Patients and sample collection
A total of 590 respiratory samples (thrice a week) and 432 serum samples (twice a week) were obtained from 150 COVID-19 patients with informed consent hospitalized in the Leiden University Medical Center (LUMC), from March 2020 to Dec 2020. The study was approved by the Medical Ethical Committee Leiden Delft Den Haag (NL73740.058.20) and registered in the Dutch Trial Registry (NL8589). From each patient the daily disease severity was scored. The severity score (range 0–17) includes the following parameters: respiratory rate, peripheral oxygen saturation on room air, p/f ratio2, oxygen flow, FiO2, Glasgow coma scale score, urea, and C reactive protein (Extended data Table 5).
SARS-CoV-2 RT-PCR and definition of viral clearance
After extraction of nucleic acids from 200 µl sample using a MagNa Pure 96 instrument (Roche Diagnostics), ten microliters extract was used for SARS-CoV-2 E-gene detection by real-time reverse-transcription PCR on a CFX96 PCR instrument (Bio-Rad): 50˚C for 5 min, followed by 95˚C for 20 s and then 45 cycles of 106 95˚C for 15 s, 55˚C for 10 s, and 72˚C for 50 s using primers and probe described by Corman et al (2020). Viral clearance was defined as the last positive PCR available (Ct > 40), either ≤ 21 or > 21 days post onset of symptoms, which was considered delayed viral clearance25.
SARS-CoV-2 antibody assays
Semi-quantitative detection of SARS-CoV-2 anti-nucleocapsid (N) protein IgG and anti-RBD of the S protein IgM antibodies was performed on the Abbott Architect platform26,27. In this antibody chemiluminescent microparticle immunoassay (CMIA) test, the SARS-CoV-2 antigen coated paramagnetic microparticles bind to the IgG respectively IgM antibodies that attach to the viral nucleocapsid protein in human serum samples. The Sample/Calibrator index values of chemiluminescence in relative light units (RLU) of 1.40 (IgG assay) respectively 1.00 (IgM assay) and above were considered as positive per the manufacturer’s instructions.
SARS-CoV-2 IgG antibody responses against the N-terminal (N-NT, position amino acids 1-246) and C-terminal (N-CT, position amino acids 181–419) antigen were analyzed in a microparticle immunoassay. N full-length of human coronavirus 229E was included as a specificity control (N-229E). In brief, as previously described previously28,29, viral gene fragments (gBlocks; IDT, San Jose, CA, USA) were cloned into pGEX-5x-3 vector (GE Healthcare Life Sciences, Chicago, IL, USA) expressed in E.coli and coupled to independent colour-coded magnetic bead sets (Bio-rad Laboratories, Hercules, CA, USA), to allow distinction between the antigens. Serum samples, 1:100 diluted, were pre-incubated with the antigen bead-mix for 1 hr in blocking buffer and subsequently incubated with the antigen-coated bead sets for 1 hr. For detection, biotinylated goat anti-human IgG (H + L) was used, followed by streptavidin-R-phycoerythrin. The phycoerythrin signal on each bead set was analyzed in a Bioplex 200 analyzer (Bio-Rad Laboratories, Hercules, CA, USA) and expressed as median fluorescence intensity (MFI).
Quantitative detection of SARS-CoV-2 anti-S1/S2 IgG antibodies was performed using the DiaSorin LIAISON platform. The CLIA assay consists of paramagnetic microparticles coated with distally biotinylated S1 and S2 fragments of the viral surface spike protein. RLUs proportional to the sample’s anti-S1/S2 IgG levels are converted to arbitrary units (AU)/milliliter based on a standardized master curve.
Semi-quantitative detection of SARS-CoV-2 anti-RBD IgM antibodies was performed using the Wantai IgM-ELISA (CE-IVD) kit (Sanbio)30. Briefly, the IgM u-chain capture method was used to detect IgM antibodies using a double-antigen sandwich immunoassay using mammalian cell-expressed recombinant antigens containing the RBD of the spike protein of SARS-CoV-2 as the immobilized and horseradish peroxidase-conjugated antigen. Sample/Cut-off index OD values of 1 and higher were considered positive per the manufacturer’s instructions.
Semi-quantitative detection of SARS-CoV-2 anti-S1 IgA antibodies was performed using the Euroimmun IgA 2-step ELISA31. Ratio values of 1.1 and higher were considered positive per the manufacturer’s instructions.
