Immunophenotyping assessment in a COVID-19 cohort (IMPACC): A prospective longitudinal study

Sample Collection and Processing

Sample collection was designed to meet minimal risk guidelines for blood collection for hospitalized adults, and sample-sparing assays are employed when feasible. Blood samples (10 ml per timepoint) and nasal swabs (mid-turbinate) are collected at each specified timepoint and blood is processed within six (6) hours of collection according to the IMPACC standardized operating procedure. Whole blood and peripheral blood mononuclear cells (PBMCs) are collected in order to identify distinct immune cell populations and quantify changes in cell populations, gene expression, and activation markers (e.g., Cytometry by Time of Flight (CyTOF), bulk RNA transcriptomics) over the course of COVID-19 and convalescence. DNA is collected from whole blood at a single timepoint for genetic analyses (e.g., whole exome sequencing). Serum is used to characterize SARS-CoV-2 specific antibodies, including virus neutralization, both serum and plasma are used for proteomics and metabolomics, and plasma is used to measure soluble inflammatory mediators (e.g., cytokines, chemokines) using oligonucleotide-linked antibody detection (Olink). RNA from the nasal swab is used to assess SARS-CoV-2 viral load and genomic sequence, and to evaluate changes in immune-related upper airway epithelial gene expression (i.e., bulk transcriptomics). Additionally, endotracheal aspirates (EAs) are collected from intubated patients and processed within two (2) hours of collection according to the IMPACC standardized operating procedures. EA cells are assessed by CyTOF and bulk transcriptomics to identify and quantify changes in gene expression and activation state of distinct immune cell populations in the lower respiratory tract. Processed samples are barcoded and centrally tracked on a laboratory information management system (LDMS, Frontier Science). All supplies necessary for sample collection and sample processing are centrally procured and supplied to the participating sites. Sample collection, processing, and storage procedures (Fig. 3) are standardized across sites and samples are transported to centralized core laboratories in batches for testing and analysis. The complete sample processing manual of procedures is included as Supplementary Methods.

Core Laboratories and Technologies

IMPACC is using a systems immunology approach for simultaneous immune profiling on the same patient sample, using a wide range of assays, during COVID-19 disease and resolution. Profiling assays were chosen that are sample sparing and provide a comprehensive, unbiased assessment of immunologic changes in the airways and circulation during disease progression and resolution (Fig. 3). Core immune profiling laboratories have been established for in-depth sample analysis, using optimized assays. The Core Labs work in close collaboration with the IMPACC clinical sites and their respective sample processing labs to ensure uniformity in sample collection and processing using the standardized manual of procedures. The shared methods and reagents promote high-quality assessment across all sample types. Each assay is performed at a single expert Core Lab or two harmonized Core Labs using rigorous standardized procedures, validated instruments and reagents, relevant controls, sample and batch randomization, assay timetables, and data sharing. The Core Laboratory assays have been chosen to provide comprehensive immune assessment while minimizing the amount of sample needed per assay.

Virus Detection by Real-Time Polymerase Chain Reaction (SARS CoV-2 RT-PCR) Assay

Viral shedding is increased and prolonged in severe compared to mild COVID-19 (8), but the precise relationship between viral kinetics and key clinical outcomes is not fully understood (9). Relating longitudinal kinetics of viral shedding to both host immune responses and clinical outcomes will provide important insights into mechanisms of severe disease. For patient comfort and consistency of collection, mid-turbinate swabs (rather than nasopharyngeal) are collected for serial viral quantification. Swabs are collected and placed in 1 ml of Zymo-DNA/RNA shield reagent (Zymo Research) and shipped to the Core lab at Benaroya Research Institute (BRI). RNA is extracted using the quick DNA-RNA MagBead kit (Zymo Research) following the manufacturer’s instructions. RT-PCR for SARS-CoV-2 is performed on RNA extracts using the SARS-CoV-2 (2019-nCoV) CDC qPCR Probe Assay with target genes 2019-nCoV_N1, 2019-nCoV_N2, and control human RNase P (Integrated DNA Technologies) (10, 11). Viral levels will be modeled longitudinally in relation to the kinetics of host immune responses and clinical outcomes.

