Elevated ATG13 in serum of pwME stimulates oxidative stress response in microglial cells , 2022, Gottschalk et al

from Dynamic association of the ULK1 complex with omegasomes during autophagy induction (formatting mine)

I still can't find out how atg13 gets from cells into the blood.

I am aware that I'm getting a bit carried away on the basis of barely understood snippets of a preliminary paper. The complexity of how cells work though - it's amazing stuff.

I felt a bit better while on a ketogenic diet. I was losing weight on the diet, so I assume it was triggering autophagy.

Source

When my son (also with ME/CFS) seriously over does things, he sleeps for long periods (e.g. the worst was sleeping 20 hours a day for a month) and then gets back to his baseline. At these times, his already fairly low food intake decreases further. I assume the hypersomnia is protective; perhaps the reduced food intake is as protective as the sleep? Members here have said that they feel better when they don't eat (obviously that can only be taken so far...).

 

Autophagy and calorie restriction is seen in some, not all, animal studies. There's a lot of hype surrounding fasting, sprinkled with the old morality of how much better of a person you are if you can restrict yourself from eating.

Comparing studies on fasting is a nightmare, as protocols differ (alternative day fasting? Eight hour eating window? Four hours? Ten hours? Fasting a set number of days a week? What is allowed in the fasting period? Water? Broth? Food < a set number of calories? Are other changes made to the diet, in addition to the fasting element?). Additionally, many/most studies are weightloss studies and if you have a control group that also loses weight the "magic" behind fasting comes from the weight loss, not something else. A small pet peeve of mine is that fasting studies seldom cares about female hormones. Fasting could potentially mess with the menstrual cycle, but why bother with that, it's not like those hormones do anything important

NTNU in Norway is doing a fasting study on pregnant women right now (10 hour eating window, and participants are encouraged to include high intensity interval training in their daily life), where weight loss is not part of the study design for obvious reasons, will be interesting to see the results in a few years time

Other reasons fasting fails could be that we don't have the metabolic flexibility required The body needs to be able to adapt to the stress of maintaining blood glucose and other functions, that might simply be too much for pwME.

 

Reduced food intake can be protective in that the body then requires less energy for digestion and dealing with potential post-prandial inflammation (possibly higher in pwME, especially after exertion given the increase seen in LPS in the blood), but at the same time it is very easy to become deficient in nutrients and energy, which is the opposite of protective when ill. There are guidelines for how to tweak the diet to be easier to digest (though these could potentially be low on nutrients/compounds not found in food composition tables which might have other benefits such as improving the integrity of the gut lining etc).

 

It's annoying not being able to see the paper. But the preprint linked in the first post of this now-merged thread gives some insights.

On the negative values, there's this:

So, I imagine the x axis is increasing concentrations of the serum, and the y axis is the concentration of the target protein. And so, for some of the subjects, the proteins might only be detectable at a full strength or half strength serum dilution, whereas for others, there might be identifiable amounts at a 1: 16 dilution. So, they fitted curves to the data points for each subject. And then they used the protein concentration at a specific dilution (perhaps 1:4), even though for those with barely detectable levels at full strength that meant using a negative value estimated by the fitted curve.

 

Thanks to those who offered to share the paper with me. There are a lot of potentially interesting findings in this paper.

Finding 1: Strong tendency for protein aggregation in the serum of ME/CFS (compared to controls) (Section 3.2)

This was done using a thioflavin T assay. On the face of it, this result looks pretty compelling as a preliminary result to me.

Figure 1 shows the individual graphs of the first two case-control pairs e.g.

This shows the change in protein aggregation over time. The steeper the slope, the faster the aggregation. And the higher the final result on the y axis (i.e. at 100 minutes), the higher the amount of aggregation.

Figure 1C gives the slopes of the aggregation for a further 7 case-control pairs. It's a remarkable difference.



Questions and comments I have about this:

It's not a specific result, this alone doesn't necessarily tell us much about autophagy related proteins.

I don't know about this assay. Is it reliable? Perhaps it hasn't been used in serum samples before for a reason?

There's a sample size calculation to support the use of only 7 case-control pairs. I think it's a bit, I don't know, 'unreasonably applying precision to unknowable things'? to do a calculation like that. I think with preliminary studies, researchers should usually be using sample sizes of at least 20. It's a quibble, I know.

The paper says 'results were confirmed after 3 independent experiments'. I find that the opposite of reassuring. Where's the data? Which experiment produced the data for the chart shown in Figure 1c?

 

(please correct anything I get wrong)

Finding 2: increased autophagy proteins in ME/CFS serum (section 3.3)

They used a commercially available autophagy antibody array (i.e. an Elisa method where you have antibodies that bind to the target protein). They started with 1 case control pair, and found a number of autophagy proteins in the serum, and confirmed it using different methods. This is where they adopted that dilution series approach to an Elisa analysis - making protein concentration estimates at different dilutions, which they then applied a curve to (with the result of the artefact of negative concentration estimates). Why not just report a 0 if there is no detectable protein at a particular dilution and do more experiments at different dilutions to finesse things, rather than extrapolate? I suppose, maybe the cost of multiple analyses. But yes, as @LarsSG commented above, calculated means should treat negative values as 0.

They also used a 'novel near- infrared-based Elisa method' which they suggest is much more sensitive to low-abundance proteins. So far, this is all just with one case-control pair. Then they extend the analysis to another case-control pair. And then 12 case-control pairs. As far as I can tell from Table 1, these 12 are different to the first two pairs.

Whereas they had found a few autophagy proteins elevated in the first two ME/CFS patients, only ATG13 was significantly elevated in the larger sample. Table 1 with the data we have already seen (the one with the negative values) gives the results, as does Figure 3a.



But, the results in the chart and in the table look different. Table 1 has 12 patients (although one only has SFN, not ME/CFS) and 12 controls. The table reports a value of 45 for a ME/CFS patient and 15.4 for another - the chart's y axis only goes to 15 and all of the dots are well below it. One of the patients has a value of -9 in the table, but all of the patient dots on the chart are above 0.

The result in the chart is a lot less impressive than the results in the table, but still a bit interesting.

The caption for the chart suggests that 11 case control pairs were analysed - but there are only 10 dots for each of the controls and cases on the chart. The caption does suggest that the values are the mean from 3 experiments, so perhaps that explains the difference with the values in Table 1. However, the paper doesn't explain the difference.

I found the paper confusing when it came to understanding what particular method was used to produce particular results- I think the standard Elisa approach was used for the larger sample.

I still have questions about how and why the autophagy proteins which operate intracellularly might end up in serum. I think there was a couple of references provided on that.

 

I'm not on twitter, please do follow up. Yes, it feels a bit like they give us a blow by blow account of everything they tried, and the detailed disease history of the two initial cases who aren't actually included in the larger sample. Which all gets confusing. It might have been better if they had worked out their approach with a couple of case-control pairs, and then just given us the data for the large sample.