[EV] proteomics uncovers energy metabolism, complement system, and [ER] stress response dysregulation postexercise in males with [ME/CFS], 2025,Glass+

[forestglip]

Extracellular vesicle proteomics uncovers energy metabolism, complement system, and endoplasmic reticulum stress response dysregulation postexercise in males with myalgic encephalomyelitis/chronic fatigue syndrome

Katherine A. Glass, Ludovic Giloteaux, Sheng Zhang, Maureen R. Hanson

[Line breaks added]

Background
Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) is a debilitating illness characterized by post-exertional malaise (PEM), a worsening of symptoms following exertion. The biological mechanisms underlying PEM remain unclear.

Extracellular vesicles (EVs) play a key role in cell–cell communication and may provide insight into ME/CFS pathophysiology post-exertion. Emerging evidence suggests similarities between ME/CFS and Long COVID, including PEM and overlapping immune and metabolic dysfunctions, highlighting the need for deeper mechanistic understanding.

Methods
This study explores the EV proteome response to exercise in 10 males with ME/CFS and 12 well-matched sedentary male controls. Participants underwent a maximal cardiopulmonary exercise test, and plasma samples were collected at baseline, 15 min, and 24 h postexercise.

EVs were isolated from plasma using size-exclusion chromatography and characterized with nanoparticle tracking analysis. EV protein abundance was quantified with untargeted proteomics (nanoLC-MS/MS). Comprehensive analyses included differential abundance, pathway enrichment, protein–protein interaction networks, and correlations between EV protein dynamics and clinical or exercise physiology data.

Results
ME/CFS patients exhibited many significantly altered EV proteomic responses compared with controls, including downregulation of TCA cycle-related proteins and upregulation of complement system proteins at 15 min postexercise.

Changes in proteins involved in protein folding and the endoplasmic reticulum (ER) stress response during recovery were highly correlated with PEM severity, highlighting their potential as therapeutic targets. EV protein changes postexercise were also associated with disease severity and unrefreshing sleep.

Correlations between EV protein levels and the exercise parameters VO₂ peak and ventilatory anaerobic threshold were observed in controls but were absent in ME/CFS patients, suggesting disrupted EV-mediated physiological processes.

Conclusions
ME/CFS patients exhibit a maladaptive EV proteomic response to exercise, characterized by metabolic impairments, immune overactivation, and ER stress response dysregulation. These findings provide insight into the molecular basis of PEM and suggest promising targets for improving recovery and energy metabolism in ME/CFS.

Key points

• EVs were isolated from plasma of ME/CFS patients and healthy controls at baseline, and 15 min and 24 h postexercise.
• Untargeted proteomics revealed dysregulation in energy metabolism, the complement system, and the endoplasmic reticulum stress response.
• Changes in EV protein levels postexercise are associated with post-exertional malaise.
• These findings suggest promising therapeutic targets for post-exertional malaise and ME/CFS pathophysiology.

Link | PDF (Clinical and Translational Medicine) [Open Access]

 

[Utsikt]

Participants underwent a maximal cardiopulmonary exercise test, and plasma samples were collected at baseline, 15 min, and 24 h postexercise.

It’s a shame they didn’t get samples from 48 and 72 h as well.

 

[V.R.T.]

It’s a shame they didn’t get samples from 48 and 72 h as well.

Agreed, and small sample size is disappointing but the findings look interesting.

 

[Utsikt]

Agreed, and small sample size is disappointing but the findings look interesting.

Probably budget limitations at the time of the testing, as usual.

 

[John Mac]

BMI-matched sedentary male controls participated in this study

Good to see

 

[jnmaciuch]

Oh this is exactly the type of study I’ve been hoping for. I wish they’d had a 48 hour time point as well though. Quick skim through the methods doesn’t give me anything to complain about so far, everything lines up with the checks and balances I would expect from a robust analysis.

Only thing I’m always slightly concerned about is the amount of missingness and imputation. If they provide the data pre-imputation I’d like to manually check the how many of the top hits were from proteins with a lot of imputation.

Will probably come back to this after the initial excitement wears off and I have more time to dig into it.

 

[V.R.T.]

Probably budget limitations at the time of the testing, as usual.

Hopefully a bigger replication attempt is on the cards, especially if they think they've found therapeutic targets.

 

[ME/CFS Skeptic]

This publication is only on male ME/CFS patients. An earlier paper contained the data on female ME/CFS patients:

Giloteaux L, Glass KA, Germain A, Franconi CJ, Zhang S, Hanson MR. Dysregulation of extracellular vesicle protein cargo in female myalgic encephalomyelitis/chronic fatigue syndrome cases and sedentary controls in response to maximal exercise. J Extracell Vesicles. 2024;13(1):12403.

