Thursday, July 11, 2024

Protons (74) Arne Astrup and the formerly obese

I have a soft spot for Arne Astrup. Back in the days of the depths of the Danish fat taxation stupidity, he was one of the voices of reason speaking out against the tax. It was a near miss for sanity. Academics have since argued that the tax was repealed too soon (and the sugar tax never got started) and that it was actually "working", at least among those who couldn't hop over the border to buy their (Danish?) butter in Germany. Had it been allowed to continue to "work" we might have successfully forced a whole nation to avoid fat, especially saturated fat. Where might that have led? If you wish to compose the answer on a postage stamp it is just three letters long, which will fit neatly on to even the smallest stamp.

Anyhoo. People may have noticed that I like this paper from the Astrup led lab

Fat metabolism in formerly obese women

mostly because Table 3 confirms all of my biases by showing formerly obese women are exquisitely insulin sensitive, which is pure Protons:

The rest of the paper is more difficult.

The formerly obese are, as expected, only deriving around 35% of their energy from fat oxidation at time zero on the graph below while the never-obese controls are deriving just under 80% of their energy needs from lipid oxidation, time zero again. These values are while sitting still on a bicycle ergonometer, after an over night fast. Not quite basal metabolic rate or resting energy expenditure but pretty close:

It is also worth noting that performing exercise at 50% of VO2 max (previously individually measured) completely normalises energy production derived from fat oxidation. Those are time points 15-60min. All we need here is for AMPK to instigate oxidation of the fatty acids available while suppressing their formation. Then the pathological insulin sensitivity is bypassed. There are several posts possible on AMPK but again, here is not the place to explore the control of insulin signalling by AMPK and vice versa. Both happen.

Finally, the really strange thing is that these formerly obese women have modestly *elevated* FFAs, both at rest and throughout exercise, consistently around 300µmol/l greater than controls.  If the formerly obese have all this extra lipid available, why don't they oxidise it?

We can say, quite conclusively, that these FO women have normal electron transport chains. Under exercise they oxidise lipids exactly as well as control women do. My assumption is that there has to be a signalling problem which is inhibiting fatty acid oxidation but can be over-ridden by AMPK activation.

We know, from a mass of rodent and human studies, that when you allow a subject access to carbohydrate food (or an OGTT) after an extended fast, they perform de novo lipogenesis, giving an RER > 1.0, for about an hour. The duration is interesting. Insulin-induced insulin resistance (which is complex and probably involves the glycerophosphate shuttle) usually comes in to effect at around about an hour in many models. This will reduce insulin signalling from its peak action under "hungry" conditions to a more moderate "fed" signal.

So peak insulin signalling after a fast, but before establishment of some physiological limitation, is a potential major driver of de novo lipogenesis with storage as triglyceride and, as we shall see, an effective inhibitor of fat oxidation.

Mechanistically we have to look briefly at the Randle Cycle.

Two of the many actions of insulin are to activate the pyruvate dehydrogenase complex and the acetyl-CoA carboxylase complex. This generates malonyl-CoA which inhibits CPT1 mediated transport of fatty acids in to mitochondria. Hence FFAs are available but not oxidised. But there is no problem with the mitochondrial ETC itself, all that is needed is for the insulin signal to be reduced.

During that initial refeeding period both the RQ of >1.0 and the inability to oxidise lipid can be viewed as manifestations of marked insulin sensitivity. Carbohydrate uptake is enhanced by insulin and the products of this carbohydrate catabolism are diverted, by insulin, to metabolites which inhibit fat oxidation and away from the Krebs Cycle and the electron transport chain.

Aside: My interest is in ROS based control systems so I have tended to ignore such details. But here the downstream effects of excessive insulin signalling on the Randle Cycle do matter. My bad. End aside.

My premise is that obesity is cause by a pathological sensitivity to the hormone insulin, mediated by linoleic acid. If this is correct then we would expect pathological lipid synthesis/storage to be combined with an inhibition of fatty acid oxidation. The normal "one hour" of peak insulin sensitivity is extended or even becomes continuous by using linoleic acid as a significant energy source (pax uncoupling intakes).

Here we have the formerly obese who are, without a doubt, destined to become obese again in the future. We also have people with obese parents who are not yet obese themselves. Both show the accentuated insulin sensitivity in combination with depressed fatty acid oxidation, both at rest and post prandially. All that is required to do this is to allow insulin to continue to act at peak efficacy under conditions where a functional limitation should have been imposed. Linoleic acid replacing palmitate/stearate under the Protons hypothesis provides exactly this.

So, in Astrup's particular group of formerly obese subjects described in the current study, it has proved possible to have inhibited fatty acid oxidation with sufficient severity that it leads to elevated plasma FFAs, because they cannot be used for energy generation. All as a consequence of augmented insulin signalling.

This particular group of FO subjects do appear to be a rather extreme example. Other FO people assessed by Astrup's group in previous studies do not feature the elevated FFA or profoundly depressed fasting insulin aspects, though the inability to oxidise lipid to produce adequate energy is a consistent feature over many studies. I think this current group of women are probably outliers who give an insight in to mechanisms. That's good.

It's also clear that these formerly obese women have a metabolic rate under fasting conditions significantly lower than that of the never obese controls. This is not surprising. Fat oxidation is being largely inhibited by elevated malonyl-CoA and a significant portion of glucose is being diverted from energy production to form that malonyl-CoA and its derived and stored lipid.

The FO women's resting energy expenditure is 3.77kJ/min, ie 0.9kcal/min, 1296kcal/24h. Never obese women expend 4.88kJ/min at rest, 1.2kcal/min, 1728kcal/24h. Except of course the FO women are not oxidising fat, because they are unable to effectively oxidise FFAs. They are using glucose.

So they are 461kcal/24h "hungrier" than never obese controls and are running on limited supplies of glucose from glycogen. The obvious solution is to access more glucose, which insulin has actively locked in to the liver/muscle stores of glycogen.

Traditionally this is solved in the real world by raiding the fridge at 3am. For something sweet. Much of which will be diverted to storage.

Weight gain.

It happens.


