Identifying Neural Substrates of Behavior in Drosophila Melanogaster

National Institute Of Mental Health

Investigators

Linked publications & trials

Abstract

Insect ecdysis sequences represent a simple, robust, and tractable model for studying the neuromodulatory mechanisms that govern behavior. Because initiation of an ecdysis sequence involves a profound shift in behavioral priorities, study of these sequences offers the opportunity to understand the neuromodulatory mechanisms that govern changes in behavioral state. In addition, because ecdysis behaviors are inherently sequential, they permit the systematic investigation of how motor programs are assembled and serially executed by the nervous system. Finally, the study of ecdysis sequences promises insight into how neural circuits can be variably configured to generate immensely different behaviors. In Drosophila, for example, the motor sequences performed at pupal and adult ecdysis are completely different. This is because of the profound differences in the pupal and adult body plans. Despite these anatomical differences, the two behavioral sequences are governed by a common set of neuromodulatory/hormonal inputs. By analogy to computing, these inputs can be regarded as instructions written in a higher programming language that are then compiled into different outputs. Exposing the mechanisms of neural compilation in ecdysis is likely to deeply inform our understanding of how neuromodulators contribute to neurocomputation by reconfiguring the activity of neural networks. To investigate these issues, my laboratory seeks to elucidate the circuitry that governs both the pupal and adult ecdysis sequences in Drosophila. Our efforts over the last year have been more or less evenly divided between study of these two circuits. With regard to pupal ecdysis, the work conducted over the past year has focused on describing the pupal ecdysis sequence at the highest possible level of resolution, namely the activity of each of the individual muscles of the animal. This work was motivated by the realization that if we are to understand in detail how the fly brain generates a pupal ecdysis sequence, we must first understand in detail what, exactly, it is generating. While comprehensively monitoring single muscle activity in behaving animals is generally impossible, we leveraged several advantages of the Drosophila pupal preparation to do so. Muscles in the pupa, for example, are relatively well-separated and ordered, enabling their identification during activity. More importantly, the pupal ecdysis sequence is performed within the confines of the so-called puparium, the protective casing in which the animal undergoes metamorphosis. The puparium can be clarified to permit imaging during behavior, and the animal remains in the field of view during execution of the entire sequence. Muscle activity can thus be comprehensively recorded using genetically encoded calcium indicators. By imaging animals from the dorsal, lateral, and ventral aspects we were able to fully document muscle activity underlying the pupal ecdysis sequence in scores of animals. Among our several discoveries was a previously undescribed phase of pupal muscle activity preceding the onset of ecdysis behavior. Activity during this novel phase consists of stochastic combinations of muscle contractions that result in small twitches rather than coherent movement. This phase of activity, which we call P0, is driven by nervous system activity and becomes coherent only with the release of ETH, which initiates the pupal ecdysis sequence. We showed that suppression of a subset of neurons that respond to ETHnamely those that express the B-isoform of the ETH receptorimpairs the initiation of pupal ecdysis by preventing coordinated activity of specific muscle groups. In fact, our analysis showed that these and other groups of muscles that exhibited coincident activity were common throughout the ecdysis sequence. We identified eight such groups and found that they are assembled in various temporal combinations to compose most pupal ecdysis movements. These movements could also be categorized into eight elementary classes, which are sufficiently stereotyped to be recognized from their patterns of single muscle calcium activity by a trained convolutional neural network (CNN). Together, the eight movements account for all organized pupal ecdysis behavior. Although pupal ecdysis behavior exhibits stereotypy at higher levels of behavioral description, we also found considerable evidence of stochasticity. This was evident in the frequent idiosyncratic muscle contractions observed during the pupal ecdysis sequence. In addition, the order of recruitment of muscle activity into co-active groups, or of these groups into movements, was highly variable. A major observation was that release of ecdysis hormones reduces variability of muscle activity and increases behavioral coherence, as illustrated by the example of ETH terminating P0 and initiating ecdysis. Because muscle calcium activity directly reflects the output of the motor system, our results provide a template onto which nervous system activity can be mapped to better understand how behavior is generated. Our results have recently been published in the journal eLife (eLife 2021;10:e68656). Our study of the adult ecdysis circuit over the past year has focused primarily on understanding the dual roles of the hormone Bursicon in wing expansion. Our previous work had shown that Bursicon released from a single pair of neurons (called the BSEG) was important in helping flies decide whether to expand their wings after adult ecdysis depending on environmental conditions. Expanding under adverse conditions risks permanently damaging the wings and flies will generally delay wing expansion and seek better conditions rather than risk damage. Using the Trojan exon technology that we developed in 2015 (Diao et al., 2015, Cell Rep. 10:1410-21) to investigate targets of Bursicon, we have now found that a key component of the decision-making circuitry is a set of cholinergic neurons that express the Bursicon receptor. These neurons signal back to the BSEG via a positive feedback loop to initiate and maintain wing expansion in an environmentally sensitive manner. Importantly, glutamatergic motor neurons that express the Bursicon receptor are also critical for producing the behaviors required for wing expansion. The hormone Bursicon thus targets neurons at two different circuit levels to motivate and execute wing expansion behavior. At the higher level, a positive feedback loop promotes robust secretion of Bursicon to provide drive, and at the lower level this drive (represented by high levels of Bursicon release) directly promotes motor execution. This work has recently been submitted for publication. In technology development, my laboratory has continued to refine methods that exploit the use of split inteins. We are actively adapting our SpaRCLIn technique, which is based on the use of a split Cre recombinase and which was published in 2020 in eLife (eLIFE: Apr 14;9:e53041), for use with the Split Gal4 method. The latter method was introduced by my laboratory in 2006 (Neuron 52:425-436) and has since become a workhorse for circuit-mapping efforts in Drosophila. Last year, we published a comprehensive review of the Split Gal4 technology and its various applications (Front Neural Circuit 14:603397). In summary, we have made good progress during the last year in advancing research on the principal questions of interest to the laboratory. At the same time, we have continued to develop technology that supports not only our own circuit mapping efforts, but also those of other members of the Drosophila research community. As we use these tools to extend and refine our analysis of the circuitry underlying ecdysis sequences, our work should provide insight into the principles that govern the development and function of behavioral circuits in all organisms, including humans.