Autoantibodies against type I IFNs in patients with life-threatening COVID-19

The genetics underlying severe COVID-19 The immune system is complex and involves many genes, including those that encode cytokines known as interferons (IFNs). Individuals that lack specific IFNs can be more susceptible to infectious diseases. Furthermore, the autoantibody system dampens IFN response to prevent damage from pathogen-induced inflammation. Two studies now examine the likelihood that genetics affects the risk of severe coronavirus disease 2019 (COVID-19) through components of this system (see the Perspective by Beck and Aksentijevich). Q. Zhang et al. used a candidate gene approach and identified patients with severe COVID-19 who have mutations in genes involved in the regulation of type I and III IFN immunity. They found enrichment of these genes in patients and conclude that genetics may determine the clinical course of the infection. Bastard et al. identified individuals with high titers of neutralizing autoantibodies against type I IFN-α2 and IFN-ω in about 10% of patients with severe COVID-19 pneumonia. These autoantibodies were not found either in infected people who were asymptomatic or had milder phenotype or in healthy individuals. Together, these studies identify a means by which individuals at highest risk of life-threatening COVID-19 can be identified. Science, this issue p. eabd4570, p. eabd4585; see also p. 404

INTRODUCTION: Interindividual clinical variability is vast in humans infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), ranging from silent infection to rapid death. Three risk factors for life-threatening coronavirus disease 2019 (COVID- 19) pneumonia have been identifiedbeing male, being elderly, or having other medical conditions-but these risk factors cannot explain why critical disease remains relatively rare in any given epidemiological group. Given the rising toll of the COVID-19 pandemic in terms of morbidity and mortality, understanding the causes and mechanisms of life-threatening COVID-19 is crucial.
RATIONALE: B cell autoimmune infectious phenocopies of three inborn errors of cyto-kine immunity exist, in which neutralizing autoantibodies (auto-Abs) against interferon-g (IFN-g) (mycobacterial disease), interleukin-6 (IL-6) (staphylococcal disease), and IL-17A and IL-17F (mucocutaneous candidiasis) mimic the clinical phenotypes of germline mutations of the genes that encode the corresponding cytokines or receptors. Human inborn errors of type I IFNs underlie severe viral respiratory diseases. Neutralizing auto-Abs against type I IFNs, which have been found in patients with a few underlying noninfectious conditions, have not been unequivocally shown to underlie severe viral infections. While searching for inborn errors of type I IFN immunity in patients with life-threatening COVID-19 pneumonia, we also tested the hypothesis that neutralizing auto-Abs against type I IFNs may underlie critical COVID-19. We searched for auto-Abs against type I IFNs in 987 patients hospitalized for life-threatening COVID-19 pneumonia, 663 asymptomatic or mildly affected individuals infected with SARS-CoV-2, and 1227 healthy controls from whom samples were collected before the COVID-19 pandemic.
RESULTS: At least 101 of 987 patients (10.2%) with life-threatening COVID-19 pneumonia had neutralizing immunoglobulin G (IgG) auto-Abs against IFN-w (13 patients), against the 13 types of IFN-a (36), or against both (52) at the onset of critical disease; a few also had auto-Abs against the other three individual type I IFNs. These auto-Abs neutralize high concentrations of the corresponding type I IFNs, including their ability to block SARS-CoV-2 infection in vitro. Moreover, all of the patients tested had low or undetectable serum IFN-a levels during acute disease. These auto-Abs were present before infection in the patients tested and were absent from 663 individuals with asymptomatic or mild SARS-CoV-2 infection (P < 10 −16 ). They were present in only 4 of 1227 (0.33%) healthy individuals (P < 10 −16 ) before the pandemic. The patients with auto-Abs were 25 to 87 years old (half were over 65) and of various ancestries. Notably, 95 of the 101 patients with auto-Abs were men (94%).
CONCLUSION: A B cell autoimmune phenocopy of inborn errors of type I IFN immunity accounts for life-threatening COVID-19 pneumonia in at least 2.6% of women and 12.5% of men. In these patients, adaptive autoimmunity impairs innate and intrinsic antiviral immunity. These findings provide a first explanation for the excess of men among patients with lifethreatening COVID-19 and the increase in risk with age. They also provide a means of identifying individuals at risk of developing life-threatening COVID-19 and ensuring their enrolment in vaccine trials. Finally, they pave the way for prevention and treatment, including plasmapheresis, plasmablast depletion, and recombinant type I IFNs not targeted by READ  M ycobacteriosis, staphylococcosis, and candidiasis can be driven by monogenic inborn errors of interferon-g (IFN-g), interleukin-6 (IL-6), and IL-17A and IL-17F, respectively, or they can be driven by their genetically driven autoimmune phenocopies, with the production of neutralizing autoantibodies (auto-Abs) against these cytokines (1)(2)(3)(4)(5)(6)(7)(8). Type I IFNs, first described in 1957, are ubiquitously expressed cytokines that contribute to both innate immunity (through their secretion by plasmacytoid dendritic cells and other leukocytes) and cell-intrinsic immunity (in most if not all cell types) against viral infections (9)(10)(11)(12)(13). Their receptors are ubiquitously expressed and trigger the induction of IFN-stimulated genes (ISGs) via phosphorylated STAT1-STAT2-IRF9 trimers (STAT, signal transducers and activators of transcription; IRF, interferon regulatory factor) (14). Neutralizing immunoglobulin G (IgG) auto-Abs against type I IFNs can occur in patients treated with IFN-a2 or IFN-b (15) and exist in almost all patients with autoimmune polyendocrinopathy syndrome type I (APS-1) (16). They are also seen in women with systemic lupus erythematosus (17).