Saturday, June 01, 2024

Protons (73) Spanish vs Canadian studies or 3% LA vs 10.3% LA

I hope no one has forgotten the "Spanish Study"

Distinctive postprandial modulation of beta cell function and insulin sensitivity by dietary fats: monounsaturated compared with saturated fatty acids

which was very, very carefully set up to demonstrate post prandial insulin resistance following the ingestion of the saturated fats from butter while also demonstrating a progressively improving insulin sensitivity using oils with increasing content of (mostly) linoleic acid.

A brief look at reference 15 methods section confirms they measured the test meal lipids by gas chromatography.

Obviously, to anyone lacking the Protons perspective, the clear cut message is that saturated fat causes insulin resistance. Insulin resistance is BAD. Saturated fats are BAD.

What Protons actually predicts is that resisting insulin by ingesting saturated fats limits insulin-facilitated obesity, so eliminates the subsequent adipocyte distension derived release of FFAs (which cannot be suppressed by insulin), which would lead to metabolic syndrome.

Resisting insulin prandially resists obesity and so resists obesity-derived insulin resistance. Read that very carefully.

Aside: Resisting insulin in the immediate post ingestion period is a short term effect for a few hours. It's physiological. Constant presence of FFAs secondary to increased basal lipolysis from distended adipocytes is present 24/7 irrespective of what you eat. These fatty acids supply calories and if you are eating carbohydrate then you must resist cellular glucose ingress to take in to account that FFA supply of calories. This insulin resistance is different (still the correct physiological response to FFA availability) because it follows on from pathology related to adipocyte lipid droplet size. The third type of insulin resistance is much more complicated. So is the fourth. Here is not the place to discuss them. End aside.

Sooooo. I really, really like this:

This is *not* demonstrating saturated fat induced pathological insulin resistance. Here the insulin "resistance" will simply stop you getting fat. I would define this as the "normal" response to an high fat meal, assuming a low linoleic acid based fat.

What it *is* demonstrating is pathological insulin sensitivity following the ingestion of 10.3% (actually slightly under this but close enough) of calories as linoleic acid. Here as little as 140pmol/l of insulin will rapidly clear your plasma of calories and leave you hungry. The meal is largely lost in to your adipocytes. You WILL want to eat again, and soon.

This is fundamental and simply falls out of the Protons hypothesis.

Now, there are problems with the study and the authors are to be congratulated on the result generated (though not on their conclusions of course), especially the composition and size of the meals they had to design to get there. But here we are looking at a dynamic response to a single meal. Is it possible to examine their important findings under more steady state conditions? Would the relationship of PUFA ingestion to pathological insulin sensitivity still hold?

What would we find if we kept plasma FFAs forcibly elevated for 24h using repeated small oral fat loads supplying 2430kcal over a 24h period (but no other food) instead of a single oral ingestion of 800kcal as one mostly fat meal?

Then, instead of tracking the insulin response to a small amount of starch/protein along side the fat of the 800kcal meal, we could assess insulin sensitivity during an hyperglycaemic clamp at 20mmol/l of glucose in plasma. The more glucose needed to achieve this level, the greater the insulin sensitivity.

We are now well away from normal physiology but we are asking essentially the same question under more constrained conditions.

I'm assuming people have realised that I'm now describing Xiao's study from the last post

which produced this chart from the above protocol:

To me, the results of Xiao's study and the Spanish study concur beautifully.

Not everyone will agree with me that these reflect a core reality, I wouldn't expect that. But in my NSVHO this is how physiology works. Linoleic acid is insulin sensitising.

However, when you have confirmation bias as badly as I do, and you find two non related studies which neatly corroborate each other while confirming your biases, you know you are trapped. That's me.

You have been warned.


Addendum: Obviously palm oil at around 8% of calories as LA is already insulin sensitising, it's in the same ball park as the 10.3% LA arm of the Spanish Study. A true SFA arm to the study would need to have LA at around 2% of calories and I would predict a GIR well under 40μmol/kg/min.

Thursday, May 30, 2024

Protons (72) Humans: 8% LA vs 74% LA by sustained oral ingestion

This study in humans is very different to the previous rat study. People were fed repeated oral fat loads (you could call this Bulletproof Cocoa rather than Bulletproof Coffee), once an hour for 12 hours then every two hours overnight until the start of an hyerglycaemic clamp at 24h, through which the oral fat loading continued. Glucose was infused to a stable 20mmol/l and insulin allowed to respond as best it might. Insulin sensitivity was determined by the glucose infusion rate in the last 30 minutes of the clamp. In some ways this is more physiological than the hyperinsulinaemic euglycaemic clamp, which is considered the gold standard. The oral fat ingestion was slightly above calculated 24h caloric requirements for these subjects.

Differential effects of monounsaturated, polyunsaturated and saturated fat ingestion on glucose-stimulated insulin secretion, sensitivity and clearance in overweight and obese, non-diabetic humans

Over a 24h period the ingestion of safflower oil, with linoleic acid providing in the region of 70% of total calories, ought to demonstrate the initial Protons predicted insulin sensitising effect, which would only be later replaced by the uncoupling effect if the study had been continued for a week or two. 

Again we can assess insulin sensitivity by how much glucose was needed to be infused during the last 30 minutes of the clamp to maintain an hyperglycaemia of 20mmol/l. This is what happens:

I don't think I have to make any qualifications here. SFA oral ingestion for 24h causes a very similar degree of insulin resistance to oral ingestion of a minimal calorie supplying control chocolate drink. Tallow rather than palm oil would have accentuated the effect.

Ingesting 70% of your calories as linoleic acid over a 24h period is insulin sensitising compared to ingesting SFA, p less than 0.001. Or ingesting virtually nothing at all, p < 0.05.

Linoelic acid is insulin sensitising.

This is BAD. When fasting you *must* resist even basal insulin or that insulin will lower fasting glucose, lower fasting FFAs and you will be hungry. And raid the fridge at 2am. And get fat.



Protons (71) Rats: 13% LA vs 61% (mostly) LA by infusion

I have a certain, very specific, idea of how linoleic acid produces obesity. It seems as though relatively few people share this point of view. That is absolutely fine. Bright people have their own ideas and, eventually, if the core process is consistent, all views of the development of obesity and its associated insulin resistance will eventually converge. I spend a great deal of time thinking about whether linoleic acid enhances insulin sensitivity -> directly causing obesity or whether linoleic acid causes insulin resistance directly -> reactive hyperinsulinaemia -> obesity. The data make me favour the former.

