Neuronal energy consumption

Question 1

Which part of the AP is responsible for the most energy consumption?

Answer 1

The overlap where Na is still entering and K is leaving the neuron, because this is a futile cycle
It’s unclear why there’s such heterogeneity in AP costs i.e. overlap size. THe overlap can be minimised by altering kinetics of the channels (Na-activated K+ channels, if sufficiently fast, could minimise the overlap. So why don’t they?)

Question 2

What do we need in order to work out neuronal energy consumption

Answer 2

Resistance, capacitance, currents, reversal potentials, avogadro and faraday constants. Work out how many Na ions have entered the cell, then divide by three, because of stoichiometry of Na/K ATPase

Question 3

Energy consumption of a neuron is dependent upon:

Answer 3

Mean spike rate
Diameter of neuron
Length of neuron
Myelination (the cost of myelination could easily be repaid in reducing transmission costs except for the fact of maintaining oligodendrocyte resting potential - therefore myelin evolved to increase transmission speed, not to reduce costs.)
Synaptic transmission

Question 4

Direct vs indirect consumption

Answer 4

Direct - e.g. Na/K ATPase, Proton pump
Indirect - e.g. Glutamate reuptake, calcium extrusion (or choline reuptake, removal of acetyl Co-A from Krebs cycle for cholinergic neuron)
Indirect costs rely on the previously set up ion gradients (and reduce these). Different synaptic transmission costs depending on which transmitter is used

Using reaction rates for these, we can determine an energy budget

Question 5

Which processes and receptors account for the most energy usage?

Answer 5

AP>NMDA receptors>non-NMDA receptors>Glu recycling>presynaptic Ca>mGlu receptors>exo/endocytosis

So second messenger systems are way less costly - because they’re slower.

In rat grey matter at 4Hz firing rate, 40% of energy consumption is APs, 47% of energy consumption is synaptic transmission, 13% is resting potential. The fact that 13% is basically to counterract the leak current raises the question - why haven’t we evolved out of it?

Voltage insensitive sodium leak current is via the NALCN channel (Lu et al 2007). It’s a nonselective cation channel that controls the excitability of neurons (perhaps prevents a residual sodium effect similar to the residual calcium?) and is required for normal respiratory rhythm. Think of it like an overflow valve?

Question 6

Oxygen supply to insect brain

Answer 6

Air-filled tracheae ramify through the brain, carrying oxygen which neurons cannot store. Glucose comes via hemolymph, which bathes all the organs. This places a limit on insect brain size, because of SA:V ratio affecting diffusion

Question 7

Oxygen supply to vertebrate brain

Answer 7

Blood-filled capillaries ramify through the brain, carrying oxygen and glucose. They’re covered in pericytes and astrocytes (more than 99% covered by astrocyte foot processes), which coordinate blood flow to neuronal requirements, via glissandi (Kuga et al 2011), large-scale Ca waves that travel between astrocytes, emerge depending on neuronal activity, and decrease RBC flow.

Question 8

BOLD - what is it? how is it produced?

Answer 8

Blood Oxygen Level-Dependent Contrast Imaging
Relies on blood releasing oxygen to active neurons at a higher rate than inactive
Shows relative differences in blood oxygenation rather than absolute
An increase in deoxyhaemoglobin, which is paramagnetic, causes greater dephasing of the magnetic field.

Question 9

BOLD - limitations

Answer 9

Hard to infer neural activity
Low resolution (because responds to many neurons)
Slower changes than neural activity changes, i.e. low temporal resolution

Question 10

Influences of energy consumption on neural behaviour

Answer 10

The relationship between resting and signalling costs helps determine coding pattern - low resting costs favour large populations of rarely active neurons, i.e. population/gnostic coding.
Reducing time constant to increase speed requires increasing conductance, which is costly - hence myelination instead.
Reducing noise (intrinsic or extrinsic) requires amplification (costly) or averaging across neurons/vesicles/ionchannels, still costly.
The existence of noise means neurons use higher spike rates than would be predicted by an efficiency-maximising model (i.e. more efficient would be exponential distribution), because when you get to v low frequencies it's indistinguishable from noise.
Ballistic control e.g. used to throw a ball is more efficient, since you don't need to encode information than can be predicted.
Sensory systems are designed to extract only the specific useful information (e.g. contrast detectors in the retina), rather than encoding a broadband signal. This saves energy in higher processing centres too.

