Large-Scale Biophysically Realistic Model of Mouse Olfactory Bulb
by Justas Birgiolas, Ph.D., M.B.A
The model represents a piece of neural tissue of the dorsal mouse olfactory bulb. It realistically replicates the morphology and electrophysiology of three cell types: mitral, tufted, and granule. The cell models and glomeruli are placed within reconstructed mouse olfactory bulb layers and connected by excitatory and inhibitory synapses. Model output is assessed by measuring local field potential (LFP) oscillations.
This model reproduces the gamma range LFP oscillations observed in rodent olfactory bulbs. Computational modifications of the model reveal the mechanisms behind the oscillations.
The model can be extended to model other bulbar phenomena, while the methods described in the dissertation can be used to build models of other parts of the brain.
Each cell type was modeled by five different cell models.
Conductance-based ion channel models were inserted into reconstructed membrane morphologies of Tufted, Mitral, and Granule cells. Ion channel conductance combinations were identified that resulted in cell model electrophysiology properties that closely matched those observed in mouse cells.
Example action potential propagations in each cell model can be observed in the following video.
Model cells are realistically distributed within the granule, mitral, external plexiform, and glomerular layers of the olfactory bulb. The layer shapes were reconstructed from mouse olfactory bulb slices.
Cells were connected by electrical and chemical synapses. Based on proximity, lateral dendrites of mitral and tufted cells formed reciprocal excitatory-inhibitory synapses with apical dendrites of granule cells.
Within glomeruli, tufted dendrites of mitral cells formed chemical synapses with glomerular sibling mitral cells. Tufted cell tufted dendrites were similarly connected to other tufted cells.
Mouse glomerular activations observed in response to real odors were used to stimulate the tufted dendrites of mitral and tufted cells.
The output of the model was assessed by measuring the local field potential (LFP) caused by currents within synapses and ion channels.
The model reproduced the stereotypical, bi-modal gamma range LFP oscillation pattern observed in rodent olfactory bulb LFPs.
Resulting Network Model
To investigate the mechanisms underlying the gamma oscillation pattern (“the gamma fingerprint”), the network model was manipulated computationally.
The experiments found that blocking electrical synapses between glomerular sibling mitral or tufted cells, equalizing the glomerular input strength to mitral and tufted cells, or blocking inhibition by granule cells abolished the bi-modal character of the gamma fingerprint.
Mammalian Bulbar Gamma Fingerprint Hypothesis
The experiments suggest that the first cluster of the fingerprint is formed by synchronized excitation of tufted cells. The tufted cells excite granule cells, which inhibit sibling mitral cells. Continued stimulation of mitral cells overcomes the fading granule cell inhibition, resulting in synchronized mitral cell firing, forming the second cluster of the fingerprint.
Downloads and Documentation
The model is open-source and can be downloaded from GitHub. Documentation on how to run and modify the model and data used to build and constrain it are provided as well.
- Arizona State University: Interdisciplinary Graduate Program in Neuroscience Fellowship
- National Institute on Deafness and Other Communication Disorders: NRSA F31 F31DC016811
- National Institute of Health: R01MH1006674 to Dr. Sharon Crook
- Google: Summer of Code Program
- Institute for Conscious Automaton Research