This is the first paper I have come across where various fatty acid mixtures were assessed, in rats, for their acute effects on insulin sensitivity in vivo. In particular I was interested in the effect on glucose utilisation under hyperinsulinaemic euglycaemic clamp conditions. The higher the infusion rate, the more insulin sensitive the rat is.

They infused intravenous oil emulsions continuously for five hours and then continued throughout the exogenous hyperinsulinaemia over the following two hours, while clamping glucose at around 6.5mmol/l. So this is looking at normoglycaemia combined with fasting levels of FFAs until the clamp period. The rate of deliver per hour was roughly comparable to a 24h intake of calories for a rat of this size, averaged to an hourly rate.

Everything is fairly physiological until you add in the insulin/glucose infusions for the clamp while maintaining the lipid supply. Then you are looking at the situation where FFA supply cannot be suppressed by insulin, so you have a model for metabolic syndrome.

The results are quite clear. Whole body insulin responsiveness is suppressed by any fatty acid availability.

Clearly the glucose infusion rate, representing whole body insulin sensitivity, is lower in the SATU group (lard oil) compared to the PUFA group (soybean oil) but this is not remotely statistically significant (p = 0.2849). However there is no suggestion that linoleic acid is uniquely triggering insulin resistance compared to saturated fats, bearing in mind that modern (2015) Canadian lard is higher in insulin sensitising LA at 15% than my preferred fats such as beef tallow or suet which are around 2% LA (correctly ignoring any CLA content).

So the Protons concept could be suggested to have earned some marks here, there is more insulin resistance in the saturated fat group (GIR 43micromol/kg/min) when compared to the less insulin resistant linoleic acid infused group (GIR 73 micromol/kg/min). But not statistically significant.

However, there is no suggestion that linoleic acid per se causes enhanced insulin resistance, so causing obesity via secondary hyperinsulinaemia. In fact the trend is in the reverse direction.

In these rats.

Humans next.


Wednesday, May 29, 2024

Foie Gras (11) Hepatocyte mitochondria

Another tidy up, this time related to 

Fat Quality Influences the Obesogenic Effect of High Fat Diets

and the paradox of mitochondrial uncoupling in section B of Fig 4:

It is patently obvious from this plot that mitochondria extracted from the liver tissue of lard fed rats (consuming an obesogenic level of linoleic acid) do have an higher uncoupled oxygen consumption at all values of membrane potential when compared to the level of oxygen consumption in those rats fed the high safflower oil diet.

That is exciting and paradoxical.

We know from Figure 1 that the safflower oil fed rats were more uncoupled overall than the lard fed rats. They were synthesising much more UCP-1/cell in their brown adipose tissue and they had a greater absolute mass of brown adipose tissue by the end of the study.

They were also actively expending more energy per day at the end of the study compared to the lard fed rats. This is stated in the legend to Figure 2:

"Percent contribution of lipids, proteins and carbohydrates to total daily energy expenditure (lard = 380 ± 15, safflower-linseed = 410 ± 25 kJ/day x kg0.75) in rats fed lard or safflower-linseed high fat diet."

I would expect all rats/mice fed high safflower oil diets to use this technique and so eventually normalise their weight to that of chow fed rats/mice on a long term basis, as was found (in mice) here:

Prevention of diet-induced obesity by safflower oil: insights at the levels of PPARalpha, orexin, and ghrelin gene expression of adipocytes in mice

Okay, let's summarise:

Safflower oil induced an initial obesity by increasing insulin sensitivity which was, by day 14, in the process of being reversed by UCP-1 reducing that excessive insulin sensitivity in WAT, assisted by activating BAT.

No one would expect hepatocytes to express UCP-1, they just don't do this. The liver deals with excess calories by sequestering them as triglycerides under the influence of insulin, sequestering them as triglycerides under the influence of succinate derived from peroxisomal omega oxidation or by signalling to BAT using FGF21 as a mediator to increase UCP-1 expression so as to bulk off-load calories as heat. But not in the liver.


Safflower oil (~70% linoleic acid) produces whole-body uncoupling in the rats in the current study, apparently with the exception of within liver tissue.

Hepatocytes *do* use UCPs, they definitely synthesise UCP-2 and UCP3, but not for bulk lipid oxidation. Current thinking is they are used to fine tune their inner mitochondrial membrane potential while other signals deal with bulk caloric overload.

So the paradox is that lard fed rats have more uncoupled mitochondria than safflower fed rats. That's what the graph at the top of the page shows, ie hepatocytes are doing the opposite of what the whole rat is doing...

They measured delta psi of isolated mitochondria with a dye (safranin O) calibrated back (through 4 layers of references) to the standard technique which gives us our best estimation (don't ask) of membrane potential. They then fed isolated mitochondria in the presence of oligomycin (to block ATP synthesis) and rotenone (to prevent RET through complex I). At this point all oxygen consumption is from uncoupling. If you add increments of malonate to progressively inhibit complex II you can progressively lower the delta psi and look at the degree of uncoupling at a given titrated delta psi.

On the face of it it looks very much as if the liver really is doing the opposite to the rest of the body, which seems counter intuitive:

The degree of uncoupling is being assessed at a fixed potential, here the group chose to use 150mV (the blue line) for their example, giving an uncoupled oxygen consumption of 41.9 in lard fed vs 22.2ngatoms/(min x mg protein) of oxygen if safflower oil fed.

But is this the case in vivo? The lard fed rats are chronically underfed and have lipid locked in to adipocytes by excessive insulin sensitivity so what little lipid is being released is via augmented basal lipolysis. It is completely plausible (but also completely made-up) that they might be running a membrane potential, in vivo, as low as 120mV. Like this, blue line:

At 120mV you are not going to making a lot of ATP so uncoupling would be actively disadvantageous. In this example the uncoupled oxygen consumption would be low, in the region of 19ngatoms/(min x mg protein) of oxygen, red line.

The safflower oil fed rats went through an initial hypocaloric episode during their initial weight gain phase, but now they are uncoupling in WAT which will blunt insulin signalling and release a surfeit of FFAs, enough to supply large amounts of FFAs the liver mitochondria and (in parallel) accumulate as lipid droplets to the point of cellular damage occurring.