These patients do not seem to suffer from unusually severe viral infections, although human inborn errors of type I IFNs can underlie severe viral diseases, both respiratory and otherwise (18). In 1984, Ion Gresser described a patient with unexplained auto-Abs against type I IFNs suffering from severe chickenpox and shingles (19,20). More recently, auto-Abs against type I IFNs have been found in a few patients with biallelic, hypomorphic RAG1 or RAG2 mutations and viral diseases including severe chickenpox and viral pneumonias (21). Our attention was drawn to three patients with APS-1, with known preexisting anti-type I IFN auto-Abs, who had life-threatening coronavirus disease 2019 (COVID-19) pneumonia (22) (see detailed case reports in Methods). While searching for inborn errors of type I IFNs (18,23), we hypothesized that neutralizing auto-Abs against type I IFNs might also underlie life-threatening COVID-19 pneumonia.
Auto-Abs against IFN-a2 and/or IFN-w in patients with critical COVID-19 We searched for auto-Abs against type I IFNs in 987 patients hospitalized for life-threatening COVID-19 pneumonia. We also examined 663 individuals infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) presenting asymptomatic infection or mild disease and 1227 healthy controls whose samples were collected before the COVID-19 pandemic. Plasma or serum samples were collected from patients with critical COVID-19 during the acute phase of disease. Multiplex particle-based flow cytometry revealed a high fluorescence intensity (FI) (>1500) for IgG auto-Abs against IFN-a2 and/or IFN-w in 135 patients (13.7%) with life-threatening COVID-19 (Fig. 1A). We found that 49 of these 135 patients were positive for auto-Abs against both IFN-a2 and IFN-w, whereas 45 were positive only for auto-Abs against IFN-a2, and 41 were positive only for auto-Abs against IFN-w.
We also performed enzyme-linked immunosorbent assay (ELISA), and the results obtained were consistent with those obtained with Luminex technology (fig. S1A). We found that 11 and 14 of 23 patients tested had low levels of IgM and IgA auto-Abs against IFN-w and IFN-a2, respectively ( Fig. 1B and fig. S1B). Auto-Abs against type I IFNs were detected in two unrelated patients for whom we had plasma samples obtained before SARS-CoV-2 infection, which indicates that these antibodies were present before SARS-CoV-2 infection and were not triggered by the infection. As a control, we confirmed that all 25 APS-1 patients tested had high levels of auto-Abs against IFN-a2 and IFN-w ( fig. S1C). Overall, we found that 135 of 987 patients (13.7%) with life-threatening COVID-19 pneumonia had IgG auto-Abs against at least one type I IFN.
The auto-Abs neutralize IFN-a2 and IFN-w in vitro We then tested whether auto-Abs against IFN-a2 and IFN-w were neutralizing in vitro. We incubated peripheral blood mononuclear cells (PBMCs) from healthy controls with 10 ng/mL IFN-a2 or IFN-w in the presence of plasma from healthy individuals or from patients with auto-Abs. A complete abolition of STAT1 phosphorylation was observed in 101 patients with auto-Abs against IFN-a2 and/or IFN-w (table S1). The antibodies detected were neutralizing against both IFN-a2 and IFN-w in 52 of these 101 patients (51%), against only IFN-a2 in 36 patients (36%), and against only IFNw in 13 patients (13%) at the IFN-a2 and IFN-w concentrations tested (Fig. 1, C and D). IgG depletion from patients with auto-Abs restored normal pSTAT1 induction after IFN-a2 and IFN-w stimulation, whereas the purified IgG fully neutralized this induction ( Fig. 1C and  fig. S1D). Furthermore, these auto-Abs neutralized high amounts of IFN-a2 (fig. S1E) and were neutralizing at high dilutions ( We also analyzed ISG induction after 2 hours of stimulation with IFN-a2, IFN-b, or IFN-g in the presence of plasma from healthy individuals or from patients with auto-Abs. With plasma from eight patients with auto-Abs against IFN-a2, the induction of ISG CXCL10 was abolished after IFN-a2 stimulation but maintained after stimulation with IFN-g (Fig. 1F). We then found that plasma from the five patients with neutralizing auto-Abs neutralized the protective activity of IFN-a2 in Madin-Darby bovine kidney (MDBK) cells infected with vesicular stomatitis virus (VSV) (table S2). Overall, we found that 101 of 987 patients (10.2%)-including 95 men (94%)-with life-threatening COVID-19 pneumonia had neutralizing IgG auto-Abs against at least one type I IFN. By contrast, auto-Abs were detected in only 4 of 1227 healthy controls (0.33%) (Fisher exact test, P < 10 −16 ) and in none of the 663 patients with asymptomatic or mild SARS-CoV-2 infection tested (Fisher exact test, P < 10 −16 ).