Viral Sequencing

During the global spread of SARS-CoV-2, several dominant lineages have emerged that are characterized by distinct mutation patterns and are tracked in related nomenclatures by Global Initiative on Sharing All Influenza Data (GISAID) (12) and others (13). Although most genetic changes in SARS-CoV-2 are expected to be neutral, selected mutations alter viral properties such as ability to infect cells or evade host antibody responses (14–16). Treatment with antiviral drugs can also create selective pressure for the emergence of variants that escape drug effects. The availability of complete viral genomes for all participants will allow an assessment of host responses and outcomes in the context of specific SARS-CoV-2 strains and mutations. SARS-CoV-2 viral load will be assayed directly from mid-turbinate swab samples at each timepoint, and viral sequencing will be performed for the first nasal swab sample per participant with sufficient quantity of virus, at the Icahn School of Medicine at Mount Sinai (ISMMS; New York, NY). The team will use a combination of tiling primer designs to specifically amplify the SARS-CoV-2 genome and generate paired-end 2×150 nt sequencing data on the Illumina platform to an average depth of >1,000-fold. If needed, supplemental data may be generated on other platforms such as Ion Torrent to maximize the ability to obtain complete genomes for patients with low viral loads. The availability of deep sequencing data from each genome will also facilitate analyses of intra-host variants.

Serology

Antibody responses to SARS-CoV-2 are thought to be beneficial as they can neutralize viral entry and clear infected cells through effector functions. Antibodies may also protect from reinfection, although this hypothesis is still under investigation. However, the timing and magnitude of the antibody response have been linked to disease severity (9) and certain types of antibody responses may potentially be harmful as has been hypothesized with regard to disease-enhancing antibodies in the context of dengue infection (17).

The IMPACC Serology Core lab at ISMMS is quantifying the antibody response, including different isotypes, to the Spike protein of SARS-CoV-2 and against the receptor binding domain (RBD), which is the part of the Spike protein that interacts with angiotensin-converting enzyme 2 (ACE2) on host cells. This analysis employs well-established enzyme-linked immunosorbent assays (ELISA) (18, 19). ELISA titers will be reported as endpoint titers. Anti-spike, and especially anti-RBD antibodies, have been linked to virus neutralization. To assess the functional specificity of the antibody response, neutralization assays with authentic SARS-CoV-2 are being conducted (20). The readout for this assay is the 50% inhibitory dilution (ID₅₀), which is calculated based on virus neutralization compared to a negative control.

In addition to protein-based ELISAs, the IMPACC serology Core Lab at UCSF is measuring serum antibody responses with a programmable phage display library (i.e., VirScan (21)) containing 38 amino acid overlapping peptides tiling across the SARS-CoV-2 and other human coronavirus (HuCoV) proteomes, including SARS-CoV-1 (NC_004718), beta coronavirus England 1 (NC_038294), HuCoV 229E (NC_002645), HuCoV HKU1 (NC_006577), HuCoV NL63 (NC_005831), HuCoV OC43 (NC_006213), Infectious Bronchitis virus (NC_001451), and MERS CoV (NC_019843) (22). The VirScan assay measures the specificity of antibodies induced by SARS-CoV-2 infection, as well as HuCoV antibodies present early in the disease course likely as a result of past infections with other HuCoVs. These data may help determine whether the presence of pre-existing HuCoV antibodies affects COVID-19 disease severity and whether specific aspects of the adaptive immune response to SARS-CoV-2 infection correlate with disease outcome. The oligonucleotides are cloned into T7 bacteriophage so that the viral peptides are displayed on the phage surface. The phage display library is then incubated with patient sera, and immunoglobulins are immunoprecipitated by magnetic protein A and G beads along with antibody-bound phage, which are sequenced to generate viral peptide counts whose fold enrichment is calculated relative to bead-only negative controls and pre-pandemic healthy control sera.