Under this level of direct hepatic caloric excess the mitochondrial membrane potential is likely to be high. If we run another thought experiment (ie make up) a potential of 160mV, just under that 170mV threshold for marked ROS generation, this would give us an uncoupled oxygen consumption of 29ngatoms/(min x mg protein) like this:

So, if the membrane potential differs between groups in vivo, so would the level of uncoupling. It is completely plausible that (safflower) lipid overloaded mitochondria are running an high delta psi, so need more uncoupling. Mitochondrial will never have a fixed delta psi of 150mV. It is absolutely possible that, in vivo, the safflower oil fed rats had more uncoupled hepatic mitochondria compared to the lard fed rats.

I feel much more comfortable with having hepatocytes uncouple *more* with safflower oil than with lard. The whole study is bias confirming of multiple aspects of the Protons hypothesis. Things have to make sense.

I have no problem with the mitochondrial preparation the group developed here and how they have used it. It's no better/worse than any other mitochondrial preparation. What is crucial is how you interpret the data it provides you with in the light of what must be happening physiologically. Then extrapolate backwards to the most plausible in-vivo situations, with caveats.

I have my biases.


Late addendum.

The mitochondrial uncoupling curve I have been discussing was generated from mitochondria treated with FFAs to facilitate uncoupling. Of course, if you argue that the lard fed, hypocaloric rats had lower levels of FFA in the fed state in vivo (in the fasted state there is no difference in FFA level) then there would be much less uncoupling than that discussed above, emphasising the point. Without FFA supplementation, at 160mV in the lard fed rats uncoupled oxygen consumption was as low as ~5ngatoms/(min x mg protein) in section A of Figure 4. In reality it would be somewhere between this value and the FFA supplemented value. As much as any mitochondrial prep reflects reality.

Foie Gras (10) Liver

Just to tidy up my thoughts on

TLDR: I suppose all I really have to say is that the title is incorrect and the scrutineers are completely incompetent.

We have these data for mRNA production from "pro-inflammatory" genes in liver tissue:

which we know, from their section of adipose tissue, have absolutely zero correlation with active inflammation, which they assessed in adipose tissue using the activity of the myeloperoxidase system.

The liver is full of resident macrophages, known as Kuffper cells, which are very good at activating their myeloperoxidase system. The group has an assay for this activity. They didn't use it on liver tissue. Why not?

I have no idea whether the liver macrophages actually used any of the reported mRNA products to generate inflammation. 

Go figure.

The group did, for some reason, measure plasma CRP levels, CRP being an acute phase protein produced by the liver in response to any inflammation, *anywhere* in the body. It might have been raised in response to the activity of the myeloperoxidase in the adipose tissue of the HF fed mice, that might be logical. We'll never know because they omitted to measure CRP in the plasma of the HF fed mice.

Go figure. I have not edited this chart in any way:

So we know essentially nothing about inflammatory changes in the liver and we know nothing about the levels of CRP produced (or not) in the plasma of HF diet fed mice. Which did have inflamed adipose tissue and *might* have had inflamed liver tissue.

Oh, their one interesting finding was that saturated fat is suppressive of "pro-inflammatory" gene expression in liver tissue. But not in adipose tissue.

Would this be protective against inflammatory liver damage? There is no way you can assess this from this study, but the idea is nice.

But ultimately the liver section of the paper is complete dross.

I said it before, these people are rank amateurs.


Saturday, March 30, 2024

Foie gras (9) Adipocyte ROS

Time to look at the Vaughan mouse weight gains and actual diet compositions. Here is a recap of the weight gains:

and here is Table 1 with the percentages of energy from linoleic acid added in red by myself:

It's clear that the saturated fat group were only fed 6% of their calories as insulin sensitising linoleic acid. In addition to this relatively low amount of LA, the stearic acid in the saturates has an high F:N ratio and will in part offset the low F:N ratio of the LA component.

If adipocytes resist insulin they stay small.

The oleic acid group were fed 7% of calories as LA but with only 12% of calories as saturated fat there is nothing to oppose the insulin sensitisation effect of LA so they gained a significant amount of weight.

The high (35% of calories) LA diet is uncoupling, produced low levels of ROS as a consequence, and so limited weight gain. Given long enough it would normalise bodyweight.

The high omega 3 diet contained an obesogenic level of LA coupled with an uncoupling level of alpha linolenic acid, though putting a number to the level of ALA needed to uncouple is difficult, but it is lower than the level of LA needed. This is another combination which I expect, given long enough, would normalise body weights. 

I feel that some explanation is needed as to why the LA/ALA 18:3n3 diet mice, and to some extent the high LA diet mice, weigh more than the saturated fat diet mice, to the point that the 18:3n3 diet looks obesogenic.

There are, initially, two effects of fatty acids with multiple double bonds. The first is the reduction in RET through complex I which itself has two effects. Under peak insulin action there is reduced negative feedback so adipocytes distend more than they should. Second is that during fasting, when fat is the primary fuel, RET should occur to limit glucose utilisation in order to spare glucose for the brain/hypothalamus and so limit hunger. With the blunted RET from LA/ALA more glucose is used by muscle etc so more food must be eaten, so breaking the fast, in order to keep glucose levels adequate to limit hunger. Much of this extra food then gets stored.

These two together put fat in to adipocytes and demand more food intake. This is the classic situation under D12492.

This is also likely to be the initial situation when using 35% LA or a mix of 11% LA with 23% ALA in the period before uncoupling becomes established.

The second effect is via uncoupling.

There will be weight loss in these latter two diets, but only once UCP-1 is activated in white adipocytes to lower delta psi and so reduce insulin signalling. At that point adipocyte FFAs can either be oxidised to release heat in situ (beiging of WAT) or transferred to BAT where high levels of UCP-1 can oxidise them to release heat in bulk.

This concept suggests that, by 14 days, the mice on the "18:2n6" diet would be in weight loss and should have low ROS generation due to uncoupling, after an initial weight gain.

The effect should be more marked in the "18:3n3" group, ie an higher initial weight gain, then incipient weight loss by 14 days.

