Data-driven analysis of motor activity implicates 5-HT2A neurons in backward locomotion of larval Drosophila

Jeonghyuk Park,Shu Kondo,Hiroshi Kohsaka,Hiromu Tanimoto,Akinao Nose

doi:10.1038/s41598-018-28680-8

Abstract

Rhythmic animal behaviors are regulated in part by neural circuits called the central pattern generators (CPGs). Classifying neural population activities correlated with body movements and identifying the associated component neurons are critical steps in understanding CPGs. Previous methods that classify neural dynamics obtained by dimension reduction algorithms often require manual optimization which could be laborious and preparation-specific. Here, we present a simpler and more flexible method that is based on the pre-trained convolutional neural network model VGG-16 and unsupervised learning, and successfully classifies the fictive motor patterns in Drosophila larvae under various imaging conditions. We also used voxel-wise correlation mapping to identify neurons associated with motor patterns. By applying these methods to neurons targeted by 5-HT2A-GAL4, which we generated by the CRISPR/Cas9-system, we identified two classes of interneurons, termed Seta and Leta, which are specifically active during backward but not forward fictive locomotion. Optogenetic activation of Seta and Leta neurons increased backward locomotion. Conversely, thermogenetic inhibition of 5-HT2A-GAL4 neurons or application of a 5-HT2 antagonist decreased backward locomotion induced by noxious light stimuli. This study establishes an accelerated pipeline for activity profiling and cell identification in larval Drosophila and implicates the serotonergic system in the modulation of backward locomotion.

Highlights

The neural circuits generating rhythmic behaviors such as walking and breathing are called the central pattern generators (CPGs)[1,2,3]
Since the motor activity pattern in the x-t image is similar to images such as handwritten digits[24], we expected that convolutional neural network (CNN), known to give superior performance in image classification tasks[24,25,26], may be used to efficiently categorize the motor patterns
We tested if a CNN model, VGG-16 pre-trained on the ImageNet dataset[27,28], which is widely used in transfer learning[29], can classify neural activities from the calcium imaging data

Summary

Introduction

The neural circuits generating rhythmic behaviors such as walking and breathing are called the central pattern generators (CPGs)[1,2,3]. An isolated CNS can generate fictive motor outputs such as coordinated propagation of motor activity along the body axis, which resembles forward and backward locomotion of the animal, and left-right asymmetric bursts in anterior neuromeres which likely correspond to turning[14]. We present a new methodology for classifying neural activity patterns in larval Drosophila that utilizes a convolutional neural network (CNN) and unsupervised learning This method successfully classified forward and backward waves as well as synchronous activities in anterior- and posterior-most neuromeres from large activity data derived from different sub-populations of central neurons. We identified cells associated with the classified motor activity patterns by voxel-wise correlation mapping We applied this method to a population of neurons targeted by 5-HT2A-GAL4, which we generated by CRISPR/Cas9-mediated gene knock-in[23], and successfully identified two classes of interneurons that are active during fictive backward locomotion. Our study provides an accelerated pipeline that efficiently identifies neurons correlated with motor patterns from wide-volumetric functional imaging data

Objectives

Methods

Results

Conclusion