Plasma proteomics show altered inflammatory and mitochondrial proteins in patients with ... post-acute sequelae of SARS-CoV-2, 2023, B. Hanson et al

[SNT Gatchaman]

Plasma proteomics show altered inflammatory and mitochondrial proteins in patients with neurologic symptoms of post-acute sequelae of SARS-CoV-2 infection
Hanson; Visvabharathy; Orban; Jimenez; Batra; Liotta; DeLisle; Klausner; Cohen; Padhye; Tachas; Koralnik

Persistent symptoms of COVID-19 survivors constitute long COVID syndrome, also called postacute sequelae of SARS-CoV-2 infection (PASC). Neurologic manifestations of PASC (NeuroPASC) are particularly debilitating, long lasting, and poorly understood.

To gain insight into the pathogenesis of PASC, we leveraged a well-characterized group of Neuro-PASC (NP) patients seen at our Neuro-COVID-19 clinic who had mild acute COVID-19 and never required hospitalization, to investigate their plasma proteome. Using the SomaLogic platform, SomaScan, the plasma concentration of >7000 proteins was measured from 92 unvaccinated individuals, including 48 NP patients, 20 COVID-19 convalescents (CC) without lingering symptoms, and 24 unexposed healthy controls (HC) to interrogate underlying pathobiology and potential biomarkers of PASC. We analyzed the plasma proteome based on post-COVID-19 status, neurologic and non-neurologic symptoms, as well as subjective and objective standardized tests for changes in quality-of-life (QoL) and cognition associated with NeuroPASC.

The plasma proteome of NP patients differed from CC and HC subjects more substantially than post-COVID-19 groups (NP and CC combined) differed from HC. Proteomic differences in NP patients 3-9 months following acute COVID-19 showed alterations in inflammatory proteins and pathways relative to CC and HC subjects. Proteomic associations with Neuro-PASC symptoms of brain fog and fatigue including changes in markers of DNA repair, oxidative stress, and neutrophil degranulation. Furthermore, we discovered a correlation between NP patient’s lower subjective impression of recovery to pre-COVID-19 baseline with an increase in the concentration of the oxidative phosphorylation protein COX7A1, which was also associated with neurologic symptoms and fatigue, as well as impairment in QoL and cognitive dysfunction. Finally, we identified other oxidative phosphorylation-associated proteins correlating with central nervous system symptoms.

Our results suggest ongoing inflammatory changes and mitochondrial involvement in Neuro-PASC and pave the way for biomarker validation for use in monitoring and development of therapeutic intervention for this debilitating condition.

Link | Paywall (Brain, Behavior, and Immunity)

 

[SNT Gatchaman]

Discovery Research Proteins (DRP) were defined as any proteins which obtained stringent statistical significance by Bonferroni correction of p < 0.05 when correcting for the total number of SomaMER comparisons.

Seven proteins met the DRP criteria, as previously described, and were significantly decreased in patients experiencing brain fog including FASLG, PTK2B, BROX, HUS1, RAD51C, SRXN1, and NTAN1.

Three proteins met the DRP criteria and were significantly elevated in NP [NeuroPASC] patients with fatigue: PSPN, CRISPLD2, and PRTN3.

NP patients showed higher overall concentrations of COX7A1 relative to CC [Covid convalescent] and HC groups. We then compared the concentration of COX7A1 across all neurologic and non-neurologic symptoms within the NP group and found that COX7A1 was significantly higher in patients with the PNS neurologic symptoms of numbness/tingling, and pain, as well as the non-neurologic symptom of fatigue.

Further comparisons against subjective PROMIS QoL continuous measures showed that both worse cognition and worse fatigue were significantly associated with increased levels of COX7A1. Additional inverse correlations with COX7A1 were found with the objective NIH Toolbox measures of cognition of processing speed, and attention. In all instances and consistent with the previous findings COX7A1 levels were increased in patients with worse outcomes.

 

[SNT Gatchaman]

Analysis of the core enrichment proteins from the Oxidative Phosphorylation pathway proteins from NP and CC groups show that proteins with the strongest correlations in both groups come from diverse mitochondrial compartments, Fisher’s exact analysis indicates that no mitolocation is over- or under-represented in either CC or NP core enrichment proteins (p=0.37, data not shown). Additionally, Complex I protein NDUFA2, Complex IV protein COX5A, and Complex V proteins ATP5PF and ATP5PO show strong inverse correlations with COX7A1 in NP as compared to strong direct correlations in the CC group. By contrast, Complex IV protein COX7A2L and Complex V protein ATP5F1B show direct correlations in both groups.

