EBioMedicine 66 (2021) 103339 Contents lists available at ScienceDirect EBioMedicine journal homepage: www.elsevier.com/locate/ebiom Research paper Epigenome-wide association study of COVID-19 severity with respiratory failure Manuel Castro de Mouraa,1, Veronica Davalosa,1, Laura Planas-Serrab, Damiana Alvarez-Erricoa, Carles Arribasa, Montserrat Ruizb, Sergio Aguilera-Albesae,  s Troyah, Juan Valencia-Ramosi, Valentina Ve Jesu lez-Santamariab,c, Agustí Rodríguez-Palmerob,d, Judit Villar-Garciag, Juan P. Horcajadag, Sergiu Albuf, Carlos Casasnovasb,c, Anna Rullj, Laia Revertej, Beatriz Dietlk, David Dalmaul, Maria J. Arranzm, -Caroln, Anna M. Planaso, Jordi Pe Laia Llucia rez-Turp, Israel Fernandez-Cadenasn, Paula Villares , Jair Tenorio , Roger Colobrant, Andrea Martin-Naldau, Pere Soler-Palacinu, q r,s Francesc Vidalj, Aurora Pujolb,s,w,*, Manel Estellera,v,w,x,* a Josep Carreras Leukaemia Research Institute (IJC), 08916 Badalona, Barcelona, Catalonia, Spain b Neurometabolic Diseases Laboratory, Bellvitge Biomedical Research Institute (IDIBELL), 08908 L'Hospitalet de Llobregat, Barcelona, Catalonia, Spain c Bellvitge University Hospital, L'Hospitalet de Llobregat, Barcelona, Catalonia, Spain d University Hospital Germans Trias i Pujol, Badalona, Barcelona, Catalonia, Spain e Navarra Health Service Hospital, Pamplona, Spain f Institut Guttmann Foundation, Badalona, Barcelona, Catalonia, Spain g Hospital del Mar - IMIM Biomedical Research Institute, Barcelona, Catalonia, Spain h Infanta Leonor University Hospital, Madrid, Spain i University Hospital of Burgos, Burgos, Spain j Hospital Universitari de Tarragona Joan XXIII, IISPV, Universitat Rovira i Virgili, Tarragona, Catalonia, Spain k Servei de malalties infeccioses Hospital Universitari MutuaTerrassa, Universitat de Barcelona, Barcelona, Catalonia, Spain l MutuaTerrassa Research and Innovation Foundation, HIV/AIDS Unit Hospital Universitari MutuaTerrassa, University of Barcelona, Barcelona, Catalonia, Spain m Fundacio  Docencia i Recerca Mutua Terrassa i Hospital Universitari Mutua Terrassa, Barcelona, Catalonia, Spain n Stroke Pharmacogenomics and Genetics Group, Sant Pau Institute of Research, Sant Pau Hospital, Barcelona, Catalonia, Spain o Department of Brain Ischemia and Neurodegeneration, Institut d'Investigacions Biomediques de Barcelona (IIBB), Consejo Superior de Investigaciones Científicas (CSIC), Area of Neurosciences, Institut d'Investigacions Biomediques August Pi i Sunyer (IDIBAPS), Barcelona, Catalonia, Spain p Institut de Biomedicina de Valencia-CSIC, CIBERNED, Unitat Mixta de Neurologia i Genetica, IIS La Fe, Vallencia, Spain q Internal Medicine Department, Hospital HM Sanchinarro, HM Hospitales, Madrid, Spain r INGEMM-Instituto de Genetica Medica y Molecular, Hospital Universitario La Paz, Madrid, Spain s Center for Biomedical Research on Rare Diseases (CIBERER), ISCIII, Madrid, Spain t Immunology Division, Genetics Department, Hospital Universitari Vall d'Hebron, Vall d'Hebron Research Institute, Vall d'Hebron Barcelona Hospital Campus, UAB, Barcelona, Catalonia, Spain u Pediatric Infectious Diseases and Immunodeficiencies Unit, Hospital Universitari Vall d'Hebron, Vall d'Hebron Barcelona Hospital Campus, Barcelona, Catalonia, Spain v Centro de Investigacio n Biomedica en Red de Cancer (CIBERONC), Spain w Institucio Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia, Spain x Physiological Sciences Department, School of Medicine and Health Sciences, University of Barcelona (UB), Catalonia, Spain A R T I C L E I N F O A B S T R A C T Article History: Background: Patients infected with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), Received 5 January 2021 responsible for the coronavirus disease 2019 (COVID-19), exhibit a wide spectrum of disease behaviour. Since Revised 16 March 2021 DNA methylation has been implicated in the regulation of viral infections and the immune system, we per- Accepted 26 March 2021 formed an epigenome-wide association study (EWAS) to identify candidate loci regulated by this epigenetic Available online 15 April 2021 mark that could be involved in the onset of COVID-19 in patients without comorbidities. * Corresponding authors. E-mail addresses: apujol@idibell.cat (A. Pujol), mesteller@carrerasresearch.org (M. Esteller). 