Tuesday, September 14, 2021

Modelling energy intake (2): Corn oil

This post has been waiting around for some time so I thought I'd just put it up before settling down to read

but I'm also looking at

and this one is pure Protons and ruminants

ditto this

and there might be at perhaps two more post on canagliflozin. Maybe.

When on earth I'll get to post on these I have no idea. Working on it!

Anyhoo. Back to today:

I thought it might be interesting to very, very crudely apply Kevin Hall's mathematic model to a much more interesting study. This one came my way via Jacob in comments quite a few weeks ago.

Response of body weight to a low carbohydrate, high fat diet in normal and obese subjects

This graph is an example of one single individual out of a total of five people in 1973, so we are talking about near-anecdotal data, but fascinating never the less.

The diet contained a fixed 168g/d of carbohydrate and 64g/d of protein plus a variable amount of fat over time. This is the weight change curve for subject "1", as far as I can make out:

This person lost around 2kg in the first 10 days on the diet then regained just a small amount over the following 25 days. For comparison if we now go on to look at Table 3 from Hall's

How Strongly Does Appetite Counter Weight Loss? Quantification of the Feedback Control of Human Energy Intake

we can see that a standard "lifestyle intervention" (Weight Watchers and exercise perhaps????) established an enforced caloric deficit of around 700kcal per day, which was eroded by hunger ("appetite") at something like an exponential rate, approaching the re establishment of baseline caloric intake with persistent ongoing hunger:

In the first month or so this caloric deficit triggered something around 2kg of weight loss. So if we took the graph for subject 1 at the top of the post we might reasonably assume an initial deficit of in the region of 700kcal with a rapid onset of hunger which would try to erode this weight loss effort. Something like:

Absolute CI decrease + CO increase = hunger, ie real life.

Except the subject in the graph at the top of the post was not in a weight loss study. He/she was in an overfeeding study. Carbohydrate and protein were kept fixed (and non-ketogenic) and progressively more fat was added to the diet, in steps, every five days. Here is the graph with the calorie intakes illustrated for this particular individual.

Yes, that is just under 6000kcal/d at a stable reduced weight.

Two things come to mind.

First is that it is stated that all subjects consuming over 2700kcal/d of fat felt warm all of the time and sweated easily. Second is that DLW measurement of TEE would certainly pick this up perfectly well and estimate a massive "calories out". Those would be as heat. Reversing-engineering weight loss to estimate changes in food intake is clearly completely out of its depth here. What happened?

The fat was corn oil.

Linoleic acid -> 4HNE -> activates uncoupling to blunt insulin signalling and causes insulin resistance per se -> hot, sweaty weight loss.

It takes a significant amount of linoleic acid to do this, well in excess of that needed to augment fat storage.

This effect appears to apply just as well to humans as it did to those mice in The ginger paradox (3), even when overfeeding is exogenously enforced. Clearly the mice which actively lost weight "effortlessly" (ie mice never do the human "appetite" battle unless they are exogenously semi-starved) on safflower oil used uncoupling to blunt insulin signalling and so increase lipolysis and adipocyte derived calorie supply.

Subject 1, on corn oil, had a peak of around 84% of calories from fat which put the linoleic acid percentage in the region of 40% of 6000kcal, well in to uncoupling levels. Corn oil in the 1970s was suggested to be around 45% LA.

Now, what would be expected to happen if we massively over fed with a lower LA, less uncoupling fat? The estimate for LA in olive oil in the 1970s was 7%. In this next graph the maximum LA percentage was 6% of calories, which is more akin to an obesogenic dose than to an uncoupling dose. Overfeeding olive oil does this:

That is 9kg weight gain in 40 days, and still going.

Now, what might we expect if we tried the same thing with beef fat? My expectations:

Weight gain would be even greater.

Metabolic syndrome would develop rapidly.

It should be much harder to sustain a 6000kcal diet of mostly beef fat than it was from either an uncoupling or  an obesogenic fat source.

The group didn't try overfeeding beef fat, sensibly.

There are a number of studies which I have picked up over the years which suggest that the uncoupling effect of double bonds kicks in at essentially all levels of their metabolism. At low levels the effect is over-ridden by the effect of failing to limit insulin signalling in adipocytes as per the Protons concept, leading to weight gain ie insulin still signals perfectly well and it does so more than is physiologically appropriate, especially in the immediate post prandial period. As uncoupling comes to predominate the ability of a low mitochondrial membrane potential to markedly suppress ROS generation becomes progressively more and more dominant, so insulin signalling becomes profoundly blunted. It will never get to high enough levels where insulin-induced insulin resistance should have kicked in, so the Protons concept becomes irrelevant. Under uncoupling, mitochondrial metabolism is functionally hypoinsulinaemic, it should resemble that of reduced insulin gene dose mice in Jim Johnson's lab where reduced insulin signalling was simply the end result of reduced insulin production, 24/7. It should also resemble ketogenic, hypoinsulinaemic eating.

