Monday, September 21, 2020

Protons (64) The miracle of fish oil (6)

Preamble: I started this current series of post about the ability of fatty acids with multiple double bonds to limit weight gain. To me, this is a paradox. Paradoxes are, without a doubt, the most productive sources for the development of an idea. Even as I started this current post I had no idea where it was going to end up and was bit surprised at where the metabolism took me. So be it. Let's begin.

Beta oxidation in peroxisomes does not consume oxygen and does not produce CO2.

The first step of oxidation of saturated fats runs like this:


In peroxisomes this is followed by

FADH2 + O2 -> FAD + H2O2


2xH2O2 -> Signalling -> Catalase -> 2xH2O + O2

The energy from FADH2 is released as heat and the oxygen is regenerated.

The NADH from beta oxidation is of no immediate use in a peroxisome and has to be transferred to mitochondria before it can be utilised. I suppose it could be phosphorylated to NADPH for anabolism but I have no data on that. It's not clear how reducing equivalents might be transferred from peroxisomes to mitochondria. There is speculation about something along the lines of the malate-aspartate shuttle used to import cytoplasmic NADH in to mitochondria.

It's also something of a truism that peroxisomes cease beta oxidation at C8 and then export this (by uncertain mechanism) to mitochondria for completion of oxidation. Digging back through the reference trail leads to the origin of this as the finding that isolated peroxisome preparations happily oxidise lauric acid but won't oxidise caprylic acid (much). Clearly oxidising DHA will never produce caprylic acid directly because there are double bonds within the residual eight carbon atoms. What exactly happens to truncated DHA at the C8 length appears to be an unasked question.

And there will be no CO2 production in peroxisomes as they do not have the TCA, that's limited to mitochondria.

So beta oxidation in peroxisomes produces heat, NADH, acetyl-CoA and signalling H2O2. And perhaps some caprylic acid from any saturated fatty acids being oxidised. It requires no oxygen consumption and results in no CO2 production, which gives an unchanged respiratory exchange ratio (RER).

Going back to

The round symbols are the fish oil fed groups. Average VO2 through 24h is reduced by fish oil from about 3500ml/kg/h to about 3000ml/kg/h, ie that's a just under 15% reduction.

Here are the RER figures, still fish oil as circles. As expected high fat diets show a low RER, low fat diets show the converse. The reduced O2 consumption is exactly balanced by a reduced CO2 production and the RER is still largely set by the dietary carbohydrate-fat ratio.

Clearly, under fish oil, approximately 15% of calories are being used to generate heat and anabolic substrate without consuming oxygen or being transferred to the ETC.  Provided there is enough fish oil to stimulate peroxisomal proliferation the changes are quantitively independent of the absolute amount of fish oil. 

So with fish oil at as low as 10% of calories, not all of which are PUFA, VO2 is dropped by 15% suggesting that the peroxisomes are activated and are oxidising more fatty acids than just the PUFA from the diet. Presumably on the low fat fish oil diet the peroxisomes are also metabolising palmitate and oleate derived from carbohydrate by de novo lipogenesis too.

If we go to this paper:

Peroxisomal and Mitochondrial Oxidation of Fatty Acids in the Heart, Assessed from the 13C Labeling of Malonyl-CoA and the Acetyl Moiety of Citrate

we can see, by clever carbon 13 labelling, that peroxisomal derived acetyl-CoA in cardiac muscle (and I would guess most other extra-hepatic sites) does not enter mitochondria, it all stays in the cytoplasm as malonyl-CoA.

These data are from perfusing hearts with docosanoate, a C24, fully saturated, fully peroxisome targeted fatty acid. We get lots of labelled malonyl-CoA in the cytoplasm, minimal labelled citrate in the mitochondria.

The next fascinating paper (HT to Peter Schmitt for the link) used erucic acid, another peroxisome targeted fatty acid.