1 These authors contributed equally. https://doi.org/10.1016/j.ebiom.2021.103339 2352-3964/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 Methods: Peripheral blood samples were obtained from 407 confirmed COVID-19 patients  61 years of age Keywords: and without comorbidities, 194 (47.7%) of whom had mild symptomatology that did not involve hospitaliza- Coronavirus tion and 213 (52.3%) had a severe clinical course that required respiratory support. The set of cases was SARS-CoV-2 divided into discovery (n = 207) and validation (n = 200) cohorts, balanced for age and sex of individuals. We COVID-19 analysed the DNA methylation status of 850,000 CpG sites in these patients. Epigenetics DNA methylation Findings: The DNA methylation status of 44 CpG sites was associated with the clinical severity of COVID-19. Of these loci, 23 (52.3%) were located in 20 annotated coding genes. These genes, such as the inflammasome component Absent in Melanoma 2 (AIM2) and the Major Histocompatibility Complex, class I C (HLA-C) candi- dates, were mainly involved in the response of interferon to viral infection. We used the EWAS-identified sites to establish a DNA methylation signature (EPICOVID) that is associated with the severity of the disease. Interpretation: We identified DNA methylation sites as epigenetic susceptibility loci for respiratory failure in COVID-19 patients. These candidate biomarkers, combined with other clinical, cellular and genetic factors, could be useful in the clinical stratification and management of patients infected with the SARS-CoV-2. Funding: The Unstoppable campaign of the Josep Carreras Leukaemia Foundation, the Cellex Foundation and the CERCA Programme/Generalitat de Catalunya. © 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) the viral arena, it is known that the activity of DNA and RNA viruses, such as HPV, HBV, EBV, KS and HIV, is regulated by DNA methylation 1. Introduction changes [18-21]. Most importantly, in the case of COVID-19, the rec- ognition that the activity of the adaptive immune system, including The severe acute respiratory syndrome coronavirus 2 (SARS-CoV- that of B- and T-cells, is significant for the provision of pre-existing 2) virus was identified in Wuhan, China, in late 2019 [1], and has pro- immunity to SARS-CoV-2 [22-24]. This means that epigenetic modifi- voked an ongoing global pandemic of the resulting illness, COVID-19. cation is a potentially powerful mechanism that determines the At the time of the writing (March 15th, 2021) there have been more development of severe symptoms. In this regard, the DNA methyla- than 120 million confirmed cases worldwide and 2.6 million deaths tion landscape is central to the homoeostasis of the immune system (https://origin-coronavirus.jhu.edu/map.html. COVID-19 has a wide and, in the clinical setting, DNA methylation changes in cancer affect spectrum of clinical manifestations, with the majority of infected sub- the success of immunotherapy [25,26]. Examples for the latter sce- jects showing only mild symptoms or being asymptomatic [2,3]. The nario are the observation that DNA methylation signatures are associ- principal group of patients with high mortality rates comprises those ated with the clinical response to PD-1 checkpoint blockade in with severe respiratory failure associated with acute respiratory dis- human tumors [27,28]. Similarly to the way GWAS studies were tress syndrome (ARDS) and interstitial pneumonia. These high-risk developed for COVID-19 [8,9], and our previous analyses of DNA patients require early and prolonged support by mechanical ven- methylation risk variants for other common human diseases [29,30], tilation to compensate for their respiratory failure [4,5]. The rea- we have performed a comprehensive epigenome-wide association sons for the varied clinical repertoire of COVID-19 are largely study (EWAS) in COVID-19 patients to identify candidate DNA meth- unknown. Only three risk factors have been consistently related ylation loci linked to the severity of the disease, particularly with to life-threatening COVID-19-associated respiratory failure: male respect to respiratory failure. sex, old age and concomitant medical conditions, such as diabe- tes, obesity, hypertension and cardiovascular pathology [6,7]. However, even considering all these components, there is great 2. Methods inter-individual variability in each demographic and epidemiolog- ical group. Thus, given the immense impact of COVID-19 on mor- 2.1. Study design and participants bidity and