Uncoupling and weight loss

I've spent the last three posts making the point that fatty acid oxidation (supplemented by ketosis) increases the amount of ATP (and energy yield of ATP hydrolysis) available per unit oxygen consumed. This is particularly clear under the conditions of extended, intensely hypocaloric eating described by Phinney, where exercise can be sustained for longer, at a lower VO2, than on a mixed diet.

Now, oxygen consumption is a surrogate for caloric output. How many calories you "spend" per unit oxygen consumption is a complex calculation and depends on your fat to carb ratio.

But we don't run on calories. We run on ATP (mostly), or rather we run on the energy yielded from ATP hydrolysis.

To make that absolutely clear: We know, from Phinney, that under pure fat oxidation, we can generate enough ATP energy (physical treadmill load) to sustain moderate exercise by using less calories (ie lower VO2) on fat oxidation than on mixed diet oxidation. The increase in ability shows as a 25% drop in VO2, ie a 25% drop in calories needed to get enough ATP energy to move at 70% VO2 max.

That, to me, is pure survival adaptation. It's elegant, neat, cool etc.

It's not providing a metabolic advantage for weight loss.



Now, once again, I must wander off in to rodent studies.

If you take a rat or a mouse and feed it a genuine ketogenic diet you get some interesting effects. Let's look at this small study in rats. Here's heat output. Red is chow fed, grey is ketogenic:

Day or night, energy output is lower for the ketogenic rats compared to the chow fed rats. Phinney got a 25% drop in VO2 on his treadmill, the rats have calorie output down by an average of 11%, at a similar RQ. Running on fat (+/-ketones) requires less calories to generate adequate ATP levels.

Note, these are not real heat outputs in the rats. No one measured heat flux in any way. They're calculated from the VO2. They're done using the software built in to a CLAMS device around well accepted values of calories used per litre of oxygen consumed. This drop in calculated heat output, in itself, is not a surprise in view of Phinney's work.

What is surprising is that VO2 actually increased to generate this reduced heat output:

The rats should be using less oxygen per minute to produce their whole-body ATP energy requirement running on fat, according to Phinney. And me. And the chart. They're not, they're using more, in absolute terms.

The conclusion here is that the VO2 has gone the wrong way. So we have to ask: What is the difference between a fasting, exercising human on an RQ of 0.66 and a ketogenic rat slumming around its cage with a very similar RQ of 0.7?


The rats are uncoupled. They pump protons through complexes I, III and IV but a significant number of those protons drop straight back in to the mitochondria through open uncoupling proteins. Calories and oxygen are used (at the same RQ as any other more productive oxidation) but no ATP is produced from any protons which do not use ATP synthase. VO2 moves in two directions. It goes down (and so do calories used) due to switch from glucose oxidation to fat oxidation. It goes up due to uncoupling. The overall effect, up or down, on VO2 depends on the relative effects of RQ change, uncoupling, gluconeogenesis, NEAT and actual exercise.

Phinney's treadmill walkers had high FFAs and high ketones but absolutely no suggestion of any sort of uncoupling. Why?

To get any further we have to go to the Protons thread back in 2014.

Uncoupling proteins are kept closed by cytoplasmic ATP. And there is always enough cytoplasmic ATP in a functional cell to keep UCPs closed. There is one particular way (of several) to open them. The inhibition from cytoplasmic ATP can be overcome by an excess of mitochondrial ATP. Mitochondrial ATP, obviously, enters the UCP from the opposite end to cytoplasmic ATP and gets in the way of the latter's binding. Mitochondrial ATP cannot reach far enough into the UCP to induce the closed conformation itself, so the pore opens. The blog post has nice images and a more thorough description. Here's my fave picture:

How do we keep mitochondrial ATP levels low?

Phinney had six week starved humans on a treadmill showing every probability of low mitochondrial ATP and UCPs closed tighter than the proverbial monkey's @rsehole.

On the opposite front we have rats in a cage whose biggest effort is to move over to the hopper of ketogenic pellets and have a munch. These animals uncouple like mad while eating to satiety. They also either maintain low fat reserves or lose fat reserves if previously made obese from fat/sucrose feeding. We've all read this mouse study even if today's rat epic is very inaccessible (thanks Mike).


It seems to me that it is possible to maximise the efficiency of energy usage to ensure survival under near starvation conditions. However fat based your metabolism, you are not going to uncouple your oxidative metabolism unless you have adequate ATP within the mitochondrial matrix.

It's very clear that an ad libitum ketogenic diet allows uncoupling and metabolic inefficiency down to a lean bodyweight, certainly in rodents. This is not arguable. Here's the graph. No mouse was forcibly calorie restricted:



Days 1-4 after switch to ketosis they ate less, by day eight after the switch to ketogenic eating they were eating more calories (ns) than other groups but staying weight stable.

The question to me is: By how much do you have to deliberately restrict the calories of a ketogenic diet fed human to eliminate the uncoupling effect? Or, more simply, turn the question round: How do you get a human to lose weight most effectively on a ketogenic diet? This is easier to answer.

As Amber O'Hearn suggests:

Eat meat.
Not too little.
Mostly fat.

Perhaps someone should tell Dr Hall this. Better still, make it his epitaph as science progresses.

Peter