Overview

Understanding how the collective activity of neurons gives rise to behavior is one of the great challenges for biology in this century. This task is especially daunting because it is difficult to both monitor the activity of the relevant neurons in behaving animals and quantify behavior precisely. We are addressing this problem by developing microscopy tools and experimental techniques with which to study the model organism with the simplest and best-characterized nervous system, Caenorhabditis elegans. We hope that this will allow us to develop key insights into the control of behavior by neural circuits and reveal the genes that are required for the function of these circuits. It is our expectation that these lessons can then be applied to larger organisms, where the technical challenges are greater.

Population-wide Behavioral Quantification

C. elegans exhibits a variety of simple movement behaviors, including withdrawal from aversive chemical, mechanical, and thermal stimuli; taxis to favorable chemicals and temperatures and away from unfavorable ones; regulation of movement based upon quality of food; and variable search strategies when removed from food.

These behaviors are typically scored by eye. Some behaviors are also amenable to population assays; chemotaxis, for example, can be scored by leaving worms on a plate and counting the number of individuals who end up in a target zone around a favorable or unfavorable stimulus as compared to those in a control zone. However, other behaviors, including habituation to noxious stimuli and strategies for taxis and search, can only be detected by continuously following the movement behavior of individual worms. This approach is too time-consuming to permit screening for mutants, even if aided by single-worm automated tracking and analysis systems.

In collaboration with the Rankin Lab at the University of British Columbia in Vancouver, we have developed a high-speed (30Hz+) tracking system, the Multi-Worm Tracker (MWT), that can simultaneously analyze the movement behavor of dozens of worms on an individual plate. By using real-time processing of video from a high-speed, high-resolution camera, we can capture movements as small as a few microns and time events to tens of milliseconds for arbitrarily long periods of time.

We are tailoring the system to identify genes that are involved in habituation to tap: the goal is to place a plate of F3 progeny of mutagenized worms, or a plate of worms on RNAi feeding bacteria, on the system, hit one button, and have the system report the habituation rate for that population in about ten minutes. In addition, we are collaborating with the Simpson Lab to track Drosophila movement behaviors.

Once the system is complete and we have finished appropriate software for browsing the worm positions, we plan to release the software under an open source license and make the specifications for the hardware available. Our hope is that this systems will help find a comprehensive list of genes involved in learning, memory, and behavior.

Monitoring the Activity of the C. elegans Brain

Adult C. elegans have only 302 neurons, half of which are located in the animal's head. To relate behavior to neural activity, we must monitor the activity of these neurons. Calcium imaging has allowed recording of neural activity in individual neurons in intact worms which are physically restrained but otherwise "awake" and "behaving". However, the approximately 160 neurons in the C. elegans head are closely spaced and distributed in three dimensions around the pharynx. Thus, it is nontrivial to simultaneously monitor multiple neurons using this technique.

We are developing hardware and software that allows us to rapidly image the entire population of neurons in the worm's head. The first technical challenge, capturing the volume at sufficient speed to monitor neural activity, is a largely solved problem; various line-scanning or swept-field confocals can gather enough photons, as can fast Z-scanning widefield systems. However, no existing system meets the second challenge: to capture enough photons from the sample to obtain an accurate report of neuronal activity for behaviorally relevant duration. We are pursuing two ways to solve this problem. First, in collaboration with Eric Betzig's lab, we are converting a fast Z-scanning widefield system to use selective plane illumination (SPIM) instead of epi-illumination. This restricts the excitation volume tenfold to a narrow sheet about the focal plane of the microscope, thus allowing us to increase the duration of experiments by an equivalent amount (tenfold). Second, we hope to use improved indicators from Loren Looger's lab to increase the signal-to-noise ratio and thus allow us to use dimmer illumination and extend our imaging duration. These two approaches should enable us to image the entire head with signals as good or better than the current best for single-neuron imaging.

Successful imaging does not necessarily result in usable data. The third technical challenge is to distinguish and identify the imaged neurons and convert the raw volume data into a calcium trace for each neuron. Initially, we will develop and use rapid 3D viewing and region-of-interest creation software to mark neurons by hand; once marked in one volume, the neuron(s) will be tracked automatically (a 3D analog of the 2D Jmalyze software Rex wrote for single-neuron imaging). However, as a long-term solution we will collaborate with Gene Myers' lab to automatically identify the neurons and segment them from each other. Due to the close packing of the neurons and trade-off between imaging resolution and acquistion speed, it may not be practical to separate or identify all neurons simultaneously. In this case, we will drive our calcium indicators with promoters that target neurons known to be important for the sensation or behavior being tested. To avoid missing important activity in currently unidentified neurons, we would expand the set of labeled neurons to include all neurons in a common circuit motif (as categorized by the Chklovskii Lab) if at least one of those neurons is known to be important for the sensation.

Although there are a wide variety of interesting C. elegans behaviors, we feel that it is premature at this stage to engage in hypothesis-driven study of nervous system function. We know so little about the broad pattern of neural activity in the worm that the first experiments should just be to observe. Once we have an idea of patterns of neural activity in response to different behaviors--and the specific behaviors will be chosen once we are aware of the limitations of the imaging technology--we will then begin to develop hypotheses regarding how the nervous system functions to give rise to behavior. We expect to make full use of the wide variety of tools that are now available to aid in hypothesis testing, ranging from traditional single-cell ablation through to activation or inhibition of individual neurons using light-activated channels. We also hope to collaborate with the wider C. elegans community to take advantage of their collective knowledge of behavior and nervous system function.