Collectively these data provide additional confirmation that the presence of COX7A1 is unlikely to be a result of overall mitochondrial leakage into the plasma in NP patients and that the relationship between COX7A1 and other mitochondrial respiration chain protein subunits is unique in NP patients as compared to CC subjects. This further suggests that mitochondrial involvement is indeed associated with Neuro-PASC.

The exact significance of detecting intracellular proteins in the peripheral blood remains to be determined. It is possible that the increased protein concentration in the plasma reflects the overall production of those proteins, either through increased expression in one or more cell types, or through expansion of specific cells which express those proteins. It is also possible that increased plasma levels are secondary to damage to the cells or tissues in which the protein originates either through intrinsic cell death pathways, through viral cytopathic effect, or through phagocytic/cytotoxic immune interactions. It is probable that a variety of mechanisms occurs.

The possibility that inflammatory tissue damage is ongoing, combined with changes in T-reg associated proteins ASNS and GBP5, suggests that epitope-spreading and loss of self tolerance may be a mechanism in subsets of NP patients. Indeed, autoantibodies have been found to be increased following COVID-19 in subjects with and without PASC symptoms.

 

[SNT Gatchaman]

Proteomic alterations specific to the NP group which have a known association with neurologic function include PTGDS which is highly expressed in the CNS and functions as a neuromodulator regulating sleep and maintaining BBB integrity. PTGDS and GLDN both enhance the interactions of Schwann cells with myelinated neurons within the PNS.

Finally, our DRP findings which linked COX7A1 plasma concentrations to patient-reported impression of recovery as compared to their pre-COVID-19 baseline was of significant interest. We subsequently found increased plasma concentration of COX7A1 in NP patients as compared to HC and CC subjects, in more symptomatic NP patients, and in NP patients with decreased QoL and cognitive dysfunction. COX7A1 is the muscle isoform of the VIIA subunit in the Cytochrome C Oxidase enzyme, which is the terminal enzyme in the mitochondrial respiration chain responsible for oxidative phosphorylation. The SomaScan platform does not measure COX7A2, which is the non-muscle isoform.

Pathway analyses showed an inverse relationship of Hallmark oxidative phosphorylation pathway proteins to COX7A1 concentration in NP patients, but a direct relationship in CC subjects. Although these results are tantalizing, the exact mechanism leading to increased concentration of COX7A1 in plasma remains to be fully elucidated. COX7A1 may be released upon damage to muscle tissues but extracellular release of whole damaged mitochondria has also been observed under conditions of stress in cell culture.

Mice with COX7A1 deletions show exercise intolerance as well as cardiomyopathy. Exercise intolerance is a common complaint among PASC patients, therefore damage to COX7A1-expressing muscle cells would be consistent with PASC symptomatology. However, either of those mechanisms would likely lead to consistent increases in levels of other muscle tissue proteins or mitochondrial proteins in addition to COX7A1, which was not observed in our NP population, but may be consistent with the CC population based on COX7A1 GSEA analysis. Isolated plasma elevations of COX7A1 in the absence of other specific mitochondrial proteins may rather be a marker of mitochondrial function, or inefficiency in the case of elevation, as opposed to cell death/energy, with increases paralleling what is seen in rigorous exercise.

While COX7A1 is muscle-derived, our findings of pathways associated with multiple neurodegenerative diseases included other mitochondrial respiration complex proteins in the core enrichment groups.

Altogether, these data suggest that mitochondrial dysfunction may be contributing to PASC.

 

[SNT Gatchaman]

 

[Hutan]

The Hanson in the author list is Barbara Hanson of North-Western University, Chicago.
@DMissa, one in your wheelhouse.

From the posted excerpts, it looks interesting. The separations aren't perfect, but I guess there could be quite a bit of daily variation in the levels of the proteins depending on activity levels and other things, even in the healthy people.

 

[Hutan]

Forty-six (46) of the NP patients were evaluated by an attending neurologist in the Northwestern University Medical Center Neuro-COVID-19 clinic

Just worth noting this clinic, may be of interest to people with Long covid. They are doing studies and seem to have people assessed by specialists, so even with just that, they are well ahead of many clinics. They don't appear to assess for PEM though.

Objective cognitive function evaluation was performed using the National Institutes of Health (NIH) Toolbox v2.1 instrument, which assessed processing speed (pattern comparison processing speed test); attention (inhibitory control and attention test); executive function (dimensional change card sort test); and working memory (list sort working memory test). 18-21 All tests were administered by clinical or research staff in the Neuro-COVID-19 clinic. Both PROMIS and NIH Toolbox results are expressed as T scores with a score of 50 representing the normative mean/median for the US reference population with a standard deviation of 10.

