Depression and Potentiation of Pyramidal Gamma Rhythms by
Depression and Potentiation of Pyramidal Gamma Rhythms by External Gamma Rhythms Effects of Noisy Drive on Rhythms in Networks of Excitatory and Inhibitory Neurons Börgers and Kopell (2005) Cortical Gamma Rhythms Modulate NMDAR-Mediated Spike Timing Dependent Plasticity in a Biophysical Model Lee, Sen, and Kopell (2009)
Pyramidal Inter-neuronal Gamma Rhythms (PING) • Observed during sensory processsing + memory tasks
Gamma 30 -80 Hz
Pyramidal Inter-neuronal Gamma Rhythms (PING) • • • Generated by networks of I cells and E cells drive and synchronize the I cells gate and synchronize the E cells
PING model • • • Assume all-to-all connectivity Assume no heterogeneity Use Theta model
Theta model of a spiking neuron • • Most currents negligible during PING So, use Theta model (simplification of HH)
Synapses • • Excitatory synapses: AMPAR Inhibitory synapses: GABAA
Connectivity
Intrinsic frequency of the I cells • f. I < f. E for PING
How can PING rhythms be destroyed? • • Phase walkthrough Supression
Phase walkthrough • I cells fire >1 between I cell spikes
Supression • I cells overinhibit the E cells, stopping PING
Phase walkthrough occurs when • • connectivity parameters are not within the phase walkthrough boundary highly stochastic spiking of E cells causes phase walkthrough
Phase walkthrough boundary • • frequency of I group needs to be < freq of E otherwise, I cells improperly gate E cells
Phase walkthrough boundary • • so, f_I = f_P defines the phase walkthrough boundary f_i and f_P are determined by connectivity values
Phase walkthrough occurs when • • • External current to E (I_E) cells is too small I->E connections (g_IE) are too strong Exernal drive to I cells (I_I) is too large
I_E needs to be larger for PING
I_E needs to be larger for PING
g_IE needs to be smaller for PING
g_IE needs to be smaller for PING
I_I needs to be lower for PING
I_I needs to be lower for PING
Phase walkthrough is harder with I->I synapses
Phase walkthrough is harder with I->I synapses
Phase walkthrough occurs when • • connectivity parameters are not within the phase walkthrough boundary highly stochastic spiking of E cells causes phase walkthrough
Stochastic spiking of E cells -> phase walkthrough
Stochastic spiking of E cells -> phase walkthrough
Stochastic spiking of E cells -> phase walkthrough
How can PING rhythms be destroyed? • • Phase walkthrough Supression
Suppresion
Suppresion occurs when. . . • • Parameters are not within the suppression boundary Highly stochastic spiking of I cells causes suppression
Suppresion boundary • Parameter space that leads to E-cells NOT being dominated by inhibition
Suppresion occurs when. . . • • External drive to E cells (I_E) is too small External drive to I cells (I_I) is too large I->E connection (g_IE) is too strong I->I connectiviy (g_II) is too weak
I_E must be larger for PING
I_E must be larger for PING
g_IE must be smaller for PING
g_IE must be smaller for PING
I_I should be smaller for PING
g_II should be larger for PING
g_II should be larger for PING
Suppresion occurs when. . . • • Parameters are not within the suppression boundary Highly stochastic spiking of I cells causes suppression
Stochastic I-cell spiking causes supression
Stochastic I-cell spiking causes supression
Stochastic I-cell spiking causes supression
Stochastic I-cell spiking causes supression
Bistable region • • where, depending on init. conditions, can get PING or suppression
Bistable region
Sparse connectivity and heterogeneity
Frequency range for which PING is robust • • Gamma frequencies, thus "PING" As we saw, too low-freq oscillations weren't robust and using the parameter choices we found to lead to stable oscillations, can't really get stable oscillations for frequencies > 90 Hz
I_I
I_E I_I
g_IE I_I
g_II I_I
g_EI I_I
g_EE I_I ? ? ?
Gamma rhythms found during learning + memory tasks
Gamma 30 -80 Hz
Gamma rhythms found during learning + memory tasks • suggests gamma rhythms important for memory
Basic Oscillator with external rhythm input
2 -cell PING oscillator
Input rhythm • • • Regular, discrete spikes varied frequency of input over trials input spikes activated AMPA + NMDA currents
STDP via glutamatergic model
External inputs had Potentiating and depressing effects on PING oscillators
Strict rhythm in external input required for effect
Assymetry in STDP rules also required
Assymetry in STDP rules also required
Assymetry in STDP rules also required
Multiple input rhythms • • • With multiple input rhythms, higher coherence -> more potentiation but slightly incoherent inputs -> more potentiation than perfectly coherent ones
Multiple input rhythms
Multiple input rhythms
Theta rhythms • • Theta rhythms didn't seem to have an effect But pauses in gamma rhythm at theta durations had potentiating effect
Mechansim for entraining oscillators • • • Authors think this mechanism may be used during tone-training in the auditory system Also, a similar mechanism though to be involved in olfaction (Dmitry Rinberg's lab) Suggests similar mechanisms may be important throughout the brain
Limitations? • • • Only used 2 -cell network No heterogeneity Bistable regions? Didn't test init condits. . .
- Slides: 75