Central circuits underlying leg proprioception

Proprioception is perhaps the most poorly understood of the primary senses. In particular, we know very little about how proprioception is processed within circuits of the central nervous system: these circuits are embedded within the vertebrate spinal cord or the analogous invertebrate ventral nerve cord (VNC), where they are difficult to access and record from, we don’t know much about how these sensorimotor circuits are organized, and we lack quality genetic tools that label the relevant cell-types. To fill these gaps in our knowledge, I took advantage of the precision of Drosophila genetic tools and new methods for in vivo recordings from proprioceptive circuits to dissect the neural computations that underly proprioception.

I identified and characterized several novel cell-types that receive input from different leg proprioceptor subtypes. By combining electrophysiology with pharmacological manipulations, I mapped the proprioceptive inputs to each cell-type and constructed a functional circuit diagram for how proprioceptive signals are integrated by second-order neurons. Surprisingly, several of these central neurons integrate signals from multiple proprioceptor subtypes, resulting in complex representations of tibia movement and position. This rapid integration of proprioceptive information may be necessary for rapid feedback control of fast motor behaviors.

Next, I assessed how these central neurons mediate motor output by optogenetically activating each cell-type and measuring leg joint kinematics in stationary and walking flies. I found that each cell-type drives a distinct motor response, such as postural reflexes or abrupt cessation of walking. These experiments are the first to directly link central pathways for limb proprioception to motor control of behavior. This is also the first study to investigate central proprioceptive circuits in Drosophila, establishing Drosophila as an important model system for understanding circuit principles of proprioceptive neural computation.

This work was completed in the lab of Dr. John Tuthill at the University of Washington.