mortality, there is an unmet medical need to discover endogenous cellular and molecular biomarkers that predict the Whole blood samples and clinical data from 407 patients with expected clinical course of the disease. In this regard, striking confirmed COVID-19 were retrospectively collected between March recent data from genome-wide association studies (GWAS) in 7th 2020 and September 14th 2020 from fourteen Hospitals in Spain. large populations of COVID-19 patients and genome sequencing Patients were eligible if they did not present the risk factors of in particular subsets with life-threatening pneumonia indicate comorbidities (obesity with a BMI  30, diabetes, hypertension, auto- that loci at the chromosome 3p21.31 region [8,9], antiviral immune disorders, and chronic cardiovascular or lung diseases), restriction enzyme activators clustered at the chromosome smoking habit or advanced age (> 61 years). Clinicopathological 12q24.13 locus [9] and the ABO blood-group system are linked to characteristics of the COVID-19 patients studied are summarized in COVID-19 susceptibility [8], and inborn genetic defects of type I Table 1. Our study was designed to find epigenetic biomarkers associ- interferon (IFN) immunity are associated with extreme severe ated with COVID-19 severity in patients  61 years of age and with- pneumonia [10,11]. out comorbidities, thus, we did not recruit all consecutive cases and Here we have explored another layer of biological information it was not intended to be representative of the spectrum of the dis- that has not been comprehensively addressed so far in the COVID-19 ease as a whole. Following the general guidelines of the National field: the impact of epigenetic variation on disease severity. Epige- Institutes of Health for COVID-19 clinical staging (https://www.cov netics, defined as the study of changes in gene function that are heri- id19treatmentguidelines.nih.gov/overview/clinical-presentation/), table and that do not entail a change in DNA sequence, plays a major the patients were categorized in two groups: those asymptomatic or role in tissue homoeostasis. In particular, DNA methylation, the most with a paucisymptomatic clinical status who were non-hospitalized widely studied epigenetic mark, is altered in many human diseases (Group G3) compared with those who required hospitalized oxygen [12,13]. This knowledge has been applied clinically in the field of therapy, such as patients with nasal mask or nasal prongs non-inva- oncology, which uses DNA methylation biomarkers for the clinical sive ventilation or high-flow oxygen ventimask (Group G2) and those management of gliomas [14,15] and cancers of unknown primary ori- requiring mechanical ventilation; additional organ support such as gin [16], and for the pathological classification of brain tumors [17]. In vasopressors; or extra corporeal membrane oxygenation (Group G1). M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 3 Briefly, the methylation data pre-processing performed with the Research in context ChAMP package in R included normalization with the ssNoob proce- dure, filtering probes with a detection P-value greater than 0.01, Evidence before this study NoCG Start, probes with SNPs, multihit start probes and XY chromo- The clinical spectrum of COVID-19 ranges from asymptomatic some probes. Confounding covariates in the methylation dataset individuals and patients with mild symptomatology not requir- were identified by the single-value decomposition (SVD) method ing hospitalization to those admitted to the ward for respira- [32]. The genomic analysis presented in the study was performed tory failure due to pneumonia that may eventually led to acute using the GRCh37  hg19 human genome reference build, as respiratory distress syndrome (ARDS). Risk factors for worse described in the Illumina manifest file associated with the DNA meth- clinical course include older age, male sex, and comorbidities ylation EPIC microarray. CpG methylation status was further vali- such as diabetes, obesity and hypertension. Genetic susceptibil- dated by pyrosequencing using PyroMark Q48 system. Primers were ity loci, such as blood group A, and inborn errors of type I IFN designed using Qiagen's PyroMark Assay Design 2.0 software. Primer immunity have more recently been recognized as risk factors. sequences are shown in Supplementary Methods. To our knowledge, this is the first epigenome-wide