And also noting the NIH Toolbox for assessing cognitive function.

Plasma samples were analyzed by SomaScan v4.1 by SomaLogic Operating Company (Boulder, CO) as previously described.22 Briefly, DNA based aptamers, known as SOMAmers, bind to specific protein epitopes to yield a fluorescent signal for over 7000 protein targets, detectable over a 10-log dynamic range.23 Protein targets are available at https://somalogic.com/somascan-panels/. Heparinized plasma samples were normalized against plasma EDTA standardized samples, with high levels of consistency observed.

This is the method for identifying specific proteins. There's a normalising process.

Study participant demographics are shown in Table 1. NP and CC groups were of similar age, and CC and HC age were also comparable. The mean age for HC subjects, however, was significantly lower than the NP patient group thus direct comparisons between NP and HC groups, without consideration of CC, were not independently considered. All populations were similarly female predominant, but the HC population had more frequent Asian subject enrollment.

So, some differences in age between the people who had had Covid (both with and without persisting symptoms) and the healthy controls. I had wondered about sex ratios, but they report these were similar. I guess one major difference between the NP (neuroPASC) and the other two groups might be activity levels, but, off the top of my head, I can't think why lower activity levels would cause an increase of mitochondrial proteins being expelled into the blood.

 

[Hutan]

They found more depression and anxiety in the NeuroPASC cohort, but it's worth noting that these patients are being assessed 6 months into their illness, so they won't have had time to adapt and come to terms with the abrupt negative change in their lives.

Objective measures of cognition obtained by NIH Toolbox showed only attention (p<0.0001) with a statistically significant NP group deviation from the normative demographically matched population mean (Figure 2 B and C). However, NP patients had a median of 1 domain with a performance T score of > 1 SD worse than the normative population, with most (54.8%) meeting this threshold for attention. Other domains, though not significantly altered for the NP group overall, were more than one standard deviation worse than the mean for 31.0% for executive function, 30.9% for processing speed, and 14.3% for working memory.

So, attention was the most affected cognitive measure. Interesting that processing speed wasn't significantly different. I'm not quite sure what they mean by that last sentence and haven't bothered to understand it; perhaps the fluctuating nature of cognitive impairment in people with mild PASC is relevant. Figure 2b gives the data, so you can draw your own conclusions. The 50% line is the population average given the individual's age (and maybe other things).


According to Fig 2A and 2C, only 34 of the NeuroPASC patients reported their subjective cognition, and only 35 patients reported their subjective fatigue. So, these samples are quite a lot lower than the 46 cohort size that is reported. That raises questions about the demographic characteristics of these smaller samples. Why were these patients surveyed while other patients in the 46 were not? The smaller samples becomes important when it comes the Figures 3A and 3B

Seven proteins met the DRP criteria, as previously described, and were significantly decreased in patients experiencing brain fog (Figure 3A) including FASLG, PTK2B, BROX, HUS1, RAD51C, SRXN1, and NTAN1. All proteins identified were significantly decreased in NP patients experiencing Brain Fog.

Three proteins met the DRP criteria and were significantly elevated in NP patients with fatigue (Figure 3B): PSPN, CRISPLD2, and PRTN3. Table 4 contains protein description, localization, function, and Human Protein Atlas Immune cell expression clusters for brain fog and fatigue altered proteins.

To understand this, we need to understand what is being compared. For example, Figure 3A is, if I'm understanding it right, making comparisons within the 34 NeuroPASC patients. So, from Table 2c, we know that 2/3 of patients reported cognitive problems that were 1 SD worse than the normalised population average, and so something like 22 patients are in the group described as 'experiencing brain fog' and around 12 were not experiencing brain fog. So, figure 3A compares the proteins in the brain fogged NeuroPASC people with the un-brain fogged Neurological PASC people. That's all very well, but I don't think it's really the comparison we want.

I think it would have been better to compare the proteins in the subset of the NeuroPASC with cognitive dysfunction, or fatigue (and PEM would have been good too) against the healthy controls and the convalescent controls. I think the comparisons within the NeuroPASC group start to get too small, and there are the inaccuracies in measuring subjective symptoms like fatigue (e.g. people still trying to do their usual activities will report more fatigue than people who have given up work).

So, I'm inclined to not pay much attention to Figures 3A and 3B.

(I've got to take a break, but I don't want to leave things suggesting I'm dismissing the whole study. The Cox7A1 finding sounds interesting, and Figure 3c looks convincing, so I do want to poke into it some more.)