Insulin resistance (09) NAC and body weight

 Let's start here with:

N-acetylcysteine Protects Mice from High Fat Diet-induced Metabolic Disorders

in particular with Fig 1:

That's correct. N-acetylcysteine is virtually completely protective against D12942 (linoleic acid) induced obesity.

Think about that. One of the most effective scavengers we have for ROS is hugely protective against LA induced obesity. The presumption almost has to be that excess ROS are causal in obesity. Something with which I profoundly disagree.

What is going on?

The obesity from D12942 stems from pathological insulin sensitivity due to linoleic acid's failure to generate an adequate ROS signal to resist adipocyte distention. If we go back to Czech's work

Evidence for electron transfer reactions involved in the Cu2+ -dependent thiol activation of fat cell glucose utilization

with its image illustrating the role of ROS (here from hydrogen peroxide) in mediating a putative insulin signal in to glucose uptake and oxidation by adipocytes:

Now, this is a model. A very tightly controlled and far from reality model. It's useful because it conceptualises certain functions of ROS. The ROS signal is not a simple function of insulin exposure. Many features of cell metabolism contribute to ROS, especially concurrently oxidised fatty acids, NOX enzyme activities, the glycerophosphate shuttle and probably several other sources.

But in its most basic form we can assume that the levels of ROS added to the system, from a bottle of hydrogen peroxide bought from Sigma Aldridge, are equivalent to the end result of caloric ingress in some sort of proportion to insulin exposure. Pax fasting.

So I've taken the right hand graph from above, smoothed it and added what might be the level of insulin associated ROS generation which might result in 0.01mM H2O2 generation (here supplied from the bottle, no actual insulin involved):

And we can do the same for higher added levels of H2O2, considered to represent results of moderate and high levels of insulin exposure:

We can assume that the highest level of response to peroxide, here 0.3mM, represents the peak of insulin's ROS mediated ability to activate its signalling cascade.

If we go back to the bottle and increase ROS levels above 0.3mM, to 1.0mM or even 4.0mM of peroxide, we have a decreasing level of activation of the insulin signalling cascade. This nicely represents/models insulin-induced insulin resistance.

We can illustrate it like this:

But those "supra-physiological ROS" came out of a bottle, not from metabolism and certainly not from a response to endogenously secreted insulin.

What seems far more likely in vivo is that something like this happens:

in which physiology makes every attempt possible to limit ROS generation to within a given target range, going back to the origin of life or at least to LECA (last eukaryote common ancestor). As that alpha proteo-bacterium snuggled down inside its archaebacterial host it carried on using its normal growth signal (superoxide rather than peroxide) while actually inside the archaebacterium., where ROS generation concentrations became quite a problem. See Dave Speijer's work.

Anyway, I'd like to take this, my interpretation of Czech's seminal work, and drop a dose of n-acetylcysteine on to it. And see what might happen. Theoretically.

Okay. Let's imagine a dose of NAC is given to a mouse which removes, from all cells, about 0.2mM of H2O2. This shifts the onward insulin signal to the left. All of the ROS generated by a low level of insulin will be eliminated and the onward signal to effect insulin's action will be reduced (but not eliminated) at the higher levels of insulin:

It's also clear that the level of ROS which would have resisted additional insulin signalling (ie 0.3mM H2O2) has now been reduced to 0.1mM H2O2.

So there is no mechanism to resist insulin signalling and calories will continue to enter the cell until a rapidly rising delta psi raises mitochondrial ROS generation high enough to hit that magic 0.3mM of H2O2 to resist insulin using this "correct" level of ROS. That will look like the dashed line:

The question is this. Do the adipocytes in the live mice featured in the weight graph above ever get their insulin signalling high enough to actually generate the 0.3mM H2O2 to stop caloric ingress?

The answer is yes if you choose the correct dose of NAC, Like this:

This is what happens to the chow fed mice with NAC supplementation. They *appear* to be normal. All pretty straight forward.

Now we have to think about the D12942 fed mice. I will again have to simplify things to a point which strains the limits of accurate description, so forgive me.

Here's that normal model on chow reaching peak insulin signalling, using 0.3mM H2O2 to represent the limit of onward insulin signalling, which allows a peak of 20nmol/of labeled glucose per 10*7 cells to be oxidised to labelled CO2:

 

Next we have to imagine the same situation but now include some fatty acids in to the sources we imagine are generating the H2O2 signal. Palmitate or stearate do this correctly. Linoleate doesn't. At a given level of substrate oxidation the linoleate component gives too few ROS for its level of oxidation. The ROS line is again, without antioxidants, slightly shifted to the left all the way through until it finally hits that 0.3mM value that stops caloric ingress:

With the curve shifted to the left more calories enter the cell before that magical 0.3mM of peroxide is achieved. So our 20nM glucose/10*7 cells becomes 22nM glucose/10*7 cells. Insulin, as well as oxidising glucose, will also store fat. That extra 2nM glucose/10*7 cells is a surrogate for a few extra nanograms of fat stored. In a human this just might add up to half a kilo over a period of a year....

