Exaggerated IFN-I Response in Long COVID PBMCs Following Exposure to Viral Mimics, 2025, Humer et al

forestglip

forestglip

Moderator
Staff member
Exaggerated IFN-I Response in Long COVID PBMCs Following Exposure to Viral Mimics

Humer, Bart; Berentschot, Julia C.; van Helden-Meeuwsen, Cornelia G.; Bek, L. Martine; de Bie, Maaike; Defesche, Tobias M.; Boly, Chantal A.; Drost, Manon; Hellemons, Merel E.; Dik, Willem A.; Versnel, Marjan A.

Purpose
Long COVID (LC) is a long-term debilitating disease of which the exact pathophysiology is unknown. A dysregulated immune response resulting in hyperresponsive immune cells is hypothesized as a key mechanism in the development of LC. Several studies suggest that acute infections can leave lasting epigenetic changes, which result in heightened immune reactivity. Upon stimulation, these primed immune cells may exhibit exaggerated responses. This form of epigenetic memory can contribute to altered immune dynamics, particularly in response to induction of type I Interferons (IFN-I) pathway activation using a viral mimic. Therefore, we investigated if LC patients exhibit a hyperresponsive response towards viral mimics in comparison with healthy controls (HC).

Methods
PBMCs of two distinct LC cohorts, characterized by a different disease course and duration, were collected and transfected using Lyovec with the cGAS and RIG-I agonists G3-YSD and 3p-RNA followed by measurement of IFN-I bioactivity with a reporter cell line.

Results
Transfection of PBMCs of LC patients with the cGAS and RIG-I agonists resulted in increased IFN-I bioactivity in comparison with HC. Unsupervised hierarchical clustering revealed two distinct clusters, each predominantly composed of either patients or HC. In addition, a moderate correlation between RIG-I stimulation with 3p-RNA and fatigue severity scores was found.

Conclusion
These data show a hyperresponsive phenotype of immune cells of LC patients upon stimulation with viral mimics. The current availability of biologicals and small molecule inhibitors that interfere with aberrant IFN-I pathway activation underscores the importance of pursuing future investigations into this phenomenon.

Web | DOI | PDF | Journal of Clinical Immunology | Open Access
 
I'd be lying if I said I wasn't excited by the abstract, so I'll try to be extra skeptical looking through the data.

Two LC cohorts, one from earlier in the pandemic (called "Immunofatigue", described in the paper in this thread), and one newly recruited ("Microx"). Both were age and sex-matched.

The "Immunofatigue" cohort is comparing hospitalized COVID-19 cases with persistent fatigue to COVID-naive healthy controls. The "Microx" cohort seems to be non-hospitalized, and the controls are recovered convalescent.

The methods state that the Microx cohort had PEM as part of the inclusion criteria, but don't mention how PEM was defined or assessed. Microx LC appears to have a longer disease duration and to be much more severely affected.
1765561004407.png1765560858881.png1765560952518.png

The experiment was testing PBMC reactivity to synthetic DNA/RNA molecules (G3-YSD and 3p-RNA), meant to stimulate two different methods that cells use to detect viral entry and mount an anti-viral response. They also had a Lyovec control--it's the reagent that the DNA/RNA has to be mixed with to enter cells (and can cause a cellular response on its own).

Figure 1:
Testing differences in IFN production after stimulation. There's a lot of variability in the Immf LC group and the outliers skew the scale quite a bit. P-values were derived from Kruskal-Wallis which is more appropriate for data with such a wide range, but a log transformation would still be recommended. P-values are uncorrected. At the very least there's no significant differences between the two HC groups, which makes me more confident that differences between the LC groups are biological (i.e. due to duration of illness and/or hospitalization) and not due to batch.

1765561484006.png

Since difference in IFN-production could also be due to differences in proportions of interferon producing cells, they also compared frequency of monocyte subsets and found no differences. I would have liked to see comparisons of DCs as well since those are going to produce the bulk of interferon, but they only make up a small percentage of PBMCs so I understand why they weren't measured. (These are all supplemental figures, I won't copy them here).

