Neuroenergetics and the Kinetic Design of

Excitatory SynapsesPaper by: David Attwell and Alasdair Gibb

Nature Reviews Neuroscience Vol 6 (2005)

DSI Artificial Cognitive Memory Journal Club09 February 2011

Presented by: An Jingzhi

Content Overview of the Paper

Analysis presented in the Paper

- Energy Budget of the Brain - AMPA Receptor Affinity - NMDA Receptor Affinity - Glutamate Removal - Glutamate Receptor Signaling Bandwidth

Conclusions and Perspectives Presented

1. Overview 2. Analysis Presented 3. Conclusions

The Big Picture

1. Overview 2. Analysis Presented 3. Conclusions

Processing Power is Limited by ENERGY!!

Metabolic Energy – ATP - Food

Multitude of Uses in Brain

Consequences

Ways to Deal with IT Limited Size

Be Thrifty 节流• Efficient Wiring e.g. smaller neuron size, dist btw

• Efficient Coding e.g. sparse coding

Gain More开源• Increase Blood Supply• Extract More Resources from the Supply (i.e. glucose and O2)• Evolve denser vascularisation

Brain’s Power to Process Info

Speed of Processing by Individual Neurons

Dendrite -Subthreshold Sypnatic Potential

Axon – Action Potential

“ … To investigate how the brain’s energy supply limits the maximum rate at which brain can compute, and how the molecular components of excitatory synapses have evolved properties that are matched to the information processing they perform ”

The Big Picture

1. Overview 2. Analysis Presented 3. Conclusions

Processing Power is Limited by ENERGY!!

Proof of Limitation using Theoretical Energy Budget

Chemical Kinetics Discussion of the Biological Design of A Excitatory Glutamate Synapse

• Synaptic Design Links Apparent Disparate Parameters

Apparently independent aspects of the brain’s design, such as energy supply, receptor kinetics and affinity, synaptic bouton anatomy and transporter properties are intimately related to each other

• Mechanism of Information Retention and Extraction at Post-Synaptic Membrane

Proposed the collaborative role of various post-synaptic receptors in extracting the temporal components from the glutamate concentration increase caused by pre-synaptic action potentials

Energy Budget of Brain

1. Overview 2. Analysis Presented 3. Conclusions

1. Brain Typically Processes Information on Millisecond ScaleSpeed of Processing by Individual Neurons

Dendrite - Subthreshold Sypnatic Potential Axon – Action Potential

Typical τm ≈ 1- 20ms

Neuronal dendrite cut-off ≈ 200Hz (5ms)

Mean in-vivo firing ≈ 4 Hz (250ms)

Max in-vivo firing ≈ 100-300 Hz (3.33 – 10 ms)

>> Speed of Information Processing is matched at dendrite – axon level

Energy Budget of Brain

1. Overview 2. Analysis Presented 3. Conclusions

2. The Energy Supply to Brain is Limited… so speed of processing cannot be faster than ms scale

Neurons are already designed to minimize Cm and it Cannot keep decreasing- Need membrane- Increased sensitivity to noise

More FlexibleManipulated by channel insertion… But increases energy demand to reset ionic balance!

Energy Budget of Brain

25 %Housekeeping

75 %Signaling Related

10%Maintenance of

Resting Membrane Potential of Neuron

3%Maintenance of

Resting Membrane Potential of Glial

87%Scales with the

average Firing Rate

Power Action Potential

Pre and post synaptic flux

Transmitter Recycling

x 10 = 100 %

x 1.15= 100 %

Rate of ATP consumption is inversely proportional to Rm

Synapses

Chemical Electrical

Inhibitory Excitatory

Glutamate AcetylcholineReceptors

MetabotropicIonotropic

Secondary Messenger System

mGluR

AMPA

NMDA

kainate

A Little Refresher…

Non-NMDA

1. Overview 2. Analysis Presented 3. Conclusions

AMPA Receptor Affinity

1. Overview 2. Analysis Presented 3. Conclusions

Need to have kinetics that match up to the millisecond time scale, require fast glutamate unbinding and low glutamate affinity.Decay time constant of AMPA current = effect of glutamate unbinding + effect of kinetics of channel gatingAmplitude weighted decay time constant = ~0.84ms