Virus neutralization assay
Neutralization assays against live SARS-CoV-2 were performed using the microneutralization assay previously described by Algaissi and Hashem32. Vero-E6 cells [CRL-1580, American Type Culture Collection (ATCC)] were grown in Eagle’s minimal essential medium (EMEM; Lonza) supplemented with 8% fetal calf serum (FCS; Bodinco BV), 1% penicillin-streptomycin (Sigma–Aldrich, P4458) and 2 mM L-glutamine (PAA). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO2. Clinical isolate SARS-CoV-2/human/NLD/Leiden-0008/2020 was isolated from a nasopharyngeal sample and its characterization will be described elsewhere (manuscript in preparation). The next-generation sequencing derived sequence of this virus isolate is available under GenBank accession number MT705206.1 (https://www.ncbi.nlm.nih.gov/nuccore/1864563703) and shows one mutation in the Leiden-0008 virus spike protein compared to the Wuhan spike protein sequence resulting in Asp > Gly at position 614 (D614G) of the Spike protein. In addition, several non-silent (C12846U and C18928U) and silent mutations (C241U, C3037U, and C1448U) in other genes were found. Isolate Leiden-0008 was propagated and titrated in Vero-E6 cells using the tissue culture infective dose 50 (TCID50) endpoint dilution method and the TCID50 was calculated by the Spearman-Kärber algorithm as described33. All work with live SARS-CoV-2 was performed in a biosafety level 3 facility at Leiden University Medical Center.
Vero-E6 cells were seeded at 12,000 cells/well in 96-well tissue culture plates 1 day prior to infection. Heat-inactivated (30 min at 56°C) serum samples were analysed in duplicate. The panel of sera were two-fold serially diluted in duplicate, with an initial dilution of 1:10 and a final dilution of 1:1280 in 60 µL EMEM medium supplemented with penicillin, streptomycin, 2 mM L-glutamine and 2% FCS. Diluted sera were mixed with equal volumes of 120 TCID50/60 µL Leiden − 0001 virus and incubated for 1 h at 37°C. The virus-serum mixtures were then added onto Vero-E6 cell monolayers and incubated at 37°C in a humidified atmosphere with 5 % CO2. Cells either unexposed to the virus or mixed with 120 TCID50/60 µL SARS-CoV-2 were used as negative (uninfected) and positive (infected) controls, respectively. At 3 days post-infection, cells were fixed and inactivated with 40 µL 37% formaldehyde/PBS solution/well overnight at 4°C. The fixative was removed from cells and the clusters were stained with 50 µL/well crystal violet solution, incubated for 10 min and rinsed with water. Dried plates were evaluated for viral cytopathic effect. Neutralization titer was calculated by dividing the number of positive wells with complete inhibition of the virus-induced cytopathogenic effect, by the number of replicates, and adding 2.5 to stabilize the calculated ratio. The neutralizing antibody titer was defined as the log2 reciprocal of this value. A SARS-CoV-2 back-titration was included with each assay run to confirm that the dose of the used inoculum was within the acceptable range of 30 to 300 TCID50.
Circulating leukocyte counts by flow cytometry
Absolute counts of the main circulating leukocyte subsets (Extended data Fig. 3) were obtained based on an adapted standard protocol for peripheral blood sample processing for flow cytometry measurement (for detailed protocol see www.EuroFlow.org 34) using an optimized combination of markers for surface staining (Primary Immunodeficiency Orientation Tube: PIDOT, Cytognos, Salamanca, Spain, https://www.cytognos.com/products/pidot-primary-immunodeficiency-orientation-tube35,36).
For a better separation of the circulating plasmablasts, CD38 was added to the PIDOT combination (Extended data Table 6 for details of antibody clones used). In summary, the procedure consisted in the bulk lysis of erythrocytes in fresh (same day) peripheral blood samples and staining of 2,5*106 leukocytes with reconstituted PIDOT lyophilized antibody cocktail (containing CD8 FITC, IgD FITC, CD16 PE, CD56 PE, CD4PerCPCy5.5, CD19 PeCy7, TCRgd PE-Cy7, CD3 APC, CD56 APC-C750) and drop in antibody cocktail (containing per test: 2 µL CD27 BV421, 2,5 µL CD45RA BV510, 2 µL CD38 BV605, 0,6 µL pure CD38) (Extended data Table 6) in a final 100 µL staining volume. The data (at least 1 million events) were acquired on a 3-laser Cytek®Aurora instrument (Cytek Biosciences, Fremont, CA, USA) at the Flow cytometry Core Facility (FCF) of Leiden University Medical Center (LUMC) in Leiden, Netherlands (https://www.lumc.nl/research/facilities/fcf). For data analysis, the Infinicyt software (Cytognos SL, Salamanca, Spain) was used. The cell subtypes identified and their expression profiles reflecting the gating strategy used are presented in Extended data Fig. 3. The absolute counts per µL fresh blood were determined by a double platform approach, using the absolute fresh leukocyte counts determined prior sample processing with hematological analyzer (Sysmex) to the Statistics Configure tool of the Infinicyt software.
