(
A) Schematic of light path for widefield calcium imaging of femur muscles. Infrared illumination is used to track leg movement. GCaMP emission is reflected by a longpass dichroic mirror to a second camera. (
B) Tracking the femur and tibia with DeepLabCut. A network was trained to detect six points on the femur and six points on the tibia in the IR illuminated videos of the leg. We calculated the center of the tibia from the six detected points. The elliptical arc of the centroid across frames allowed us to estimate the angle of elevation of the tibia relative to the plane of the metal holder (black background). (
C) K-means clustering of GCaMP activity in muscle fibers during unloaded movements of the tibia. Clusters were similar across five flies, including the ventral distal cluster 1, the proximal cluster 2, the more dorsal cluster 3, and a thin dorsal cluster 4. Clusters 5 and 6 were more variable in shape and location. (
D) Example epoch from Fly one showing the femur-tibia angle and the fluorescence of each cluster. (
E) Averaged fluorescence across all frames in which the tibia was extended (top) vs. flexed (bottom), for Fly 1, normalized to the maximum ΔF/F. Note the lack of signal during extension. (
F) K-means clustering with different numbers of clusters for Fly 1, shown in
Figure 1. With fewer than five clusters, the proximal cluster tended to be much larger, incorporating much of the region labeled as cluster 3. With more than six clusters, the smallest and least modulated clusters tended to divide, not providing any further information. When k = 6, pixels in the extension region clustered together, but we did not see large increases in fluorescence with extension (
Figure 1).