This is the problem which NAC corrects.

So here is the ROS curve from the linoleate doodle treated with NAC which shifts the curve down and to the left (in purple), exactly as we discussed above for the chow fed situation:

Of course this doesn't actually happen, there is no insulin resistance at 0.1mM H2O2 ROS. Calories continue to enter the cell until, at an higher level of ingress, ROS finally reach 0.3mM and insulin resistance kicks in, like this:

The dose rate of NAC needed to nicely balance the effect of D12492 in Bl/6 mice has to be worked out by trial and error or be taken from other publications. Obviously to 100% normalise end body weight would need slightly more than 2g/l of NAC in the drinking water but the graph at the top of the page is perfectly good enough.

That's what is happening.

Of course there are some glaring problems with NAC which the above discussion hints at but I've not gone in to because the post is way too bloody complicated as it is. Next time.

Peter

Insulin resistance (08) UCP2

Another accidental find from my hard drive, probably saved somewhere around 2020.

Differential effects of saturated versus unsaturated dietary fatty acids on weight gain and myocellular lipid profiles in mice

These are the weight gain profiles with the linoleic acid compositions of the diets added.

A couple of comments. Obviously the low LA cocoa butter diet was indistinguishable from the low LA low fat diet, which probably also contained 35% of calories as sucrose (D12450B). I have to say I'm no fan of sucrose but one thing it does not do is to make mice fat. Some people really do think it makes people fat. Shrug.

Second point is that "olive oil" has an LA content varying from 3-21% LA as it comes from the freshly squashed olive, depending on the cultivar and the weather that year. This may increase if cut with rapeseed oil before it was purchased by Research Diet Services in the Netherlands. The study lab had, and used for tissue analysis, a gas chromatography machine, but did not use it on the diets the check their fatty acid composition. So no one knows what the mice ate in the olive oil group. I'd guess just under 4.5% of energy as LA.

All of which is pretty straight forward.

The group which I like best is the one which ate 35% of their calories from linoleic acid. That's the safflower oil group

Now this is nothing new. I've discussed very high intakes of linoleic acid as a weight loss intervention before. The current study merely replicates:

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

discussed here (and yes, these folks did measure the dietary fatty acids).

What is new is that we have both hepatic and intramuscular triglycerides measured. This is what we get in the liver:

We have to bear in mind that the low fat group, on 35% of E from fructose, should have some degree of hepatic triglyceride accumulation awaiting export as VLDLs under fasting conditions. The liver samples where obtained at 14.00h when most respectable mice would be sleeping but not hungry.

The two high LA fed mouse groups gave the two high liver triglyceride samples as you would expect and the cocoa butter fed mice resisted insulin's signal to store liver fat. Nothing exceptional.

But 35% linoleic acid in your diet is very protective against triglyceride accumulation within your liver, in addition to be moderately protective against obesity. Before we go on to mechanisms let's also look at muscle triglycerides. They look like this:

with the one problem of the cocoa butter based diet. Despite complete protection from obesity and hepatic lipid accumulation, stearic acid does not protect against lipid build up in the muscle cells. I find that difficult to understand.

I would guess one of two things. Perhaps the data are wrong. I very much doubt this, the group seem quite innocent. Second, running your metabolism on 45% fat vs 10% fat might require some degree of triglyceride storage in muscle cells, facilitated by insulin, which seems more plausible. Though why the various levels of LA did not produce a differential effect is problematic. File under "think about it".

So, on to the real question. Why is a diet containing 35% of energy as linoleic acid completely protective against lipid accumulation in the liver?

We can start with UCP1. The liver does not, as far as I can find out, ever express the gene for UCP1. Given a caloric overload to hepatocytes they just off-load excess caloric substrate, especially fatty acids, to BAT using FGF21 and let the BAT get on with generating a warm, enhanced oxygen consumption environment so the hepatocytes only see a caloric supply they are happy to deal with. That's the job of UCP1. And thermogenesis is nice for mice at 20 degC.

UCP2 is used in many, many tissues, but is not constitutively expressed in the liver. And it's not thought to be used for thermogenesis. When you supply the liver with FFAs for 24h, the gene is expressed whether the FFA supplied is oleate or linoleate, so I'd guess this is a generic property of FFA exposure.