SARS-CoV-2 specific T cells
Blood was collected in CPT tubes (BD Biosciences, cat#362753) from which PBMCs were isolated using Ficoll-Isopaque. After cryopreservation, 1x106 PBMCs were cultured in IMDM (Lonza, cat#BE12-722F) supplemented with 10% FCS (Sigma Aldrich, cat#F7524), 1.4% L-glutamine (Lonza, cat#BE17-605E), and 1% Pen/Strep (Lonza, cat#DE17-602E). T cells were stimulated with 1 µg/mL SARS-CoV-2 peptide pool covering nucleocapsid (Miltenyi, cat#130-126-699), membrane (Miltenyi, cat#130-126-703), and immunodominant regions of the spike protein (Miltenyi, cat#130-126-701). 1% DMSO (Merck, cat#1029311000) or 1µg/mL CMV pp65 peptide pool were used as negative and positive control, respectively. After one hour incubation, 5 µg/mL Brefeldin A (Sigma Aldrich) was added and after 16-hour incubation, the PBMCs were stained using Zombie-Red (biolegend, cat#423110), CD4-Pe-Cy7 (Beckman Coulter, cat#737660), and CD8-APC-H7 (BD Biosciences, cat#560179) for 30 minutes at 4°C. This step was followed by an 8 minute 1% paraformaldehyde fixation at room temperature and subsequent permeabilization with 0.1% saponin (Sigma Aldrich) for 20 minutes at 4°C. Intracellular staining was performed with an antibody mix containing the following antibodies: CD137-APC (BD Pharmingen, cat#550890), CD154-Pacific Blue (Biolegend, cat#310820), and IFNγ-BV711 (BD Biosciences, cat#564039) in 0.1% saponin for 30 minutes at 4°C. After staining, the cells were resuspended in 0.1% saponin and measured on a 5-laser Cytek Aurora. The analysis was performed using FlowJo V10.7.1. (BD Biosiences), in short: The lymphocytes were gated based on FSC-A and FSC-H. Zombie-Red negative cells were considered alive and T cell subsets were defined by expression of CD4 or CD8. Finally, activated CD4 + and CD8 + T cells were gated based on CD154 + CD137 + of total CD4 + and CD137 + IFNγ + of total CD8 + T cells. SARS-CoV-2 CD4 + or CD8 + frequencies were calculated by subtracting the background (DMSO) and taking the sum of the three stimulations (=(activated in S% - activated in DMSO%) + (activated in M% - activated in DMSO%) + (activated in N% - activated in DMSO%)).
Cytokine/chemokine measurements
Cytokines and chemokines were measured in serum by bead based multiplex assays using the BioPLex 100 system for acquisition as previously described37. Standard curves were provided with kits and a pooled sample of 4 Covid-19 patients was included as internal reference in all assays. Four commercially available kits were used Bio-Plex Pro™ Human Cytokine Screening Panel 48-plex; Bio-Plex Protm Human Chemokine Panel 40-Plex; Bio-Plex Protm Human Inflammation Panel 1, 37-
Plex; Bio-Plex Protm Human Th17 panel (IL-17F, IL-21, IL-23, IL-25, IL-31, IL-33) (all from Bio-Rad, Veenendaal, The Netherlands).
Statistical analysis
Median with inter-quartiles range (IQR) were used to report continuous variables visualized in violine plots. Groups (early versus late clearance, non-ICU versus ICU) were compared with non-parametric Mann-Whitney U test. Correlations were analysed using Pearson’s correlation coefficient. Imputation for missing data (death or transfer within 21 days) was not performed, cases discharged to home within 21 days were considered early clearance and included in rank correlation analyses. Tests with p-values ≤ 0.05 were considered statistically significant. Hierarchical clustering of correlation values were produced for all immune parameters (maximum level per patient) in relation to each other, and for the groups early versus delayed viral clearance. Multiple testing correction of confidence values was done using false discovery rate (FDR, Benjamini–Hochberg correction38). Statistical analysis were performed using statistical libraries of R (Pearson correlation coefficients, heatmaps with hierarchical clustering, violine plots) and Perl (Mann-Whitney U test).