This is why I like the Schwartz data, daily resolution of food intakes and fat mass changes allow you look at things more mechanistically. It would have been nice to have these data from the current pro-linoleic acid study but thats not what the study was all about.

Here are the mRNA data for inflammatory gene expression in adipocytes in vivo (or immediately post euthanasia!), which I am taking to be a surrogate for ROS generation:


From the ROS perspective the SFA adipocytes are generating ROS by RET, so are limiting insulin signalling-induced lipid droplet distension. The mice are slim. And healthy.

The 18:2n6 mice are uncoupled, have low ROS due to this and are actively losing weight after an initial gain. Same for the 18:3n3 mice.

The HF fed mice (9.6% LA, obesogenic, failing to limit insulin signalling) have low ROS because they are failing to generate enough of them via RET to limit caloric ingress, ie have "pathological" insulin sensitivity. "Healthy" insulin sensitivity, through healthy ROS, is shown by the SFA group. The HF group are simply sequestering calories in to lipid droplets without oxidising them. Here weight gain is on-going but there is no issue with high ROS because they are effective at sequestering calories. Except...

Now the HF fed mice are really interesting. They have levels of inflammatory gene mRNA expression comparable to all of the other groups, including the SFA group (p>0.05 for the comparison) but look at their MPO activity, an indicator of active inflammation. I've rearranged chart G so all columns are on the same scale:

All groups of mice have comparable levels of inflammatory gene mRNA expression (pax the SFA fed group) but only the HF group have actively inflamed adipose tissue.


We can say that generating mRNA from pro-inflammatory genes alone is not sufficient to activate the inflammatory cascade to the extent of activating the myeloperoxidase system.

I have to ask myself what, exactly, is the function of these genes we are looking at, within physiology, at the most basic level.

I would suggest that they might be to deal with normal ROS generated from normal metabolism. The SFA diet induces high levels of healthy ROS via RET. It generates a large, effective response in ROS mitigating genes. All other groups, at the 14 day mark of the study, have low levels of RET derived ROS, so low levels of mRNA from inflammatory (or rather mitigating) genes. Including the HF diet group.

What is different about the HF diet group is that they are, through Protons, unable to limit caloric ingress. As much of the excess calories as possible will be rendered in to harmless stored triglycerides but all that is needed to generate frank inflammation is the generation of a delta psi in excess of 170mV. This leads to ROS which are only in a small part derived from RET, ie are mostly pathologically derived.

I think the HF diet fed mouse adipocytes are doing this. There is tissue damage occurring and lipid peroxides are produced at levels which signal danger of serious injury and so macrophages move in to sort out the damage. Probably making incorrect assumptions about the source of the damage, leading to pathology. This appears, in this study, to be independent of the expression of what are considered, in this study, to be pro-inflammatory genes.

Activation of the myeloperoxidase system, as observed in the current study, is not a simple consequence of activating mRNA generation of inflammatory genes.

It just strikes me that expressing a gene and using its product may be greatly influenced by factors this study doesn't address.

So there are at least three descriptions possible for the state of ROS generation in the adipocytes of these mice. There are no simple linear relationships. You need some sort of framework to understand what is going on.


Which makes me happy.


Sunday, March 24, 2024

Foie Gras (8) Vaughan's macrophages

Okay, back to the Vaughan paper

and a look at their macrophages in cell culture. The pattern is consitent so I'll just run through MCP-1 gene mRNA expression. As in the last post I am going to work on the basis that these genes are, in the absence of trauma or infection, going to respond to mitochondrial/NOX ROS production. It's a graph of relative expression, the control represents the response to the ROS being produced by supplying glucose at around 30mmol/l. That will be in response to NOX-2 being activated through calmodulin kinase II as we would expect. Even without insulin, these ROS will activate the phosphorylation of AKT and translocate GLUT4s to the cell membrane. It's the black column:

Adding 50 μmol/l of stearic acid (white column) will generate ROS by reverse electron transport though complex I, via an increase FADH2:NADH ratio, physiologically to limit insulin facilitated glucose ingress. Even if there is no insulin present, it still generates the ROS, which will limit the facilitation provided by NOX-2 derived ROS. Saturated fatty acids do this at low delta psi levels, there is no multi-enzyme/complex ROS generation, they just generate neat and tidy, fully physiological RET. The cell responds appropriately with superoxide dismutase and catalase and a coincidentally extended lifespan. Okay, I can't resist it, here he is again:

Oleic acid is another whole series of posts, so I'll mostly gloss over it. Oleic acid is designed to facilitate insulin signalling as a physiological balance to palmitic acid. It generates some ROS but not enough to significantly resist insulin signalling. There are many concepts stemming from this and here is not the place to discuss them.

Linoleic acid (hatched column) utterly fails to generate enough RET to limit glucose ingress, so glucose continues to enter, while at the same time the LA provides calories in excess of what the cell needs. The problem is that LA cannot shut down glucose ingress because it fails, through its double bonds, to generate the RET facilitated ROS to limit that glucose ingress.

The cell is full of ATP and depleted of ADP so ATP synthase cannot turn. Delta psi rises and above 170mV large numbers of ROS are generated. These come from a combination of complex II, complex III, electron transporting flavoprotein dehydrogenase and mitochondrial glycerol-3-phosphate dehydrogenase. Oh, and I guess some from RET. All secondary to high delta psi.

All stemming from inadequate RET derived ROS from LA's surfeit of double bonds. A few prophylactic ROS, produced without an high delta psi, stops an awful lot of problems. I think the phrase is that a stitch in time saves nine...

I think here is a good place to point out that ROS generate a response in inflammatory gene mRNA expression, probably at any level of exposure. Low levels will result in normal physiology, high levels will result in frank inflammation,  but also will induce inflammation limiting gene expression, in which I would include multiple uncoupling protein genes. Failure to limit high levels of ROS generation clearly results in apoptosis or, given a catastrophic failure of the ETC secondary to 4-HNE and associates, necrosis.

This paper is quite unique in view of the low dose of fatty acids used. If they had chosen the standard route of 1000μmol/l FFAs in addition to glucose at 25mmol/l (utterly non physiological) then RET would have dominated, giving apoptosis/necrosis levels of ROS, meaning stearic acid would be bad (usually palmitic is used), oleic less so and LA would be almost harmless. There is an infinite supply of such papers to reinforce the stupidity of saturophobia.