Peroxisomal oxidation of erucic acid suppresses mitochondrial fatty acid oxidation by stimulating malonyl-CoA formation in the rat liver

In the liver peroxisomal oxidation of fatty acids generates acetate but that this is still converted to acetyl-CoA and then malonyl-CoA without entering mitochondria. We know from the Randle cycle that malonyl-CoA is an inhibitor of fatty acid oxidation so it should come as no surprise that erucic acid feeding to peroxisomes inhibits fatty acid oxidation in mitochondria. So we end up with lipid accumulation within the liver, progressing to fatty liver and NASH. I have mention before that in rodent models of alcoholic fatty liver disease fish oil is one of the most effective generators of alcohol induced liver damage...

But perhaps the best line from this last paper is:

"Peroxisomal metabolism of erucic acid also remarkably increased the cytosolic NADH/NAD+ ratio..."

It seems very, very unlikely that fish oil will be any different.

We find ourselves in a situation where peroxisomal oxidation of fatty acids generates benign heat combined with large amounts of anabolic substrate and a high NADH:NAD+ ratio without requiring oxygen while simultaneously inhibiting mitochondrial fatty acid oxidation while shifting metabolism to glucose.

Does that look like a recipe for cancer?

It does to me.

I had no idea that there is a large literature looking at the role of peroxisomes in all sorts of cancer types. Woohoo, they are a drug target! Perhaps avoiding peroxisome activating fatty acids and their derivatives might be a better approach. Apart from accepted Bad Things like drinking erucic acid or 4-HNE (a superb peroxisome activator) we might ask serious questions about drinking bulk fish oil.

Perhaps don't.


Addendum: I recall this study (observational but not a food frequency questionnaire in sight), which I was fairly uncertain about back in 2013

Plasma phospholipid fatty acids and prostate cancer risk in the SELECT trial

Now I'm more convinced...

Sunday, September 20, 2020

Protons (63) 4-hydroxy-2-nonenal (2)

Brief aside for a one-liner-ish post.

This study gives an idea of what happens when you drink 4-HNE:

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

Here are the diets. The rats on heated soybean oil (HO) ate relatively little so all of the other groups were partially starved to the caloric intake of the HO group.

and here are the weight gains (green circles).

Note especially how thin the 4-HNE fed rats (blue rectangle) were and how the fish oil fed rats (red rectangle), even under caloric restriction were still the fattest, fatter even than the fresh soybean oil fed rats. On the same calories.

This diet had no sucrose, it was starch based. Like the rodent chow in the last 4-HNE post which gave the greatest weight gain on DHA or EPA. Starch diets seem worse than sucrose diets when mixed with fish oil...

Back to 4-HNE. These rats were mildly glucose intolerant too. Glucose was ns different from controls throughout an OGTT but the area under the curve was greater under 4-HNE.

In this case we have, I would speculate, 4-HNE causing insulin resistance within adipocytes, limiting fat storage (ie reducing loss into adipocytes), so limiting hunger by actually reducing fat gain. 

And doing god only knows what other damage along the way to these un-arguably thin rats. As the authors suggest, the HO diet could even be destroying the beta cells of the pancreas...


Addendum HT to raphi for this link in the comments. Well, that's cool

Thursday, September 17, 2020

Protons (62) The miracle of fish oil (5)

Okay, its time to look at this paper:

We have four diets, two based around low fat (10% fish oil or 10% lard), each with a little soybean oil thrown in. The other two were 60% of calories from fat, either from fish oil or lard, both with generous soybean oil added. Fatty acid composition was measured by gas chromatography. I've replaced the percentage of lipid as linoleic acid with percent of total calories as LA written in red.

Notice the study altered protein levels in addition to switching starch for fat. Talk about failing to control your variables.

