WASF3 disrupts mitochondrial respiration and may mediate exercise intolerance in myalgic encephalomyelitis/chronic fatigue syndrome, 2023 Hwang et al

chillier said:
This figure is the bit that really matters for sure. They have taken muscle biopsies from these 10 controls and 14 ME/CFS patients (from the intramural study i think so should be diagnosed well) and done western blots on them for WASF3, MTCO1, COX17, PERK and BiP. Here's the rest of the figure showing the western blots themselves and the quantified bar plots (just quantifying how large the blobs are from the western blot).

View attachment 20151

They see elevated WASF3, reduced ComplexIV proteins MTCO1 and COX17, increased endoplasmic reticulum (ER) stress marker PERK, and reduced protein-folding chaperone BiP. I really don't like when this type of data is represented in bar plots without showing the individual datapoints as a strip style scatter plot. We have no idea how much the groups overlap and they also don't tell us the effect size. Our only indication of variance is the error bar which as you point out represents the standard error which is the wrong metric to use - it should be standard deviation. The standard error of the mean is always smaller than the standard deviation so it looks better too. They do at least show all of the blots for all these proteins for the controls and patients (some hidden in the supplementary) so theoretically we could make these plots properly ourselves.

They go on to see the same set of proteins showing the same phenotype when they chemically induce ER stress in a separate experiment which is potentially interesting.

Their overall model for what is happening is that ER stress leads to an increase in PERK expression (a marker for ER stress), a reduction in BiP expression (an ER protein which helps proteins fold as they're being made) and in someway therefore an increase in WASF3 expression (being translated directly into the ER co-translationally I think). WASF3 translocates over to the mitochondria and binds to oxidative phosphorylation complex 3 preventing it from forming a complex with complex 4, thereby inhibiting oxidative phosphorylation activity.

According to the text S1 had all sorts of different diseases and that she had chronic fatigue and exercise intolerance - though no specific mention of ME/CFS diagnosis.


So yeah as always it needs to be replicated in a much bigger cohort with sedentary controls carefully selected and with disease controls - possible it could be a broadly applicable phenotype not specific to ME/CFS.
Click to expand...

I've done the densitometry myself and quantified relative to GAPDH as they would presumably have done (a "housekeeping" gene there to normalise for total protein levels). Bear in mind I'm working with the snippets of the blot membranes they are showing in the paper and not the membrane itself, and it is split across different figures. Hopefully the contrast has been maintained between the different crops they've done. This also made it tricky to get a good read of the background signal which you need to normalise properly. That being said:
hwang_blot_quant.png

P-values (two-tailed unpaired student's t test) and effect sizes (cohen's D):
Gene: WASF3
P-value: 0.0859249
Cohen's d: 0.7596534

Gene: MTCO1
P-value: 0.0007785032
Cohen's d: 2.143763

Gene: COX17
P-value: 0.00237593
Cohen's d: 1.633564

Gene: PERK
P-value: 0.02902655
Cohen's d: 0.9279234

Gene: BiP
P-value: 0.004006675
Cohen's d: 1.298791

Weirdly WASF3 doesn't actually make the 0.05 threshold in my attempt, but it is close so I can see how it might have squeezed past the threshold when they did it. Some of the others though scream through with tiny p values and enormous effect sizes.
 
The ER stress marker PERK was significantly higher while BiP was lower in the ME/CFS muscle samples (Fig. 5 A and B). This discordance between PERK and BiP levels in ME/CFS samples suggested impairment of the canonical ER stress pathway, termed “ER Stress Response Failure,” which has been proposed to result in metabolic disorders (31)
Click to expand...

From what I understood PERK and BiP are supposed to go up together during ER stress but in ME/CFS PERK is high and BiP low. This sounds a bit like the "blunted/absent physiological response to exercise" finding in some recent studies.

Figure 6 shows immunoblots of patient S1 fibroblasts treated with ER stress inhibitors TUDCA and salubrinal. Sadly the effect of TUDCA is modest. It would have been nice if it worked better because it's already available.

The effect of salubrinal on oxygen consumption of patient S1's fibroblasts seems meaningful.

PS: some are going to find fault with this paper but I love it. We've never had anything like this published before. In particular a mouse model that appears to replicate at least some of the reduced stamina that patient suffer from. They have shown that interesting discoveries can be made. Even if it's not going to be THE explanation it's still valuable. Now we have to figure out what's going on with the ER stress response.
 
Last edited:
DMissa said:
There is a paucity of absolutely critical information in this paper that makes it hard for me to draw any conclusions.
Click to expand...

Did you check the supplemental materials for some of the info you're looking for? You could also write Paul Hwang and ask for it.
 
B_V said:
Yes, he's looking at Relyzrio, approved as an ALS drug a few years ago.
Click to expand...

