Neuron fuel and function

In the comments following a previous post Dustin linked to this rather lovely paper from the early 1970s, back when I was still at school and marathon racing my kayak.

This is one of the nicest figures:

The lower section is the interesting part. CJ had not eaten for 50 days and had a (very) fasting blood glucose of 4.0mmol/l when he received around 0.1 IU/kg of insulin by intravenous injection. Please don't try this at home. You can see that a) he is still insulin sensitive and b) his blood glucose bottomed out at around 0.5mmol/l by 60 minutes. Throughout this period he was asymptomatic. No hypo.

CJ was not running his brain on glucose. The upper section shows a rapid and sustain increase in B-OHB extraction (the dark hatching between the arterial line and jugular bulb line for B-OHB)  by his brain through this period of time. Ketones, at levels in excess of 11.0mmol/l, can sustain apparently normal brain function. Given an alternative fuel source this would seem to put the level of blood glucose needed for normal neural function at some (non determined) value of less than 0.5mmol/l.

Drenick et al did not look at fatty acid extraction by the brain. That's a pity, but understandable. No one expects the brain to metabolise palmitic acid. Well, perhaps we should say that neurons should not metabolise palmitic acid. Astrocytes do. Astrocytes are ketogenic and are in a perfect position to supply ketone bodies to neurons using the monocarboxylate transporters ubiquitous on them. I hope to come back to this by the end of the post.

Of course, a 50 day starvation period is not exactly the normal human predicament and we could argue that the normal human brain is glucose dependent. This too, I think may be a very debatable point.

Work has been done with humans under hypoglycaemia with brain function supported by either lactate or pyruvate by intravenous infusion. They're fairly effective, not perfect, but there are limits in how far you can push the metabolism of human volunteers.

Rats are not so fortunate.

If we go to Figure 1 from the rather nice paper emailed to me by Edward

we can see that, under the influence of a massive 20 IU/kg of insulin, there is an almost complete loss of plasma glucose and a slightly more complete loss of brain response to limb stimulus in the insulin-only, profoundly hypoglycaemic group, top row of  section A. This has occurred by two and a half hours. The next row has had the hypoglycaemia corrected with glucose (i.e. it's essentially a control group) and has a normal response to stimulus at 4 hours. The lower row shows the effect of lactate in supporting brain function during four hours of persistent, profound, uncorrected hypoglycaemia. You have to note that progress from left to right is time in milliseconds after stimulus and that there is a clear cut delay of about 10 ms in the response time under pure lactate compared to under glucose. This is reiterated in section D, where the response can be seen to be delayed and blunted when compared to the glucose supported data of curve C.

This has led the authors to speculate that, heresy of heresies, there may actually be an absolute need for some glucose by the brain! Strange I know, but... They're not sure of this, just speculating. There are other potential explanations.

Now, most people do not walk around with a blood lactate of 9.0mmol/l. Perhaps most people really do run their brain on glucose?

This seems very unlikely. Or, rather, it seems very unlikely that the neurons in the brain run on glucose. Astrocytes certainly do. But one of the main functions of astrocytes appears to be to manufacture lactate from glucose (directly or from stored glycogen) and deliver it to neurons as a one step conversion fuel giving pyruvate, which can enter the TCA as acetyl-CoA without any messy glycolysis. There is an awful lot of information in this paper.

I put up this nice illustration previously, in the Protons thread:

Neurons are spared glucose. Why?

Mitochondrial glycerol-3-phosphate dehydrogenase. Glucose, during glycolysis, is quite able to input to the electron transport chain through an FADH2 based input at mtG3Pdh, which can reduce the CoQ couple and set the ETC up for reverse electron flow through complex I, with the generation of superoxide as this occurs. Modest superoxide is a Good Thing, especially if you want to signal for mitochondrial biogenesis or cell division. Excess superoxide is a potent signal for apoptosis. Apoptosis is verboten for CNS neurons because any information stored in their synaptic connections will be lost along with the cell. Replacing the cells will hardly replace the memories and appears equally forbidden. There is a need for immortality without reproduction in neurons, for their survival in excess of the lifespan of an organism needing a functional memory in a learning brain.

Feeding lactate through pyruvate to acetyl-CoA does not drive CoQ pool reduction "ahead" of the throughput of electrons coming in from complex I. Neurons do not want to generate insulin resistance. Avoiding glycolysis looks (to me) to be the way they do this. Generating hyperglycaemia looks like a way to overcome the normal lactate shuttle and of forcing glucose directly in to neurons. Enough apoptosis and eventually neural loss just might show as memory loss.

Hyperglycaemia and Alzheimer's...

Generating large amounts of superoxide in astrocytes during glycolysis is not damaging to the neurons supported by the derived lactate. Astrocytes are certainly replaceable, although there seems to be some debate about cell division vs stem cell recruitment. Astrocytes are also able to divide unreasonably rapidly and form various grades of brain tumour. They are common and frequently aggressive. Neuron derived tumours are much rarer and are usually derived from embryonic cells giving medullablastomas rather than being derived from mature neurons. That seems to fit the metabolic arrangements in the brain rather neatly.

This takes us back, eventually, to palmitate as a ketogenic energy supply to the brain via astrocytes. Again, an FADH2 input through electron transporting flavoprotein dehydrogenase can couple with hyperglycaemia to generate reverse electron flow through complex I giving excess superoxide generation. I consider this to be why free fatty acids are excluded from neurons. It's not that FFAs generate excess superoxide per se, they don't. But combined with hyperglycaemia they certainly do, especially palmitate and the longer of the saturated fatty acid series. You really don't want this happening in a cell whose remit is immortality.

Ketones and lactate do not drive reverse electron flow through complex I. Glucose can. Palmitate certainly can. What you want from a metabolic fuel depends on the remit of your cell types. Neurons within the brain preserve information by their continued existence. This is best done by burning lactate or ketones. NOT glucose and, of course, not FFAs. Anyone who claims that glucose is the preferred metabolic fuel of the brain has not though about what a neuron has to do and what an astrocyte actually does do. Or much about the electron transport chain.

Peter