association study (EWAS) of COVID-19 considering the severity of respira- 2.4. Computational analyses tory failure. Our search of PubMed on March 15th 2021, limited to articles in English, but not by date, using the terms “COVID- The EWAS analysis was performed by identifying the probes dif- 1900 , “severity”, “respiratory failure”, “EWAS”, “epigenomics”, ferentially methylated between asymptomatic/paucisymptomatic “DNA methylation, “biomarker” and “prediction”, yielded no and severe COVID-19 cases. This involved deriving a linear model studies addressing this exact topic. with the limma package in R, adjusted by the age and gender covari- ables, using the methylation beta values of the discovery dataset. We Added value of this study checked the data to ensure that the assumptions for the methods used (such as linearity) were met (Supplementary Methods). CpG Our findings demonstrate the existence of differential DNA probes with a Benjamini-Hochberg adjusted-P value less than 0.05 methylation sites that distinguish COVID-19 patients with pau- and an absolute difference in methylation beta greater than 0.10 cisymptomatic clinical status from those who will require hos- were selected for further analysis. An additional feature selection pitalized oxygen therapy, including mechanical ventilation and step involved applying the elastic net method to the differentially additional organ support measures. The epigenetic loci identi- methylated markers in order to select the best markers for predic- fied were mostly located within genes associated with the tion. A ten-fold cross-validation was used to identify the best alpha interferon response pathway. Using these DNA methylation tuning parameter and the minimum lambda value. Gene Ontology biomarkers, we obtained an epigenomic signature that we have (GO) terms associated to the candidate genes derived from EWAS, called “EPICOVID” that showed great accuracy in predicting were examined using BioMart data-mining tool (Ensembl) to classify COVID-19 severity. the genes according to biological processes. Significant GO terms were determined according to the hypergeometric test. The signifi- Implications of all the available evidence cantly differential DNA methylation sites were used to generate a This study has identified new biomarkers associated with predictive model of the COVID-19 clinical outcome. A supervised COVID-19 clinical severity that can be assessed for particular classification model based on the combination of six machine-learn- DNA methylation sites or included within a designed overall ing algorithms was used to predict asymptomatic/paucisymptomatic epigenomic signature. The determination of the DNA methyla- and severe COVID-19 progression status. The classification methods tion status of these sites can be easily added to other molecular, used were random forest (rf), adaptive boosting (adaboost), multivar- cellular and clinical parameters to enable the more accurate iate adaptive regression splines (earth), k-nearest neighbour (knn), prediction of the COVID-19 patients who might suffer the worst radial kernel support vector classifier (svmRadial) and logistic regres- clinical outcomes, so that early intervention strategies may be sion model (glmnet). The caretList function from the caretEnsembl devised to improve the cure rate and relieve the burden on the package was used to fit the different models with the same resam- healthcare system. pling parameters (ten-fold cross-validation and five repeats). The final glm-weighted meta-model was generated from the six predic- tive models using the caretStack function from caretEnsembl. Model performance was assessed in terms of AUC, accuracy and kappa val- The study protocol and the statistical analysis plan is described in ues. Finally, the prediction model was tested in the validation cohort Supplementary Methods. and the corresponding confusion matrix was obtained. The hierarchi- cal clustering analysis was performed using the Ward.D2 clustering 2.2. Ethics method with Euclidean distances in the pheatmap package in R. The protocol of this retrospective study was approved by the insti- 2.5. Role of funders tutional ethics review boards of the participating institutions. Writ- ten informed consent was obtained from all participants. This work was mainly funded by the Unstoppable campaign of the Josep Carreras Leukaemia Foundation, the Cellex