A scalable method for automatically measuring pharyngeal pumping in C. elegans
We describe a scalable automated method for measuring the pharyngeal pumping of Caenorhabditis elegans in controlled environments. Our approach enables unbiased measurements for prolonged periods, a high throughput, and measurements in controlled yet dynamically changing feeding environments. The automated analysis compares well with scoring pumping by visual inspection, a common practice in the field. In addition, we we observed overall low rates of pharyngeal pumping and long correlation times when food availability was oscillated.
The nematode Caenorhabditis elegans feeds by drawing bacteria suspended in liquid into its pharynx, a neuromuscular organ that functions as a pump. The cycle of contraction and relaxation that draws food from the environment and filters bacteria from liquid is referred to as pharyngeal pumping. About one out of four pharyngeal pumps is followed by a posterior moving peristaltic contraction that transports food past the pharyngeal isthmus. This is referred to as isthmus peristalsis and, while coupled to pumping, it is distinctly regulated (Avery 1987, Raizen 1994, Avery 2012, Song 2013). The rate of pumping is thus the primary indicator of food intake (Avery 1993).
The rate of pumping depends on feeding history, quality of food, and the familiarity of food (Shtonda 2006, Song 2013a, Hobson 2006). Counting the stereotypical motion observed in the terminal bulb of the pharynx enabled detailed analyses of neuronal and molecular mechanisms that regulate pumping. However, these regulatory pathways were predominantly examined in a stationary environment, containing a saturated, high abundance of (familiar) bacteria on a standard agar plate. Moreover, traditional feeding assays rely on manual scoring of the mean number of pumps over brief (typically 30 sec) intervals (Avery 1989, Raizen 1995, Song 2013a). The reduction of a potentially complex time-series to a single average rate may result in loss of pertinent information or even unintended bias.
Bacterial lawns on a standard cultivation plate are typically highly concentrated and large compared to the worm. Under these conditions, reported pumping rates are 4-5 Hz and the duration of a single pump (constrained by the physiology of the pharynx) is approximately 170 ms (Avery 1993). The pharynx is separated from the rest of the body by the basal lamina, an extracellular formation of connective tissue. It contains 20 muscle cells and 20 neurons. Of these, only the MC cholinergic motoneurons were found to be individually required for rapid pumping in the presence of food (Avery 1989, Raizen 1995). The cholinergic neurons M2, M4, MC, and I1 form a degenerate network, excitatory for pumping and robust, where I1 can activate both MC and M2 (Trojanowski 2014). In addition, the glutamatergic M3 neurons regulate the termination of a pump and M4 regulates isthmus peristalsis (Avery 1989, Avery 1993, Raizen 1995). Under standard conditions, M3, M4, and MC are sufficient for supporting nearly normal feeding and growth (Raizen 1995, Avery 2012). The functions of additional pharyngeal neurons are poorly understood. In part, this may be due to the challenges of characterizing the phenomenology of pharyngeal pumping more systematically.
Here we describe an affordable and scalable method for automatically assaying pharyngeal pumping. Our method combines a previously described microfluidic device (Kopito 2014), low cost educational microscopes, and an image analysis pipeline implemented using widely available open source tools and libraries (Scholz 2016). The advantages of our approach include precise control of conditions such as the quality, uniformity, and concentration of available food, the possibility of prolonged measurement durations (hours, if required), unbiased automatic detection of pumping events, and the possibility to assay feeding conditions that change dynamically in a controlled manner. Manual scoring of pumping is arduous and limited to brief measurement periods. Automatic scoring of pumping on a high quality microscope is throughput-limited due to the cost of the imaging equipment. We found that lower quality imaging can be compensated for by rapid sampling and improved analysis without compromising the quality of the data. Using three microscopes we were able to assay up to 50 animals per day, i.e., 6 animals per objective, each for a full hou