UCP2, a Member of the Mitochondrial Uncoupling Proteins: An Overview from Physiological to Pathological Roles

We know that FFA oxidation will automatically generate ROS without needing high delta psi if the delta psi is above a certain value. It looks like UCP2 has a function (among many) of reducing the generation of these fatty acid oxidation derived ROS, at least in liver tissue. Three quotes from the authors:

"The UCP2 activity is increased in the presence of ROS in a manner dependent on fatty acid oxidation. As a result, the UCP2 acts to revert ROS production by decreasing the membrane potential of mitochondria through a mechanism of H+ leak that could be different from UCP1."

"The role of UCP2 as a regulator of mitochondrial ROS production is corroborated by results in the presence of nucleotide GDP that blocks the UCP2 activity, causing mitochondrial membrane polarization and ROS production [46]."

"We can assert that the H+ leak activity of UCP2 is a “relief valve” for the polarized IMM (Figure 2) to avoid the formation of superoxide anion in conditions in which the mitochondrial electron carriers of the respiratory chain are in a reduced state [4,68]."


The question is: Does UCP2 lower delta psi enough to effectively limit ROS mediated activating insulin signalling?

The normal limitation applied to insulin signalling, from the Protons perspective, is a surfeit of ROS. However it is equally possible to limit insulin signalling by reducing the generation of ROS or by scavenging the normally produced ROS using drugs such as N-acetyl cysteine or Mito-TEMPO, as in this study:

Insulin-stimulated lipid accumulation is inhibited by ROS-scavenging chemicals, but not by the Drp1 inhibitor Mdivi-1

Major caveats are that is is in vitro and uses those highly malleable 3T3-L1 adipocyte-like cells with which you can show just about anything. There is an in-vivo rodent study using NAC which probably deserves its own post some time.

Anyway, the un-answerable question is whether activating UCP2 within hepatocytes drops the mitochondrial delta psi to limit the generation of ROS to the point that insulin signalling fails and the drive from insulin to store intra hepatocyte triglyceride is limited.

We know that BAM15 can do this, it is largely studied as a management for what used to be called NAFLD. However BAM15 is a pharmacological uncoupler and has no limit to how significantly it can drop delta psi. The beauty of UCP2 is that it is inhibitable by guanosine diphosphate, used pharmacologically as an analogue of ADP. Increasing ADP above a certain level shuts down the uncoupling activity of UCP2. Both BAM15 and UCP2 can increase the ADP:ATP ratio, and so activate AMPK, but there are strict limits placed on this for UCP2 in normal physiology. 

I would posit that linoleate is an excellent agent for the induction of UCP2 gene expression in the liver and, once the protein is in place, it will be a significantly better activator of the uncoupling process than either oleate, palmitate or stearate. As in

Polyunsaturated fatty acids activate human uncoupling proteins 1 and 2 in planar lipid bilayers

probably limited by rising ADP levels, or falling ATP levels if you prefer.

Bear in mimd I am not considering a degree of uncoupling which would measurably increase whole body oxygen consumption, merely a small, targeted lowering of delta psi to reduce ROS generation and so reduce insulin signalling. It's not a "calories out" scenario.

Reduced insulin signalling results in reduced hepatic triglyceride storage, reduced intramuscular triglyceride storage and reduced adipose triglyceride storage. All resulting from LA exposure if the intake is high enough, ie with LA at 35% of daily energy intake. By 45% of calories there is evidence of uncoupling through UCP1 in BAT. As in:

Voluntary Corn Oil Ingestion Increases Energy Expenditure and Interscapular UCP1 Expression Through the Sympathetic Nerve in C57BL/6 Mice

Nobody should doubt that linoleic acid, at obesogenic levels (6-20% E), drives both adipocyte and hepatocyte lipid storage and mediates hypometabolism by accentuating insulin signalling. At around 35% E it activates UCP2 in hepatocytes (and probably in adipocytes) so limits ROS mediated insulin signalling which reverses hepatic lipid accumulation (and obesity) with a minimally detectable increase in metabolic rate. At 45% of E it activates UCP1 in BAT via FGF21 giving a detectable rise in metabolic rate in addition to avoiding pathological lipid accumulation.

It would be nice if linoleic acid metabolism effects were simpler.

They're not.

Peter

Insulin resistance (07) Astrup's FFAs in the post-obese

Preamble/caveat. You have to be careful with Astrup's work over the years. When you get in to the fine print some of his control groups are perhaps not quite as normal as you might like and some of the post weight loss subjects might not be quite as slim as you might like. His work spans many years and what the subject might have been eating when free living will have changed and the lead-in food served from the hospital kitchens may well have changed too. So caution. Having said that several of his papers have picked out various features of the pre/post obese. This is one such.

I wrote this post about a year ago but never hit publish. Here we go.

Okay. It's time to look at the people in this study from Astrup's group

Insulin sensitivity in post-obese women

and in particular I'd like to reconstruct this part of figure 2. It illustrates nicely what happens when you track the FFA levels under an hyperinsulinaemic euglycaemic clamp:

Astrup discussed the right hand data points at 105 minutes, as below, with their statistically significant difference in free fatty acid levels, lower in the post-obese, whom I identify as identical to pre-obese, if they were to eat to satiety using the food choices which made them obese in the first place.