Okay, the in vivo mice next.


Foie Gras (7) What do you mean by inflammation?

This paper contains a very superficial and bullet point overview of the inflammatory cascade. It omits swathes of anti inflammatory pathways and many of the boxes on the chart could be massively expanded. The paper is interested in using phytotoxins (mis-termed phytonutrients) as anti inflammatory agents, so all I take from the paper is their doodle

Animal Models of Inflammation for Screening of Anti-inflammatory Drugs: Implications for the Discovery and Development of Phytopharmaceuticals

A legible version of the diagram is in the paper if anyone would like to actually read it.

We can simplify the process slightly with the red arrow. In addition I've circled in blue the component that is being used as a marker of inflammation in the Vaughan paper:

We can now consider the basic process a little more carefully. This massive review (no, I've not read it all) brings to prominence the central role of ROS in inflammation

and has this nice illustration featuring mitochondria as sources of ROS. Superoxide generation is never accidental. It is a tightly controlled and highly specific process to achieve certain ends. Mitochondria and NOXs are not making mistakes or having accidents:

BTW MPO is the myeloperoxidase we've seen recently in Vaughan's paper. HOCl is bleach, it's an inflammatory tool. We use it to kill germs.

So this lets us redraw this doodle of inflammation with a small modification:

Right up at the top of the inflammation doodle is a small red circle labelled PLA2, phospholipase A2. Its job, in the event of tissue injury, signalled by ROS and derivatives, is to release arachidonic acid from lipid membranes which then allows the generation of a raft of inflammatory mediators using cyclooxygenase and lipoxygenase enzymes.

Corticosetroids are potent suppressors of PLA2. They also releases both glucose and FFAs from the liver and adipocytes in to the blood stream. Physiological cortisol does this to deal with the sort of situation where you and a bunch of your mates are about to go and drive an angry mammoth into a bog, to kill it and butcher it. But high FFAs and high glucose together generate metabolic ROS, which are pro inflammatory.

You want raw energy, not to be crippled by an acute inflammatory reaction. So cortisol also down regulates the inflammatory cascade which *should* have been triggered, via ROS from the ETC, while flooding mitochondria with large amounts of substrate in anticipation of incipient need.

So cortisol floods your metabolism with calories, simultaneously controlling inflammation, and your mitochondria convert those calories in to the ability to successfully kill a mammoth, ie obtain food for a month for you and your tribe. Those calories get used successfully. Nothing is overloaded, nothing is damaged. Glucose and FFAs get used up during the hard work involved, facilitated by AMP-kinase rather than insulin. There is no generation of bulk inflammation, and what minor ROS mediated lipid peroxidation does occur has the routine inflammatory response suppressed. 

Evolution sets the levels correctly. It may be blind but it recognises functionality when it sees it.

Now imagine you're an Homo modernus visiting your bank manager in the 1960s to get a mortgage with a borderline adequate income and a just passable deposit, for the house of your dreams. The manager is a local petty Hitler and, even if he intends to grant the mortgage, he's going to really make you suffer, mentally, for it. Because he can.

You're stressed pre interview. So cortisol floods your system. Glucose and FFAs are made plentiful to allow you to fight at your hardest during a necessary mammoth kill. But there are consequences to killing the bank manager and most people, understandably, decide not to do so. Much as they might wish to. Especially if he refuses to grant you the debt you can barely afford. For your safety. There is no energy usage, no AMPK activation, just a serious availability of calories.

If the levels are high enough they will flood your mitochondria and provide substrate well in excess of what is needed for current ATP demand. If ATP synthase refuses to turn because ATP is high and ADP is low, while NADH and FADH2 keep supplying electrons to the ETC at the CoQ couple, delta psi will rise above that safe figure in the region of 170mV and ROS will be produced in large amounts from multiple sites, giving both insulin resistance and tissue damage.

You can use pharmaceuticals to inhibit phospholipase A2 as much as you like to avoid any response to this damage, classically using prednisolone or dexamethasone, but there is still the flood of calories generating ROS at tissue damaging levels. You might feel okay because the whole inflammatory cascade is suppressed, but cells are still dying from high ROS levels. If these happen to be endothelial cells lining your coronary arteries you are in trouble. Some will be.

So I would posit that the increased risk of CVD under corticosteroids could be (in part, there are many other issues) an effect of acute ROS injury. Under corticosteroids the inflammatory response is suppressed but the excess calories are still generating excess ROS. The damage is still done.

Of course the same applies to metabolic syndrome.

I define metabolic syndrome as the inability to shut down FFA availability in the presence of insulin and elevated glucose. FFAs are high from basal lipolysis at the same time as glucose from the diet causes hyperglycaemia. If substrate supply is high enough ROS will be generated and damage will be done. This time there will be an actual inflammatory response, no corticosteroid involved, and people can make statements like "obese people are chronically inflamed". They are producing mitochondria mediated ROS to physiologically resist insulin. If they are forced to continue to accept calories despite resisting insulin, the ROS become simply damaging. Inflammation follows. 

If you do not link this "inflammation" back to ROS then you would, logically, treat it with dexamethasone. Which would be a booboo, as we say in the UK. You would suppress inflammation while supplying even more ETC input, so worsening ROS mediated damage. Those poor coronary artery endothelial cells. In the absence of slaughtering a mammoth of course.

It's also worth thinking about the role of 4-HNE. Just a skim of the abstract of this book chapter is worth a moment:

4-HNE is, in itself, a product of ROS damage. It is also a generator of ROS per se. Much of the inflammatory cascade is self amplifying (which is why we have systems to turn inflammation off as well as on). Taking exogenous 4-HNE has profound effects on ROS mediated signalling in the insulin cascade, at low dose by injection to enhance, at high dose (with other lipid peroxides) orally to inhibit. It matters in the inflammatory cascade as much as in the insulin cascade.

My interest, getting back to Vaughan's obese mice without hepatic inflammation, is rooted in the factors which generate 4-HNE in situ around the ETC and which factors stop this. I'm talking about ROS generation.

Back to those mice and their non-inflamed hepatocytes and adipocytes.