Then we just have to feed ad-lib for eight weeks and look at the weights. In fact we can actually look at the fat mass in addition to weight, which is much nicer. We can also look at the energy efficiency, weight gain per unit calories absorbed. Note the columns have changed order, I've again added the linoleate percentages of calories in red, highlighted the energy efficiency in blue and circled the results of interest in green:

I did a rough back-of-the-envelope calculation for the number of mg/kg/d of DHA consumed by the mice fed the 10% fish oil diet. It works out as around 1250mg/kg/d on a semi-purified diet background, so probably in the peroxisome activating level.

Obviously the amount of fat stored roughly follows the LA content of the diet outside the green circle anomaly.

Notice that the "average energy absorbed" is lowest in the fattest group of mice. These animals are not making fat out of nothing. To understand this you have to go back to this image from a long time ago:

The top line is HFD (D12492, fed to Long Evans rats). All of the "hyperphagia" needed due to rapid weight gain occurs during the first 10 days and is only statistically elevated during the first 6 days. If the averaged food intake of the mice in the current study is lower than the brief "hyperphagic" phase this would explain the low overall calorie intake. The effect might be exacerbated in part due to the strain of mouse used, in this case the Swiss mouse, which is not prone to obesity in the way that many rodent strains are.

I started to look at these mice in terms of energy budgets. The mice are fed ad lib so are all going to eat exactly as much food as they need. No more, no less.

The high lard, high LA fed mice need 77.85kJ/d to meet basal metabolic rate, thermogenesis, cage exploration and this will also include a modest loss of calories in to their already distended adipose stores. They can do all of this on 77.85kJ/d. This is what they need.

The high fish oil fed mice should need very slightly less than 77.85kJ/d because they are lighter, so have less tissue to support metabolically, ie have minutely lower BMR and, again being lighter, it takes less calories to move themselves around their cage. They are also barely losing any fat in to their adipocytes. Despite all of these small combined contributions to decreased caloric needs they still have to absorb more calories (83.42kJ/d) than the fat mice to support a significantly lower weight and fat mass.

Clearly they are losing some of those extra calories but in this case the calories end up in peroxisomes rather than in adipocytes. I would predict that these mice will have a normal body temperature (this is tightly controlled in awake mice) but they will be physiologically adapted to dissipate excess waste heat. At 20degC this is easy for a mouse, it normally spends more calories on thermogenesis than it does on BMR at this temperature.

So the thin mice are unable to meet their basic caloric needs without having to "over-consume" food to make up the deficit induced by heat generation. In peroxisomes. They are not overeating and then using heat generation as a technique to stay slim. They are eating enough food to meet their needs but the lipid sources in their food are intrinsically and wastefully heat generating.

Just as the fat mice are fat because they lose calories in to adipocytes due to linoleic acid, so the skinny mice are skinny because they lose heat calories through their skin due to DHA in peroxisomes.

Ad-lib fed mice never over or under consume calories. They eat to meet their needs. Exactly.


Note that the mice fed 10% of calories fish oil "over-consume" to total 92.51kJ/d. They need more daily calories than the 60% fish oil fed mice because they are losing some calories in to adipocytes. That poses some more questions. As does the low oxygen consumption in the fish oil fed mice. An interesting paper. More to think about.

Tuesday, September 15, 2020

Protons (61) The miracle of fish oil (4)

There is a huge body of work on the requirement for DHA, its lipid peroxides, its use in the body but almost nothing about its bulk disposal. As in what happens to the excess DHA when you drink 60% of your total daily calories as fish oil, for weeks. As a rat.

I've been chasing tenuous leads as to whether DHA is catabolised in peroxisomes, in mitochondria or in both. I'd like a nice clear cut answer, but you can't always have what you want. It's clear that DHA can only be synthesised in peroxisomes because it requires elongation from ALA to eventually form a 24 carbon PUFA which is then shortened by beta oxidation to the C22 DHA. Only peroxisomes appear to deal with the beta oxidation of C24 fatty acids. For C22 and especially C20 its not quite so clear cut.

