Astrocytes as Modulators of Synaptic Information Transfer

Herbert Levine UCSD/CTBPWith V. Volman and E. Ben-Jacob, Tel-Aviv U.

Help from S. Nadkarni (UCSD) and P. Jung (Ohio)

Organization of Neuroglia tissue

One type of glial cell, astrocytes

– are more numerous than neurons

– have processes which associate with synapses – seem to be required for various aspects of neural development, maintenance and neural dynamics

Nedergaard et al., TINS 26, 523 (2003)

Astrocytes

Astrocytes are connected by gap junctions thereby forming a syncytium that is able to propagate signals for large distances

V. Parpura, UC-Riversideglial fibrillary acidic protein (GFAP)tagged with antibody.

The tripartite synapse

From Fields, Scientific American

Tsodyks-Uziel-Markham Synapse

Basic mechanism: depletion of pre-synaptic resources can occur if spikes arrive too frequently

Presynapic firing(actual data)

Postsynaptic current(TUM model)

0 sec 4

This model ignores any possible role for the astrocyte!

SYNAPTIC VESICLE RELEASE

Tsodyks-Uziel-Markham Synapse

Basic mechanism: depletion of pre-synaptic resources can occur if spikes arrive too frequently

Presynapic firing(actual data)

Postsynaptic current(TUM model)

0 sec 4

This model ignores any possible role for the astrocyte!

Reciprocal Communication

• Astrocytes detect released neurotransmitter, e.g. glutamate, by appropriate receptors

• This leads to an activation of the calcium signaling pathway– IP3 levels go up

– Calcium is released from internal stores– May propagate intra/intercellularly

Intracellular Calcium handling

Positive feedback via CICR - excitable dynamics

Calcium waves in glia

QuickTime™ and aVideo decompressor

are needed to see this picture.

Newman Lab, rat retina

Can use mechanical stimulus, but data also exists for neural stimulation

Here, we focus only on the local increase in Ca in one domain

Li-Rinzel model

We use a standard model for Ca dynamics in the glial domain

where the channel current is given by

This term couples to synapse!

(Can also treat stochastically)

IP3 dependence

Glutamate released in the synaptic cleft binds to glia mGluR’s

Leads to an increase in the IP3 messenger

Can lead to Ca oscillation -deterministic-stochastic

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Here, we will focus on an excitation pf a local microdomain of the astrocyte

Typical response should be composed of several elementary “puffs”

Xenopus oocyte, From I. Parker

Effects on the synapse

• Reduced amplitude of EPSC’s• Increased number of spontaneous releases

– Depends on presynaptic mGluR and AMPA receptors– These are connected through the existence of a vesicle pool (that

can become depleted)– Modifies depletion terms in the TUM model

• Increases fidelity of “weak” synapses– Studied by Nadkarni and Jung

• There are also effects on the soma which will not be taken into account here, but are needed for a treatment of the full network.

Enhanced Spontaneous Activity due to Astrocytes

Liu et al. , PNAS, 2004

•Astrocyte is loaded with caged Ca2+

•Upon photolysis, astrocyte released glutamate causes enhanced spontaneous activity.

•Leads to probability of current events (y axis) to shift to the left.

Effects on the synapse

• Reduced amplitude of EPSC’s• Increased number of spontaneous releases

– Depends on presynaptic mGluR and AMPA receptors– These are connected through the existence of a vesicle

pool (that can become depleted)– Modifies depletion terms in the TUM model

• Increases fidelity of “weak” synapses– Studied by Nadkarni and Jung

• There are also effects on the soma which will not be taken into account here, but are needed for a treatment of the full network.

Modulated synapse

Input spike train

Glia calcium level

Synapse coupling factor

Postsynaptic current

From Volman, Ben-Jacob and Levine, Submitted to Neural Comp.

Consequences?

• Given this new level of biophysical detail, can we say something non-trivial about how this might be important for network dynamics?

