National Center for Behavioral Genomics, Department of Biology, Volen Center for Complex Systems, Brandeis
This article is a part ofJoVE General. If you think this article would be useful for your research, please recommend JoVE to your institution's librarian.Recommend JoVE to Your Librarian
Current Access Through Your IP Address
Current Access Through Your Registered Email Address
Slawson, J. B., Kim, E. Z., Griffith, L. C. High-Resolution Video Tracking of Locomotion in Adult Drosophila Melanogaster. J. Vis. Exp. (24), e1096, doi:10.3791/1096 (2009).
Part 1: Feeding and Management of flies
Part 2: Setting up the tracking system in a controlled environment
Part 3: Delivering flies to the locomotion chamber
Part 4: Running the Locomotor Assay
Part 5: Analysis of Videos
Figure 1 shows representative traces from wild type Canton S organisms (Figure 1, left panel) and flies null for the dCASK gene which were generated by crossing two large overlapping deletions (Df(3R)x307 and Df(3R)x313) (Figure 1, right panel). dCASK null flies have previously been shown to have locomotor problems using Buridan’s Paradigm3, and in our paradigm, compared to Canton S, they show greatly decreased locomotion. We always run wild type organisms first to make sure that conditions and behavior are within the normal range on a given day. We have observed deviations from standard responses in less than 5% of testing days. These differences can usually be attributed to problems with parameters of our controlled environment.
Figure 1. DIAS-generated traces of both Canton S wild type flies (above, left) and Locomotor-deficient dCASK Null flies generated from overlapping deletions (above,right). The traces represent what is seen in the recorded videos. Both pictures contrain traces from 8-10 flies run for 30 seconds in the video tracking assay following an air pulse.
Figure 2. Wild type flies were run in the locomotor assay (above) following an air pulse (purple bars) and without an air pulse (blue). In all parameters calculated by the analysis program, there were no significant differences between the two conditions (determined with two-tailed student t-test). This demonstrates that following an air pulse, flies locomote normally in our setup.
In our setup, a moderately strong air pulse transiently stops fly movement. When the air pulse ends, the flies are released from this stationary state and locomote normally, as shown in figure 2. Because this air pulse effectively synchronizes locomotion of the population, we use it to begin the trial so that we can also study the onset of movement following a momentum-breaking stimulus. The assay can, however, be done without an air pulse since locomotor parameters following initiation are unaffected by the air pulse (Figure 2).
One frequently occurring problem seen with automated analysis of locomotion is the issue of collisions. Tracking programs are for the most part notoriously bad at dealing with flies that collide. These programs often lose “sight” of an object momentarily during a collision, and upon finding this object label it as a new object. The result can be an output that has many more objects traced than actually exist in the chamber. Many people solve this problem by tracing single flies. The obvious problem with this, however, is that single fly behavior may be very different than population behavior due to the role of social cues in Drosophila behavior and locomotion4. This is, of course, not necessarily a bad thing, depending on what you are trying to study, but for our purposes, we prefer to use populations of flies to increase our statistical power. Because of this, we deal with collisions in two ways. First, we only use 8-10 flies per trial (in our 56 mm sq chamber), so that collisions will be minimal. Second, within our thirty second trials, we only analyze traces that are at least 18 seconds in length. By doing this, we are always recording enough time to capture multiple bouts of motion in their entirety. This also ensures that our sample size always reflects the number of flies in a chamber, as smaller fragments of motion are discarded and the total length of the video is only 30 seconds.
Data analysis can be the most difficult part of this whole process. The most important step of analyzing tracking data is to determine the difference between noise and motion. DIAS (like many other programs) will almost never output a speed of 0 mm/s, even if the fly has stopped moving. Because of this, it is important to watch the videos while looking at the data output frame by frame to see when the organisms are actually in motion, and when they are not. In our setup, all speeds below 1 mm/s seem to be noise, which is consistent with what other groups using DIAS have found with adult flies1. For looking at the dynamics of bout structure, one also needs to define a bout. We define activity as 3 or more consecutive frames of speed above 1 mm/s, and inactivity as 3 or more consecutive frames below this threshold.
The data must also be properly smoothed to eliminate artifacts from transient light flickering and camera distortions, which occasionally occur. There is no standard method for data smoothing, but it is important that the smoothing process does not change the general trend of the data too drastically, or one can generate smoothing artifacts. We smooth twice with a Tukey window of “5,15,60,15,5”, because it seems to eliminate any large jumps or changes in speed that are clearly wrong, but does not change the overall trend or nature of the data.
It is important to recognize that different setups and environments will produce different behavioral outcomes. Our recommendation to anyone setting up a tracking assay is to experiment with all of these issues until you find a method which produces data that matches what can be seen visually, and then be as consistent as possible in treating all data in the same manner.
Selection of the right tracking program is also important. Although we use DIAS 3.2 (www.solltechnologies.com), this is by no means the best or the only system available. For tracking of single flies, Dan Valente (Mitra Lab, CSHL) has developed a program called Ftrack. For multiple flies locomoting together (i.e., where collisions might occur), Kristin Branson (Dickinson/Perona Labs, Caltech) has developed a program called Ctrax. Ftrack is available at www.chronux.org, while Ctrax is available at www.dickinson.caltech.edu/Research/Mtrax. Ethovision software (Noldus, Netherlands) is another powerful option available for video tracking of both single and multiple flies, but it is only commercially available, and is quite expensive.
If you have set up a tracking assay and cannot get reliable recordings, there are a few things to consider. Circadian rhythm issues are often a big problem. When running behavioral assays, one should always make sure NOT to run flies during their midday siesta. Since we are currently interested in genes that can produce decreases in locomotor behavior, we prefer to run flies near the late afternoon activity peak (ZT 8-10). Another problem can also arise from inconsistent air flow from the source. Before starting trials, be sure to test the air flow with the flowmeter, so that it will not fluctuate greatly over the 15 second air pulse. Lastly, genetic background can play a role in all behavioral tasks, so some of the parameters of this assay may need to be optimized for a particular background.
This work was supported by National Institutes of Health Grant R01 GM54408 awarded to L.C. Griffith. We would like to thank Fred Wolf for all his help in designing our square chamber and setting up our assay, Frank Mello at the Brandeis University machine shop for building our chamber, and Dan Valente and Tim Lebestky for helpful conversations regarding analysis issues.
|Square Chamber||Tool||Machine Shop||N/A||Design from Wolf et al. 20021|
|Digital Camera||Camera||Sharp||ViewcamZ VL-23|
|Charcoal Filter||Tool||Fisher Scientific||09-744-37|
|DIAS 3.2||Software||Soll technologies||N/A||www.solltechnologies.com|