Auto-Abs against all 13 IFN-a subtypes in patients with auto-Abs to IFN-a2
We investigated whether patients with neutralizing auto-Abs against IFN-a2 only or those with neutralizing auto-Abs against IFN-a2 and IFN-w also had auto-Abs against the other 15 type I IFNs. ELISA showed that all patients tested (N = 22) with auto-Abs against IFN-a2 also had auto-Abs against all 13 IFN-a subtypes (IFN-a1, -a2, -a4, -a5, -a6, -a7, -a8, -a10, -a13, -a14, -a16, -a17, and -a21), whereas only 2 of the 22 patients tested had auto-Abs against IFN-b, 1 had auto-Abs against IFN-k, and 2 had auto-Abs against IFN-e ( Fig. 2A). The auto-Abs against IFN-b had neutralizing activity against IFN-b (Fig. 1D). We confirmed that all of the patients had auto-Abs against all 13 subtypes of IFN-a by testing the same samples using luciferase-based immunoprecipitation assay (LIPS) (Fig. 2B). For IFN-b, we also screened the whole cohort in a multiplex assay. We found that 19 of 987 (1.9%) patients had auto-Abs against IFN-b and that all of them were in our cohort of severe COVID-19 individuals with neutralizing auto-Abs against IFN-a and/ or IFN-w. Of these patients with auto-Abs against IFN-b, only two were neutralizing against IFN-b ( Fig. 1, D and F).
Ten of the 17 genes encoding type I IFNs (IFN-a2, -a5, -a6, a8, -a13, -a14, -a21, -b, -w, and -k), have undergone strong negative selection, which suggests that they play an essential role in the general population. By contrast, the 3 of 12 other seven IFN loci in the human genome often carry loss-of-function alleles (24). Moreover, the 13 IFN-a subtypes and IFN-w are more-closely related to each other than they are to the other three IFNs (IFN-b, IFN-e, and IFN-k), which are structurally and phylogenetically more distant (Fig. 2C). Thus, all patients with neutralizing auto-Abs against IFN-a2 that we tested (N = 22) had auto-Abs against all 13 IFN-a subtypes, and 3 of the 22 patients tested (14%) had auto-Abs against 14 or more type I IFNs.
The auto-Abs neutralize IFN-a2 against SARS-CoV-2 in vitro and IFN-a in vivo Plasma from eight patients with neutralizing auto-Abs against type I IFN also neutralized the ability of IFN-a2 to block the infection of Huh7.5 cells with SARS-CoV-2 (Fig. 3A). Plasma from two healthy controls or from seven SARS-CoV-2-infected patients without auto-Abs did not block the protective action of IFN-a2 ( Fig. 3A and fig. S3A). These data provide compelling evidence that the patients' blood carried sufficiently large amounts of auto-Abs to neutralize the corresponding type I IFNs and block their antiviral activity in vitro, including that against SARS-CoV-2.
We also found that all 41 patients with neutralizing auto-Abs against the 13 types of IFN-a tested had low (one patient) or undetectable (40 patients) levels of the 13 types of IFN-a in their plasma during the course of the disease (Fig. 3B) (25,26). Type I IFNs may be degraded and/or bound to the corresponding circulating auto-Abs. The presence of circulating neutral-izing auto-Abs against IFN-a is, therefore, strongly associated with low serum IFN-a levels (Fisher exact test, P < 10 −6 ). Consistently in patients with neutralizing auto-Abs against IFN-a2, the baseline levels of type I IFNdependent transcripts were low, whereas they were normal for nuclear factor kB (NF-kB)dependent transcripts (Fig. 3C and fig. S3B). Overall, our findings indicate that the auto-Abs against type I IFNs present in patients with life-threatening COVID-19 were neutralizing in vitro and in vivo. IFNs. This proportion of males was higher than that observed in patients with critical COVID-19 without auto-Abs (75%; Fisher exact test, P = 2.5 × 10 −6 ), and the proportion was much higher than that in male patients in the asymptomatic or pauci-symptomatic cohort (28%; Fisher exact test, P < 10 −6 ) ( Table 1, Fig. 4A, and fig. S4A). Further evidence for X-chromosome linkage was provided by the observation that one of the seven women with auto-Abs and life-threatening COVID-19 had X chromosomelinked incontinentia pigmenti (IP), in which cells activate only a single X chromosome (cells having activated the X chromosome bearing the null mutation in NEMO dying in the course of development) (27). The prevalence of auto-Abs against type I IFNs in the general population was estimated at 0.33% (0.015 to 0.67%) in a sample of 1227 healthy individuals-a value much lower than that in patients with lifethreatening COVID-19 pneumonia, by a factor of at least 15.
The patients with auto-Abs were also slightly older than the rest of our cohort (49.5% of patients positive for auto-Abs were over 65 years of age versus 38% for the rest of the cohort; P = 0.024), which suggests that the frequency of circulating anti-type I IFNs auto-Abs increases with age (Table 1 and Fig. 4B). However, auto-Abs were found in patients aged from 25 to 87 years ( fig. S4B). Principal components analysis (PCA) was performed on data from   (Fig. 4C). Large-scale studies will be required to determine the frequency of such auto-Abs in humans of different sexes, ages, and ancestries. Finally, the presence of auto-Abs was associated with a poor outcome, with death occurring in 37 of the 101 patients (36.6%) (table S1).
Neutralizing auto-Abs to type I IFNs are causative of critical COVID-19 There are multiple lines of evidence to suggest that the neutralizing auto-Abs against type I IFNs observed in these 101 patients preceded infection with SARS-CoV-2 and accounted for the severity of disease. First, the two patients for whom testing was performed before COVID-19 were found to have auto-Abs before infection. Second, three patients with APS-1 known to have neutralizing auto-Abs against type I IFN immunity before infection also had lifethreatening COVID-19 (22) (supplementary methods). Third, we screened a series of 32 women with IP and found that a quarter of them had auto-Abs against type I IFNs, including one who developed critical COVID-19 ( fig. S1C). Fourth, there is a marked bias in favor of men, which suggests that the production of auto-Abs against type I IFNs-whether driven by germ line or somatic genome-may be X chromosome-linked and therefore preexisting to infection. Moreover, IFN-a subtypes were undetectable during acute disease in the blood of patients with auto-Abs against IFN-a, which suggests a preexisting or concomitant biological impact in vivo. It is also unlikely that patients could break self-tolerance and mount high titers of neutralizing IgG auto-Abs against type I IFN within only 1 or even 2 weeks of infection. Finally, inborn errors of type I IFNs underlying life-threatening COVID-19 in other previously healthy adults-including autosomal recessive IFN-a/b receptor subunit 1 (IFNAR1) deficiency-have also been reported in an accompanying paper (18). Collectively, these findings suggest that auto-Abs against type I IFNs are a cause and not a consequence of severe SARS-Cov-2 infection, although their titers and affinity may be enhanced by the SARS-CoV-2driven induction of type I IFNs. They also provide an explanation for the major sex bias seen in patients with life-threatening COVID-19 and perhaps also for the increase in risk with age.