>> Matched to information processing speed of brain

Glutamate Removal

1. Overview 2. Analysis Presented 3. Conclusions

To prevent receptors from switching off by desensitization instead of deactivation; and the lost of information at synapses

>> lower glutamate concentration in synapse on ms scale

Diffusion- One bouton releases glutamate- Need small synapse diameter (<1um diffusion time is <1ms)

Glutamate Transporter- Large synapses- High frequency AP- Multiple bouton release

>> glutamate transporter need to work on the time scale of 1ms

MechanismOverall cycle time

of glutamate transporter is

~70ms;

but initial removal step occurs at

~1 to 3ms;

High Transporter Density200/um3 EAAT4 neuronal transporter

20800/um3 glial GLAST + GLT1 transporter15200/um3 glial transporters

Within 1ms of release, 4000 molecules of glutamate will encounter ~ 8000 – 12000

transporters

NMDA Receptor Affinity

1. Overview 2. Analysis Presented 3. Conclusions

Unbinding rate constant = 5 s-1; is 400x slower than AMPAConsequently, dissociation constant is also 400 times slowerAmplitude weighted decay constant of 150ms>> much slower than ms scale, factor other than energy usage is important

NMDA Receptor Affinity

1. Overview 2. Analysis Presented 3. Conclusions

>> COINCIDENCE MEDIATION (10s of ms) high-affinity receptor to temporally integrate information from low-affinity AMPA receptors

Glutamate Removal

1. Overview 2. Analysis Presented 3. Conclusions

“ design of transporter is set by the need for transporters to have sufficient accumulative power to lower the extracellular glutamate concentration below the range that will tonically activate or desensitize glutamate receptors”

>> stoichiometry of glutamate transporters is determined by the timescale over which NMDA receptors mediate coincidence detection

Signaling Bandwidth

1. Overview 2. Analysis Presented 3. Conclusions

e) Includes also background study on kainate and glutamate; normalizing responses to body temperature using Q10=2.5;

>> existence of several receptor types with different kinetics allow neurons to carry out different functions according to the duration of incident elevations in glutamate concentration.

>> combination of AMPA, NMDA and mGluR receptors provides fairly efficient sampling of the entire duration range from 0.033ms to 20s

a-d) Response of glutamate receptors to increases in glutamate concentration of different durations. (akin to the arrival of high freq train of A.P.)Line:: response at the end of each stepCircles:: peak response produced by each step>> occurrence and duration of a glutamate elevation can be encoded by AMPA receptor from ~ 0.1-10.0 ms>> NDMA has incomplete desensitization and is responsive up to ~ 300ms of glutamate elevation

Summary

1. Overview 2. Analysis Presented 3. Conclusions

Energy constraint limits speed of information processing in brain to millisecond scale

AMPA receptors may have therefore evolved to function on the millisecond scale

Consequently, glutamate removal from AMPA must work on a similar time scale

NMDA receptors have kinetic properties catered to its role in synaptic plasticity

Glutamate transporters have ionic stoichiometry set by the demand of NMDA; to avoid tonic activation of NMDA that might result in cell death

Combined response of all glutamate receptors decomposes glutamate concentration increase triggered by the incoming action potential stream into different temporal components

Proposed Explanations

1. Overview 2. Analysis Presented 3. Conclusions

Energy constraint limits speed of information processing in brain to millisecond scale:: why not faster mechanism? - limited capability to increase capillary density - limitation in energy supplied in food - co-evolution with musculoskeletal system

NMDA receptors have kinetic properties catered to its role in synaptic plasticity:: why this particular time scale (<100 ms)? - minimize temporal jitter btw info converging from diff pathways

Combined response of all glutamate receptors decomposes glutamate concentration increase triggered by the incoming action potential stream into different temporal components; why not finer division?

Future Work

1. Overview 2. Analysis Presented 3. Conclusions

Investigate different information processing speeds for receptors at different brain region and how they relate to the local energy supply

To perform a similar analysis on inhibitory synapses

Thank You.

1. Overview 2. Analysis Presented 3. Conclusions