On that basis I'm willing to go with peroxisomes as the main site of DHA catabolism, grudgingly and without hard data. Peroxisomal degradation is particularly difficult to justify from the simple FADH2:NADH ratio because DHA has so many double bonds that it's not going to drive reverse electron transfer through complex I. But it might be too fattening of course...

While searching I came across this study:

Enhanced Peroxisomal beta-Oxidation Is Associated with Prevention of Obesity and Glucose Intolerance by Fish Oil-Enriched Diets

which provides a number of points which need discussion, but for today I'm looking at following the reference trail back through DHA and peroxisomes.

The trail is good at level one backwards, with a nice paper on reagent grade DHA gavaged in to rats, but beyond that it drifts off into partially hydrogenated fish oil (goodness only knows what that contains but it undoubtedly induces peroxisome proliferation) and beyond that in to very long chain mono unsaturated fatty acids which do the same thing but neither helps me with DHA/EPA catabolism.

So I'll just start with the DHA gavage paper today

Docosahexaenoic acid shows no triglyceride-lowering effects but increases the peroxisomal fatty acid oxidation in liver of rats

The biggest problem with it is that for a lot of the work they were using group sizes of three rats. You don't do stats with n=3 group sizes, so I see it more of a proof of concept paper.

Rats were ad-lib fed semisynthetic diets (experiment I) +/- added cholesterol (experiment II). They were also gavaged with 500mg/kg, 1000mg/kg or 1500mg/kg of pure DHA daily for 10 days. Controls got nothing or palmitate 1500mg/kg/d. I'll come back to experiment III later.

Control rats grew at normal rat growth rate. DHA at 500mg/kg increased weight gain over the 10 days, 1500mg/kg did not, giving a comparable growth rate to controls. Using 1000mg/kg/d varied in effect but that's probably due to n=3 group sizes. Palmitate at 1500mg/kg/d is not obesogenic (well, whodathunkit?).

These are the numbers, relevant weight gains outlined in red:

Experiment III is even more interesting. Here they fed the rats ad-lib on a then-current 1993 style fairly high quality crapinabag, possibly something a bit like 5001. They gavaged EPA as well as DHA and had palmitate as control, all at 1000mg/kg/d, there was no untreated control group. This time we have n=5 rats. From the blue square palmitate gave 33g weight gain, DHA 52g and EPA 58g over 10 days. I particularly like this as DHA might be peroxisomaly directed but EPA, being shorter, less so. I get the impression this is not "all or nothing". DHA looks as if it might simply go through mitochondria if there is just a little of it around. If there is a lot it around it induces peroxisome proliferation and peroxisomal beta oxidation. Putting double bonds through mitochondria should produce fat gain, through peroxisomes less so. If you have my biases. Of course we have no idea re fat gain vs muscle gain in these rats.

The thing which struck me is how neatly you can control weight gain by choice of dose of DHA and by choice of background diet. I like it. Rats are so like people.

It's also interesting to look at Table 5

The column of interest here is outlined in red again. This is the ability to oxidise palmitoyl-CoA in the presence of potassium cyanide. Because KCN completely blocks the respiratory chain at complex IV any oxidation of palmitate in its presence is exclusively within peroxisomes. Increasing doses of DHA increase peroxisomal palmitate oxidation. Given high enough DHA ingestion peroxisomal activation appears able to over ride the weight gain effect of low dose DHA.

Summary: Low dose DHA causes increased weight gain in growing rats at 500mg/kg/d. At 1500mg/kg/d it doesn't, almost certainly through peroxisomal activation.

Adding DHA or EPA to a particularly healthy low fat/high complex carbohydrate diet might make the weight gain worse. In a rat.

Does this mean anything for humans?

Perhaps if you are going to take fish oil, take lots. Or, better still, none at all.


Oh, and, as far as I can see, no one has ever taken radio-labelled DHA and fed it to isolated peroxisomes or isolated mitochondria and looked at labelled metabolite or CO2 production. The test fatty acid has always been palmitate. Which is odd.

