Authors: Thomas Lu1,5, Aarohi Nadkarni2,5, Sophia Ma3,5, Sreeanvitha Emani4,5, Dr. Marianne Bezaire5
This is the code repository for the project, "Decreased Calcium Concentrations Lead to Hyperexcitability in Computational Network Model of the Dentate Gyrus", conducted as part of Boston University's RISE program in the summer of 2020. Starting code is adapted from Santhakumar et al. (2005): https://senselab.med.yale.edu/ModelDB/ShowModel?model=51781#tabs-1 Our poster is located on our website: https://anviemani.github.io/rise-group-a-2020/
Seizures are a dangerous consequence of hyperexcitable neuronal networks in the brain and affect over 50 million epileptic patients around the world. Hypocalcemia (low blood calcium concentration) is one factor found to contribute to seizures, but this phenomenon is highly counterintuitive given calcium’s role in driving neuron firing. Additionally, hypocalcemia-induced seizures and their underlying mechanisms are still relatively underexplored in both experimental and computational settings. However, identifiable molecular mechanisms exist based on previous experimental observations, such as calcium interacting with leaky sodium channels (NALCN) and voltage-gated sodium channels (VGSC) to increase neuron excitability. We used a previously studied, 527-cell computational model of the dentate gyrus, a brain region in the hippocampus, to analyze network excitability due to lowered calcium concentrations. In particular, using experimental observations in the literature, we implemented code to relate external calcium to 1) NALCN conductance, and 2) VGSC voltage sensitivity. We then simulated the network under external calcium concentrations between 0.1 mM and 2.0 mM and generated analyses of network and neuron activity, including spike rasters and duration of network activity. Low calcium concentrations were found to have a significant enhancing effect on network excitability, which is consistent with the literature. Additionally, both models (VGSC and NALCN) saw an increase in excitability with decreasing calcium concentration, with the VGSC model's excitability increasing more rapidly. However, VGSC excitability greatly decreased for concentrations around 0.2 mM and below, while NALCN maintains excitability; we believe this provides an explanation for the experimental observation in Lu et al. (2010) that only NALCN contributes to hypocalcemic excitability, as those evaluations were performed in relatively lower calcium concentrations. This study provides a first computational insight on mechanisms and dynamics of hypocalcemia-induced seizures. Future studies can build on our approach and findings to further understand the specific molecular mechanisms underlying this phenomenon.
There are three folders in this repository:
- dentategyrus2005: A clone of the 2005 repository, with various additions for our project. Most notable are main.py (a python wrapper for running the hoc script, A-DG500_M7.hoc), plots.py (python file for plotting results), and various bash scripts for running on the shared computing cluster (including m_qsub.sh and run_on_scc.sh).
- vs: Stands for "voltage shift" or, alternatively, "voltage gated sodium channel". A copy of dentategyrus2005, but with a modified hoc file (A-vs.hoc) and modified ichan2.mod file. Incorporates voltage shift in Hodgkin-Huxley parameters found in Hille 2001 through adding shift to v term in alpha and beta expressions. New RANGE variable vshift added.
- glna: Stands for "conductance (g) of leak Na+". A copy of dentategyrus2005, but also with a modified hoc file and modified ichan2.mod. Incorporates new leak conductance term for representing NALCN; il=gl*(v-el) in ichan2 has been expanded to il=glna*(v-ena)+glo*(v-elo), where ena is reversal potential for sodium (calculated as 55 mv in this case) and glo and elo represent conductance and reversal potential for all sources of leak other than NALCN. Base glna is calculated from NALCN current observed in Lu et al. 2010, their voltage conditions, and an area approximation from the mossy cells in our model, and the scaling of glna based on external calcium is derived from the same study (we fitted a logistic equation to their data, with log(cao) as the input variable). glo and elo is calculated based on the base glna through solving il=ilna+ilo in the Ca=2.0 mM situation.
1Thomas Jefferson High School for Science and Technology, 6560 Braddock Rd, Alexandria, VA 22312 2Unionville High School, 750 Unionville Rd, Kennett Square, PA 19348 3Phillips Academy, 180 Main St, Andover, MA 01810 4Massachusetts Academy of Math and Science at WPI, 85 Prescott St, Worcester, MA 01605 5Boston University, Boston, MA 02215