Conclusion
We report here that at least 10% of patients with life-threatening COVID-19 pneumonia have neutralizing auto-Abs against type I IFNs. With our accompanying description of patients with inborn errors of type I IFNs and lifethreatening COVID-19 (18), this study highlights the crucial role of type I IFNs in protective immunity against SARS-CoV-2. These auto-Abs against type I IFNs were clinically silent until the patients were infected with SARS-CoV-2a poor inducer of type I IFNs (28)-which suggests that the small amounts of IFNs induced by the virus are important for protection against severe disease. The neutralizing auto-Abs against type I IFNs, like inborn errors of type I IFN production, tip the balance in favor of the virus, which results in devastating disease with insufficient, and even perhaps deleterious, innate and adaptive immune responses.
Our findings have direct clinical implications. First, SARS-CoV-2-infected patients can be screened to identify individuals with auto-Abs at risk of developing life-threatening pneumonia. Such patients recovering from life-threatening COVID-19 should also be excluded from donating convalescent plasma for ongoing clinical trials, or at least they should be tested before their plasma donations are accepted (29). Second, this finding paves the way for preventive or therapeutic intervention, including plasmapheresis, monoclonal Abs depleting plasmablasts, and the specific inhibition of type I IFN-reactive B cells (30). Finally, in this patient group, early treatment with IFN-a is unlikely to be beneficial; however, treatment  with injected or nebulized IFN-b may have beneficial effects, as auto-Abs against IFN-b appear to be rare in patients with auto-Abs against type I IFNs.

