key: cord-0310758-fohl5qfa
authors: Jackson, H.; Rivero Calle, I.; Broderick, C.; Habgood-Coote, D.; D'Souza, G.; Nichols, S.; Gomez Rial, J.; Rivero-Velasco, C.; Rodriguez-Nunez, N.; Barbeito-Castineiras, G.; Perez-Freixo, H.; Barreiro-de Acosta, M.; Cunnington, A.; Herberg, J.; Wright, V.; Gomez-Carballa, A.; Salas, A.; Levin, M.; Martinon-Torres, F.; Kaforou, M.; PERFORM consortium,; group, GEN-COVID study
title: Characterisation of the blood RNA host response underpinning severity in COVID-19 patients
date: 2021-09-21
journal: nan
DOI: 10.1101/2021.09.16.21263170
sha: 5b7ee524a75eb7eb6bf89bc7415654d10b5b9746
doc_id: 310758
cord_uid: fohl5qfa

Infection with SARS-CoV-2 has highly variable clinical manifestations, ranging from asymptomatic infection through to life-threatening disease. Host whole blood transcriptomics can offer unique insights into the biological processes underpinning infection and disease, as well as severity. We performed whole blood RNA Sequencing of individuals with varying degrees of COVID-19 severity. We used differential expression analysis and pathway enrichment analysis to explore how the blood transcriptome differs between individuals with mild, moderate, and severe COVID-19, performing pairwise comparisons between groups. Increasing COVID-19 severity was characterised by an abundance of inflammatory immune response genes and pathways, including many related to neutrophils and macrophages, in addition to an upregulation of immunoglobulin genes. Our insights into COVID-19 severity reveal the role of immune dysregulation in the progression to severe disease and highlight the need for further research exploring the interplay between SARS-CoV-2 and the inflammatory immune response.

host response to different pathogens and severity of disease 12,13 . Targeted 6,14 and untargeted 15-18 112 transcriptomic profiling of whole blood from SARS-CoV-2-positive hosts has already been undertaken. 113

To the best of our knowledge, severity of COVID- 19 has not yet been explored in whole blood with 114 non-hospitalised SARS-CoV-2-positive cases included as a comparator group, nor has it been explored 115 across a range of severities. The study of transcriptomic profiles from individuals with varying 116 severities will be an essential tool for improving our understanding of the course of disease resulting 117 from SARS-CoV-2 infection. 118

We have analysed the whole blood transcriptomes obtained from individuals with different levels of 119 COVID-19 severity to explore how the host blood transcriptome changes with increasing COVID-19 120 severity, aiming to identify the key biological processes and genes underpinning differences in 121 severity. 122

Adult patients with COVID-19 were recruited through the GEN-COVID study group (www.gencovid.eu) 125 at Hospital Clínico Universitario de Santiago de Compostela (Galicia, Spain) between March 2020 and 126

May 2020. COVID-19 was defined according to the current national guidelines in Spain 127 (https://www.mscbs.gob.es/profesionales/saludPublica/ccayes/alertasActual/nCov/documentos.ht 128 m). 129

Subjects granted informed consent for their participation in the study. If this was not possible at the 130 time of sampling, deferred consent was allowed, and subjects were approached for consent at the 131 Patients with COVID-19 were categorised as having mild, moderate, or severe disease. Mild patients 135 were those who were always outpatients; emergency department attendance was the ceiling of care, 136 and they were not admitted to hospital (WHO score 1-2). Moderate patients were those admitted to 137 hospital, for whom ward-based therapy was the ceiling of care with supportive care limited to oxygen 138 delivery; no intensive care unit (ICU) admission (WHO score 3-4). Severe patients were those who 139 were admitted to ICU at any time throughout the course of their disease (WHO score 5-7). Supportive 140 care included high flow oxygen (>16 litres/ minute), non-invasive ventilation (NIV), invasive 141 ventilation, inotropic support, renal replacement therapy, and extracorporeal membrane oxygenation 142 (ECMO). The severe category also included those patients who died (WHO score 8), in the emergency 143 department, ward or ICU. 144

