Viruses have long been shown to transiently perturb the immune system, leading to its dysregulation and the generation of autoantibodies [17, 18]. Molecular mimicry and cross-reactivity between viral and tissue antigens, direct cytopathic effects of viral replication and apoptosis, and dysregulation of B-cells have all been proposed as pathomechanisms for their development [19, 20]. To mention few examples, infection with Epstein-Barr virus has been associated with higher frequency of autoantibodies reacting with cardiolipin and cytoskeletal proteins. Autoantibodies directed against liver and kidney tissues were also described in the context of infection with viral hepatitis [21]. On the other hand, many patients infected with the human immunodeficiency virus (HIV) were found to have a wide range of autoantibodies, reacting against platelets, and erythrocytes, smooth muscle, in addition to antinuclear antibodies, rheuma factor, and anti-phospholipid antibodies [22]. More recently, antinuclear/extractable-nuclear antibodies and autoantibodies reacting with type I IFNs, Anti-Platelet Factor 4, were associated with COVID-19 [3,4,5,6,7,8,9,10,11], although causation or correlation has not yet been established with the infection.

Here, we tested the appearance of lung-specific autoantibodies in severe COVID-19 patients during their hospital care. More than half of these patients featured one or more clones of autoantibodies, recognizing human pulmonary proteins. Mortality increased among patients with at least 3 autoantibody clones recognizing different pulmonary proteins, similar to previous observations linking autoimmunity to disease severity [23]. In addition, multi-producer patients were elder than the mean age of the patients enrolled which is in good correlation with previously published data [4]. Of note, the study represents correlations between age and the number of autoantibody clones that cannot be considered an established causative relationship, nevertheless, other studies have already established that autoimmunity is more likely among the elderly (e.g., [24]).

These results show that anti-pulmonary autoantibodies are frequently developing in acute COVID-19 patients, over 50% of the study population. It appears that excessive tissue damage and hyper-inflammation in severe acute COVID-19 promote autoantibody production. It is not clear whether these autoantibodies are only biomarkers of general organ damage in severe acute COVID-19, or whether they also contribute to the deterioration of physiological functions and worsening of the patient’s condition.

Studies have already identified multiple targets of autoantibodies as anti-cardiac autoantibodies [14], autoantibodies to antigens related to systemic autoimmune rheumatic conditions [10], interferons [11], and other targets [3,4,5,6,7]. In our study, we also observed a multitude of autoantibodies against the lung tissue and importantly, the distribution of autoantigens was not random, a definite accumulation of some specific targets was seen in the case of pulmonary autoantibodies. In other words, some of the pulmonary proteins proved to be better immunogens. This observation is in contrast to our previous study on cardiac autoantibodies as, apparently, there are no apparent preferred targets in the case of the cardiac tissue [14].

Importantly, the presence of anti-lung autoantibodies showed correlation with mortality and worse clinical picture of the disease, in contrast to anti-cardiac autoantibodies that did not associate with mortality [14]. Moreover, we identified a specific ~ 50 kDa antigen in the pulmonary tissue that was highly associated with mortality. This suggests that there may be a molecular mimicry between the pulmonary protein and the SARS-CoV-2 (facilitating autoantibody production), or we have prevalence of patients with this specifically targeted protein. This question cannot be decided based on the available data, but it is intriguing that individuals with the ~ 50 kDa antigen may have more severe symptoms, therefore, presenting at the emergency care unit, from where most of our patients were recruited.

In a previous publication [14], we performed a similar analysis than that in the present manuscript on heart tissue homogenates. We found a similar level of autoimmunity (more, than half of the patients harbored anti-cardiac autoantibodies). Nonetheless, we noted three important differences: first, in most cases, the anti-lung and anti-cardiac antibodies recognized different molecular targets (different sizes of autoantigens); second, the presence of anti-cardiac autoantibodies did not correlate with mortality; and third, anti-cardiac autoantibodies did not recognize a cluster of similarly sized autoantigens. All these differences suggest that the anti-pulmonary autoantibodies are targeting specifically the lungs.

There are limitations to this study. It represents patients infected by the delta strain, and it is most likely that different strains of SARS-CoV-2 induce different immune responses. This is a retrospective study performed in a population with severe COVID-19; of note, about half of the involved patients died. We do not have information on the titer of the SARS-CoV-2 at the time of blood sampling, making us unable to correlate the autoimmunity with the severity of infection. This did not allow identifying certain variables, such as the time of infection and the time of hospitalization relative to that. The lack of sufficient human biological samples precluded the identification of the antigens. Another limitation of the study is that with the available methodology, we cannot determine the affinity of the autoantibodies, as the signal on blots does not necessarily correlate with binding affinity but is affected by the quantity of the autoantibodies in serum and the quantity of the antigens in the lung tissue homogenate used as bait.

Independently of their role in mortality, novel autoantibodies generated during COVID-19 may re-activate after the acute phase of the disease and can cause symptoms related to the targeted organ, contributing to long (or post-) COVID syndrome. The occurrence of anti-pulmonary autoantibodies (54% among the severe COVID-19 patients) correlated well with the observed occurrence of long-COVID syndrome in severe (up to 83%) [25] and in mild (up to 35%) COVID-19 patients [26] during the same phase (delta strain) of COVID-19 pandemic. It appears that autoimmunity may be an important contributor to long COVID-19, and may require an adequate, targeted therapeutic approach to treatment.