Astrup talks, correctly, about how we know nothing about what factors are influencing the FFA levels to produce this difference. This would need tracer studies which do not appear to have been done, certainly not by Astrup's group. So I am going to speculate, as I do, about how much explanatory power the Protons hypothesis has in this situation.

The first thing to do is to remove the lines for both groups from time -15m to time +45m. This is valid because Astrup's diagonal hash mark indicates that these lines were never intended to indicate what the FFA levels were while the clamp was being set up and started. I've also stuck in a line for time zero. So now we have:

I'm going to assume that FFA levels from time -15m to time zero didn't change by enough for me to bother about, giving this:

which is non controversial. What is more dubious is that I'm now going to guesstimate what the FFA levels did between time zero and time +45m. I'm going to suggest that the lines follow some sort of exponential decay. Like this for the control group in red:

Next I want to torture my scarce data points in to a slightly different curve, which accepts that FFA levels have probably plateaued out at ~300μmol/l, like this:

Now let's do the same for the post-obese group. The initial curve in blue looks like this:

and it takes only a minimalist eye of faith to extend it to this longer line:

We can now remove Astrup's hard data points and just look at a concept of what might actually be happening in the both groups, as viewed from the Protons perspective. Like this:

In the control group being exposed to insulin at just over 1000pmol/l with glucose clamped at 5.0mmol/l the FFAs bottom out at 300μmol/l. In the post-obese they are down to ~100μmol/l and are still dropping at the 105 minute time point.

Insulin at 1000pmol/l is around the peak level experienced transiently by normal humans at 30 minutes after a modern low fat, high carbohydrate meal.

I'm now going to assume that the fall in FFAs under the clamp is from insulin mediated suppression of FFA release from adipocytes. There will also be enhanced uptake but never mind that.

Following the red line of the control group we have, by 30 minutes, a state where adipocytes are exposed to 1000pmol/l of insulin, which will have translocated way more GLUT4s to the cell membrane than exposure to 5mmol/l of glucose would normally generate. So glucose pours in to cells, adipocytes included. There is no fall in glucose because, well, it's a clamp. By 30 minutes the control subjects' adipocytes are getting a large amount of glucose from the insulin/glucose infusion superimposed on lipid oxidation derived from a plasma concentration of ~300μmol/l of FFAs.

The adipocytes are generating 300μmol/l of FFA derived ROS. They are receiving enough glucose in addition that that their energy needs are met completely, there is a physiologically appropriate rise in delta psi and a completely appropriate rise in overall ROS which is the actual signal to limit insulin signalling, ie demonstrate cellular "fullness". Insulin's action is inhibited.

Inhibiting insulin's action inhibits insulin's inhibition of lipolysis.

FFAs stop dropping, the adipocytes stop taking up and storing lipid, the VMH feels full and the subject stays slim.

The blue line representing the post-obese shows the same initial fall but for these people the oxidation of a significant amount of linoleic acid means that, by 30 minutes, the adipocytes and VMH cells are ATP replete but are only generating sub-physiological levels of ROS, so there is no sense of cellular "fullness" in adipocyes (or the VMH cells) so no rejection of insulin signalling.

Insulin signals, lipolysis continues to be inhibited, plasma FFAs continue to fall. In real life, outside of the clamp situation, you now have to eat. Now.

You have to eat because the cells in the VMH fail to generate an adequate ROS "fullness" signal from their linoleic acid oxidation and also because adipocytes have "stolen" and retained FFAs which the VMH now never gets to see. Both happen, plus other effects, for other posts.

Hunger is no fun.

Peter

Insulin resistance (06) VMH

I've got these three, apparently independent, origins for the concept that insulin resistance is an evolutionary conserved mechanism, essential for survival but maladaptive in the modern world. What links them is the concept that an essential feature can be maladaptive. Here are the quotes:

Insulin resistance is an evolutionarily conserved physiological mechanism at the cellular level for protection against increased oxidative stress

Erol 2007

"For this reason, insulin resistance could be a physiological mechanism activated at the cellular level in response to conditions stimulating ROS production and leading to the prevention of oxidative stress, and extension of life. Concerning the whole organism, however, IR is a maladaptive process in the long term causing a diabetic state."

Insulin resistance is a cellular antioxidant defense mechanism

Hoehn et al 2009

"For most organisms the latter [nutrient excess] is presumably quite rare or at least an intermittent phenomenon emphasizing why the current situation of constant nutrient oversupply is not easily tolerated."