The combination was approved for medical use in Canada as Albrioza, in June 2022,[1][2][8] and in the United States, as Relyvrio, in September 2022.[5][9][7]
Click to expand...

Society and culture
In the United States, healthcare insurer Cigna decided in 2023 to reverse its prior decision to cover the cost of the medication for all ALS patients, opting instead to cover "patients who meet certain clinical criteria", arguing that the drug is "experimental, investigational or unproven".[17]
Click to expand...

Controversies
The FDA approval is controversial because of the small size of the trial. The FDA Peripheral and Central Nervous System Drugs Advisory Committee voted not to recommend approval, and then in an unusual second vote recommended approval.[18][19]
Click to expand...

https://en.wikipedia.org/wiki/Sodium_phenylbutyrate/ursodoxicoltaurine
 
chillier said:
I really don't like when this type of data is represented in bar plots without showing the individual datapoints as a strip style scatter plot. We have no idea how much the groups overlap and they also don't tell us the effect size. Our only indication of variance is the error bar which as you point out represents the standard error which is the wrong metric to use - it should be standard deviation. The standard error of the mean is always smaller than the standard deviation so it looks better too.
Click to expand...

Yes. Just to reiterate this, here's some of the text from the webpage I linked to upthread that I thought set it out well:
Is it better to plot graphs with SD or SEM error bars? (Answer: Neither)
There are better alternatives to graphing the mean with SD or SEM.

If you want to show the variation in your data:
If each value represents a different individual, you probably want to show the variation among values. Even if each value represents a different lab experiment, it often makes sense to show the variation.
With fewer than 100 or so values, create a scatter plot that shows every value. What better way to show the variation among values than to show every value? ....
Click to expand...
If you want to create persuasive propaganda:
If your goal is to emphasize small and unimportant differences in your data, show your error bars as SEM, and hope that your readers think they are SD. If your goal is to cover-up large differences, show the error bars as the standard deviations for the groups, and hope that your readers think they are a standard errors.
Click to expand...
Not that I'm suggesting that the authors here were trying to suggest more about their data than was warranted. It's just - please give us the scatter plots.

chillier said:
I've done the densitometry myself and quantified relative to GAPDH as they would presumably have done (a "housekeeping" gene there to normalise for total protein levels).
Click to expand...
That's awesome. Can you explain why it is quantified relative to the GAPDH, rather than just comparisons between the actual blots? Is it to account for the hydration of each sample? I see that the blots for GAPDH do look pretty consistent between the controls and ME/CFS samples. Does the normalising make much difference? Is both the size and density of the blot important?

(Background on GAPDH: Glyceraldehyde 3-phosphate dehydrogenase is an enzyme of about 37kDa that catalyzes the sixth step of glycolysis and thus serves to break down glucose for energy and carbon molecules.)


The authors showed the blots for 5 controls and 7 people with ME/CFS in Figure 5A, and the other 5 controls and 7 people with ME/CFS in the Supplementary Figure 11. For anyone with a suspicious mind, like myself: they don't appear to have cherry-picked which blots are shown in the article itself. Which is great.
Edit - and they've given us the data for all of the participants. Not all papers do - so, also, good.
 
Last edited:
@chillier thanks for asking for detail and holding my statements accountable. Should actually happen more.

chillier said:
Out of interest is there data showing differences in seahorse results with changes in confluence level, trypsinisation time and so on? I can see how trypsinisation could be harming the cells a little bit when you passage them but it would be a while after you trypsinise and seed the cells, let them grow, change culture medium etc before you actually carry out the assay - does it really make that much of a difference?
Click to expand...

Density (contact inhibition): I don't know about published seahorse studies off the top of my head but I have seen it firsthand in the lab with seahorse, and others have demonstrated it with other methods eg: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1000514

Trypsinisation: seahorse incubation times can vary between 30 minutes to 24 hours. Since this is not described in the paper, a shorter incubation time would mean that the treatment and circumstances of the cells prior to seeding in seahorse plate would likely matter. This may not be an issue with the study if they use a longer time, but since the method is not described, who knows? I have asked in an email to Hwang. I expect that it was probably a longer time to allow for adherence. We will see!

chillier said:
Also with protein quantification it seems surprising to me that differences in expression would really affect the global amount of protein in a cell so much as to make it a poor proxy for cell mass. I would have thought it would be fairly negligible or otherwise the cells would be visibly different. I'm happy to be wrong just trying to understand.
Click to expand...

I would say this is sensible, my view is just that it opens up possibilities for new variables to influence the result, which is counter to the purpose of normalisation. Even if it was a discrepancy of only 0.5% protein content between samples, why do it? Especially when it's adding another step + technique to the assay, and using a different instrument, which opens up more chances for sources of variation or error to creep in. One can measure average cell size while counting if cell size is a concern. It won't add another step or any manual handling, most counting machines do this already.