Foundation and the 2.3. DNA methylation data CERCA Programme of the Generalitat de Catalunya. Additional sup- port was provided by the grant COVID19-PoC-6_17, Direccio  General The DNA methylation status of the COVID-19 samples was estab- de Recerca i Innovacio en Salut (DGRIS), Departament de Salut, Gen- lished using the Infinium MethylationEPIC Array (~850,000 CpG sites) eralitat de Catalunya; the Spanish AIDS Research Network “RD16/ following the manufacturer’s instructions for the automated process- 0025/000600 Co-funded by European Regional Development Fund/ ing of arrays with a liquid handler (Illumina Infinium HD Methylation European Social Fund; “A way to make Europe”/”Investing in your Assay Experienced User Card, Automated Protocol 15019521 v01), as future”; Programa de Intensificacio n de Investigadores INT 20/00031, previously described [31]. The DNA methylation beta values were Instituto de Salud Carlos III; PFIS FI18/00141 from Instituto de Salud obtained from the raw IDAT files by using the minfi package in R. Carlos III; Rio Hortega grant CM18/00145; the centre for Biomedical 4 M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 Table 1 the screening strategy used to identify epigenetic biomarkers associ- Clinical characteristics of the studied discovery, validation and entire cohorts of ated with the severity of the disease is shown in Figure S1. COVID-19 patients. Characteristics COVID-19 cohorts 3.2. Epigenome-wide association analysis in the discovery cohort Discovery cohort Validation cohort Entire cohort (N = 207) (N = 200) (N = 407) The DNA methylation analysis of the 207 COVID-19 cases of the discovery cohort identified 51 CpG sites with a differential methyla- Age (years) - Median 43 [19 - 60] 42 [22 - 61] 42 [19 - 61] tion status between the mild (G3, n = 104) and severe (G2 + G1, [range] Gender - Frequency (%) n = 103) groups (Table S1). Because male patients tend to have a Female 108 (52.2%) 114 (57.0%) 222 (54.5%) worse clinical disease course [6,7] and the male patients in our entire Male 99 (47.8%) 86 (43.0%) 185 (45.5%) cohort also exhibited more clinical severity (Pearson’s Chi-squared Oxygen therapy- Frequency (%) test with Yate’s continuity correction, P = 1.35e20), the candidate No oxygen support 105 (50.7%) 90 (45.0%) 195 (47.9%) Mask or nasal prongs 30 (14.5%) 57 (28.5%) 87 (21.4%) CpG sites were corrected for sex. In a similar manner, although we Non-invasive ventilation 30 (14.5%) 20 (10.0%) 50 (12.3%) limited our study to patients  61 years of age to reduce the effect of or high-flow oxygen the established association between older age and disease severity Mechanical ventilation 33 (15.9%) 1 (0.5%) 34 (8.4%) [6,7], our younger patients performed much better than the older ECMO, pressors 9 (4.3%) 21 (10.5%) 30 (7.4%) patients (one-tailed Wilcoxon rank sum test, P = 2.47e07), so age cor- Any oxygen therapy (not 0 (0%) 11 (5.5%) 11 (2.7%) specified) rection also featured in the analysis. Overall, after correction for sex Severity group - Frequency (%) and age in the analysis of the originally identified 51 CpG sites, we G1 (ICU) 75 (36.2%) 24 (12.0%) 99 (24.3%) found 44 CpGs whose DNA methylation status differed significantly G2 (Hospitalization with 28 (13.5%) 86 (43.0%) 114 (28.0%) between asymptomatic/paucisymptomatic and severe COVID-19 O2 support) G3 (Mild symptomatology, 104 (50.2%) 90 (45%) 194 (47.7%) patients (Table S2). The Manhattan plot of the fully adjusted values of home) P from the EWAS for COVID-19 severity in the discovery cohort is COVID-19 pneumonia - Frequency (%) shown in Fig. 1. No 93 (44.9%) 91 (45.5%) 184 (45.2%) The genomic context of these distinctly methylated 44 CpG sites is Yes 114 (55.1%) 89 (44.5%) 203 (49.9%) shown in Table S3. Fifteen of these loci were located in genomic Unknown 0 (0%) 20 (10%) 20 (4.9%) human regions with no currently described gene sequence; six were Abbreviations: ECMO = Extracorporeal membrane oxygenation; ICU = Intensive care associated with non-coding RNA loci; and the other 23 CpG sites unit. were located within 20 known coding genes (Table 2). RNA and/or protein for the studied COVID-19 patients was not available, but Research on Rare Diseases (CIBERER); “la Caixa” Banking Foundation using a collection of 62 blood derived cell lines previously character- (LCF/PR/GN18/51140001); and the Programa de Suport als Grups de ized for DNA methylation and expression