Serum Cytokines

Selected pro-inflammatory cytokines in serum or plasma correlate with or even predict disease severity in COVID-19 (1, 23, 24). In order to comprehensively measure serum cytokines in a high-throughput manner for the IMPACC study, the Olink Inflammation Panel is utilized at the co-Core Labs at ISMMS and Stanford University. The Olink multiplex immunoassay offers the advantage of being a more comprehensive predictor of biological mechanisms at play than single cytokines (1). This method uses proximity extension assay (PEA) technology, whereby two antibodies for each target protein are conjugated to complementary oligonucleotides; if the correct antibody pair binds to the same target molecule, there is annealing of the oligos, and a PCR template can be created by extension and then dissociation of the extended product. Quantitative PCR is then carried out on the Fluidigm Biomark microfluidic platform. This platform allows for rapid setup of 96 samples × 96 reactions, with a minimal sample requirement, nominally 1 microliter per sample. The Inflammation panel consists of 92 analytes including pro- and anti-inflammatory cytokines, chemokines and related molecules (https://www.olink.com/products/inflammation).

Proteomics and Metabolomics

The blood serves a major role in modulating and distributing the immune responses throughout the entire body. Thus, determining how soluble immunomodulatory molecules are affected by SARS-CoV-2 infection (25) and recovery from COVID-19 is essential for a comprehensive understanding of immunophenotypes. To this end, selective quantitative maps of the plasma proteome and metabolome are being acquired by the Proteomics/Metabolomics Core (PMC) at Boston Children’s Hospital. To support the overarching goal of generating well-founded hypotheses to inform future research, unbiased LC-MS methods are being employed for proteomics and metabolomics. Of note, our integrated sample-sparing proteomic and metabolomic work-flow has been successfully employed in populations with limited blood volumes (26).

The plasma proteomics analysis will take a two-pronged approach (27). First, the plasma proteome will be quantitatively mapped without any depletion in light of the important immunomodulatory roles of a significant fraction of the standard depletion targets such as immunoglobulins and complement pathway components (28, 29). Mapping the COVID-19-associated changes in abundance for these proteins is important for the comprehensiveness of the immunophenotyping efforts of IMPACC. This analysis will be performed in a high-throughput fashion using a triple quadrupole mass spectrometer operated in multiple reaction mode. We target 300 of the immunologically most relevant plasma proteins using a fast and highly sensitive state-of-the-art triple quadrupole mass spectrometer (LCMS-8060, Shimadzu). Then, the most abundant plasma samples will be depleted using biochemical methods that can be conducted in a high-throughput and cost-efficient manner on thousands of samples (30, 31). The resulting depleted plasma samples are processed for analysis by LC-MS in discovery mode using a high throughput sample delivery and PLC system (Evosep One) front-end and a Bruker ion mobility/quadrupole/time-of-flight mass spectrometer (timsTOF Pro) back-end to ensure robustness.

The plasma metabolomics for the IMPACC study also follows a two-pronged approach. First, discovery metabolomics will be conducted in collaboration with Metabolon using reverse-phase LC-MS/MS in positive ion mode, reverse-phase LC-MS/MS in negative ion mode, and hydrophilic interaction liquid chromatography (HILIC) LC-MS/MS in negative ion mode (32). In the subsequent step, select subsets of metabolites and metabolite families, deemed to be of relevance based on the discovery experiments, will be precisely quantified in a targeted fashion by HILIC LC/MS using high accuracy/high resolution Orbitrap mass spectrometers (Q Exactive).