Thread:
Dysregulation of extracellular vesicle protein cargo in female ME/CFS cases & sedentary controls in response to maximal exercise, 2023 Giloteaux et al | Science for ME​

 

[Hutan]

The introduction is very good - there's a summary of previous work looking at EVs in ME/CFS.

It's great to see this regular flow of well-conducted studies now (even if many are still rather too small), getting things right like well-matched controls and repeatedly pointing to similar pathways.

 

[jnmaciuch]

The differences between their male and female cohort are really quite striking. I would have expected at least a little more consistency between them.

Quick side note that their pathway analysis methodology differed between then two studies, though that probably wouldn’t drive the differences.

If the two year gap between this study and the last one was due to funding issues or time delay in cohort recruitment, then that would mean the samples weren’t all processed together. I think we ought to be mindful of likely batch effects confounding any sex effects.

 

[Hutan]

2.4 Proteomics
Detailed information can be found in Giloteaux et al.28 In brief, a TMT 10-plex shotgun proteomics analysis was performed. Nine protein-extracted EV samples from three subjects at 0 h, 15 min, and 24 h postexercise were included in each of two TMT 10-plex sets, one for the ME/CFS group and one for the control group. The remaining channel in each TMT set was used for a pooled reference of proteins from all 18 samples to bridge the results between groups. A total of four TMT experiments were conducted (Figure S1).

I didn't find this very easy to follow, and Figure S1 doesn't really help, presumably being only for one TMT experiment. I'm hoping that each subject's samples were analysed i.e 10 ME/CFS and 12 controls, although it isn't completely clear how that was done and why there would be more controls than ME/CFS subjects. Each TMT 10-plex set analysed 3 subjects... I haven't read the whole study yet, so perhaps it is clearer what was done later. If anyone is talking with the researchers, perhaps they can explain? Regardless of what was done, it sounds as though there would have been quite a bit of data adjustment to account for possible differences between experiments.

Finally, while a similar study was previously conducted in female ME/CFS patients and sedentary controls,28 a direct comparison between male and female EV proteomics postexercise is not possible, as these studies were conducted years apart. The current study in males, which was completed more recently, detected a greater number of proteins due to advancements in the analytical platform.

I wonder if the female samples could be re-analysed using the better analytical tools? (Added to the Study Idea thread)

 

[jnmaciuch]

I didn't find this very easy to follow, and Figure S1 doesn't really help, presumably being only for one TMT experiment. I'm hoping that each subject's samples were analysed i.e 10 ME/CFS and 12 controls, although it isn't completely clear how that was done and why there would be more controls than ME/CFS subjects. Each TMT 10-plex set analysed 3 subjects... I haven't read the whole study yet, so perhaps it is clearer what was done later. If anyone is talking with the researchers, perhaps they can explain? Regardless of what was done, it sounds as though there would have been quite a bit of data adjustment to account for possible differences between experiments.

There were analyzed separately—TMT is a method where you can multiplex (run multiple samples together to save resources and reduce batch effects) but all the analytes from each sample receive a unique “tag.” That allows you to demultiplex—i.e. retroactively assign readings to the samples they came from

 

[Hutan]

The text and Fig S1 suggests: (M=ME/CFS; C=control; T=time)
(Edited to also take into account a later part of the text that talks about experiments with only two ME/CFS subjects. The paper is not clear about the experiment structure - it has to be pieced together.)


TMT 1:
M1 at T1 vs C1 at T1; M1 at T2 vs C2 at T2;M1 at T3 vs C2 at T3
M2 at T1 vs C1 at T1; M2 at T2 vs C2 at T2;M2 at T3 vs C2 at T3
M3 at T1 vs C1 at T1; M3 at T2 vs C2 at T2;M3 at T3 vs C2 at T3
Mixture of M1, M2, M3, C1, C2, C3 vs Mixture of M1, M2, M3, C1, C2, C3
(i.e. 18 subject samples and 2 mixtures)

TMT2:
as for TMT 1, but for M4, M5, M6 and C4, C5, C6

TMT3:
as for TMT 1, but for M7, M8 and C7, C8, C11

TMT4:
as for TMT 1, but for M9, M10 and C9, C10, C12



Supplementary Figure 1 and the text seems to suggest that the mixtures in each experiments were only of the samples included in that TMT experiment.

The remaining channel in each TMT set was used for a pooled reference of proteins from all 18 samples to bridge the results between groups.

But, how would that help standardise for batch effects between experiments?

 

[Hutan]

They say:

Protein expression across different TMT experiments was assessed by comparing sample ratios to the pooled reference.44

So, I hope Figure S1 and the earlier text is wrong. I think the mixtures must surely have all of the samples, not just the ones in the particular experiment.

 

[Hutan]

There's really rich data collected - for example, good information on symptoms.

