Correlation of Interferons and Autoimmune Aspects in Long COVID-19 Patients, 2025, Hattori et al

[forestglip]

Correlation of Interferons and Autoimmune Aspects in Long COVID-19 Patients

Fumiyuki Hattori, Junji Nishiyama, Hideaki Hasuo

[Line break added]

Abstract
Long COVID, or post-acute sequelae of COVID-19 (PASC), represents a major global health challenge, with its underlying mechanisms remaining poorly understood despite substantial research and clinical trials. This study investigates the role of the interferon (IFN) axis in the pathogenesis of PASC, drawing parallels to systemic lupus erythematosus (SLE). The potential pathogenic role of IFNs was detected by meta-analyses of mRNA sequencing data comparing PASC patients to healthy controls.

We analyzed serum samples from 39 PASC patients and found significant correlations among multiple IFN subtypes, including IFN alpha-2, beta, gamma, lambda-1, and lambda-2/3. The biological activity of IFNs in the serum was positively correlated with levels of both total and type III IFNs. Notably, we detected the widespread presence of anti-double-stranded DNA (anti-dsDNA) and anti-Smith (anti-Sm) antibodies in these patients, with anti-dsDNA levels showing a strong correlation with IFN activity. Based on these findings, we propose a hypothetical autoimmune pathogenesis for PASC highlighting the crucial role of IFN signaling.

Link | PDF (International Immunology) [Open Access]

 

[Hutan]

Meta-analysis of the RNA sequencing data set of PASC

We utilized RNA sequencing data from the Gene Expression Omnibus (GEO) Dataset (GSE251849) (15) involving healthy controls (N = 7), recovered from an acute infection of COVID-19 (N = 5), and long COVID (general: N = 6 and brain fog: N = 5), which was processed by the web-based software GREIN. To highlight characteristics of PASC, we constructed a new control group combining healthy controls (N = 7) and those recovered from an acute infection of COVID-19 (N = 5) and compared two groups: new control (N = 12) and PASC (N = 11).

We extracted the differentially expressed gene (DEG) set by DESeq2 using web software, RNAseqChef. The DEG list (false discovery rate, FDR < 0.05), including 18026 genes (Supplementary Table 1), was processed by SRplot to perform GO- and pathway-enrichment analyses. Among the three major ontologies, biological process, cellular components, molecular function, and pathway enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) were shown.

For the comprehensive gene expressions comparing PASC with healthy controls, we conducted a meta-analysis of the public RNA sequencing data set using peripheral blood mononuclear cells (PBMCs) deposited by Greene et al. [15]

To extract the common characteristics of PASC patients, we combined the healthy group and the recovered group from the COVID-19 infection for a new control group data set (N = 12), also general PASC patients and brain fog patients for a new PASC group (N = 11), and conducted pairwise comparisons.

We extracted a significantly DEG set(Supplementary Table 1) and applied it to the gene ontology (GO) enrichment analysis and revealed alterations in type I and II IFN signaling, antigen presentation activity, and antigen processing activities (Fig. 1A). The pathway enrichment analysis revealed that the PBMCs from PASC patients showed autoimmune disease-related signatures (Fig. 1B). Pathway enrichment analysis identified gene expression alterations in MHC class II, IgG, and BCR as components of the SLE pathway (Fig. 1C), although these alone do not adequately represent SLE features.

For this sub-study, there are issues with the samples that need consideration. The authors used an existing RNA sequencing data set - Green et al Ref 15. Therefore, if their findings overlap with findings from other teams (and they do), then it might be because there is overlap in the datasets or the individuals contributing data.

Also, the number of samples was very very small for this sort of work (12 controls and 11 PASC) and the definition of PASC is broad (basically new persisting symptoms). The number of genes that were found to be differently expressed was very large, 18026. So, all that is to say that any findings have, in my opinion, a high chance of being noise.

And yet...
If you look at the Biological Processes list, the top 4 pathways in terms of enrichment score have to do with interferons. The next three are to do with antigen processing. And the next two, with very high numbers of genes are to do with the proteasome (although I am mindful of @jnmaciuch's comment that the proteasome might be a false positive as it seems to pop up in many datasets). All of those issues are being talked about in relation to ME/CFS at the moment.

There are other potentially interesting pathways mentioned in the other two lists. Some are involved in antigen presentation/binding and immune receptor activity. There is Nad dependent HDAC activity, which could be affecting a lot of things, including gene silencing.

So, the findings could easily be noise, but I thought it was worth noting them, because they tend to fit some of the ideas we are thinking about at the moment.

 

[Hutan]

Interferon levels in the serum

We used the LEGENDplex™ Human Type 1/2/3 Interferon Panel with Filter Plate (BioLegend, CA, USA Cat. 740350) to quantify human IFN-α2, IFN-β, IFN-λ1 (IL-29), IFN-λ2/3 (IL-28a/b), and IFN-γ in the serum

To confirm the biological IFN activity of each sample, we stimulated the human hepatocellular carcinoma cell line, HepG2, with PASC patient serum.

We collected serum samples from 39 PASC patients, including 21 women and 18 men. The summary of patients’ information, all data obtained by ELISA, and symptoms is shown in Supplementary Table 2 and Supplementary Fig. 1. Each normal range of IFN was provided by the manufacturer using normal human serum samples (N = 18) with the same multiplex ELISA system (Supplementary Table 3), showing lower serum IFN concentrations than those of PASC patients tested here.

So, they are suggesting that the serum of the PASC group (39 people) produced higher IFN than controls when applied to a carcinoma cell line. They seem to be relying on a manufacturer's standard using normal serum for their comparator.

They plot the levels of a range of IFNs for the PASC individuals, but, oddly, they don't show the healthy standard result on their chart. (I haven't gone to the supplementary table to see what the standard values were.)

The authors then seem to go off and run correlations between anything that isn't tied down, but, I think, without giving any context from healthy people with which to interpret the results. So, I'm a bit uneasy about the study.

 

[wigglethemouse]

In Robert Phair's Itaconate shunt webinar he stated that IFN-alpha measurements are often several orders of magnitude off. This paper has IFN-alpha2 in ug/ml but Robert Phair's measurements of IFN-alpha in units of fg/ml.