Whether it is via 4-HNE/UCPs or 2, 4-dinitrophenol, high enough levels of uncoupling will absolutely blunt insulin signalling, with subsequent increase in access to adipocyte calories and consequentially suppressed hunger, leading to adipose tissue loss without increased food intake, with a few small caveats thrown in.



raphi said...


I read that in diseases affecting the respiratory chain a higher membrane potential actually decreases ROS, a seeming exception to the norm whereby ATP synthase shutdown lowers mitochondrial membrane potential and thus increases ROS. If I got that right, are diabetics helped by Metformin because it inhibits mtG3PDH => increases mitochondrial membrane potential => lowers ROS?

Passthecream said...

Metformin has had me scratching my head lately. Mtg3pdh isn't uniformly expressed across all cell types which is one complication, and perhaps the Metformin effect on it is more involved with pancreatic cells thereby insulin production, but this mito membrane version of g3pdh is fad(h) bound/dependant so I have been trying to understand just how the effect of metformin differs from eg fatty acids with a different f/n ratio. Why is the effect of Metformin on the ETC not the same as that of pufa for instance? I briefly had an Aha! moment when I read the above about the uncoupling effects of pufa, but it is unclear to me atm.

raphi said...


"Why is the effect of Metformin on the ETC not the same as that of pufa for instance?"

I started narrating an answer in my mind but soon realized I was bloviating. I don't know. I couldn't even call upon complex 1 effects given they're not meaningful at physiological levels...

Jonathan said...

Well, I wish they had done the beef fat experiment. I would be fascinated to know for sure how it would turn out.

You hypothesize greater weight gain. But if I understand the Protons concept right, the beef fat would have made the adipocytes more insulin resistant. So is your belief that the body would have generated as much insulin as necessary to force the beef fat into the adipocytes, as it has nowhere else to go?

Monica said...
This comment has been removed by the author.
Monica said...

I guess weight gain would be the result of overfeeding, but in real life, with peripheral insulin resistance, you are not hungry so you lose weight. Yes, I would love some research on beef fat.

Peter said...

Hi raphi,

That's probably worth a post in its own right. I have thought about it on many occasions and the basic difference appears to be in the effect on the liver. Metformin suppresses hepatic glucose output from gluconoegensis based on glycerol or lactate. This is based on the rise in NADH:NAD+ ratio in the cytoplasm. Converting lactate to pyruvate requires NAD+ but lactate can still enter mitochondria for oxidation via the TCA. Working on the basis that metformin is specifically excluded from mitochondria then complex I will still supply plenty of NAD+ to convert lactate to pyruvate to enter the TCA. Also in the cytoplasm of the liver glycerol will be phosphorylated to glycerol 3 phosphate but requires NAD+ to convert the glycerol 3 phosphate to dihydroxyacetone phosphate. Metformin, by blocking the glycerophosphate shuttle, makes both reactions unfavourable.

BTW there is some evidence the metformin does actually inhibit complex I at therapeutic levels by acting on the cytoplasmic side of the membrane component of complex I, but this does not seem to be terribly relevant compared to the redox changes at therapeutic dose rates in the cytoplasm...


Peter said...

Obviously in addition to suppressing insulin secretion directly and blunting insulin induced insulin resistance...


Peter said...

Jonathan and Monica,

Absolutely. Overfeeding is not real life!


raphi said...


Thanks, I had whiteboarded your explanation and think I got it. I didn't know there were (small?) therapeutic effects on complex 1 from metformin on the cytoplasmic side of the mitochondrial membrane - good to know!

Peter said...

Found it.


To me, any complex I inhibition is an off target toxicity at any concentration. After all, the Galega officinalis plant hates all herbivores/omnivores, even those who should stick to eating fatty meat!


Passthecream said...

Galega O. is a legume relative btw. Another black mark against the bean family.

karl said...

I had missed the Galega officinalis connection up to now - interesting..

I can think of several dietary factors that drive T2D/obesity. So every time I go to the grocery store and look in other people's carts, I think of this obvious experiment to run. (This has become a very strange habit - but my unscientific observations tend to link T2D with CIAB and sugar drinks - but I bet there is more to learn)...

A checkout lane that weighs the customer - but not the cart - a LASER scanner gets a height - we now have a rough BMI - and what is purchased. Not perfect, but by combining the UPC codes scanned with food content labels - one should be able to find correlations - and my hunch is there might be surprises to be found.

These correlations wouldn't mean much on their own, but could direct further research. Some correlations could be large enough to mean something .. we really could use this data.

One could draw out total amount of PUFA vs SAT - sugar, carbs, protein. There could be particular bar codes that have high correlations - things we are not yet looking at.