Question 11

AP shape variability

Answer 11

Varies between animals and within animals (e.g. Kenyon cell has a really slow AP).
In most cases, AP uses more energy than it needs to to charge the membrane, because of the overlap
Mouse Thalamocortical Relay Neuron is almost 100% efficient, squid giant axon v inefficient
Differences in AP cost correlate with overlap Na load
All this variability means that looking only at squid giant axons skewed our view - APs are actually much more efficient than we thought

Question 12

Graded vs action potentials

Answer 12

Graded potentials are more efficient, because they convey more information per unit time and require less ion flux (esp less overlap).
Graded potentials carry more information because the timescale of APs obscures v short-term deflections in the signal, and there’s more intrinsic noise and non-linearity from the channels generating the AP.

BUT despite both these, neurons must use APs to convey info over long distances without decaying or accumulating noise.

Question 13

Soma externalisation

Answer 13

If the soma is small relative to the neurite, it’s efficient to be bipolar (because dense brains don’t have space for soma stuck on the end of neurites).
But if the soma is big relative to neurites, like in arthropods and mollusks, it’s efficient to be pseudounipolar, so the signal doesn’t have to go through and be attenuated by the soma

Hesse and Schreiber 2015 - previously it was argued that soma externalisation is for wiring costs (by having a soma layer and a neuron layer), soma access to nutrients, and the use of graded potentials. This paper was the first to suggest it was for energy conservation, because the soma has a much larger membrane to depolarise. Central location of soma, in vertebrates, is to allow recurrent connections and perhaps less space for protein synthesis is required (due to outsourcing of organelles to proximal dendrites)
Alt view = dense brains.

Question 14

Wiring efficiency

Answer 14

The extent to which neurons are connected to each other (i.e. everything connected to everything, or just to a few) affects amount of axon needed, and hence costs
The more brain regions, the more wiring needed
So bigger brains are more dense, to minimise this

Question 15

Component placement

Answer 15

Exact positioning of the different components affects amount of wire needed
Cephalisation - moving the processor nearer to the sensors means the majority of axons get shorter
Centralisation - peripheral structures move towards each other - CNS evolves
In C. elegance, almost every soma is positioned as a computer model suggests for minimum overall axon length. Outliers are for specific structures, e.g. egg sac.
The placement of regions in the PFC also minimises length of axons, suggesting optimal component placement in mammalian brains

Question 16

Matching brain size to ecology - expensive tissue hypothesis

Answer 16

Regions involved in processing sensory information is matched to the habitat of the species, e.g. hedgehog dominated by vision bc diurnal, star-nosed mole closely related but dominated by mechanosensation. So no wasted brain space.
Proposed gut-brain trade off, because both are metabolically expensive (though we now think the real trade-off was with the adipose tissue)
Higher quality diet meant less bulk (so less gut needed), more energy available (so larger brain possible) and more complex foraging behaviour (so larger brain needed). But did improved diet increase brain size, or vice versa?

Question 17

Plasticity and brain evolution

Answer 17

Sherwood and Gomez-Robles 2017 review - humans have increased their capacity for plasticity during evolution.

a) humans and chimpanzees reach adult-like dendritic spine morphology in PFC later than macaques do (so continued developmental pruning window)
b) humans reach peak myelination late in adolescence, later than other primates. White matter plasticity is important!
c) human brains show more fluctuating asymmetry than chimpanzee brains, which can indicate increased developmental plasticity
d) heritability for cortical organisation is relatively low in humans
e) human specific alterations in genes associated with plasticity, inc in FOXP2 involved in development and function of circuits for speech and language

Question 18

Reduced brain size

Answer 18

Homo floriensis lived at the same time as Homo sapiens (died out 60-70kya), but geographically isolated on Flores Island, Indonesia. They were much smaller, with smaller relative brain size. People used to argue that they were just microcephalic humans, because two closely related species couldn’t have such different brain sizes. But 3 species of pygmy hippos vs modern hippos showed a brain size reduction much greater, so disproved this argument. Myotragus is an island goat, also with smaller brain and eyes. Island rule = species isolated on an island will converge to an intermediate size (some think this is because on an island with few predators, animals can be their optimum size, with small species getting bigger and big getting smaller.
Marquet and Taper 1998 - As island area gets smaller, variation in bird size decreases.

Domesticated species often have reduced brain size relative to their wild counterparts

Question 19

Reduced visual systems

Answer 19

Fish living in pitch black caves have much smaller visual processing regions, but not a smaller overall brain. Darwin hypothesised the regions were lost due to disuse, but in fact they were a disadvantage because of energy cost.
Drosophila have smaller eyes the longer they’ve been kept in culture

Question 20

Size vs energy consumption

Answer 20

There is a general correlation, so larger brains require larger supply networks. This is why bigger brains are also more dense. There’s a Law of Diminishing Returns
but e.g. elasmobranch brains are way bigger than those of teleost fish, but with reduced Na/K pump activity, so no higher energy consumption.