Whole blood was collected at the time of recruitment into PAXgene blood RNA tubes (PreAnalytiX), 146 frozen, and total RNA (including RNA > 18 nucleotides) was isolated according to the manufacturer's 147 instructions (Qiagen). RNA samples were stored at −80°C, before undergoing an additional DNAse 148 treatment using an RNA clean & concentrator kit (Zymo Research) prior to sequencing at The 149

Wellcome Centre for Human Genetics in Oxford, UK. Material was quantified using RiboGreen 150 (Invitrogen) on the FLUOstar OPTIMA plate reader (BMG Labtech) and the size profile and integrity 151 analysed on the 2200 TapeStation (Agilent, RNA ScreenTape). Input material was normalised and 152 strand specific library preparation was completed using NEBNext® Ultra™ II mRNA kit (NEB) and NEB 153 rRNA/globin depletion probes following manufacturer's instructions. Libraries were on a Tetrad (Bio-154 Rad) using in-house unique dual indexing primers (based on 19 ). Individual libraries were normalised 155 using Qubit and pooled together. The pooled library was diluted to ~10 nM for storage and denatured 156 and further diluted prior to loading on the sequencer. Paired end sequencing was performed using a 157 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. ; https://doi.org/10.1101/2021.09.16.21263170 doi: medRxiv preprint Severity was also explored as an additive variable (mild = 0; moderate = 1; severe = 2), with full 207 methods described in the supplementary text. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 

Whole blood transcriptomic profiling through RNA Sequencing (RNA-Seq) was performed on 65 220 samples from patients recruited through the GEN-COVID study. Samples from patients in whom a 221 pathogen in addition to SARS-CoV-2 was isolated less than 5 days before or 10 days after the research 222 blood sample were not selected (n=10), as the aim was to identify gene expression changes in blood 223 that reflect COVID-19 disease and not coinfections. Of the 55 remaining COVID-19 patients, 19, 26 and 224 10 patients had mild, moderate, and severe disease, respectively. In all but two cases, the severity 225 categorisation of the patient matched their level of supportive care at the time the research blood 226 sample. For these two patients, the decision to transfer them to ICU was made within 36 hours of the 227 sample extraction, therefore they were classified as severe in our analyses. 228

Characteristics of the 55 COVID-19 patients are summarised in Table 1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. ; https://doi.org/10.1101/2021.09.16.21263170 doi: medRxiv preprint ambulatory patients, 47% (n=26) were admitted to the ward, 18.2% (n=10) were admitted to ICU, and 243 none died. Supplementary Fig. 1 shows that whilst the samples are clearly stratified by severity in 244 PC1, there is confounding between severity and sex/age, a pattern which is also clear in Table 1 . 245 Endocrine 4 (21%) 14 (53.8 %) 7 (70%)

6 (23.1%) 3 (30%) Immunosuppressed 0 (0%) 1 (3.8%) 1 (10%) Hematology 0 (0%) 0 (0%) 0 (0%) Allergic/autoimmune/inflammatory 2 (10.5%) 0 (0%) 0 (0%) Genetic 0 (0%) 1 (3.8%) 0 (0%) Organ transplant 0 (0%) 0 (0%) 1 (10%) Obesity 1 (5.3%) 12 (46.2%) 5 (50%) Pregnant 1 (5.3%) 0 (0%) 0 (0%) Other 0 (0%) 0 (0%) 0 (0%) Admission timelines Days since symptom onset (median, IQR) 11 (9.25-16.75) 15 (11-17) 10.5 (9-13.5) Presenting symptoms Fever 9 (47.4%) 22 (84.6%) 10 (100%) Respiratory 12 (63.2%) 24 (92.3%) 9 (90%) Cardiac 0 (0%) 1 (3.8%) 0 (0%) Gastrointestinal 2 (10.5%) 9 (34.6%) 4 (40%) Musculoskeletal 9 (47.4%) 16 (61.5%) 6 (60%) Rash 0 (0%) 0 (0%) 0 (0%) Sensory (ageusia, anosmia) 9 (47.4%) 8 (30.8%) 1 (10%) Headache 10 (52.6%) 4 (15.4%) 0 (0%) Shock 0 (0%) 0 (0%) 0 (0%) Ill appearance 0 (0%) 2 (7.7%) 1 (10%) Investigations