patterns [33], we observed Recerca AGAUR (2017SGR1080 and 2017SGR948). AP and ME are that for 17 of 20 (85%) coding genes the presence of hypermethyla- ICREA Research Professors. The sponsors of the study had no role in tion located in the 50 -end regulatory region was significantly associ- study design, data collection, data analysis, data interpretation, or ated with transcript downregulation (Pearson's test, P < 0.05). To writing of the manuscript. The authors collected the data, and had determine the activities of the identified 20 candidate genes derived full access to all of the data in the study. They also had the final deci- from EWAS, we data-mined the Gene Ontology (GO) terms associated sion and responsibility to submit the study results for publication. to each gene using BioMart tool (Ensembl). Significantly overrepre- sented genes (hypergeometric test, P value = 2.14e06) associated 3. Results with GO biological processes included those of “immune response”, “type I interferon signalling pathway”, “interferon-gamma-mediated 3.1. Patients and epigenotyping signalling pathway”, “antigen processing and presentation”, “defence response to virus”, “cytokine-mediated signalling pathway” and Between March 7th 2020 and September 14th 2020, we collected “inflammatory response”. These processes are highly relevant to the whole blood samples from 407 adult patients diagnosed with COVID- degree of potential response to COVID-19 infection, since they are 19 that did not present the risk factors of comorbidities (obesity with related to the capacity of the immune system to respond to viral a BMI  30, diabetes, hypertension, autoimmune disorders, and infections. In this regard, the 20 genes containing the CpG methyla- chronic cardiovascular and lung diseases), smoking habit or advanced tion variants associated with COVID-19 severity include an overrep- age (>61 years) (Table 1). These patients were divided into a discov- resentation of genes that mediate the response to interferon, a key ery cohort of 207 cases and a validation cohort of 200 cases (Table 1). pathway in the physiopathological pathway of the disease [8,9,34]. The two cohorts did not present significant differences related to This reinforces the biological plausibility of the results of the EWAS. age (Mann-Whitney U test, P = 0.673), gender (Fisher’s exact test, Overall, 35% (7 of 20) of the genes identified were effectors of P = 0.370), requirement of oxygen therapy (Fisher’s exact test, interferon signalling, implying that they are potential participants in P = 0.275), severity of the disease (G3 vs G2/G1, Fisher’s exact regulating the efficacy of the immune system to deal with viral expo- test, P = 0.321) and the occurrence of pneumonia (Fisher’s exact test, sition. The interferon-related candidate genes with DNA methylation P = 0.308). Related to ethnicity, although 72% of our samples were status associated with the disease were: AIM2 (absent in melanoma from the West-Eurasia group, Central-South America population was 2), HLA-C (major histocompatibility complex, class I, C), IFI44L (inter- higher in the validation (33%) vs discovery (18%) cohort. DNA from feron-induced protein 44-like), CXCR2 (C-X-C motif chemokine the whole blood samples was extracted in all cases and hybridized to receptor 2), KIFAP3 (Kinesin Associated Protein 3), SGMS1 (Sphingo- a DNA methylation microarray that interrogates almost 850,000 CpG myelin Synthase 1) and VIM (vimentin) (Table 2). The cases of AIM2 sites of the human genome [31]. The goal of the study was to identify and HLA-C are particularly interesting because each gene is repre- genomic loci with a differential DNA methylation status in patients sented by two differentially methylated CpG sites in our analyses, with asymptomatic or paucisymptomatic clinical status who were making it more likely that they contribute to the onset of the disor- non-hospitalized (Group G3, see Methods) compared with those who der. AIM2 is a member of the innate immune system that exerts vital required hospitalized oxygen therapy (Groups G2 and G1, see Meth- activity during viral infections, and is responsible for the assembly of ods). A graphical schema representing the population of interest and a macromolecular complex called the inflammasome that unleashes M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 5 Fig. 1. Manhattan plot from the EWAS performed in the