Transcriptional Profiling (Bulk RNA-seq)

Transcriptional profiling is a powerful approach to identify biomarkers and mechanisms of immune-mediated diseases (33, 34). Transcriptional profiling accurately reflects both dynamic changes in cellular composition and cellular response during the course of disease. Furthermore, network analysis approaches including cell deconvolution (35) and modular analysis (36) have been developed as robust computational approaches to unravel biologically coherent and insightful signatures across infectious and immunologic diseases, and are particularly powerful approaches for longitudinal analyses (37–40). Transcriptional profiling of patients with COVID-19 has shown alterations in interferon responses and inflammatory pathways which may relate to disease outcomes (35–38), sparking interest in immunomodulatory treatments directed at such pathways. The IMPACC network includes bulk transcriptomic analysis of upper (nasal – BRI Core Lab) and lower airway (EA – UCSF Core Lab) and PBMCs (Emory University and UCSF Co-Core Labs).

Airway Bulk RNA-seq

RNA extracted from nasal and EA specimens is DNase treated and human cytosolic and mitochondrial ribosomal RNA are depleted. cDNA synthesis employs a random hexamer approach to capture human coding and non-coding RNA transcripts, as well as non-human RNAs. cDNA libraries are sequenced with paired-end reads at a target depth of 50 million reads per sample using NovaSeq S4 200 cycle flow cells. Human reads are aligned to the GRCh38 reference genome and quality controlled by total counts per library and median CV coverage. Raw counts for these genes are normalized across libraries according to the “Trimmed Means of M Values” (TMM) method (41), as implemented in the edgeR package for downstream analysis. Remaining non-human reads are aligned using the NCBI nucleotide and non-redundant protein databases followed by assembly of the reads matching each taxon detected. Repeated control and participant samples are sequenced within each sequencing batch to mitigate batch effects throughout the study.

PBMC Bulk RNA-seq

RNA is extracted from PBMCs lysed in Qiagen RLT buffer using the Zymo Quick-RNA MagBead Kit on an automated liquid handling system in batches designed to balance covariates across library preparation runs. Library preparation is performed using the TECAN NuGen Universal Plus mRNA-seq KIT in combination with the Qiagen FastSelect hemoglobin and ribosomal depletion kit to produce stranded, poly(A)-enriched mRNA-seq libraries depleted of ribosomal and hemoglobin RNA. Libraries are normalized utilizing the output of shallow QC sequencing runs and pooled using appropriate dilution ratios. The targeted read depth for each sample is at least 25 million reads per sample using NovaSeq S4 200 cycle flow cells and a 100 bp, paired-end read length. Reads are aligned to a composite reference of the GRCh38 human genome and SARS-CoV-2 genome. Gene counts are generated internally with STAR, and alignments are run through RSEM for comparison and alternate abundance metrics; in parallel, Kallisto will be run to produce Transcript per Million (TPM) values (42–44). Repeated measures of healthy control and participant samples are included to track inter-site batch effects; Universal Human References RNA (UHRR) controls to assess intra-site variation, and reference PBMCs stimulated with TLR7 agonist will be included to assess intra-site batch sensitivity.

Mass Cytometry — Blood

Single cell technologies have provided techniques that can resolve disease in humans at an unprecedented level of detail, capturing the clinical and biological heterogeneity of disease. Mass cytometry or CyTOF employs rare metal isotope-conjugated antibodies for high dimensional single-cell analysis. By using heavy metal ions as labels and detection in a time-of-flight mass spectrometer, up to 50 single cell parameters can be measured simultaneously with little/no background and minimal signal overlap between channels, providing unprecedented multidimensional cell profiling. CyTOF has been applied in suspension to characterizing immune cells in autoimmunity, cancer, and infection (45).

