Antibodies against type I interferon: detection and association with severe clinical outcome in COVID‐19 patients

Abstract Objectives Impairment of type I interferon (IFN‐I) immunity has been reported in critically ill COVID‐19 patients. This defect can be explained in a subset of patients by the presence of circulating autoantibodies (auto‐Abs) against IFN‐I. We set out to improve the detection and the quantification of IFN‐I auto‐Abs in a cohort of critically ill COVID‐19 patients, in order to better evaluate the prevalence of these Abs as the pandemic progresses, and how they correlate with the clinical course of the disease. Methods The concentration of anti‐IFN‐α2 Abs was determined in the serum of 84 critically ill COVID‐19 patients who were admitted to ICU in Hospices Civils de Lyon, France, using a commercially available kit (Thermo Fisher, Catalog #BMS217). Results A total of 21 of 84 (25%) critically ill COVID‐19 patients had circulating anti‐IFN‐α2 Abs above cut‐off (> 34 ng mL−1). Among them, 15 of 21 had Abs with neutralising activity against IFN‐α2, that is 15 of 84 (18%) critically ill patients. In addition, we noticed an impairment of the IFN‐I response in the majority of patients with neutralising anti‐IFN‐α2 Abs. There was no significant difference in the clinical characteristics or outcome of with or without neutralising anti‐IFN‐α2 auto‐Abs. We detected anti‐IFN‐α2 auto‐Abs in COVID‐19 patients' sera throughout their ICU stay. Finally, we also found auto‐Abs against multiple subtypes of IFN‐I including IFN‐ω. Conclusions We reported that 18% of critically ill COVID‐19 patients were positive for IFN‐I auto‐Abs, whereas all mild COVID‐19 patients were negative, confirming that the presence of these antibodies is associated with a higher risk of developing a critical COVID‐19 form.

The anti-SARS-CoV-2 immune response has been extensively studied, and defects in antiviral mechanisms have been linked to disease severity. In particular, impairment of type I interferon (IFN-I) immunity has been reported in critically ill COVID-19 patients. Such defect can be because of either inherited genetic deficiencies in the IFN-I pathway or the occurrence of circulating autoantibodies (auto-Abs) directed against 14 or the 17 individual IFN-I. 1-4 These auto-Abs have also been detected in a third of patients from a small international cohort who had suffered from severe adverse events following yellow fever vaccination (YFV-17D). 5 These findings advocate for the development of diagnostic tools for the detection of IFN-I auto-Abs in routine laboratories, in order to identify early patients at risk of developing severe forms of COVID-19 and to analyse the prevalence of IFN-I Abs as the pandemic progresses and the virus evolves. To this aim, we tested a commercially available kit measuring IFN-I auto-Abs levels in the serum of COVID-19 patients.

RESULTS
A total of 84 critically ill COVID-19 patients, 11 patients with autoimmune polyendocrinopathy type 1 syndrome (APS-1), 10 mildly symptomatic COVID-19 healthcare workers and 76 healthy controls were included in the study. The critically ill COVID-19 patients were admitted to ICU in the Lyon University Hospital, France, between September and December 2020. The presence of anti-IFN-a 2 Abs was investigated: we first sought to determine a positive cut-off value for Abs detection by performing measurements in 76 putative control sera, that is from healthy donors retrieved before the COVID-19 outbreak. The mean value +3 standard deviation of these measurements provided a cut-off value at 34 ng mL À1 . We then assessed the presence of IFN-a 2 Abs in putative positive sera, that is sera from patients with autoimmune polyendocrinopathy type 1 syndrome (APS-1), a condition known to be associated with anticytokine auto-Abs. All APS-1 patients tested (n = 11) had high titres of circulating anti-IFN-a 2 auto-Abs (> 100 ng mL À1 ). We also evaluated the presence of anti-IFN-a 2 auto-Abs in 10 mildly symptomatic COVID-19 healthcare workers, and none of them was found positive.
We then measured anti-IFN-a 2 Abs levels in the sera from the critically ill COVID-19 patients: 21 of 84 (25%) were positive and had values above the cut-off (> 34 ng mL À1 ). The neutralising capacity of their sera against IFN-a was then evaluated as previously described. 5 A neutralising activity was observed in 15 of 21 positive sera; in other words, 15 of 84 (18%) critically ill COVID-19 patients had neutralising anti-IFN-a auto-Abs ( Figure 1a). Importantly, all sera with a titre of anti-IFN-a 2 auto-Abs above 1 µg mL À1 potently neutralised IFN-a in vitro. In addition, in most patients with neutralising IFN-I Abs, we noticed an impairment of the IFN-I response, which was determined by the measurement of (1) plasma IFN-a 2 levels using the new digital ELISA technology single-molecule arrays (Simoa) and (2) blood interferon stimulating gene (ISG) expression using the NanoString nCounter technology in blood samples collected in the first 15 days after symptom onset (Figure 1b and c). Moreover, there was no significant difference in the clinical characteristics (age, sex ratio and comorbidity) or outcome (death and O 2 support) of critically ill COVID-19 patients with or without neutralising anti-IFN-I auto-Abs (Table 1). Then, serial measurement of IFN-I auto-Abs level during ICU stay was performed for seven positive patients. The level of IFN-a 2 auto-Abs remained relatively stable across measurements performed up to 40 days apart ( Figure 1d). Finally, we assessed the presence of auto-Abs against other IFNs-I in all sera positive for IFN-a 2 auto-Abs (n = 21). Consistently with previous results, 1 we observed that sera with anti-IFN-a 2 auto-Ab titres above 1 µg mL À1 also contained auto-Abs against other subtypes of IFN-a and 10 of 12 contained anti-IFN-x auto-Abs ( Figure 2). Of note, these sera were able to neutralise IFN-x in vitro. Only one serum contained auto-Abs targeting IFN-b at a low titre, and none contained auto-Abs against IFN-ɛ or IFN-j.

