It's an interesting paradox to think about. As occasionally happens I found a non related paper which gives some suggestion of the mechanism. Here it is:
Characterizing the effects of saturated fatty acids on insulin signaling and ceramide and diacylglycerol accumulation in 3T3-L1 adipocytes and C2C12 myotubes
The paper uses adipocyte-like 3T3-L1 cells or muscle-like C2C12 myotubes. The 3T3-L1 cells might be worth another post in future, today is about the myotubes.
They looked at many things, but most interesting are the data on the ability of insulin to promote the storage of glucose in glycogen granules. I, being me, would look at this as a surrogate for un-measured lipid storage as lipid droplets in muscle cells. One of the cardinal rules of ectopic lipid deposition research is to never, ever suggest that myocyte lipid droplets might be enlarged by insulin signalling. They will be. Just like glycogen granules are enlarged by insulin.
So, practicalities. The myotubes were prepared in "low glucose" DMEM which is probably code for 5mmol/l, ie a normal physiological glucose concentration. To this medium was added a fatty acid at 0.75mmol/l, ie moderate fasting levels. Cells were incubated for 16 hours and then treated with supra-maximal insulin.
They noted that elevated pure palmitate is utterly harmless to cells in culture with normal glucose levels:
"Under these conditions, no signs of cell death were observed"
which certainly is not the case using 25mmol/l of glucose!
Here are the gels they obtained:
The top row, highlighted in red, is demonstrating the presence of phosphorylated (activated) Akt, a core insulin signalling step. We are comparing P-Akt under insulin and 5mmol/l glucose with that under insulin and 5mmol/l glucose plus 0.75mmol/l palmitate. It's clear that insulin signalling is markedly blunted by palmitate. If we look at the dashed oval we can see that the P-Akt band under oleate is comparable to that of the control.
The same applies to the level of phosphorylated glycogen synthase kinase 3 beta, a key enzyme in activating glycogen formation and outlined in blue. P-MAPK is irrelevant today but, again, might be interesting in the future.
Convincingly, palmitate causes insulin resistance and oleate doesn't. Bear in mind that this is a highly constrained experiment to make a specific point. These is no mention of ROS generation but it is worth looking at this from the Protons/ROS viewpoint.
Palmitate has an FADH2:NADH ratio of 0.484
Oleate has an F:N ratio of 0.457
In this experiment conditions are carefully controlled to give us an all or nothing response, almost like switch. The switch trips somewhere between an F:N of 0.484 and 0.457. Insulin resistant vs insulin sensitive. No double bonds vs one double bond.
It is possible to adjust the F:N ratio by smaller amounts than by adding a double bond simply by altering the length of a saturated fat. If we consider myristic acid the F:N ratio is 0.482 and for lauric acid it is 0.478.
The paper went on to look at insulin signalling using these fats and here are the gels they obtained:
At the extreme left is the control without insulin, next is control with supra-maximal insulin but no fatty acid, then rest of the bands have the indicated fatty acids added, all at 0.75mmol/l. I've put in the red line to divide insulin sensitive from insulin resistant results. Everything to the left of the line causes no insulin resistance, to the right significant insulin resistance. The switch is between myritic acid and palmitic acid, F:N 0.482 and 0.484.
I would expect C10 capric acid to be insulin signal facilitating and probably C8 caprylic acid too, although its strongly ketogenic effect and partial conversion to palmitate might make that less predictable.
Coconut oil is primarily medium chain triglycerides with F:N ratios on the insulin sensitising side of the switch. Formulating a high fat diet out of insulin sensitising fats combined with an insulogenic carbohydrate load seems like a good recipe for obesity.
It's always worth reiterating that there is no "switch" as such, there is a general integration of information about energy status and demand using ROS which can be pushed towards lipid storage or use depending on particular inputs. The F:N ratio looks like a switch in a very simple model asking a very simple yes-no question. But that's still useful information for understanding the Surwit diet.