White cells (10 9 /L) -mean (IQR) Unknown 7 (4.6-6.9) 7.5 (5.2-10) Neutrophils (10 9 /L) -mean (IQR) Unknown 4.5 (2.6-4.7) 6.9 (4.5-9.1) Lymphocytes (10 9 /L) -mean (IQR) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 

Supportive care required at time of sampling Oxygen 0 (0%) 15 (57.7%) 0 (0%) High-flow oxygen 0 (0%) 0 (0%) Severe COVID-19 (n=10) Uninfected 0 (0%) 0 (0%) 0 (0%) Ambulatory 1 -No limitation of activities 19 (100%) 0 (0%) 0 (0%) 2 -Limitation of activities 0 (0%) 0 (0%) 0 (0%) Hospitalised, Mild Disease 3 -Hospitalised, no oxygen therapy 0 (0%) 5 (19.3%) 0 (0%) 4 -Oxygen by mask or nasal prongs 0 (0%) 21 (80.8%) 0 (0%) Hospitalised, Severe Disease 5 -Non-invasive ventilation or high-flow oxygen 0 (0%) 0 (0%) 4 (40%) 6 -Intubation and mechanical ventilation 0 (0%) 0 (0%) 3 (60%) 7 -Ventilation + additional organ support-pressors, renal replacement therapy, ECMO is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Pairwise comparisons were made between the three severity groups (mild, moderate and severe) to 282 identify transcriptomic differences with increasing severity. In the moderate vs. mild COVID-19 283 comparison, there was greater concordance between the models accounting for immunomodulatory 284 treatment and immune cell proportions as the number of genes identified as SDE in both models 285 was higher than for the severe vs. mild and the severe vs. moderate comparisons (Fig. 2) . 286 

Moderate COVID-19 vs. Mild COVID-19 293 1,547 genes were SDE between moderate (n=26) and mild (n=19) patients whilst accounting for 294 immunomodulatory treatment, with 603 and 944 genes over-and under-expressed with increasing 295 severity respectively. When these genes were subjected to pathway analysis, EIF2 Signalling was the 296 most significant pathway reduced in moderate cases compared to mild cases (z-score=-5.778, B-H p-297 value=5.012×10 -29 ) and Regulation of eIF4 and p70S6K Signalling (z-score=-1, B-H p-value=1.000×10 -298 13 ) were also found to be significantly enriched (Supplementary Table 1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint increasing severity, respectively. 87 genes were SDE between severe and mild COVID-19 irrespective 323 of the model design (Fig. 2B) . 324 . CC-BY 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The pathways upregulated in severe COVID-19 were dominated by those related to the immune 325 response, notably the inflammatory immune response (Supplementary Table 2 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Table 3 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. 

The impacts of severity on the transcriptome were also explored with severity as an additive variable 378 (supplementary materials). 7,413 genes were SDE with severity whilst accounting for 379 immunomodulatory treatment, of which 55 had absolute LFC values greater than 2 and adjusted p--380 values < 0.0001 (Fig. 3A) . The 55 genes are split into two broad clusters (Fig. 3A) with the 1 st cluster 381 containing multiple neutrophil-associated genes (including LTF, MPO, BPI, ELANE) and the 2 nd cluster, 382 with two sub-clusters, containing various immunoglobulin and B cell genes (e.g., IGHV3-13, IGKV6-1, 383 IGHV3-10). 307 genes were SDE in all three pairwise severity comparisons whilst controlling for 384 immunomodulatory treatment in addition to displaying additive behaviour (Supplementary Table 4 ). 385

Of these 307, 10 genes had absolute LFC values greater than 2 and adjusted p-values < 0.0001 for 386 the additive analyses (Fig. 3B) . Genes SDE in the additive model in addition to the pairwise 387 comparisons show more granularity than those SDE just in the additive model as their levels differ 388 between each severity category in a stepwise manner (Supplementary Fig. 4) . 389