Tuesday, March 19, 2024

Foie Gras (6) inflammatory mRNAs

 Okay, time to look at this study

It presented itself to me while I was looking  for explanations as to why linoleic acid was or wasn't inflammatory in hepatocytes. Being "inflammatory" is determined, here, by the response of treating freshly isolated cells, in this case macrophages, to incubation with linoleic acid at 50 μmol/l for 12 hours, in addition to the normal culture medium. Some ferreting around on the internet suggests that the culture medium used, when fresh, contained glucose at 33 mmol/l. The concentration of LA, 50μM, is low and physiological for a FFA concentration under hyperglycaemia. The macrophages were assessed for expression of pro-inflammatory genes coding for these proteins:

Monocyte Chemoattractant Protein-1 (MCP-1)

Macrophage inflammatory protein-1 (MIP-1)

F4/80 is a gene for the surface marker of macrophages

and good old TNFα and Il6 get thrown around too.

We can ignore the anti-inflammatory gene expression Ym1 and Ym2.

This is what they found for pro-inflammatory gene mRNA relative expression, linoleic acid is tall hatched bars. It's a log scale:

So we can say LA is pro-inflammatory in this model. No one in obesity research wants to find that the cardiological darling and cholesterol lowering PUFA are pro-inflammatory. I told you they were rank amateurs. Anyway, they developed another model, in-vivo this time, which got the correct results.

They fed the mice the diets I mentioned in the last post for four weeks then looked at pro-inflammatory gene expression in both white adipose tissue and liver tissue immediately after euthanasia.

This was much more satisfactory. In both tissues the high PUFA diet was not inflammatory, with a trend towards it being anti-inflammatory in adipose tissue:

Adipose, from Fig 2:

and liver from Fig 3:

However, there is another snippet of information also available in Fig 3. The activity of myeloperoxidase system tells us slightly more than the relative mRNA for inflammatory genes do in the bulk of the figures. Myeloperoxidase tells us whether the WAT cells in Fig2 are actually using the pro inflammatory genes to produce inflammation. This is what they found. The chart is split in two because the HF (high fat diet, something like D12492) was so activating it was an order of magnitude higher than all of the others, so needed it's own scale:

I think we have to be very, very careful about what we mean by "inflammation". In fact I might just take a brief pause here and explain what I think "inflammation" means, before running through what is happening in this study. That may need some cellular physiology rather than mitochondrial physiology.

Here's my conundrum:

CVD is an "inflammatory" condition, sic.

The most potent anti-inflammatory agents available to modern medicine are the corticosteroids.

Corticosteroids make CVD much, much worse.

Somebody, somewhere, has an odd idea about what "inflammation" might be.

With apologies for the hiatus.


Monday, March 18, 2024

Foie Gras (5) An aside on how to stay slim

 Another one-liner:

Now, I like this paper. They found some weird stuff which is fully Protons compliant. That's not today's post. I just wanted to share this graph:

These people are clearly rank amateurs rather than hardcore obesity researcher, they have utterly failed to "improve" the saturated fatty acid (SFA) diet to make it obesogenic. Want to stay slim? Cocoa butter. As good as chow, and who would want to eat chow? Though I'd personally choose beef fat.

The linoleic acid diet is also not very fattening. Hmmmmm.

Even without looking at Table 1 you absolutely know what has been done, both to get the low weight gain on LA and to eliminate the inflammatory changes in the liver (which are there in cell culture).

Hard to decide whether to go through this study or start on the mitochondrial data from the hungry Italian rats. Or maybe go to Winks Meadow to see if the Green Lipped orchids are up and flowering yet. 

The sun is shining.


Foie Gras (4) RER

A quick one-liner from the Italian rats in

Fat Quality Influences the Obesogenic Effect of High Fat Diets

These are the RERs of the rats on the last day of the study. The horizontal lines are the food quotients calculated from the macros of the diets. Under stable conditions 24h RER should equal FQ. These are not stable conditions

It's clear that both groups of rats are catabolising protein for energy in excess of what is present in the diet. More so for the lard diet than the safflower/linseed. They are, undoubtedly, hypocaloric.

The lard based diet rats fail to oxidise lipid despite it being in the diet because they are still sequestering lipid in to adipocytes. Basal lipolysis may be high but hyperinsulinaemia (not measured) from incipient metabolic syndrome is recycling some of that lipid back in to adipocytes. Where it's not being oxidised. The rats are cold and hungry.

The safflower/linseed group have normal fat oxidation, ie equal to FQ, because fat is not being lost in to adipocytes, it's being lost by uncoupling so lipid oxidation on RER is normal. It may provide nothing but heat but the RER looks good. Of course these rats are warm and hungry.

I like it.


Edit, I don't like the rats being hungry! Just the RER data. End edit.

Foie Gras (3) The Japanese mice

So the interesting question about the rats in the Italian study which were fed on the Safflower/linseed oil diet is:

How calorically restricted were they?

In the absence of a control group allowed ad lib high fat intake (or even one fed chow) we will have to look elsewhere to estimate this. Japan is a good start with this paper:

Just give Bl6 mice a free choice to munch chow and/or drink corn oil and they will eat/drink progressively more corn oil over 4 weeks until they are stable at ~75% of their calories from corn oil, 25% from chow. Which gives us in the region of 38% of calories from linoleic acid. They are weight comparable to chow-only fed rats throughout, maybe slightly lighter:

Under stable conditions (weeks six to eight) they feed themselves 35% extra total calories to maintain an identical bodyweight to the chow fed mice:

This is because they were uncoupling, primarily in interscapular brown adipose tissue but (from other papers, a future post) they also do so in white adipose tissue.

Back to

Fat Quality Influences the Obesogenic Effect of High Fat Diets

If we imagine the Italian rats had been offered ad-lib access to the safflower/linseed diet we could expect them to eat somewhere in the region of 380kJ x 135%, so around 500-530kJ/d.

They got fat on just 380kJ/d. Like the lard fed rats. But by two weeks they were a bit less fat (p<0.05).

How does this work?

On day one of the safflower/linseed diet there is a marked increase in insulin sensitivity, by the standard Protons/polyunsaturate mechanism, with an associated hypocaloric episode as calories poured in to adipocytes but no additional food was provided. There is no significant increase in uncoupling immediately.

