gain meaningful insights about the resulting complexity, it helps to begin with models that are simple enough. A classic example of a simple dynamical system was provided by Christiaan Huygens, inventor of the pendulum clock, who noted in 1665 that two pendulum clocks mounted on the wall in his home would spontaneously synchronize so that their pendulums were swinging exactly opposite one another.¹  While easily demonstrable, there was no reason even the inventor of the clocks would have predicted this behavior. Observation of the two clocks together in context was key to the detection of the dynamic interaction between them.

In this issue of Anesthesiology, Awal et al.²  have employed modern calcium imaging technology in a classic model organism in neuroscience, the nematode Caenorhabditis elegans. As with most organisms, C. elegans is anesthetized by polyhalogenated ethers, which enhance inhibitory postsynaptic potentials via γ-aminobutyric acid type A receptors. Surprisingly, the authors show that isoflurane-induced loss of behavior correlates with a breakdown in firing relationships among neurons in the worm’s motor circuit, rather than with a simple suppression of firing rates, as expected from increasing inhibition in the network; indeed, initially activity seems to increase with administration of isoflurane. The prediction from our molecular understanding—that anesthetics produce loss of consciousness by suppressing firing—appears to be partial and incomplete. It is difficult to predict how anesthetic-induced changes in one neuron will affect its coupling to others (analogously, would changing a spring constant in Huygens’s clocks change their synchronization?). Hence the key to these authors’ insight is the choice of an experimental model ideal for studying the effect of anesthetics on interactions between neurons in an intact circuit, by allowing observation of multiple neurons together.

The beauty of C. elegans in part derives from its simplicity. The human brain contains approximately 100 billion neurons³  with variable connectivity patterns that are altered by experience, making it impossible to identify “the same” neuron in different individuals and frustrating attempts at circuit identification. By contrast, C. elegans has only 302 neurons, each of which can be identified in individual worms. While C. elegans has been used in previous studies of anesthetics, these authors have taken advantage of modern calcium imaging techniques that allow recording of the activity in multiple cells simultaneously and noninvasively to look for the impact of anesthetics not just on firing rates but also on interactions between neurons.

To assay neuronal activity patterns during isoflurane anesthesia, Awal et al. utilize transgenic animals that express the protein GCaMP6s, a fusion of the jellyfish-derived green fluorescent protein with the calcium-binding protein calmodulin, in neurons.⁴  As the calcium concentration in the cell rises, the calmodulin motif binds calcium. The resulting change in calmodulin conformation is transmitted through a linker to alter the conformation and hence fluorescence of the green fluorescent protein motif. Cells expressing proteins in the GCaMP family fluoresce when the intracellular calcium concentration increases, so in neurons expressing GCaMP6s, action potentials appear as green flashes. Because the worms are translucent, it is possible in vivo to obtain a noninvasive, high-resolution simultaneous readout of all of the neurons visible within a single microscopic field. The authors then compared the neuronal activity levels and firing patterns present at different concentrations of isoflurane.

The translation of neuronal activity into behavior is relatively well characterized in C. elegans.⁵  Motion is controlled by two different cholinergic motor neuron populations in the ventral nerve cord (known as A and B motor neurons, respectively) together with a γ-aminobutyric acid–mediated neuron population (D motor neurons) that ensures the dorsal musculature in one segment of the worm relaxes as the ventral musculature of that segment contracts, and vice versa, leading to sinusoidal undulations of the worm’s body. These patterns are triggered by activity in individual premotor interneurons identified by convention using a three-letter label that are responsible for “forward” (neurons AVB and PVC) or “reverse” (neurons AVA, AVD, and AVE) movement patterns. The forward and reverse interneurons are themselves interconnected by other interneurons (including neuron RIM).

Awal et al. recorded the activity present in the premotor interneurons AVA, AVB, AVD, AVE, and RIM at varying concentrations of isoflurane. Several premotor interneurons (AVA, AVE, and RIM) normally exhibit essentially binary activity—that is, they are either “on” or “off.” These reverse interneurons are highly correlated with each other and with the interneuron RIM, and strongly anticorrelated with the forward interneuron AVB. As C. elegans is exposed to increasing concentrations of isoflurane, motor behavior becomes disorganized and slowed, eventually ceasing by 4% isoflurane, which the authors liken to a surgical plane of anesthesia. Neuronal activity essentially ceases at 8% isoflurane. The authors have thus confirmed that isoflurane produces behavioral quiescence in C. elegans even at concentrations that allow persistent activity in the nervous system. Just as a patient can still have activity on their electroencephalogram even though they are unconscious and unresponsive in the operating room, the worm’s premotor neurons are still active even though the worm has stopped exhibiting any behavior.

Critically, however, interactions among neurons are radically shifted by isoflurane. The disorganization and eventual disruption of motor output with isoflurane is associated with a breakdown of the baseline correlation structure seen in the premotor interneurons. (For technical reasons, identification of AVB under anesthesia was not reliable, and hence AVB was excluded from analysis during isoflurane exposure.) The authors quantify the breakdown in the interneuron interactions with an information theoretic measure, multiinformation. Multiinformation peaks when the neuronal pattern is perfectly coordinated, like clockwork. Even though there is markedly more activity present with 4% than 8% isoflurane, neuronal activity patterns at both anesthetic concentrations demonstrated an equivalent, near zero multiinformation, consistent with independent random firing. That is, the circuit is nonfunctional and each neuron is firing as though it were in isolation.

Since Morton’s public demonstration of ether as an anesthetic,⁶  we have been searching for an understanding of how anesthetics produce unconsciousness. To that end, great progress has been made in identifying which receptors bind individual drugs. Effects of different anesthetics on individual conductances have been measured in isolated neurons and membranes with great precision, so that our understanding of the biophysics of anesthetics is fairly well developed. Shifts in the bulk firing properties of neurons as detected with electroencephalogram and functional imaging are also widely studied. Yet it is not obvious how these scales of investigation relate to one another, and it is unclear that either scale will give us a complete understanding of how anesthetic-induced loss of consciousness occurs. Awal et al. are to be commended for using modern neuroscience techniques to study the effects of anesthetics on circuit dynamics in a tractable model system that allows them to probe the disruption of well-defined circuit function, an approach that I would like to see taken more often. Their work, built on a circuit scale, spans the molecular and the macroscopic. In so doing, they suggest a line of investigation to bridge the molecular and the macroscopic effects of anesthesia, by asking how isoflurane disrupts coupling among neurons, and why this breakdown in coupling produces the characteristic changes we see in the electroencephalogram and on functional imaging.