. CC-BY 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. ; https://doi.org/10.1101/2021.09.16.21263170 doi: medRxiv preprint

We have explored host whole-blood transcriptomes from COVID-19 patients with varying degrees of 397 severity through differential expression and pathway analysis. We made pairwise comparisons 398 between three different severity groups: mild, moderate, and severe. Severity analyses revealed 399 major upregulation of genes and pathways related to the inflammatory immune response with 400 increasing severity, with notable increases in genes and pathways related to neutrophil-and 401 macrophage-mediated immunity accompanied by decreases in pathways related to T cell-mediated 402 immunity. 403

We observed considerable transcriptomic differences between moderate COVID-19 individuals who 404 received steroids to those who did not receive steroids, suggesting that this immunomodulatory 405 treatment has a profound impact on the transcriptome. The widespread use of steroids in COVID-19 406 and the transcriptional disruption we observed in patients receiving steroids support the inclusion of 407 immunomodulatory treatment in models related to COVID-19 transcriptomic analyses, in order to 408 account for their impacts on the transcriptome. 409

Immune dysregulation has been extensively discussed as a contributing factor in the progression to is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Other pathways of interest include pathways related to hypoxia that were enriched with increasing 463 severity. Hypoxia is a primary feature and major cause of mortality amongst patients with severe 464 COVID-19 53 . Various pathways related to protein production and DNA/protein damage were 465 identified as downregulated with increasing COVID-19 severity. SARS-CoV-2 has been shown to cause 466 major disruption to host protein production, for example viral protein NSP1 has been shown to bind 467 . CC-BY 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint COVID-19 severity is highly influenced by age which leads to major confounding between these two 485 variables. Although we controlled for this confounder (by including age and the interaction between 486 age and severity) when exploring transcriptomic changes with severity, it is possible that we may have 487 a) failed to identify key drivers of severity as they are confounded with age, b) inadvertently included 488 spurious genes that are really driven by age rather than severity. The sample sizes in our analyses are 489 modest for some severity groups. For example, in the severe COVID-19 group, only 10 samples could 490 . CC-BY 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. ; https://doi.org/10.1101/2021.09.16.21263170 doi: medRxiv preprint be included due to concomitant bacterial infection, because coinfections were likely to have had 491 profound transcriptional impacts and may have masked the genuine SARS-CoV-2 signal. 492

We have explored the transcriptomic impact of SARS-CoV-2 infection through evaluating the 494 transcriptomic differences between individuals with varying levels of COVID-19 severity. We have 495 observed considerable transcriptomic perturbation which offer insights into the host factors that 496 influence development of severe COVID-19. Upregulation of inflammatory immune pathways was 497 observed with increasing severity, with multiple neutrophil, macrophage and immunoglobulin-498 associated genes and pathways identified, suggesting that increased COVID-19 severity may be 499 mediated in part by neutrophil activation, which may be related to production of immunoglobulin as 500 acquired immunity develops. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted September 21, 2021. ; https://doi.org/10.1101/2021.09.16.21263170 doi: medRxiv preprint

Hyperinflammation and Immune Response Generation in 537 COVID-19

Impaired type I interferon activity and inflammatory responses in severe COVID-539 19 patients

Untuned antiviral immunity in COVID-19 revealed by temporal type I/III 541 interferon patterns and flu comparison

The trinity of COVID-19: immunity, 544 inflammation and intervention

Genome-wide host RNA signatures of 547 infectious diseases: discovery and clinical translation

Discovery and Validation of a 3-Gene Transcriptional 553 Signature to Distinguish COVID-19 and Other Viral Infections from Bacterial Sepsis in Adults; A 554 Case-Control then Observational Cohort Study

Diagnostic Test Accuracy of a 2-Transcript Host RNA Signature for 556 Discriminating Bacterial vs Viral Infection in Febrile Children

Integrated pathogen load and dual transcriptome analysis of systemic host-559 pathogen interactions in severe malaria

Transcriptomic profiling in childhood H1N1/09 influenza reveals reduced 562 expression of protein synthesis genes

A Dynamic Immune Response Shapes COVID-19 Progression

A diagnostic host response biosignature for COVID-19 from RNA profiling of nasal 567 swabs and blood