The evolutionary benefit of insulin resistance

Soeters and Soeters 2012

"If we accept that insulin resistance benefits limitation of net degradation of protein in long-term starvation and stress while facilitating PPP and anaplerosis, the role of insulin resistance in overfeeding is enigmatic."

Insulin resistance is absolutely essential on a cellular basis. All insightful people agree on this.

Also, on a cellular basis, insulin *resistance* is the hallmark of cellular "fullness". It marks out a given cell as being energy replete. That's Protons.

Evolution very rarely invents anything new. If you have an organism of increasing complexity which needs to regulate its overall food intake at the macroscopic level it is not going to "invent" a food counting mechanism to decide whether a cow has eaten enough mouthfuls of grass to sustain it for the coming few hours.

No. Evolution will take a few representative cells, feed them on a representative sample of what is available, energetically, in the blood stream, and see if that energetic supply generates enough of an ROS signal to decide whether those cells have enough substrate to generate an adequate "fullness' signal. This requires an adequate level of ROS generation to resist insulin. Evolution takes the core system of ROS and converts it in to either hungry or not hungry using a set of specialised brain cells.

Insulin *resistance* (ie ROS) within the ventromedial hypothalamic cells *is* satiety. If the VMH is "happy", the brain assumes that the rest of the cells in the body will also be getting enough metabolic substrate to also be "happy".

I laid out in too much detail how linoleate oxidation damages this system during the last post.

Obesity results from a failure of individual cells to adequately signal that they are "full" when they are actually having all of their energetic needs being met. This applies as much in the VMH cells as in those anywhere else with insulin regulated caloric ingress.

None of which excludes the related loss of calories in to adipocytes which will, in its own right, actually reduce the level of metabolic substrate being visualised by the VMH to compound the problem. Another post there.

Also, we need to think about the role of insulin resistance in established obesity. It's still functional but in a very different context.

Peter

Insulin resistance (05) Spanish butter in Noddy numbers.

Let's go back to butter vs PUFA as in the last post. I'm going to use reductio ad absurdum to try make it absolutely clear why a failure of insulin resistance is catastrophic and obesogenic. I'll use stearate and linoleate to simplify matters further, ignoring the other fatty acids in butter and soybean oil.

Let's say a cell needs to generate 100 ATP/min. When it has that 100 ATP/min, it doesn't need any more. If the cell is oxidising 100% glucose there will be low levels of ROS generated, insulin will signal and more glucose will eneter the cell until ATP generation settles at that preferred 100 ATP/min level. Trying to go above 100 ATP/min will raise delta psi and so raise ROS generation which, as it reaches 100 ROS/min, will reduce insulin signalling to a level which will allow just enough glucose to enter to maintain that 100 ATP/min generation. Excess ROS/min above this 100 ROS/min level will signal that the cell is "full" and must resist insulin's signal so no extra enters.

With fat it is different. Oxidising fat will always generate ROS, irrespective of delta psi and this ROS/min signal will act as a base-load of ROS on to which extra ROS from glucose, derived from raising delta psi, are added to generate the "full" signal, which I'll still consider to be 100 ROS/min.



Let's imagine stearate oxidation is generating 60 ROS/min out of the 100 ROS/min needed for "fullness" while generating 60 ATP/min out of the 100 ATP/min needed to run the cell. Generating 40 more ATP using glucose will generate the extra 40 ROS/min when delta psi rises enough to generate them to give the essential 100 ROS/min needed to signal "full". We end up with the necessary 100 ATP/min, a perfectly filled cell and 100 ROS/min "fullness" signal which limits excess caloric ingress by resisting insulin. All is stable.

Now let's oxidise linoleate to generate the same 60 ATP/min. It will take a smidge more than stearate because linoleate oxidation lacks the energy from 2 FADH2s. So linoleate generates its 60 ATP/min, as needed by the cell. But it doesn't generate the correct number of ROS (due to the lower FADH2 input) for the ATP it provides. Let's say it generates 55 ROS/min instead of 60 ROS/min.

As glucose tops up the ATP supply to 100 ATP/min it will generate its own delta psi dependent ROS/min signal, exactly as it did with stearate, ie 40 ROS/min. This 40 ROS/min from glucose will be added to the 55 ROS/min being provided by linoleate.

That gives 95 ROS/min. This is too low to adequately limit insulin signalling. But the cell is already ATP replete and *needs* to limit insulin signalling. Three things happen. Calories enter the cell in excess. They go in to storage. Rising delta psi, to pathological levels, adds the final 5 ROS/min to shut down insulin signalling.

ATP levels are very, very tightly controlled in a healthy cell. Calories in excess of cellular needs have entered the cell under linoleate oxidation because, despite there being the perfect 100 ATP/min, the cell was still signalling at only 95 ROS/min, ie giving a "still hungry" signal.

What happens to the excess calories which have entered the cell because insulin was allowed to signal for five extra ROS/min before the correct 100 ROS/min "fullness" signal was achieved? 