Subjects and samples
We enrolled 987 patients with proven lifethreatening (critical) COVID-19, 663 asymptomatic or pauci-symptomatic individuals with proven COVID-19, and 1227 healthy controls in this study. All subjects were recruited following protocols approved by local Institutional Review Boards (IRBs). All protocols followed local ethics recommendations and informed consent was obtained when required. COVID-19 disease severity was assessed in accordance with the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia. The term life-threatening COVID-19 pneumonia describes pneumonia in patients with critical disease, whether pulmonary, with mechanical ventilation [continuous positive airway pressure (CPAP), bilevel positive airway pressure (BIPAP), intubation, or high-flow oxygen], septic shock, or damage to any other organ requiring admission in the intensive care unit (ICU). The individuals with asymptomatic or mild SARS-CoV-2 infection were individuals infected with SARS-CoV-2 who remained asymptomatic or developed mild, self-healing, ambulatory disease with no evidence of pneumonia. The healthy controls were individuals who had not been exposed to SARS-CoV-2.
Plasma and serum samples from the patients and controls were frozen at −20°C immediately after collection. The fluid-phase LIPS assay was used to determine the levels of antibodies against the SARS-CoV-2 nucleoprotein and spike protein, as has been previously described (31).
Detection of anti-cytokine auto-Abs Multiplex particle-based assay Serum and plasma samples were screened for auto-Abs against 18 targets in a multiplex particle-based assay, in which magnetic beads with differential fluorescence were covalently coupled to recombinant human proteins. Patients with an FI of >1500 for IFN-a2 or IFN-b or >1000 for IFN-w were tested for blocking activity, as were patients positive for another cytokine.

ELISA
ELISA was performed as previously described (5). In brief, ELISA plates were coated with recombinant human interferon-a (rhIFN-a) or rhIFN-w and incubated with 1:50 dilutions of plasma samples from the patients or controls. A similar protocol was used when testing for 12 subtypes of IFN-a.

LIPS
Levels of auto-Abs against IFN-a subtypes were measured with LIPS, as previously described (32) . IFN-a1, IFN-a2, IFN-a4, IFN-a5, IFN-a6,  IFN-a7, IFN-a8, IFN-a10, IFN-a14, IFN-a16, IFN-a17, and IFN-a21 sequences were transfected in HEK293 cells, and the IFN-a-luciferase fusion proteins were collected in the tissue culture supernatant. For autoantibody screening, serum samples were incubated with protein G agarose beads, and we then added 2 × 10 6 luminescence units (LU) of antigen and incubated. Luminescence intensity was measured. The results are expressed in arbitrary units (AU), as a fold-difference relative to the mean of the negative control samples.