Blood transcriptional biomarkers of acute viral infection for detection of pre-569 symptomatic SARS-CoV-2 infection. medRxiv

A 6-mRNA host response whole-blood classifier trained using patients with 572 non-COVID-19 viral infections accurately predicts severity of COVID-19. medRxiv

Disease severity-specific neutrophil signatures in blood 575 transcriptomes stratify COVID-19 patients

El-Fahmawi & P. Piazza 578 Improved workflows for high throughput library preparation using the transposome-based 579 nextera system

Babraham Bioinformatics -FastQC A Quality Control tool for High Throughput 581 Sequence Data

MultiQC: summarize analysis results for 583 multiple tools and samples in a single report

BEDTools: a flexible suite of utilities for comparing genomic features

STAR: ultrafast universal RNA-seq aligner

The Sequence Alignment/Map format and SAMtools

featureCounts: an efficient general purpose program for assigning 592 sequence reads to genomic features

R: A Language and Environment for Statistical Computing

Moderated estimation of fold change and dispersion for RNA-598 seq data with DESeq2

Determining cell type abundance and expression from bulk tissues with 600 digital cytometry

Longitudinal characteristics of lymphocyte responses and cytokine profiles in the 603 peripheral blood of SARS-CoV-2 infected patients

Longitudinal analyses reveal immunological misfiring in severe COVID-19

Longitudinal immune profiling reveals key myeloid signatures associated with 608 COVID-19

Reduced numbers of T cells and B cells correlates with persistent SARS-CoV-2 610 presence in non-severe COVID-19 patients

Changes of hematological and immunological parameters in COVID-19 patients. Int 613

Severe COVID-19: what 615 have we learned with the immunopathogenesis?

An inflammatory cytokine signature predicts COVID-19 severity and 618 survival

Immune Response, Inflammation, and the Clinical Spectrum of COVID-19. Front 620 Immunol 11

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-622 19

Temporal profiling of plasma cytokines, chemokines and growth factors from mild, 624 severe and fatal COVID-19 patients

Inflammatory profiles across the spectrum of disease reveal a distinct role 627 for GM-CSF in severe COVID-19

Lymphopenia predicts disease severity of COVID-19: a descriptive and predictive 630 study

Lymphopenia in severe coronavirus disease-2019 (COVID-19): systematic 632 review and meta-analysis

Transepithelial migration of neutrophils into the lung requires TREM-1

Distinct and early increase in circulating MMP-9 in COVID-19 patients with 636 respiratory failure

Excessive matrix metalloproteinase-1 and hyperactivation of endothelial cells 638 occurred in COVID-19 patients and were associated with the severity of COVID-19. medRxiv

Proteins associated with neutrophil degranulation are upregulated in 641 nasopharyngeal swabs from SARS-CoV-2 patients

A single-cell atlas of the peripheral immune response in patients with severe 644 COVID-19

The predicting roles of carcinoembryonic antigen and its underlying mechanism 646 in the progression of coronavirus disease 2019

SARS-CoV-2 Receptor ACE2 Is an Interferon-Stimulated Gene in Human 649

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

HIF-656 1alpha, and COVID-19: from pathogenic factors to potential therapeutic targets

SARS-CoV-2 Disrupts Splicing, Translation, and Protein Trafficking to 659

Structural basis for translational shutdown and immune evasion by the Nsp1 661 protein of SARS-CoV-2

Coronavirus infection induces DNA replication stress 663 partly through interaction of its nonstructural protein 13 with the p125 subunit of DNA 664 polymerase δ

DNA Damage Response and Immune Defense

A Proposed Framework and Timeline of the Spectrum of 668 Disease Due to SARS-CoV-2 Infection: Illness Beyond Acute Infection and Public Health 669 Implications

COVID-19 illness in native and immunosuppressed states: A clinical-671 therapeutic staging proposal

Many thanks to the patients and their families who took part in the study that the data presented 526 here originated from. 527

The authors declare no competing interests. The funders had no role in the design of the study; in the 522 collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to 523 publish the results. 524