DISCUSSION AND CONCLUSIONS
A previous study has reported that IFN-I auto-Abs were present in 10.2% of life-threatening COVID-19 patients, undetectable in 663 individuals with asymptomatic or mild COVID-19, and detected in only 0.33% of healthy individuals. 1 Here, 18% of critically ill COVID-19 patients were positive for  IFN-I auto-Abs, whereas all mild-COVID-19 patients were negative. We noticed that only a part of auto-Abs detected were able to neutralise IFN-I in the conditions we used, which confirms previous studies in COVID-19 patients and systemic lupus erythematosus subjects. 1,6 This finding further confirms the deleterious role of IFN-I auto-Abs in the antiviral immune response and the importance of the IFN-I pathway in the defence against SARS-CoV-2 infection. Based on its antiviral properties, recombinant IFN-I has been tested as therapy for severe COVID-19, but the treatment showed little or no benefit. 7,8 Yet, the potential of such treatment may have been hindered by the presence of IFN-I auto-Abs in patient sera, and this question could therefore be revisited by determining the level of these Abs, for example using the ELISA method we used here. Moreover, patients could be treated with recombinant IFN-I that is not targeted by auto-Abs (e. g. IFN-b). Finally, the detection of anti-IFN-I auto-Abs in COVID-19 patients could be useful in routine to identify patients at risk of developing a severe form of the disease. The technique we described here is adequate for this purpose as it can rapidly provide quantitative measurements and has a cut-off correlated with neutralisation assays (1 µg mL À1 ). However, the presence of auto-Abs was not associated with poorer outcome in critically ill patients and does not explain all the severe forms of COVID-19, other causes should therefore be sought (e.g. cytokine release syndrome and presence of other risk factors such as obesity and hypertension).

Mildly symptomatic COVID-19 patients
Plasma samples and PAXgene â tubes were collected from symptomatic healthcare workers upon COVID-19 diagnosis. Written informed consent was obtained from all participants. The study was approved by the national review board for biomedical research in April 2020 (Comit e de Protection des Personnes Sud M editerran ee I, Marseille, France; ID RCB 2020-A00932-37). The study was registered on ClinicalTrials.gov (NCT04341142) where the eligibility, inclusion and exclusion criteria are previously described. 9

Healthy controls
Prepandemic serum was selected from healthy controls who were recruited among donors to the Lyon blood transfusion centre (Etablissement Franc ßais du Sang, EFS). According to French procedures, a written nonopposition to the use of donated blood for research purposes was obtained from HCs. The donors' personal data were anonymised before transfer to our research laboratory. We obtained approval from the local ethical committee and the French Ministry of Research (DC-2008-64) for handling and conservation of these samples.

Auto-Abs anti-IFN-I
The presence of anti-IFN-a 2 auto-Abs was assessed in the plasma using a commercially available kit (Thermo Fisher; Catalog # BMS217). The positive cut-off value for Ab detection was 34 ng mL À1 . The presence of auto-Abs against other IFNs-I was assessed using an ELISA technique, as previously described.  The neutralisation capacity of antibodies against IFN-a 2 and IFN-x was determined as previously described. 5 Briefly, HEK-293T cells were transfected with a plasmid encoding the firefly luciferase under the control of human ISRE promoters. Cells were then cultured in Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific) supplemented with 10% healthy control or patient serum/plasma, and were either left unstimulated or were stimulated with IFN-a 2 , IFN-x or IFN-b (10 ng mL À1 ) for 16 h at 37°C. Finally, luciferase level was measured using the Dual-Glo reagent, according to the manufacturer's instructions (Promega).

Plasma protein quantification
The concentration of plasmatic IFN-a (fg mL À1 ) was measured using single-molecule array (Simoa) using a commercial kit for IFN-a 2 quantification (Quanterix TM , Lexington, MA, USA). The assay was based on a 3-step protocol and an HD-1 Analyzer (Quanterix).