Protons: The pancreas

We've seen the concept of superoxide being used to produce insulin resistance as a means of limiting (glucose derived) energy input in to cells which really don't want it. Superoxide appears to be the primary marker of energy excess at the cellular level.

We know from isolated mitochondrial preparations that superoxide is physiologically produced by reverse electron transport through complex I and is driven, gently, by succinic acid alone working through complex II. Far more is produced when the NADH level is high as well as having a reduced CoQ couple through FADH2 input, be that from complex II or from fatty acid oxidation products. Macroscopically fat and glucose together should produce enough superoxide to show as cellular insulin resistance, rejecting glucose from the cell, while allowing continued fatty acid oxidation. That's simple and logical.

But if you are building an energy sensor, it would be a bit dumb to restrict access to the very energy molecules which you are trying to look at to judge overall energy status, especially when energy status is high: You need to decide when to store calories...

The beta cells appear to use both fatty acids and glucose to generate superoxide, but instead of signaling beta cell insulin resistance, they signal insulin secretion. Several lines of evidence fit in with this.

You can get succinic acid itself directly in to beta cells by providing it as a methyl or ethyl ester. As a metabolic fuel source this acts as a near pure complex II substrate, pushing electrons in to the ETC through the FADH2 of succinate dehydrogenase to reduce the CoQ couple and set the scene for reverse electron transport and superoxide production, especially when NADH from glucose metabolism rises. In a commonly used model of functional beta cells, succinic acid methyl ester is a marked insulin secretion potentiator, especially at higher glucose concentrations. Glucose supplies NADH, succinate supplies FADH2, they clash at the CoQ couple and the generation of superoxide signals that there is a ton of energy available. Better store it. Better secrete insulin.

Succinic acid methyl ester drives complex II. This drives insulin secretion in response to glucose. But it's a drug. There is nothing physiological about this drug. So shall we go a little more physiological?

To recap from previous posts: Superoxide generation is directly proportional to the ratio of FADH2 generated to the amount of NADH generated for a given substrate, the F:N ratio.

Here's a nice graph of insulin secretion stimulated in response to 12.5mmol glucose on a background of assorted free fatty acids from an isolated pancreas preparation:



If you can't be bothered to work out the F:N ratios (shame on you), here they are added to the graph:



Please excuse the C8 value; as we all know, MCTs are shunted directly to the liver via the portal vein. They do not seem to feature too prominently in pancreatic superoxide generation and insulin secretion. It would take a ton of reading to see why and how they are handled differently to longer chain fatty acids. For the time being let's stay looking at C16 and longer as these make a much tidier story...

So, for the four longer fatty acids tested, the amount of insulin secreted is remarkably closely associated with the F:N ratio of the fatty acid available.

Does this work in people?

Of course it does. Remember the Spanish study? I lo0ked at it in some detail here.

In particular look at this graph:



From the top downwards we have butter, high palmitic acid seed oil, refined olive oil and a mix of fish and vegetable oils as the white triangles. It is very clear that the insulin secretion here is in direct proportion to the saturation and length of the fatty acids in the meal, in an intact group of volunteers..

Aside: Obviously, there is a glaring error in the graph. All of the curves except the control use 800kcal of total food, of which 40g is carbohydrate/protein and the bulk is fat. The graph is missing a group where 800kcal was supplied as pure carbohydrate. We can all imagine where this much bulk glucose would have put the insulin curve, needless to say there is absolutely no way it would fit on to the presented graph. We would need a much taller vertical axis, which would show the mixed meals in their true context!

But the principle, that insulin secretion at a given level of glucose is elevated in direct proportion to the F:N ratio of the background fat, holds perfectly well in this carefully contrived human study.

BTW, lucky for me they didn't include a coconut oil group!

The obvious conclusion from this finding is that to lower insulin maximally we should, taking as given that replacing carbohydrate with fat is the biggest step by far, all go for vegetable oil with some fish oil. Not butter. But in the original post on the Spanish paper I went on to discuss what appeared to be happening to the lipid from the meal. It could stay in the bloodstream and be used for metabolism, as the butter did, or it could be cleared rapidly in to adipocytes allowing metabolism to return to being glucose based. At the cost of expanding the adipocyte stores of fat.

The high PUFA meal really was rapidly stored as fat in adipocytes. The F:N based explanation is because we are supplying a low F:N ratio fat and so not generating insulin resistance phyiologically; we are allowing lipid easily in to adipocytes because the lipid does not generate adipocyte insulin resistance. We are going back to glucose metabolism as rapidly as possible. PUFA facillitates fat storage and glucose based metabolism. All is fine until you can't get any fatter. Butter limits fat storage and runs metabolism of palmitic and stearic acids. Those high PUFA-fed mice generate obesity when fed their high PUFA diets from pre-conception onwards:



In the butter group there is some excess insulin. Does this matter if no one (cellularly) is listening to it?

I next want to look at the flip side, the reduction of supply of free fatty acids to the pancreas. This was done in the same paper. You can certainly do this in intact rats (and humans if you so wish). Then we might get back to the fat mice.

I think that had better be another post as this one is getting overly long and it's light enough to let the chickens out.

Peter