Wednesday, September 09, 2020

Protons (60) 4-hydroxy-2-nonenal

I've been re-reading Dave Speijer's

Being right on Q: shaping eukaryotic evolution

I cannot over emphasise how both broad and detailed this work is. This current post came from following a single link in the section on uncoupling.

Back in 2000 people were bulk manufacturing human uncoupling proteins using E. coli and assembling them within the membranes of synthetic lipid vesicles. They had problems getting the UCP1 to function correctly but eventually, by dint of an enormous amount of hard work, they found this requirement (the title says it all):

Coenzyme Q is an obligatory cofactor for uncoupling protein function

The UCP1 derived from E.coli could be activated by the addition of coenzyme Q, more specifically in its oxidised form CoQ. This is slightly counterintuitive as you might expect CoQH2 to be more of a signal that "excess" electrons were present in the ETC and that uncoupling to reduce the mitochondrial proton gradient might be a good idea.


The next snippet was provided by Brand's group

Superoxide activates mitochondrial uncoupling proteins

who used the oxidation of xanthine by xanthine oxidase to generate superoxide in-situ, to demonstrate that superoxide was, or could generate, the necessary co factor to allow UCP3 (in this case) to function. This is much more understandable because excess input to the ETC in the absence of a need for ATP is the classical situation for ROS generation and so ROS are more plausible as a signal to institute uncoupling compared to the oxidised version of CoQ.

Then comes this paper, again from Brand et al (which is the one I picked up from Dr Speijer's work):

Synergy of fatty acid and reactive alkenal activation of proton conductance through uncoupling protein 1 in mitochondria

The evil molecule 4-hydroxy-2-nonenal (4-HNE) is synergistic with fatty acids in activating UCP1. Physiological uncoupling is generally thought of as a Good Thing. 4-HNE as a Bad Thing. Perhaps we should be careful about making value judgements about molecules.

From an evolutionary perspective there is no obvious reason (to me) why UCPs might not be activated directly by superoxide itself but in this case the preferred solution appears to have been to allow superoxide to modify linoleic acid within/around the mitochondrial inner membrane into 4-HNE, which can then act as a cofactor to UCP1 to synergise, in this experiment, with free palmitic acid to dissipate the membrane potential and so to limit excess ROS production.

So UCPs in general appear to respond to an inappropriately high level of ROS generation by activating the safety valve of uncoupling the mitochondrial membrane potential. Linoleic acid derived 4-HNE is key to this process.

I have argued that the normal mechanism for limiting calorie ingress into a replete cell is for ROS to disable insulin signalling. And that PUFA fail to generate the appropriate ROS needed because they fail to deliver an appropriate supply of FADH2 to ETFdh and subsequent reduction of the CoQ couple. So PUFA allow an excessive, poorly controlled calorie supply. Eventually enough energy will be supplied that an excess of ATP combined with a paucity of ADP limits the activity of complex V, so membrane voltage will finally rise, the flow of electrons will back up and lots of ROS will finally be generated. At this stage there is still too much input, too little demand and a problem looking for a solution.

Uncoupling is one solution. Electrons can be allowed to continue to pass down the ETC and to pump protons but these protons are allowed back through the UCP, generating heat rather than ATP and reducing the membrane potential. Which will limit the ROS generation which might otherwise become too high.

Now, if you accept that PUFA are the cause of the situation and that uncoupling is the solution, which fatty acids would you expect to be the best activators of UCPs when uncoupling proves to be needed?

Correct. PUFA are the most effective protonophores when used by UCPs to reduce the inner mitochondrial membrane potential. As in:

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

Aside: It's worth reading the methods section of this paper. It gives insight in to a) how phenomenally difficult it is to set up models to look at individual protein functions in isolation and b) how far from physiological such models are. Difficult, extreme, necessary. But interpret with caution. And think about any requirement for 4-HNE. End aside.

