Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

Dynamics of neuronal action potential encoding

The ability of neuronal populations to rapidly encode varying stimuli and respond quickly to abrupt input changes are crucial for basic neuronal computations, such as coincidence detection, grouping by synchrony, and spike-timing-dependent plasticity, as well as for the processing speed of neuronal networks.  Our theoretical and computational analyses predicted [1, 2, 3, 4, 5] that these abilities are linked to the fast-onset dynamics of action potentials.  Using a combination of whole-cell recordings from rat neocortical neurons [6], analytically tractable non-equilibrium statistical mechanics models [3] and computer simulations [5, 7], we recently provided the first experimental evidence for this conjecture and proved its validity for the case of action potential initiation in the axon initial segment, typical for cortical neurons [7].  We found neocortical neurons with rapid-onset of action potentials capable of phase-locking their firing to signal frequencies up to 300 - 400 Hz [6, 7].  Populations of a few thousand neurons are capable of responding within 1 - 2 ms to subtle changes in their input current. We found that the ability to encode high frequencies and response speed were dramatically reduced when action potential onset was slowed by experimental manipulations or was intrinsically slow due to an immature action potential generation mechanism [7].  Multicompartment conductance-based models reproducing the initiation of spikes in the axon initial segment could encode high frequencies only if action potential onset was fast at the initiation site (e.g., attributable to cooperative gating of a fraction of sodium channels as previously predicted by us from biophysical observations), but not when a rapid onset of somatic action potential was produced solely by lateral currents.  These findings indicate that rapid-onset dynamics are a genuine property of cortical action potential generators.  Furthermore, it strongly suggests that they are essential for high bandwidth computation and information transfer in cortical circuits.

Fig. 1: Two ways to study in-vivo-like fluctuation-driven action potential activity under controlled conditions.  (a) Schematic representation of the ongoing synaptic drumfire to which neurons in the CNS are typically exposed.  Sparks represent active synapses.  Cortical pyramidal neurons will typically receive synaptic inputs at a rate of several kilohertz.  (b) & (c) Two alternative experimental approaches to emulate the resulting input fluctuations and register the fluctuation-driven activity in-vitro: whole cell current injection  (b) and CoDyPs (c), here depicted for a neuron cultured on a circular extracellular electrode.  In contrast to the whole cell stimulation/recording, CoDyPs offers extended recording and stimulation/recording of multiple neurons simultaneously.


Although our results clearly verify our previous theoretical predictions and establish an unexpected temporal accuracy of action potential initiation and encoding, the biophysical and molecular basis of this phenomenon remains elusive.  Studies of single neuron computation that preserve natural operation conditions so far rely on invasive stimulation methods that are severely constrained by limited recording duration and low yield.  Combining noninvasive optogenetic stimulation with multichannel multi-neuronal recordings promises to overcome these fundamental limitations and to open up an avenue to effectively investigate the encoding dynamics of neurons under different conditions.  The characterization of single neuron computation, however, requires a precise knowledge of the input in order to compute, for instance, quantities like the spike-triggered average or covariance of the input, or to describe correlation gain and firing rate adaptation in the dependence of the stimulus properties.  An optical, noninvasive stimulation approach would be instrumental for such studies, but only if (1) the induced conductances are highly reproducible with correlation times suitable to mimic fluctuating synaptic conductances, (2) waveforms can be precisely designed and predicted, and (3) conductance waveforms can be stably induced in long-term experiments.  In order to study dynamical encoding properties effectively using optogenetics, we have developed and characterized continuous dynamic photostimulation (CoDyPs), a novel approach to mimic in-vivo-like input fluctuations noninvasively in cells transfected with channelrhodopsins [8].  Even during long-term experiments, cultured neurons subjected to CoDyPs generated seemingly random, but highly reproducible action potential patterns.  In voltage-clamped cells, CoDyPs induced highly reproducible current waveforms that could be precisely predicted from the light-conductance transfer function of the channel.  When combined with non-invasive spike-detection, CoDyPs allows the acquisition of orders of magnitude larger data sets than previously possible, opening the way for studies of dynamical response properties of many individual neurons.


[1] Naundorf, Wolf, and Volgushev, Nature 440:1060 (2006)
[2] Naundorf, Wolf, and Volgushev, Nature 445: E2 (2007)
[3] Wei and Wolf, Phys. Rev. Lett. 106:088102 (2011)
[4] Tchumatchenko and Wolf, PLoS Comput. Biol. 7(10): e1002239 (2011)
[5] Huang, Volgushev, and Wolf, PLoS One. 7(5): e37629 (2012)
[6] Tchumatchenko, Malyshev, Wolf, and Volgushev, J. Neurosci. 31(34): 12171 (2011)
[7] Ilin, Malyshev, Wolf, Volgushev, J. Neurosci. 33(6): 2281 (2013)
[8] Neef, El Hady, Nagpal, Bröking, Afshar, Schlüter, Bamberg, Fleischmann, Stühmer and Wolf, in revision, 1305.7125

Contact:  Fred Wolf 

Members working within this Project:

 Fred Wolf 
 Andreas Neef 

Former Members:

 Ahmed El Hady 

Selected Publications:

R. Samhaber, M. Schottdorf, A. El Hady, K. Bröking, A. Daus, C. Thielemann, W. Stühmer, and F. Wolf (2016).
Growing neuronal islands on multi-electrode arrays using an accurate positioning-μCP device
J Neurosc Meth 257(1):194-203.

H. Arnoldt, S. Chang, S. Jahnke, B. Urmersbach, H. Taschenberger, and M. Timme (2015).
When Less Is More: Non-monotonic Spike Sequence Processing in Neurons
PLoS Comput. Biol. 11(2):e1004002. download file

V. Ilin, A. Malyshev, F. Wolf, and M. Volgushev (2013).
Fast Computations in Cortical Ensembles Require Rapid Initiation of Action Potentials
J. Neurosci. 33(6):2281-2292.

M. Huang, M. Volgushev, and F. Wolf (2012).
A Small Fraction of Strongly Cooperative Sodium Channels Boosts Neuronal Encoding of High Frequencies
PLoS ONE 7(5):e37629. download file

T. Tchumatchenko, A. Malyshev, F. Wolf, and M. Volgushev (2011).
Ultrafast population encoding by cortical neurons
Journal of Neuroscience 31(34):12171-12179.

T. Tchumatchenko, and F. Wolf (2011).
Representation of dynamical stimuli in populations of threshold neurons
PLoS Comp Biol 7(10)(e1002239).

W. Wei, and F. Wolf (2011).
Spike Onset Dynamics and Response Speed in Neuronal Populations
Phys. Rev. Lett. 106(8):088102. download file

B. Naundorf, F. Wolf, and M. Volgushev (2007).
Hodgkin and Huxley model - still standing?
Nature 445:E2-E3.

B. Naundorf, F. Wolf, and M. Volgushev (2006).
Unique features of action potential initiation in cortical neurons
Nature 440:1060-1063.