Demographic parameters included age, BMI, and for the ME/CFS group, duration of illness. Additionally, survey data that reflect physical function and disease severity were also evaluated, including the Bell Activity Scale score (ranging from 10 to 100, with 100 indicating greater activity), the SF-36 Physical Component Score (ranging approximately from 11 to 64, with higher scores indicating better physical function), the SF-36 General Health score, and the proportion of waking time spent in a reclined position.

We also examined symptom severity. For the SSS, each subject in both the ME/CFS and control groups rated their symptoms on a scale from 0 to 10 (0 = not present and 10 = very high) at three different times: (1) on average over the past month, (2) on the morning of the CPET (0 h) and (3) 24 h after the CPET (24 h). The ΔSSS are calculated by subtracting the score at 0 h from the score at 24 h. Therefore, a ΔSSS below 0 indicates an improvement in that symptom following exercise, while a ΔSSS above 0 suggests that the symptom worsened 24 h postexercise. The following symptoms were included in our analysis: fatigue, impaired memory or concentration, recurrent sore throat, lymph node tenderness, muscle tenderness or pain (myalgia), joint pain (arthralgia), headache, unrefreshing sleep, and PEM. The Multidimensional Fatigue Inventory-20 total score was used as another metric of general fatigue, but only in ME/CFS subjects (range 20–100, with higher scores indicating greater fatigue).

Additionally, ME/CFS patients scored lower on the SF-36 Physical Component Summary (28.7 ± 7.3 vs. 55.9 ± 6.1, p = 3.12 × 10⁻⁸) and Mental Component Summary (45.7 ± 6.5 vs. 55.6 ± 5.9, p = 1.60 × 10⁻3) compared with controls, indicating that the disease has a significant impact on both physical and mental health.

Each of the SF-36 components - physical and mental - is supposed to have a mean of 50 and a standard deviation of 10. The controls here scored 55.9 on the physical component and 55.6 on the mental component. I assume that is a population mean? A slightly higher physical score than a population mean would sense - these middle-aged men are probably doing better physically than the average 70 or 80 year old. It clearly demonstrates the reduced physical capacity of the ME/CFS subjects. But the mental health score of the controls suggests that they are doing slightly better than average on that front, but still close to normal. Similarly, the mean score of 45.7 in the men with ME/CFS is not far off normal. I guess what the researchers have written there is technically, statistically correct, but the data there, especially in such a small sample, doesn't support the idea that ME/CFS has had a substantial effect on mental health as measured by the SF-36.

 

[Hutan]

Vesicle size
No significant differences at any time point

Vesicle concentration

Red is controls, blue is ME/CFS.
The intriguing thing there is what is not happening in ME/CFS. In the healthy controls, there is a substantial mean increase compared to baseline in the concentration of EVs 24 hours after exercise. There is quite a spread in the results though, so it's a bit hard to know what to make of it.

 

[EndME]

Great, that the authors uploaded the data to mapmecfs.

Quite notable that study participation was concluded more than 5 years ago.

 

[Utsikt]

I guess what the researchers have written there is technically, statistically correct, but the data there, especially in such a small sample, doesn't support the idea that ME/CFS has had a substantial effect on mental health as measured by the SF-36.

Aren’t those questionnaires affected by just being ill?

 

[Hutan]

After filtering out proteins with missing values in more than one out of four TMT experiments, 865 proteins were retained for further analysis. Of these, 575 proteins had no missing data, while the remaining missing values were imputed using random forest (RF, missForest R package, normalized root mean square error .44). We chose to impute data using RF because it is recognized as the optimal method for imputation in mass spectrometry data when the cause of missingness is unknown, such as in cases of missing completely at random.55 RF also effectively manages non-linear data and outliers without requiring feature scaling.55, 56

We cross-referenced our EV protein dataset (pre- and postfiltered) with two well-established EV proteome databases, Exocarta57 and Vesiclepedia.58 Impressively, 88% and 93% of the proteins in our datasets matched those found in the databases, pre- and postfiltering, respectively (Figure 2C).

Just got to say that mass spectroscopy and the technology around it is pretty close to magic. To look at a sample and identify nearly all the proteins in it - it's absolutely fantastic. The technology is there to understand ME/CFS. The results from this tiny study look important. But researchers should not have to be faffing around with tiny and old cohorts. If a solid bit of money was thrown at studies like this, I think the clear answers would come. The problem is important enough and the tools are available.

 

[Hutan]

Aren’t those questionnaires affected by just being ill?

Yes. I guess my point was just that, despite being significantly physically affected by ME/CFS, these participants aren't scoring much different, on average, to the average person on the mental component of SF-36. So, my takeaways from that are
1. The disease pathology itself is probably not having a big direct effect on mental wellbeing, and
2. This small cohort is reporting remarkable mental health resilience. Now, they may have great support around them and they have respectable VO2 max figures, so they are clearly less disabled than many with ME/CFS. So it's not to criticise people who don't score well. But, it's just clearly possible to have ME/CFS and not have an abnormal mental health score on the SF-36.