ABSTRACT
STRUCTURED ABSTRACTHyperinflammation is frequently observed in patients with severe COVID-19. Inadequate and defective IFN type I responses against SARS-CoV-2, associated with autoantibodies in a proportion of patients, lead to severe courses of disease. In addition, hyperactive responses of the humoral immune system have been described. In the current study we investigated a possible role of neutralizing autoantibodies against antiinflammatory mediators. Plasma from adult patients with severe and critical COVID-19 was screened by ELISA for antibodies against PGRN, IL-1-Ra, IL-10, IL-18BP, IL-22BP, IL-36-Ra, CD40, IFN-2, IFN-{gamma}, IFN-{omega} and serpinB1. Autoantibodies were characterized and the antigens were analyzed for immunogenic alterations. In a discovery cohort with severe to critical COVID-19 high titers of PGRN-autoantibodies were detected in 11 of 30 (36.7%), and of IL-1-Ra-autoantibodies in 14 of 30 (46.7%) patients. In a validation cohort of 64 patients with critical COVID-19 high-titer PGRN-Abs were detected in 25 (39%) and IL-1-Ra-Abs in 32 of 64 patients (50%). PGRN-Abs and IL-1-Ra-Abs belonged to IgM and several IgG subclasses. In separate cohorts with non-critical COVID-19, PGRN-Abs and IL-1-Ra-Abs were detected in low frequency (i.e. in < 5% of patients) and at low titers. Neither PGRN-nor IL-1-Ra-Abs were found in 40 healthy controls vaccinated against SARS-CoV-2 or 188 unvaccinated healthy controls. PGRN-Abs were not cross-reactive against SARS-CoV-2 structural proteins nor against IL-1-Ra. Plasma levels of both free PGRN and free IL-1-Ra were significantly decreased in autoantibody-positive patients compared to Ab-negative and non-COVID-19 controls. In vitro PGRN-Abs from patients functionally reduced PGRN-dependent inhibition of TNF- signaling, and IL-1-Ra-Abs from patients reduced IL-1-Ra- or anakinra-dependent inhibition of IL-1{beta} signaling. The pSer81 hyperphosphorylated PGRN isoform was exclusively detected in patients with high-titer PGRN-Abs; likewise, a hyperphosphorylated IL-1-Ra isoform was only found in patients with high-titer IL-1-Ra-Abs. Thr111 was identified as the hyperphophorylated amino acid of IL-1-Ra. In longitudinally collected samples hyperphosphorylated isoforms of both PGRN and IL-1-Ra emerged transiently, and preceded the appearance of autoantibodies. In hospitalized patients, the presence of IL-1-Ra-Abs or IL-1-Ra-Abs in combination with PGRN-Abs was associated with a higher morbidity and mortality. To conclude, neutralizing autoantibodies to IL-1-Ra and PGRN occur in a significant portion of patients with critical COVID-19, with a concomitant decrease in circulating free PGRN and IL-1-Ra, indicative of a misdirected, proinflammatory autoimmune response. The break of self-tolerance is likely caused by atypical hyperphosphorylated isoforms of both antigens, whose appearances precede autoantibody induction. Our data suggest that these immunogenic secondary modifications are induced by the SARS-CoV-2-infection itself or the inflammatory environment evoked by the infection and predispose for a critical course of COVID-19.