• To do this, we will turn to recent experimental findings in culture systems done at Tel-Aviv Univ.

Cultured neuron system

Ben-Jacob lab

Mixed cell culture grown on electrode array

Individual neuron spikes measured by spike sorting

Activity in cultured neurons data from Ben-Jacob lab (TAU)

Bursting activity Burst fine-structure

“Spiker neuron”

Recordings from multi-electrode array w. spike sorting;Culture is from rat cortex; includes neurons and glial cells

Zoom in

Synchronized bursting events (SBE)

Activity in cultured neurons data from Ben-Jacob lab (TAU)

Bursting activity Burst fine-structure

“Spiker neuron”

Recordings from multi-electrode array w. spike sorting;Culture is from rat cortex; includes neurons and glial cells

Inter-spike interval increments

Red - spiker neuronBlack - regular neuron

Regular neuron has sharper break at 100 msec; spiker neuron has power-law tail up to longer timescale

Indicates sharp transitions to different phases

Spiker model

We propose that the spikers seen in the data arise via neurons with excitatory self-synapses which are coupled to astrocytic processes

Segal (1991) - Single hippocampal neuron microcultures

Morris-Lecar Neuron

Infinite-period (saddle-node) bifurcation; complexity matched to Li-Rinzel model and TUM synapse

Dynamic K channel

Autaptic oscillator - TUM synapse

We use a Morris-Lecar model biased to exhibit very low frequency firing in the absence of synaptic coupling

Cannot reproduce long time correlations seen in spiker firing, with TUM synaptic model

Autaptic oscillator - with coupling

Recovers observed behavior up to much larger increments

Autaptic oscillator - What is still missing?

Bias current should not be constant, but should arise self-consistently from the network

Adding even simple oscillation leads to a heavier tail (due to sharper bursts)- this is promising, but we need a full network model.

Functional Significance??

• Why is this extra level of complexity important for functional neuronal systems?

• Apparently, the longer-time scale dynamics available to the spiker neurons helps organize the bursting activity of the whole network - creates regulated fluctuations

• Bursts can have different internal structures and be accessed by specific excitation protocols - perhaps these are the templates for associative information processing??

• Next steps - full network model (including soma effects) and try to match to longer intervals; plus use pharmacology to test hypothesized mechanisms.

Spatiotemporal burst organization

Correlation matrix Clustered matrix

Clearly exist different types of burst patterns

Functional Significance??

• Why is this extra level of complexity important for functional neuronal systems?

• Apparently, the longer-time scale dynamics available to the spiker neurons helps organize the bursting activity of the whole network - creates regulated fluctuations

• Bursts can have different internal structures and be accessed by specific excitation protocols - perhaps these are the templates for associative information processing?? Perhaps this is the type of behavior only seen in epileptic brains

• Next steps - full network model (including soma effects) and try to match to longer intervals; plus use pharmacology to test hypothesized mechanisms.

(inter ictal)

(ictal)

)onset(

The challenge of epilepsy - from V. Towle

Roots of epilepsy may lie in oft-ignored brain cellsSeptember 5, 2005Star-shaped brain cells that are often overlooked by doctors and scientists as mere support cells appear to play a key role in the development of epilepsy, researchers say in a study.

This opens up a new vista in efforts to treat epilepsy. It might be possible to treat epilepsy not by depressing or slowing brain function, as many of the current medications do, but by targeting brain cells that have been completely overlooked," said Maiken Nedergaard, MD, PhD.

The findings were published online August 14, 2005, in Nature Medicine.

Summary/Outlook

• We have introduced a phenomenological model of the two-way glia-synapse coupling

• We have argued that this coupling may be evident in culture dynamics– Introduces longer time-scales– Regulates spontaneous activity

• These may allow for new forms of synaptic plasticity and stimulus associativity

• Next step: full network model including astrocytes, learning involving astrocytes?? etc.