UAMS Research Team Finds Potential Cause of COVID-19 ‘Long-haulers’


Senior Member
UAMS Research Team Finds Potential Cause of COVID-19 ‘Long-haulers’ by David Robinson

Sept. 9, 2021 | LITTLE ROCK — A UAMS research team has identified a potential cause of long-lasting symptoms experienced by COVID-19 patients, often referred to as long-haulers. The findings were published in the journal, The Public Library of Science ONE (PLOS ONE).

At the heart of the team’s findings is an antibody that shows up weeks after an initial infection and attacks and disrupts a key regulator of the immune system, said lead researcher John Arthur, M.D., Ph.D., professor and chief of the Division of Nephrology in the UAMS College of Medicine, Department of Internal Medicine.
As many as 30% of COVID-19 patients experience lingering fatigue, brain fog and shortness of breath. The cause of long COVID-19 has eluded scientists, but the UAMS team’s discovery sheds important new light on the molecular-level mechanisms behind it.

“Everything that we’ve found is consistent with this antibody as the instigator of long COVID, so it’s an exciting development that merits further study,” Arthur said.
The antibody creates problems for the immune system by attacking the angiotensin-converting enzyme 2 (ACE2). The ACE2 enzyme helps regulate the body’s response to the virus by metabolizing a peptide that activates the immune system. The attacking antibody interferes with ACE2’s work, which makes the antibody a prime suspect for the long-lasting illness.
Last edited by a moderator:


Senior Member
First off, let's look at their selection method. Patients with known SARS-CoV-2 infection are a much greater superset of COVID-19 long haulers. If they found 81% ACE2 Ab positive, does it mean that 81% of their post-viral participants were long haulers? Clearly, the Post-COVID syndrome is much rarer than 81%. So why is there is a discrepancy and why didn't they differentiate COVID-19 non-long-haulers from long-haulers? Either way, it's incomplete. If they only used long haulers, it's mostly meaningless to only use non-infected as control. If they used a mixed group, there is a huge potential for bias just by being a long hauler or not.

We also obtained 32 plasma samples from plasma donors to be used for convalescent plasma treatment. These samples were obtained from patients that had a known positive virus test by PCR and had been symptom free for at least two weeks prior to donation of plasma. There were no other inclusion or exclusion criteria. We refer to these groups as: Inpatient+, Outpatient+, Outpatient–and Convalescent+.

According to the study, they just used random samples without any further selection method. But they used existing ones from a laboratory. For convalescents, however, I'm not entirely sure. It sounds like they just used existing convalescent plasma samples from donors. This adds a huge selection bias. I'd assume that donors are mostly healthy and that long-haulers aren't inclined to donate their sometimes nonexistent SARS-CoV-2 antibodies. They also found the ACE2 antibodies in 93% of acute COVID-19 patients. This means that the study most likely doesn't represent Post-COVID pathology.

Imagine that these antibodies are the problem and that 81% of all infected and vaccinated have these antibodies, not just long haulers. This would be a huge deal. Antibodies against ACE2, even in low concentrations, from my perspective, are equally important as GPCR antibodies are for ME patients. We would disrupt the Antiogensin system of a huge population - if they already notice it or not - it's just a matter of time until a disruption of this system becomes a problem.

Blindly assuming it's applicable, we can still hope, however, that the prefusion of the vaccine spike proteins protects the uninfected vaccinated from building such antibodies. This might be the case because cross-reactivity doesn't apply to the prefused protein. It might also be the case that these antibodies are not cross-reactive antibodies against spike proteins but merely reactive autoimmunity to the destroyed receptors. Vaccines don't destroy these receptors, so there wouldn't be any reactive autoimmunity towards them in particular.

The third option, nonetheless, is that these antibodies also exist in vaccinated or that only some vaccines are affected. Novavax will be the most unlikely option to cause such a phenomenon because it doesn't even indirectly cause the destruction of ACE2 receptors. Vector and mRNA vaccines enter cells and these cells might be destroyed by the immune system while having ACE2 expression. There's also a fourth option I will elaborate later.

Fortunately, there are varieties of therapeutics that compensate for such a disruption. Many drugs could be repurposed for this indication and many drugs are in pipelines. It's not like the GPCR issue where the first monoclonal antibodies slowly enter the first trial phases.

It would not be a novel finding, by the way, that antibodies against a pathogen cross-react with physiological human proteins. I'm surprised that this hasn't been checked or excluded earlier because ACE2 is the most obvious candidate for a virus that binds to ACE2. Or maybe it has been checked but remained unpublished due to negative findings?

The median abundance of the group with an ACE2 antibody was 0 ng/ml (IQR 0.0–1.1) and the median value of the group that did not have an ACE2 antibody was 0.3 ng/ml (IQR 0–3.9). In contrast, the activity of soluble ACE2 in the plasma of patients that had an ACE2 antibody was lower than the activity for patients that did not have an ACE2 antibody (Fig 4, p<0.01). The median activity of soluble ACE2 in patients with an ACE2 antibody was 263 pmol/min/ml (IQR 0–1039) compared to 1056 (IQR 457–2230) for those that did not have an antibody.

So what does the data really tell? Despite the low differences in concentrations, the differences in activity were much greater. But I think the limitation, in this case, is that this might be confounded by the prior pathology of the virus itself. The samples were collected from relatively "fresh" patients without any clear questionnaire on the history and how much time had passed. This also adds a lot of selection bias.

Here is the controversy. Why do these convalescents have more ACE2 inhibition in exogenous ACE2 than their endogenous ACE2 counterparts when compared to outpatients? Something doesn't add up.

There might be an explanation for it though. It comes down to a general issue in pathology. Most of the time, recombinant antibodies, enzymes, proteins are used to measure enzyme activity and protein interaction ratios in vitro and ex vivo. There's no guarantee that these antibodies are specific enough for the selected target though. Cross-reactivity is always an option. This is why careful testing of the setup with verified samples is necessary to exclude the possibility of cross-reactivity. Positive and negative controls alone do not exclude cross-reactivity. SARS-CoV-2 reacts with ACE2, so the spike protein would be a target to control. Are the used antibodies specific enough to show ACE2 antibody interaction or are spike proteins the antibodies?

The authors do not declare in their methods that they checked for this possibility. A single assay would be sufficient to verify it. Maybe they checked it and left it out because it's trivial, I don't know. Studies require the methods section because they should contain all the information. Most of it is trivia to pathologists anyways.

To be sure that SARS-CoV-2 proteins don't confound the results, I would have preincubated the samples with knockout concentrations of SARS-CoV-2 antibodies because they are known not to be cross-reactive with Angiotensin II and it prevents any further interaction from potential residual spike proteins. Already fused spike-ACE2 complexes might not be affected by this though. It depends on how the interaction of the recombinant antibodies plays out with fused spikes and fused ACE2. Is it the same binding site or a different one? It's just a thought. I'm no pathologist.