The CyTOF workflow implemented for this study has been designed with several specific considerations to reduce sample usage, streamline sample processing and minimize experimental variability (46). Whole blood samples are first stained at the site of collection using a commercial lyophilized 30-marker panel designed to identify all major circulating immune cell subsets (Fluidigm Maxpar Direct Immune Profiling Assay, see Table S1) and are then fixed and cryopreserved (Smart Tube buffer). The cryopreserved samples are shipped to the IMPACC co-Core labs at ISMMS and Stanford for barcoded batched processing, where they are labeled with a supplemental panel of 14 additional antibodies targeting fixation-resistant epitopes to resolve additional dynamic changes in cell phenotype. To maximize reproducibility, the supplemental panel has been formulated as a cocktail and frozen in single-use aliquots for each processing batch. The labeled barcoded samples are frozen for batched acquisition. The resulting FCS files are evaluated using a centralized data processing pipeline including bead-based sample QC and data normalization and automated sample demultiplexing.

Mass Cytometry — Endotracheal Aspirates

CyTOF has recently been employed to analyze cells in induced sputum and numerous studies have validated this platform for multiparameter profiling of single cells from heterogeneous populations (47). The Yale Core lab employs CyTOF on endotracheal aspirates of COVID-19 patients to provide a higher resolution understanding of the inflammatory responses in the affected tissue.

Endotracheal aspirates (EA) from patients who require invasive mechanical ventilation are collected at the same time points as other samples. Saline is instilled (10 cc) to collect the aspirate in a 40 cc Argyle specimen trap. To maximize cell viability, the aspirate is processed within 2 hours of collection. Cells are passed through a series of filters to isolate a single cell suspension for labeling by CyTOF. For optimal detection of markers, surface antigens are labeled on fresh cells with a premade batch-prepared antibody cocktail prior to freezing at -80°C. To reduce variation, remaining intracellular labeling is conducted on batches of samples together at the Yale University IMPACC EA CyTOF Core Lab. Antibody labeling of EA includes spiked-in reference cells (48) and markers to define PMN, monocytes, dendritic cells, NK cells, subsets of T and B lymphocytes, and 15 intracellular markers to quantify functional status (see Table S2). CyTOF files from aspirates are normalized using the same centralized data processing pipeline and QC pipeline used for the CyTOF whole blood samples, followed by a standard gating strategy for airway cells (49). EA samples are analyzed with CyTOF data from whole blood samples of the same subjects, in concert with the other CyTOF Core teams.

Genomic Analysis

The mechanisms by which an infection leads to severe disease in a subset of all infected individuals is incompletely explained. Immune responses to infection can differ based on both rare and common genetic variations (50). To identify any genomic determinants of severe COVID-19 disease, the Yale IMPACC Core lab is conducting whole exome sequencing and SNP genotyping and assessing genetic variants associated with individual susceptibility to severe disease. The DNA sequencing of IMPACC study subjects includes whole exome sequencing (WES) to include 19,433 genes that are in the RefSeq coding sequences, xGen exome capture, and whole-genome genotyping at 1.9M SNP sites on the Illumina Infinium®️ Global Diversity Array (GDA). For WES, genomic DNA will be extracted from frozen whole blood of each enrolled subject with sample quality determined by spectroscopic and fluorometric methods. High quality DNA will be sheared for automated library construction incorporating unique dual indices for each sample followed by hybridization-based enrichment of the exome. Pooled libraries will be sequenced on Illumina NovaSeq6000 S4 flow cells using optimized conditions for concentrations to maximize unique read output while limiting duplicates using paired-end sequencing chemistry and a read length of 101 bases. Following real time analysis on Illumina’s CASAVA 1.8.2 software suite for converting signal intensities to individual base calls and completion of the run, raw data are evaluated for quality and samples are de-multiplexed. Individual sample level alignment to the human genome, variant calling and annotation enable downstream analyses. Whole-genome genotyping will be performed following manufacturer’s recommendations. Sequencing and array data will be available as fastq files to the analysis team accompanied by common variant association tests and rare variant gene burden tests for outcomes. Genetic sequence data will estimate population stratification and relatedness in our samples as covariates in other analyses.