Case-based evidence links increased SARS-CoV-2 diversity to impaired IFN-I response in severe COVID-19

Within-host viral diversity has been identified as a potential key factor driving the emergence of SARS-CoV-2 variants that drove successive waves of infection. This has been mostly described in highly immunocompromised patients with altered adaptive immunity, such as patients with cancer, under immunosuppressive treatment, or having received a transplantation. However, the adaptive immunity itself has not been associated with high within-host diversity in immunocompetent subjects (1). Instead, prolonged shedding of SARS-CoV-2 seems to be the major factor associated with high within-host diversity. The control of viral replication is driven by the type I interferon (IFN-I) response, a key player of innate immune response. Recent observations indicate that critically ill patients infected with SARS-CoV-2 frequently exhibit impairment of the IFN-I response (2). Consequently, we hypothesized that an altered IFN-I response could lead to uncontrolled viral replication and contribute to within-host viral diversity. The present study aimed to explore the association between innate immune response and the emergence of within-host SARS-CoV-2 genetic variants in a cohort of patients without known adaptative impairment, admitted to the Critical Care Unit (CCU) for acute respiratory failure.

Forty SARS-CoV-2 nasopharyngeal samples (NPS) collected from 19 patients between November 2022 and January 2024 were sequenced in duplicate. Following the exclusion of 16 samples based on quality criteria (1 sample) or the absence of intra-host single nucleotide variants (iSNVs; 15 samples), iSNV was investigated in a total of 24 samples from 16 patients in the present study. NPS were collected from patients with a median (interquartile range [IQR]) age of 77 (69–84) years; 75% were male (12/16), and the mortality rate was 25% (4/16). Four patients were treated with remdesivir. The median [IQR] interval between the onset of symptoms and swab collection was 8 (4–9) days.

We identified 52 iSNVs, of which 40 were non-synonymous, resulting in a median [IQR] of 2 [1–3] iSNVs per sample at a median [IQR] allelic frequency of 13% [8–25%]. The sample obtained from patient#2 exhibited the highest number of iSNV, with 10 mutations at high allelic frequencies (9.5–34.5%; Fig. 1, A and B). Patient#2 had an infection with JQ.1 lineage according to pangolin classification. Interestingly, we identified in this sample the presence of two non-synonymous mutations in the spike gene, one of which is known to be a marker of JN.1 variant (L455S).

The sample obtained from patient#2 was collected from a 79-year-old male admitted to the CCU for acute respiratory failure. The NPS containing high iSNV count was collected 1 wk after CCU admission. Prior samples could not be sequenced due to low viral load. Notably, this patient exhibited a viral rebound, reaching 5.2 log₁₀ copies/10⁶ cells, despite high anti-RBD IgG levels at inclusion (12,576 binding antibody units per ml [BAU/ml]) and the administration of remdesivir throughout the CCU stay (Fig. 1 C). Interestingly, the PaO₂/FiO₂ ratio decreased during viral rebound. After CCU discharge, the patient was transferred to the Intensive Care Unit in a nonteaching hospital where he stayed for 11 days before then being transferred to a pneumology unit where a superinfection caused by influenza A was documented by PCR.

We observed a consistently low IFN-I nasal score for patient#2 throughout the CCU stay, remaining below the positive threshold previously described in healthy volunteers (Fig. 1 D). The viral rebound was not associated with an increase in the IFN-I response, highlighting an inadequate innate immune response (Fig. 1 E). To explore a potential immune cause, we assessed plasma neutralization of IFN-I using a neutralization assay. Only plasma from patient#2 had strong neutralizing activity against IFN-β (10 ng/ml; Fig. 1 F).

To explore the relationship between an altered antiviral innate immune response and within-host viral variants, we conducted an analysis in a clinical cohort of SARS-CoV-2-infected patients admitted to the CCU with no prior history of adaptative immunosuppression. Considering all NPS included, we did not observe a high iSNV count (median of two mutation sites per sample). We identified one patient with the highest number of iSNVs, despite harboring an elevated anti-RBD IgG level. During his CCU stay, a viral load rebound associated with a decrease of PaO₂/FiO₂ ratio was observed without an associated increase of the IFN response. This patient’s viral load exceeded 4 logs 8 days after CCU admission, a level typically associated with an elevated IFN-I score in most critically ill COVID-19 patients (3). Plasma obtained from this patient had a high neutralizing capacity against IFN-β, indicative of the presence of IFN-β autoantibodies as previously identified in critical COVID-19 patients (2). This could explain the apparent absence of an IFN-I response. Of note, although anti–IFN-β autoantibodies are much less prevalent and less frequently associated with severe SARS-CoV-2 infections than anti–IFN-α and -ω autoantibodies (2), this case report highlights the role of those autoantibodies in the pathophysiology of SARS-CoV-2 infections.

Previous studies reported that SARS-CoV-2 infection in immunocompromised patients treated with remdesivir can lead to the emergence of viral variants harboring transmissible mutations conferring reduction in susceptibility to an RNA-dependent RNA polymerase (RdRP) inhibitor (https://covdb.stanford.edu/drms/rdrp/) (4). However, in both the consensus sequence and the iSNV, no mutation that is known to confer resistance to RdRP inhibitors was detected in patient#2, suggesting that the presence of anti–IFN-β autoantibodies, throughout uncontrolled replication, could also contribute to this variability. Moreover, vaccination has not been associated with an increase in the emergence of SARS-CoV-2 variants (1), suggesting that the patient’s vaccination status and the high level of anti-RBD level does not explain the high iSNV rate. Although heterotypic co-infection during the CCU stay could explain the high iSNV count, we are unable to test this hypothesis due to absence of eligibility criteria for prior sample obtained from patient#2.

