The inaugural release of the Allen Brain Observatory aims to elucidate visual coding: How are visual stimuli represented by neural activity in the mouse visual cortex in both single cells and populations? To answer this question, two-photon calcium imaging was recorded from several areas of the visual cortex in four distinct transgenic mouse lines during exposure of five rich visual stimuli.
Creating a Cortical Activity Map
To produce standardized protocols to be able to record from the same cortical area over the extent of the experiment required coordination and hand-off between the many teams performing platform development, data collection and analysis. A basic workflow is outlined below.
Creating a window into the brain
Accurately measuring the activity in a specific visual area in a reproducible and standardized manner first requires a window into the brain - replacing a small section of the skull with a clear glass cover slip that allows visual access to a 5 mm diameter region of the brain.
Being able to consistently locate a cortical region and even a specific cell requires that each system used to collect data maintains a consistent relationship between the mouse's eye and the visual stimulus being presented. To achieve consistent placement of the mouse across systems, and address the challenge of finding and returning to the same cells from day to day, the engineering team designed a custom headframe that is affixed to the mouse’s skull in a standardized location. The headframe consists of a plastic head plate, or well, and a stainless steal clamp-plate that allows repeatable positioning of the animal on each experimental setup. The clamp plate fits perfectly into the placement tool, which in turn is secured to the stereotaxic arm of the surgery rig. Before attaching the head frame, the surgeons used a leveling protocol that was anatomically referenced, so that the headframe was consistently placed over the visual cortex. The cranial window was made within the well to expose the part of the brain that was to be imaged. This ensures an optimal position of the headframe to give visual access to the desirable cortical areas.
The headframe clamp enables cross-platform registration
A standardized clamping system on each experimental platform was made to match the head frame clamp plate and ensure reproducible placement of the head and eye position on all platforms. This cross platform standardization is essential to register all the data into a common reference frame for downstream analysis.
Targeting functional areas of the visual cortex
Intrinsic signal imaging (ISI) measures the hemodynamic response of the cortex to visual stimulation across the entire field of view. This "retinotopic map" effectively represents the spatial relationship of the visual field to locations within each cortical area. Retinotopic mapping was used to delineate functionally defined visual area boundaries, and enabled targeting of the in vivo two-photon calcium imaging to functionally defined locations in primary and secondary visual cortical areas.
Accurately mapping cortical areas involves first imaging the cortical surface with green LED illumination to capture fiduciary markers (vasculature landmarks). Next, imaging of the cortical surface during presentation of a visual stimulus with red LED illumination allows for capture of the hemodynamic response to visual stimuli. For more information on how ISI was analyzed and used to target areas of the cortex, refer to the Overview whitepaper in Documentation.
Visual Stimulus and Neuronal Response
Once a standard protocol and equipment were developed, activity of individual neurons to a visual stimulus in a targeted area of the visual cortex could be conducted. Neuronal activity was measured in GCaMP6-expressing neurons from selected cortical layers from four separate transgenic mouse lines. For more information on these how neurons in specific layers and areas were targeted in these transgenic lines, see the Transgenic Line Catalog in Documentation.
Video recordings of the visual cortex were taken during presentation of the various visual stimuli. The data processing of these raw data movies involved motion correction and image segmentation to identify the sets of pixels representing distinct cells. These segmentation masks were used to extract traces from each neuron so that cellular activity over time can be analyzed. The goal was to relate the activity of each cell in the field of view back to the visual stimulus that was viewed by the mouse.