They get stored (if you are lucky, other things can happen). You get fat.

That's it.

Peter

PS obviously it's just a tiny step to transfer this concept to the brain. Maybe another post.

Insulin resistance (04) Spanish butter

I am firmly convinced that linoleic acid induces the rapid onset of insulin resistance in humans.

This is beautifully illustrated by this paper, courtesy of Tucker. In the study soybean oil was given either orally six hours before an hyperinsulinaemic euglycaemic clamp or as an intravenous infusion over the six hours leading up to the clamp. Each fat load was 900-1000kcal in total:

Mechanisms underlying the onset of oral lipid-induced skeletal muscle insulin resistance in humans

The summary figure is this one:

The M values are derived from the glucose infusion rate needed to maintain stable blood glucose during the clamp. The lower the value, the more insulin resistance is present. So iv intralipid, oral soybean oil or intravenous lipopolysaccharide all produce insulin resistance compared to the small infusion of glycerol given to the control group.

Clearly all interventions approximately halve the insulin sensitivity of the subjects. I'll come back to the fascinating insulin resistance generated by lipopolysaccharide another time.

So soybean oil, per os or iv, clearly induces insulin resistance.

What we are looking at with the oral dose is this, from a different study. I've kept with the Novotny/Shulman colour scheme for po fat and con:

At the 6 hour time mark an hypothetical clamp would produce what Nowotny and Shulman published:

No arguing. Consuming PUFA gives you insulin resistance. It's obvious by eye at one hour in the Spanish study and is still detectable using a clamp as late as six hours in Nowotny's study, by which time blood insulin and glucose are completely back to baseline in Spain and never budged for Nowotny.

But.

What would have happened if, instead of a PUFA load, there had been a butter load, of equal calories?

We know the answer because it was done in Spain and butter comes out like this:

At the one hour mark, from blood glucose/insulin levels, there appears to be about 80pmol/l more insulin needed in plasma to deal with the small carbohydrate/protein load given with the fat. This represents insulin resistance. More from butter than from the PUFA ingestion.

Had Nowotny and Shulman also done this and then clamped at the six hour mark, would they have found butter caused a lower insulin sensitivity than the PUFA load by this time?

Yes they would. Unless something very strange occurs between one hour and six hours, then butter is, absolutely, going to cause *reduced* insulin sensitivity compared to the PUFA load during the clamp.

Frankly I'm amazed that Shulman didn't do this. It would have been a great butter bashing publication. But then, I don't expect too much from Shulman's group.

So.

PUFA ingestion definitely, absolutely, causes insulin resistance.

The problem is that it doesn't cause quite enough insulin resistance.

The correct amount of fat-derived insulin resistance to normalise caloric ingress in to a given cell is provided by butter derived FFAs, or, better still, those from tallow. Or, in pre-agricultural times, from any animal fat.

Peter

Insulin restance (03) The failure to resist hat trick

I was searching my hard drive for another paper when I came across this one:

Substituting dietary saturated fat with polyunsaturated fat changes abdominal fat distribution and improves insulin sensitivity

which I picked up on X back when I used to be there.

Now, there are many, many, many problems with this paper which I don't want to dwell on here. All I want to document is that switching from a saturated fat including diet to an high polyunsaturated diet for 4 weeks *improves* insulin sensitivity in a motley collection of human beings. PUFAs increase (ie induce incorrect) insulin sensitivity.

The bottom line is that PUFA at 21g/d on a 2150kcal/d diet ie at ~9%E allows improved glucose uptake compared to PUFA at 10g/d on a 2400kcal diet, ie at ~4%E, especially during the final 20 minutes of a 120 minute hyperinsulinaemic euglycaemic clamp.

It's worth noting that while the glucose infusion rate was higher at all time points under PUFA, they only did stats on the final 20 minutes, which is quite normal for a clamp study.

Why do I like this mediocre study?

Confirmation bias.

We know that selective insulin sensitivity kicks in within an hour of a high fat meal, ie the Spanish study, and this is maintained out to 40 hours of oral fat ingestion, ie the Cocoa study. Both are discussed here.

The current study extends this out to 4 weeks.

A nice hat trick.

One entertaining quote from the abstract is their conclusion:

"If this result is confirmed in longer-term studies, this dietary manipulation would be more readily achieved by the general population than the current recommendations and could result in considerable improvement in insulin sensitivity, reducing the risk of developing Type II diabetes".

This, sadly, is a forlorn hope. 

What is going to happen to a person who increases their insulin sensitivity? What does insulin do to their adipocytes?

They get fat.

The human data suggest that having obese parents marks you out in two ways. One is that you eat, within the limits of the study assessment, more PUFA per day than a similar weight person with neither parent overweight. The second is that you are, on fasting insulin/glucose levels, very, very insulin sensitive. I would suggest that enhanced insulin signalling is going to make you fat.