But now we have an insulin sensitive liver and the standard response of the liver to ingress of excess non-carbohydrate calories is to signal, using FGF21, to BAT to induce uncoupling giving thermogenesis and calorie disposal.

The time scale for onset of uncoupling could be estimated if we had daily food intakes and body weights, but we don't, so let's just guess at around a week.

As uncoupling in white adipocytes kicks in they will become poorly able to respond to insulin with the correct ROS signal, so insulin signalling decreases and they release FFAs. At the same time as this suppressed insulin signalling occurs BAT will be disposing of bulk calories by thermogenesis and the caloric drive for the pancreas to secret insulin also drops, again assisting lipolysis.

By two weeks there is active, on going weight loss from an obese baseline. Lipid is being lost at an excessive rate accentuated by hypocaloric eating and the liver is dealing with this excess, under hypoinsuliaemia, in part via BAT and in part by the peroxisomal mechanism described in the previous post, the cost of which is, in rats (and mice), of hepatic lipid accumulation.

Would the lipid accumulation have occurred if the Italian rats had not been calorically restricted?


Uncoupling made the Japanese rats hungry, they ate extra to stop pathological weight loss. The extra calories include some carbohydrate and would have slowed lipolysis by raising insulin secretion to a level where lipolysis did not overload the liver. There are several papers to cover this in future posts.

Given long term ad-lib access to an uncoupling diet based on PUFA the rats would have eventually and gradually lost weight until they matched bodyweight with their non existent control group. Assuming these mice are anything to go by, who did it over a period of 10 weeks. From here:

I plotted the numbers for the body weights in Table 2, by eye, in PowerPoint. The dashed lines are the ones to follow. Obese on ad-lib lard, back to control (low fat, 35% of calories from sucrose) mouse weight on ad-lib safflower oil based diet giving LA at 35% of calories:

Okay, that will do for today.

I would not, in any way, endorse drinking either varnish or safflower oil, even if they produce weigh normalisation by what are, to me, metabolically convincing/plausible mechanisms.

Better not to make your adipocytes pathologically insulin sensitive, then you wouldn't need to address the obesity with potentially pathological double bonds in your food.

Probably safflower oil and hepatic inflammation next.


Saturday, March 16, 2024

Foie Gras (2) Lard fed rats

There is too much in the Italian rat study for a single post. Here's the easy bit discussing the rats fed the lard based diet.

I guess everyone is familiar with this graph in Figure 1 of this paper from the Schwartz lab:

Obesity is associated with hypothalamic injury in rodents and humans

which provides this gem. These are the *daily* caloric intakes of rats on bog standard lab chow, in grey, or during the sudden onset of feeding lard based D12492, in black:

These are Long-Evans rats weighing 300-350g. I drag this up because Protons suggests that on day 1 of exposure to D12492 the rats immediately sequester approximately 70kcal of energy in to adipocytes (ignoring processing losses) while still maintaining the basal (when chow fed) intake of 75kcal which are needed to run the rat's metabolism.

This immediate significant fat gain on day one (seen in section F of the same figure) raises adipocyte diameter, so increases basal lipolysis, so reduces hunger, such that on day two the extra food intake needed is lower and eventually, by day seven, excess food intake is no longer significantly increased and by day 14 it is normalised. Simple, yes?

Now let's return to the current study of interest discussed by Tucker, which I think of as the "Italian" study:

This used Sprague-Dawley rats weighing 250g, so probably more actively growing than the rats in the Schwartz lab study. They were measured as consuming 90kcal/d (380kJ/d) of chow assessed over the time before the study started. So we can plot the food intake of rats in the Italian study on a modification of graph H of the Schwartz study. I've left the chow fed line from the Schwartz study as an imaginary chow fed control line which was omitted by this group and I've renumbered the y axis to reflect the values of caloric intake actually reported in the Italian study:

We know that the red line for the lard fed group is close to correct from the methods section:

"Rats were divided in two groups with the same mean body weight (250 ± 5 g) and were pair fed with 380 kJ [90kcal] metabolisable energy (ME)/day (corresponding to the spontaneous energy intake of the same rats, that was assessed [on chow] before the start of the experiment) of a lard-based (L) or safflower-linseed oil based (S) diet for two weeks."

The Italian rats, without any access to the luxurious amounts of D12492 provided to the Schwartz rats, still got fat. 

If you view this from my perspective this is not surprising. The Italian rats still lost calories in to adipocytes but, without access to extra food, had an hypocaloric crisis. They ended up with a reduced percentage of protein mass in their carcass, despite an increased fat mass. Obese and sarcopaenic.

However, by two weeks on a diet which sequesters lipid in to adipocytes at the cost of reduced growth, a few of the rats will have achieved a sufficient increase in basal lipolysis to normalise hunger, as per the Schwartz rats, at the "cost" of obesity. It is very simple to multiply 380kJ/d by 14 days and get 5320kJ of offered food over 14 days. The mean of the actual food intake over the study was measured as 5286kJ in total for the lard fed rats. We don't have individual data but any rat still hungry on day 14 will have eaten all of its 380kJ, but no extra, because none was on offer. Adequately obese rats will have, via increased basal lipolysis, not needed to eat all of the 380kJ offered, ie these rats will have slightly reduced the mean total energy intake, by 34kJ, probably in the last few days of the study.

These rats are in a difficult position, they must maintain an adequate fat mass for increased basal lipolysis to offset increased insulin mediated lipid sequestration induced by the linoleic acid component of their diet. As they grow they will have to increase fat mass to maintain adequate basal lipolysis to function.

Running on basal lipolysis derived FFAs at a time when you have access to dietary glucose is the basic definition of metabolic syndrome.

If you keep adipocytes small by forcibly keeping insulin low (ie caloric restriction) you will completely side step increased basal lipolysis, side step insulin resistance, side step or delay many diseases and *increase* the duration of the miserable, hungry existence which will be your extended life.

There is no way in which you can transfer sufficient the FFAs from insulin sensitive but non-adequately distended adipocytes to hepatocytes as is needed to maintain a fatty liver. Caloric restriction is highly protective. Just ask any obesity researcher, the cure for fatty liver is hunger. Oops, I mean weight loss rather than hunger, in obesity doublespeak.

Safflower/linseed oil next.