Figure 2:
Same data as Fig. 1, doing unsupervised hierarchical clustering. I like the idea of doing unsupervised clustering, given the heterogeneity you expect in any LC cohort. The Immf LC participants cluster together, but the Microx LC participants are split. The main concern is that the clustering is mostly driven by hospitalization (Immf LC was the only group hospitalized for LC).
1765562091266.png

Figure 3:
Checking if fatigue survey scores are correlated with responsiveness to the 3p-RNA RIG-I agonist. There's a significant correlation in the Microx but not the Immunofatigue LC group. Note that the fatigue scores were higher overall in the Microx LC group, and that 97% reported fatigue as a symptom (so really it's a correlation with fatigue scores in the range of moderate to really bad).

I would have liked to see the same correlation with the G3-YSD cGAS agonist, especially since IFN-1 production to G3-YSD seemed to be the main differentiating factor for Microx LC in cluster 1 vs. cluster 2. The question is whether there is any meaningful clinical difference between the Microx LC participants in those 2 camps, and I'm not sure if this one correlation answers that question adequately.
1765562419799.png
Also, noting there's a discrepancy in the p-value reported for Fig 3A in the test (p = 0.001) vs. in the figure (p = 0.01). If it is p=0.01, I'd need a p-value correction.

[Edit: whoops, cross-posted Fig 1 with @Utsikt]
 
Utsikt said:
No real separation, and it looks the same for most of the other graphs.

If this was a core mechanism, why are the high HCs not experiencing the same symptoms?
Click to expand...
I agree--too much overlap to be a core mechanism driving disease. If the differences are still significant after a log transformation and p-value correction, it's about the level of separation I'd expect to see for a plausible downstream consequence.
 
It would do for downstream consequence but I also think it would do for an indirect measure of what similar cells might be doing somewhere more difficult to access. The data look interesting to me. I have not had time to look through the paper but it would be nicer to have some more detail on subpopulations, as mentioned by others.
 
The researchers have a Dutch grant for ME/CFS research:

Blog: Een behandeling op maat met immuunhandtekeningen voor ME/CV

Hebben patiënten met ME/CVS een ontregeld afweersysteem? En zijn er onderlinge verschillen? Wim Dik, medisch immunoloog bij het Erasmus MC, acht de kans groot d
www.zonmw.nl www.zonmw.nl

jnmaciuch said:
They also had a Lyovec control--it's the reagent that the DNA/RNA has to be mixed with to enter cells (and can cause a cellular response on its own).
Click to expand...
Don't quite understand why this also showed an interferon difference between groups, given that it was mainly meant as a control condition.
 
poetinsf said:
The question I always ask: how does IFN-I hypersensitivity create PEM with 12-48 hour delay?
Click to expand...
Well, the classical immune cellular hypersensitivity response (so-called Type IV) takes just that - 12-48 hours. This is more likely to be an innate response but with some cell recruitment and RNA transcription/translation (maybe for two or more steps in sequence) the time frame is in the ball park.
 
Jonathan Edwards said:
Well, the classical immune cellular hypersensitivity response (so-called Type IV) takes just that - 12-48 hours. This is more likely to be an innate response but with some cell recruitment and RNA transcription/translation (maybe for two or more steps in sequence) the time frame is in the ball park.
Click to expand...
I take that you are saying IFN-I response to virus (or viral mimic in this case) takes 12-48 hours? I was wondering more about how the IFN-I hypersensitivity (to viral infection) results in worsening of symptoms 12-48 hours after exertion. Maybe they share the same mechanism leading to IFN-I response?
 
poetinsf said:
The question I always ask: how does IFN-I hypersensitivity create PEM with 12-48 hour delay? Che/Lipkin paper also talks about hyperactive immunity (and chronic inflammation) but doesn't clearly link it to PEM other than many of the abnormalities worsens after exercise.
Click to expand...
Agree with Jonathan, that’s the typical timeframe. ~16 hours is the expected peak for IFN-b production from macrophages stimulated by polyI:C (a similar agonist), if I’m remembering old conversations correctly
 
Last edited:
poetinsf said:
Maybe they share the same mechanism leading to IFN-I response?
Click to expand...
That’s why I’m interested in mtDNA release in neurons and muscle cells under calcium flux. Stimulates the exact same pathway as viral DNA, is known to get released during exertion (though it doesn’t trigger a very strong interferon response in healthy people), happens in the right time frame, consistent with PEM symptoms.
 
You must log in or register to reply here.
Back
Top Bottom