You know the paper.

Being pre-obese allows the insulin sensitivity to show. Becoming overweight normalises the insulin sensitivity. Becoming seriously overweight leads to impaired glucose tolerance then eventually to frank type II diabetes.

I could throw in non PUFA derived rodent models.

The ventromedial hypothalamic lesion model, the neonatal MSG model, the gold thioglucose model. All present with pathological insulin sensitivity, so long as you measure it before obesity sets in.

Eventually obesity supplies FFA from distended adipocytes which, when oxidised, signal that it is essential to limit insulin signalling mediated caloric ingress in to cells on a pro rata basis to normalise the overall caloric ingress, accounting for that provided from FFA oxidation. Each cell needs to regulate caloric ingress to meet its correct needs.

All of the above rodent models eventually become obese and insulin resistant. But the insulin resistance *requires* the obesity. Caveat: insulin resistance requires that enough cells are large enough to release FFAs via accentuated basal lipolysis. Having a tonne of small adipocytes makes you obese without becoming insulin resistant. Until enough of those small adipocytes become large enough of course...

So increasing PUFA to 10% of energy intake will undoubtedly make you insulin sensitive. It will also make you pre-obese.

Just add time to delete the "pre-" part.

Peter

Insulin resistance (02) those Rhesus monkeys

I wrote this post several months ago but never got round to publishing it. Here we go.

I don't like this paper.

Targeting LDL improves insulin sensitivity and immune cell function in obese Rhesus macaques

and I am very grateful to Tucker for posting it on X and also for sending me a copy when I lost both the paper and the tweet.

Ignoring everything apart from the results section it has to be said that it does pose some interesting findings from the ROS hypothesis perspective.

Here we go.

They made rhesus monkeys fat using some unspecified "high fat" diet plus some fructose in their drinking water. This generated a model of metabolic syndrome without diabetes. The monkeys had normal glucose tolerance test results but at the cost of marked hyperinsulinaemia.

Next they developed/bought an antibody against highly oxidised LDL.

This significantly reduced the degree of insulin resistance and subsequent withdrawal of the antibody injections allowed a return of the insulin resistance. Like this:

The two upper lines are pre and post treatment and the lower small-dash line is while under treatment with the oxLDL binding antibody. If anyone would like to see how a slim monkey fed standard chow would respond to an IV glucose bolus, it's approximately the blue line which I've extrapolated from the data in panel G of the same figure to give this:

which makes the result slightly less impressive but still very real.

Some of us view the development of insulin resistance as an adaptive and an absolutely correct response to calories being supplied to a cell (or population of cells) without the mediation of insulin, with the result that insulin facilitated calories need to be rejected on a pro rata basis. This is the fundamental message from this paper:

Insulin resistance is a cellular antioxidant defense mechanism

which sets out that case and also integrates ROS as the oxidant stress signal which mediates it. Pure Protons.

So the obvious explanation for the improvement from scavenging highly oxidised LDL particles from the blood stream is that this, mechanistically, *must* be reducing the delivery of lipid to somatic cells, so reducing the necessity to resist insulin's signal. Which could be simply explained if oxLDL itself was a lipolytic agent in much the same way as fructose is lipolytic

Metabolic fate of fructose in human adipocytes: a targeted 13C tracer fate association study

and 4-HNE is lipolytic

Prevention of 4-hydroxynonenal-induced lipolytic activation by carnosic acid is related to the induction of glutathione S-transferase in 3T3-L1 adipocytes

Aside: while I would in no way suggest people prioritise oxidised frying oil (group HO) as a source of calories, it's one hell of a weight loss intervention:

A high oxidised frying oil content diet is less adipogenic, but induces glucose intolerance in rodents

The insulin sensitivity of these rats and mice is worth discussing separately but IMNSVHO they are exquisitely insulin sensitive but cannot secrete adequate insulin. Not good but absolutely NOT obesogenic. An insulin tolerance test would be fascinating and probably fatal. End aside.

So I spent some time down this lipolysis rabbit hole on PubMed and I might even be correct, there is certainly a hint that oxLDL is an ROS generator in an "adipocyte-like" cell model, bearing in mind you can show just about whatever you like in 3T3-L1 adipocyte models (caveat, I've not read the full paper):

Oxidized LDL and lysophosphatidylcholine stimulate plasminogen activator inhibitor-1 expression through reactive oxygen species generation and ERK1/2 activation in 3T3-L1 adipocytes

so maybe it's that simple, but probably not. More likely is that oxLDL delivers 4-HNE to adipocytes and the ROS developed in response to 4-HNE triggers the lipolysis cited above.

