Calcium regulation of HCN supports persistent activity associated with working memory: A multiscale model of prefrontal cortex

TitleCalcium regulation of HCN supports persistent activity associated with working memory: A multiscale model of prefrontal cortex
Publication TypeConference Paper
Year of Publication2014
AuthorsNeymotin, S. A., Mcdougal R. A., Hines M. L., & Lytton WW.
Conference NameSociety for Neuroscience 2014 (SFN '14)
KeywordsSFN, Society for Neuroscience

``Bump attractors'' are hypothesized to enable short-term memory via persistent activation in prefrontal cortex (PFC). They have been primarily assessed in terms of electrical mechanisms, without attention to molecular events. To assess this contribution, we developed a multiscale model in NEURON going from molecular to network levels, assessing contribution of calcium (Ca) release from endoplasmic reticulum (ER) to alteration in hyperpolarization-activated cyclic-nucleotide gated channels (HCN) thought to provide continued activity via rebound. The network had 800 neurons arranged in 6 cortical layers. Neurons included pyramidal (E) and 2 interneuron (I) types, with Na, K, Ca, and HCN channels. Cells connected with AMPA/NMDA/GABAA synapses using data from M1. Metabotropic glutamate receptors (mGLUR) produced inositol triphosphate (IP3). Intracellular components included: Ca, Ca buffers, ER Ca stores, IP3, ER IP3 receptors (IP3Rs; release ER Ca), sarco/ER Ca-ATP-ase pumps (SERCA; pump Ca into ER), Ca extrusion pumps, E cell HCN regulated by Ca-bound protein kinase. 18 s of reaction-diffusion simulation ran on 24 Intel XEON CPUs in 7 minutes. Stimulus-induced depolarization led to Ca influx via NMDA and L-type channels. After a delay, mGLUR activation led to ER Ca release via IP3Rs. These factors increased HCN conductance and firing (0.5-8 Hz), lasting 5-10 s. During this time, alpha oscillations decreased, and beta/gamma increased. Non-stimulated cells were suppressed from more inhibition via extra drive from activated to I cells. The network encoded stimulus strength in the ratio of firing rates of stimulated vs non-stimulated neurons (firing-rate distinction; FRD). The network supported stimulus-induced switching between 2 populations (P1, P2) with sensitivity to inter-stimulus delays. Short delays suppressed P2's response, due to P1's dominance. Free Ca regulated FRD and was manipulated via parameter changes, e.g.: 1. Ca extrusion pump time constant (tau) had an inverted-U relationship with FRD: slow tau caused Ca to saturate all neurons; fast tau prevented Ca from having time to regulate neurons. 2. Lowering concentration or binding rate of Ca buffers caused Ca to saturate all neurons, reducing FRD. 3. ER Ca stores modulated network excitability: both SERCA rate and priming time of ER stores correlated positively with FRD, since both provided ER with more Ca (released after mGLUR stimulation). The network therefore supported nonsynaptic plasticity, a Ca-dependent memory trace of neuronal excitability modulation. Our model demonstrates nanoscale electrochemical interactions may lead to persistent activity associated with working memory.