In conclusion, host factors contributing to intra-host variant emergence are multiple, and we underlined that impairment of IFN-I response could be a significant contributor.

The INTERFERICUS cohort is a prospective multicenter study investigating the impact of IFN-I autoantibodies on viral load in critically ill patients. The trial was registered on https://ClinicalTrials.gov (NCT05536219), where inclusion and exclusion criteria are detailed. The study protocol was approved by a research ethics committee (Comité de Protection des Personnes Sud-Ouest et Outre-Mer II, 09/27/2022) under the IDCRB_2022-A01824-39, and opposition to use of personal data was sought from the patients or their representative.

NPS collected from patients with acute respiratory failure induced by SARS-CoV-2 were selected for the present study. The SARS-CoV-2 load was determined from the NPS using SARS-CoV-2 R-gene kit (bioMérieux) as previously described (3). All NPS positive for SARS-CoV-2 with a cycle threshold ≤28 were sequenced in duplicate, using the Artic V4.1 NCOV-2019 Panel as previously described (5). Samples with ≥1,000,000 read count, ≥90% coverage of the consensus sequence, and mean depth at each position ≥500 were included in the present study. A threshold of 5% minor allele frequency for both replicates was used to identify iSNVs.

RNA from NPS was extracted in duplicate using the automated MGISP-960 workstation using MGI Easy Magnetic Beads Virus DNA/RNA Extraction Kit (MGI Tech) for the first extract and EMAG platform (bioMérieux) for the second extract. SARS-CoV-2 Artic V4.1 primers were tested on duplicate extracts from NPS at 10 µM final concentration. cDNA synthesis, amplification, and libraries were prepared using the COVIDSeq-Test (Illumina). Samples were sequenced with 100-bp paired-end reads using the NovaSeq 6000 Sequencing system SP flow cell.

Although contamination during sequencing could introduce artefacts, the quality controls (negative template) and the duplicate conditions performed minimize this possibility and cannot explain the elevated iSNV rate.

The IFN-I response was assessed using a FilmArray IFN-I pouch prototype, allowing the measurement of four IFN-stimulated genes and three housekeeping genes for signal normalization (this prototype has not been submitted to any regulatory agency for review at the time of writing). A total of 100 μl of diluted NPS was tested using the prototype IFN-I pouch, and the nasal IFN-I score was calculated as described previously (3).

The blocking activity of anti–IFN-α2, -β and -ω AAbs was determined using a HEK293T reporter cell line constructed to react specifically to IFN-I by induction of luciferase synthesis. Cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco) supplemented with fetal bovine serum at 10%. Sera were decomplemented (15 min/56°C). Subsequently, sera diluted 1/10 in duplicate were added to IFN-α2 (ref_130-093-874; Miltenyi), IFN-β (ref_300-02BC; PeproTech), or IFN-ω (ref_SRP3061; Sigma-Aldrich) at whether 10 or 1 ng/ml concentration to HEK293T cell (1.7 × 10⁵ cells/ml) in 96-well plate. After 24 h of incubation (37°C, 5% CO₂), the firefly luciferase substrate (Bright-Glo, Promega) was added into the wells (100 μl), and luciferase activity was measured at least 5 min later on BioTek Synergy HTX (Agilent) using a 100-ms integration time.

Anti-RBD IgG titers were measured using the VIDAS SARS-CoV-2 IgG diagnostic kit (ref_423834; bioMérieux) and expressed in BAU/ml using the conversion factors provided by the manufacturer.

Quantitative variables were expressed as median [IQR] and qualitative variables as number (percentage). Statistical analyses were performed using R version 4.3.2.

Requests for data should be made to the corresponding author. The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. The sequencing data are available under the BioProject accession number PRJNA1189494.

We thank the Clinical Research Associates for their excellent work. We thank all the technicians from the Hospices Civils de Lyon Biobank (BB-0033-00046) whose work made it possible to obtain all these data. We also thank Philip Robinson (Direction Recherche en Santé [DRS], Hospices Civils de Lyon) for his help in manuscript preparation. Lastly, we thank all the participants, patients, and their family for their participation to the study.

This study received funding from the Agence Nationale de la Recherche (ANR) (ANR-RHU Program ANR-21-RHUS-08 [COVIFERON], ANR-23-CHIN-0001) through France 2030 program.

Author contributions: D. Bussy: data curation, formal analysis, investigation, methodology, project administration, validation, visualization, and writing—original draft, review, and editing. K. Trepat: investigation, visualization, and writing—original draft, review, and editing. H. Regue: data curation, formal analysis, and software. V. Cheynet: investigation. L. Generenaz: investigation. L. Chauvelot: conceptualization, investigation, resources, validation, and writing—review and editing. J.-C. Richard: conceptualization, investigation, methodology, and writing—review and editing. INTERFERICUS study group: investigation, resources. A. Bal: conceptualization, investigation, supervision, validation, visualization, and writing—review and editing. S. Trouillet-Assant: conceptualization, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, and writing—original draft, review, and editing.