Here's what the Rhesus monkey researchers think is going on based on their very clever experiments:

There are essentially two processes going on in this macrophage. Everything down the left side of the image is pro-inflammatory and, if these are detected, the macrophage activates NF-kB mediated inflammatory cytokine release and associated insulin resistance, shown on the right. Down the (right hand brach of the) central section is the oxLDL/antibody complex activating a negative feed back mechanism to inhibit NF-kB activation and subsequent cytokine mediated insulin resistance.

This looks quite simple. In the presence of an acute infection cells of the white blood lines (mostly macrophages and polymorphonucleocytes) will attack the invader using a respiratory burst of extracellular ROS from NADPH oxidase and produce a localised soup of dead organisms (lipoplysaccharide), damaged local lipids (oxLDL) and secreted long distance danger signals (ILs) to get the body in to peak defence mode. All activate the innate immune system. You can't afford to wait around for the 10 days it takes for the adaptive immune system to generate the antibodies needed to take over.

Indeed, once the cavalry arrive in the form of effective antibodies, it is actually time to turn down or off the innate immune response. So, while oxLDL floating around as a free entity indicates the need to get in to fight mode, when oxLDL is bound to an antibody this complex signals that the adaptive (antibody based) immune system, finally, has the situation under control. Innate immunity, ie inflammatory cytokines and insulin resistance, is no longer needed. There is negative feedback and NF-kB calms down.

Which is all well and good.

It just leaves open the question as to why an acute inflammatory response should generate insulin resistance. The trite answer that failure of an organism to develop insulin resistance during a severe infection could be serious. What happened to those organisms which, in the past, didn't do this? As Holly once said 

Why insulin resistance is essential to survival during an inflammatory process is a whole other question. Equally interesting, and related, is why corticosteroids, the most potent anti-inflammatory agents we have, also induce (equally essential) insulin resistance. Mediated by ROS.

It all goes back to Hoehn's paper

Insulin resistance is a cellular antioxidant defense mechanism

and the Protons view of fatty acid oxidation automatically generating ROS.

Nothing makes sense without it.

Peter

Insulin resistance (01) Dr Erol in Turkey

This paper

Insulin resistance is a cellular antioxidant defense mechanism

presents one of the most fundamental concepts necessary to understand what insulin resistance actually means. I may have mentioned this (many times) before.

If you read nothing else from this paper, the discussion is essential. It's short. Without understanding this concept there is no hope of understanding what insulin resistance really is and you are left with bizarre concept of trying to "cure" it. As the authors say:

"In summary, the fact that mitochondrial O2•− is upstream of IR is of major significance suggesting that IR may be part of the antioxidant defense mechanism to protect cells from further oxidative damage. Thus, IR may be viewed as an appropriate response to increased nutrient accumulation as originally suggested by Unger (32), representing part of the cells attempt to return to an energy neutral situation. This concept potentially changes our thinking concerning therapeutic modes of treating metabolic disease."

Aside: the plural possessive of cells is correctly cells'. I merely reproduce the error in the quote, along side my own many transgressions. Grind your teeth at will. End aside.

Especially:

"representing part of the cells attempt to return to an energy neutral situation"

Insulin resistance is adaptive. Almost always.

The paper was written in 2009.

The multiple authors represent a large, two-university group, with the to ability to finance lovely, detailed laboratory work. Where did so many people get this unlikely idea that insulin resistance is an adaptive defense against oxidative stress?

Well it certainly wasn't from me, back in 2009. Nowadays, maybe.

Could it have been from Adnan Erol?

Who?

He's the chap who wrote this paper:

Insulin resistance is an evolutionarily conserved physiological mechanism at the cellular level for protection against increased oxidative stress

in Bioessays, published in 2007, two years before the above paper.

It's hard to see who might have precedence here. The amount of lab work in the 2009 paper suggests that that work may have been ongoing before Erol's 2007 paper was published, but it's hard to say. I cannot find anywhere that Erol's work is cited in the 2009 paper, but that doesn't mean they haven't read it.

So what else has Erol published? It turns out that Erol is quite a common surname in Turkey. There are 270-ish hits if you just click on the author link in Erol's PubMed abstract. These include many different christian names, or perhaps I should say "given" names, in view of the part of the world from which Dr Erol hails.

Searching on "Adnan Erol" gives 27 hits. All appear to be the same author and all, at a glance, look deeply insightful. They are all single author papers.

He thinks for himself.

I get the impression that Adnan Erol is a medically educated/qualified chap in Mansia, Turkey, with a keyboard and access to PubMed. In possession of enquiring mind.

I empathise with him.

I also disagree with him in places. His concept that insulin resistance at the whole organism level is maladaptive is one such. Also, because he (understandably) lacks the Protons hypothesis of obesity, he has to fall back on the old thrifty genome concept to explain obesity through "over eating".

But his overall basic concept, yes. Ten out of ten.

Peter