This summer, I am working in Dr. Gong’s Biomedical Engineering research lab on a project focused on examining place cells in the hippocampus of mice. Place cells play an important role in spatial and episodic memories and the current project looks at how visual spatial manipulations affect how place cells are fired.
A 1-photon microscope was used to record the firing of the neurons. First, an Adeno-associated virus that carries the protein GCaMP6f was injected in the hippocampus of the mice. This particular virus was selected because GCaMP6f has a unique ability to bind to calcium ions and fluoresce. As a result, when the concentration of calcium is high, more of the GCaMP6f is bound to Ca2+, resulting in increased fluorescence that can be picked up by the 1-photon microscope. Since Ca2+ increases during an action potential, we can correlate the increases in fluorescence to periods of neural activity.
The physical experimental set-up consisted of a styrofoam cylinder that served as a linear treadmill, on top of which the mouse was placed. A virtual reality display was shown on monitors in front of the treadmill and the mouse controlled its movements in the virtual reality through its actions on the wheel. The purpose of the virtual reality was to allow for manipulation of what the mouse sees in ways that couldn’t happen outside of a computer screen. After a training period, the mice ran through randomized trials that consisted of either running on an unaltered track or on a track with a visual spatial manipulation.
After running the experimental trials, an algorithm called constrained non-negative matrix factorization was used to identify areas of fluorescence that may represent neurons. However, the computer is not always accurate and may outline a group of pixels that are not neurons. For example, the outline might circle two neurons, the space between two neurons, only part of the neuron, or part of a neighboring neuron. Currently, my primary job is to identify whether the computer-generated outlines are actually neurons or not. To do this, I pay attention to nearby pixels that brighten and dim together. If the outline encapsulates a group of pixels that fulfill those criteria, then it is most likely a neuron and I select it.
After all this data is processed, we will be able to identify place cells and determine their function by correlating their respective calcium transients to locations in space. With this information, we will be able to better understand how place cells react to visual manipulations. This information will be useful in looking at how vision corrects for errors, in better understanding how the brain uses vision, and in further understanding how the brain works in general, allowing future researchers to compare this data to disease models.
A special thanks to my mentor Emily Redington for her help with this blog post!