It's also coming after a seahorse assay which is not only involving biological changes due to inhibition but also mechanical manipulation by mixing and probing. It's quite the ordeal. What's coming out by the very end is not matching what went in.

Everything I have done over literally tens of thousands of seahorse samples (including microscopy assessment of wells after each run), suggests that careful manual handling and a sufficient number of replicates produces reliable data that is not confounded by loss of cells during handling.

There is one possibility which is that overnight incubation is employed and that variable rate of proliferation of cells between samples is intended to be controlled for. But I'm not aware of a way to mess with them with a protein assay in the seahorse well before the seahorse assay that wouldn't be problematic. What one could do there is prepare a growth curve in a previous experiment and calculate an effect based on incubation time. Or do some quick bright-field imaging-based well scans to quantify cell surface area per well.

Hutan said:
Can you explain why it is quantified relative to the GAPDH
Click to expand...

GAPDH is a commonly used housekeeping gene that is considered indicative of total protein levels within a cell. Proteins of interest can move up or down with total protein even if they are not dysregulated so measuring them on their own is a problem, you have to measure total protein alongside your protein of interest in some way (this is all meant in the context of westerns).

Hutan said:
(Background on GAPDH: Glyceraldehyde 3-phosphate dehydrogenase is an enzyme of about 37kDa that catalyzes the sixth step of glycolysis and thus serves to break down glucose for energy and carbon molecules.)
Click to expand...

This may raise alarm bells as a housekeeping gene if glycolysis stands out as a potential problem area in some people's theories, but my whole-cell proteomics shows only a 2.5% average difference (ie: nothing) between healthy and pwME so I would say it's a safe pick. (Housekeeping genes used in a disease context should be known to be unaffected. People use the same old housekeeping genes every time without checking for effect of disease, GAPDH is one of them, but I think this paper dodged this bullet.)

B_V said:
Did you check the supplemental materials for some of the info you're looking for? You could also write Paul Hwang and ask for it.
Click to expand...

Hey, yeah I did, but unfortunately it's not in there. I have written to him :). Make no mistake, this is not negativity on my part. There is a lot of work in this paper with an impressive approach and study design, and the parts unaffected by my comments are saying very interesting things. I just cannot conclude much from some components of the cell-based work without knowing the full context of how the experiments were performed.

Hutan said:
It's just - please give us the scatter plots.
Click to expand...

In the past I have been guilty of multiple of the presentational things you have outlined in the comment (albeit inadvertently, just doing things by what I thought was convention as a student). Especially this part about scatter plots. So just saying that I hear you loud and clear and will improve my efforts. ;)
 
Last edited:
This article gave a great explanation of the study for a wider audience:

Science Alert This Protein Could be Responsible For The Exhaustion in Chronic Fatigue Syndrome

Quote:

Baffling perhaps to medical doctors who have long dismissed ME/CFS, but not so much to researchers who have steadily been building a picture of what triggers this debilitating illness, nor to those who live with its unrelenting exhaustion every day.


Inside every cell are energy-making machines, the mitochondria, which power our cells, replenish our brains, and keep our muscles moving.


Now a new study from a team of US researchers adds evidence to a growing theory that malfunctioning mitochondria might be one potential cause of energy-limiting illnesses such as ME/CFS and long COVID.
 
article
This Protein Could Be Responsible For The Exhaustion in Chronic Fatigue Syndrome
Scientists have just found a protein that might underpin one of the most baffling illnesses there is: myalgic encephalomyelitis, also known as chronic fatigue syndrome or ME/CFS.

Baffling perhaps to medical doctors who have long dismissed ME/CFS, but not so much to researchers who have steadily been building a picture of what triggers this debilitating illness, nor to those who live with its unrelenting exhaustion every day.

Inside every cell are energy-making machines, the mitochondria, which power our cells, replenish our brains, and keep our muscles moving.
Click to expand...
The suspect protein in this case, called WASF3, had been linked to chronic fatigue syndrome before, in a 2011 meta-analysis that no one had followed up on. A study of the woman's blood supported suspicions that symptoms of extreme fatigue were linked with an overexpression of the protein.
Click to expand...
https://www.sciencealert.com/this-p...or-the-exhaustion-in-chronic-fatigue-syndrome
 
Last edited by a moderator:
Paul Hwang sent me a couple of lovely responses, seems like a really nice guy. I didn't get the full picture because he's out of the office and my questions are very technically detailed. I may get more info later from another author. When I have concrete info I may share here if appropriate - (while preserving the intrinsic assumed confidentiality of email). Would only be very brief dot points pertaining to relevant topics from the paper. I want the body of work to be as meaningful as it can possibly be!
 
You must log in or register to reply here.
Back
Top Bottom