Let's go up a level from the ETC to the cell plasma membrane and insulin signalling. If you are a cell and you are swamped with incoming calories but can only signal using ROS by the time that ongoing incoming calories are continuously too high, what other strategies might you apply?

How about augmenting the PUFA-inadequate insulin resistance by using 4-HNE to generate a few of the necessary extra ROS? As in:

The lipid peroxidation by-product 4-hydroxy-2-nonenal (4-HNE) induces insulin resistance in skeletal muscle through both carbonyl and oxidative stress

Additional cellular insulin resistance, supplied by 4-HNE, is a logical solution to a situation where insulin resistance is needed but is not happening appropriately. The role of 4-HNE can be viewed as being protective by uncoupling at the level of the mitochondrial membrane and also protective by augmenting insulin resistance at the cell surface membrane.

And to re-iterate again: Insulin resistance in adipocytes is synonymous with decreased fat storage and/or increased lipolysis.

I think it is very reasonable to assume that our physiology knows all about PUFA and how to deal with them. The end result may not always be what we want, but it will be adaptive. I think the context in which we are exposed to them is very important, especially the level of insulin, the rate of beta oxidation (which beaks down 4-HNE and related molecules) and the total quantity of linoleic acid in the diet. Getting 1% from mammoth fat is perfectly oaky. Getting much more on a ketogenic diet can be dealt with. Margarine on your baked potato might be a no-no.

I also think that bulk ingesting aged corn oil from a deep fat fryer might not provide a particularly physiological supply of 4-HNE.

But clearly, given the correct experimental set up, we can arrange that diets based around safflower oil can be less obesogenic than those based around lard, despite the very much higher linoleic acid content of the safflower oil. It provides a tool to understand papers like this one:

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

(HT to Amber O'Hearn for resurfacing the paper which has been on my "think about it" list for a long time)

which uses these diets

How the authors describe the diets is unimportant, all that matters is the PUFA content. Here are the weight graphs:

The minimum weight gains are the LF_PO at 1% PUFA (low total fat), over lain by the HF_CB (high total fat) but just over 1% of calories as PUFA. HF_PO is worst due to 4.5% of calories as PUFA. HF_OO diet is almost as bad with the same PUFA percentage.

Yet another aside: I would never argue that there is no influence of the lipid species available to be incorporated in to adipocyte triglycerides. All-palmitate would turn adipose to candle wax, all-linoleic acid into a liquid. So there are decisions made re storage vs oxidation taken at several layers above the ETC with are not unimportant but are not my forte. End aside.

But the safflower oil based diet, despite over 35% of calories as PUFA, is almost as weight gain limiting as the two low PUFA diets.

If you wanted to explain findings like this you would need to look at the level of heat generation, the level of 4-HNE production, the rate of oxygen consumption and possibly the level of insulin signalling in the post prandial period. But there are mechanisms to support a possible explanation.

Is linoleic acid a potential adjunct to weight loss? Mostly "no" is the short answer. But it appears to depend on how carefully you set up your study and what result you would like to get. Possibly how long you run the study for. Not that there any biases involved. It might also rather depend on how close you want to get to eating F3666 high PUFA ketogenic rodent food. And how many double bonds you might be willing to accept into your inner mitochondrial membrane lipids.

Personally, no thanks.



In the first paper CoQ probably works by generating the 4-HNE needed by UCP1 while CoQH2 doesn't. I'd speculate that because CoQ is an electron acceptor, which normally accepts electrons from the terminal FeS cluster of complex I, it might be looking to accept electrons from other sources in a lipid bilayer preparation. In the synthetic lipid by bilayer there are molecules of linoleic acid. Under conditions of available oxygen I see no reason why CoQ might not accept/steal a pair of electrons from a double bond in linoleic acid which would leave behind a reactive lipid radical which is a good candidate for combining with oxygen and eventually forming the 4-HNE needed by UCP1 to work efficiently. Just a guess.