Functional evaluation of anti-cytokine auto-Abs
The blocking activity of anti-IFN-a and anti-IFN-w auto-Abs was determined by assessing STAT1 phosphorylation in healthy control cells after stimulation with the appropriate cytokines in the presence of 10% healthy control or patient serum or plasma.
We demonstrated that the IFN-a and IFN-w blocking activity observed was due to auto-Abs and not another plasma factor, by depleting IgG from the plasma with a protein G column Without eluting the IgG, the flow-through fraction (IgG-depleted) was then collected and compared with total plasma in the phospho-STAT1 assay.
For the neutralization of ISG induction, PBMCs were left unstimulated or were stimulated for 2 hours with 10 ng/mL IFN-a or 10 ng/ mL IFN-g in a final volume of 100 mL. Realtime quantitative polymerase chain reaction (RT-qPCR) analysis was performed with Applied Biosystems Taqman assays for CXCL10, and the b-glucuronidase (GUS) housekeeping gene for normalization. Results are expressed according to the DDCt method, as described by the manufacturer's kit.

Phylogenetic reconstruction
Protein sequences were aligned with the online version of MAFFT v7.471 software (33), using the L-INS-i strategy (34) and the BLOSUM62 scoring matrix for amino acid substitutions. Phylogenetic tree reconstruction was performed by the neighbor-joining method (35) with the substitution model (36). Low-confidence branches (<50%) are likely to be due to gene conversion events between IFNA genes, as previously reported (24,37). The tree was then visualized (38). Very similar results were obtained with the corresponding DNA sequences (37,39).

Statistical analysis
Comparison of proportions were performed using a Fisher exact test, as implemented in R (https://cran.r-project.org/). PCA was performed with Plink v1.9 software on whole-exome and whole-genome sequencing data with the 1000 Genomes (1kG) Project phase 3 public database as a reference.

Simoa
Serum IFN-a concentrations were determined with Simoa technology, as previously described (40,41), with reagents and procedures obtained from the Quanterix Corporation.

VSV assay
The seroneutralization assay was performed as previously described (42). In brief, the incubation of IFN-a2 with MDBK cells protects the cultured cells against the cytopathic effect of VSV. The titer of anti-IFN-a antibodies was defined as the last dilution causing 50% cell death.

SARS-CoV-2 experiment
SARS-CoV-2 strain USA-WA1/2020 was obtained from BEI Resources and amplified in Huh7.5 hepatoma cells at 33°C. Viral titers were measured on Huh7.5 cells in a standard plaque assay. Plasma samples or a commercial anti-IFN-a2 antibody were serially diluted and incubated with 20 pM recombinant IFN-a2 for 1 hour at 37°C (starting concentrations: plasma samples = 1/100 and anti-IFN-a2 antibody = 1/1000). The cell culture medium was then removed and replaced with the plasma-or antibody-IFN-a2 mixture. The plates were incubated overnight, and the plasma-or antibody-IFN-a2 mixture was removed by aspiration. The cells were washed once with phosphate-buffered saline (PBS) to remove potential anti-SARS-CoV-2 neutralizing antibodies, and fresh medium was then added. Cells were then infected with SARS-CoV-2 by directly adding the virus to the wells. Cells infected at a high multiplicity of infection (MOI) were incubated at 37°C for 24 hours, whereas cells infected at a low MOI were incubated at 33°C for 48 hours. The cells were fixed with 7% formaldehyde, stained for SARS-CoV-2 with an anti-N antibody, imaged, and analyzed as previously described (43).

Nanostring
For the NanoString assay, total RNA was extracted from whole blood samples collected in PaxGene tubes. The expression of selected genes was determined by NanoString methods and a 28-gene type I IFN score was calculated (44). of the Imagine Institute (with the support of the Fondation Bettencourt-Schueller). We thank the Association "Turner et vous" for their help and support. Sample processing at IrsiCaixa was possible thanks to the crowdfunding initiative YoMeCorono. D.C.V. is supported by the Fonds de la recherche en santé du Québec clinician-scientist scholar program. K.K. was supported by the Estonian Research Council grant PUT1367. We thank the GEN-COVID Multicenter Study (https://sites.google.com/dbm.unisi.it/ gen-covid). We thank the NIAID Office of Cyber Infrastructure and Computational Biology, Bioinformatics and Computational Biosciences Branch (contract no. HHSN316201300006W/ HHSN27200002 to MSC, Inc.), the Operations Engineering Branch for developing the HGRepo system to enable streamlined access to the data, and the NCI Advanced Biomedical Computational Science Requests for materials derived from human samples may be made available, subject to any underlying restrictions on such samples. J.-L.C. can make material transfer agreements available through The Rockefeller University. This work is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. To view a copy of this license, visit https:// creativecommons.org/licenses/by/4.0/. This license does not apply to figures/photos/artwork or other content included in the article that is credited to a third party; obtain authorization from the rights holder before using such material.