discovery cohort. The plot shows the result of the COVID-19 severity association test over 735.349 CpG positions. The blue dashed line indicates the genome-wide significance threshold of a Benjamini-Hochberg corrected P value lower than 0.05. Black dots represent the 44 CpGs signature labelled with their corresponding genes. caspase-1 and triggers pro-inflammatory cytokines such as IL-1b and genes involved in host-cell recognition and binding (Table S4). Thus, IL-18 [35]. HLA-C is another example of a candidate for mediating there is no evident association between epigenetic events at these COVID-19 severity since specific HLA subtypes have been associated genes and the clinical course of COVID-19, reinforcing the specificity with susceptibility for many viral infections, including HIV, HBV, of our approach for identifying susceptibility to DNA methylation loci H1N1, HCV and HPV [36]. From a functional standpoint, HLA-C is for COVID-19 severity that can easily be complemented by the addi- involved in both innate and adaptive immunity, but most impor- tional genetic biomarkers. tantly, HLA-C haplotypes have also been linked to the clinical course of COVID-19 [37-40]. 3.3. Testing EWAS markers in the validation cohort and development of Only one gene, in addition to AIM2 and HLA-C, was represented the EPICOVID signature by two CpG sites in our COVID-19 EWAS: PM20D1 (peptidase M20 domain-containing 1). PM20D1 has most often been categorized as a The predictive value of the single CpG sites whose DNA methyla- metabolic disease-associated gene, particularly in relation to obesity tion status was associated with COVID-19 severity in the EWAS of and diabetes [41], but genetic and epigenetic variants have also our discovery cohort was confirmed in the validation cohort. Overall, emerged as being a risk factor for neurodegenerative disorders such of the 44 CpGs whose DNA methylation status differed significantly as Parkinson’s [42] and Alzheimer’s disease [43,44]. Thus, its addi- between asymptomatic/paucisymptomatic and severe COVID-19 tional link as a potential modulator of the severity of SARS-CoV-2 patients after correction for sex and age, 37 (84.1%) were also signifi- infection, described here, suggests that this peptidase lies at the cantly associated with worse clinical outcome in the validation crossroads of many cellular and pathological pathways. cohort (Table S2). Of these 37 CpG sites, 19 loci were located in the We also examined in detail whether the DNA methylation status aforementioned 20 gene-containing regions (95%) (Table 2). The of asymptomatic/paucisymptomatic patients and those that eventu- Manhattan plot of the fully adjusted values of P from the EWAS for ally developed the severe symptomatology differed with respect to COVID-19 severity in the validation cohort is shown in Figure S2. the genes that previous studies have identified as being candidates Most importantly, when we analysed all the cases as a whole cohort, for COVID-19 infection susceptibility and/or prognosis. We did not comprising the discovery and validation sets (n = 407), 95.5% of the observe any significant DNA methylation content in the 40 genes CpG sites (42 of 44) still appeared to be associated with COVID-19 studied (Table S4), including the well-characterized ACE2 receptor severity when controlling for sex and age (Table S2). This was the and TMPRSS2 protease [45]; genes derived from GWAS [8,9]; genes case even when we included all the 20 candidate genes identified obtained from the screening of inborn errors of type I IFN immunity (Table 2). The Manhattan plot of the fully adjusted values of P from in patients with life-threatening COVID-19 [10,11,34]; and other the EWAS for COVID-19 severity in the entire cohort is shown in 6 Table 2 Epigenetic susceptibility loci in genes associated with severe COVID-19 with respiratory failure. CpG ID Chromosome Gene name Discovery Cohort Validation Cohort Entire Cohort Location P value Odds Ratio P value Analysis Odds Ratio P value Odds Ratio P value Odds Ratio P value Odds Ratio P value Odds Ratio corrected for Analysis Analysis Analysis Analysis Analysis age and sex corrected for corrected for corrected for corrected for corrected age and sex age and sex age and sex age and sex for age and sex M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 cg24145401 1:159,047,177 AIM2 2.70£1010 6.17£107 1.31£105 3.34£102 3.57£1021 7.87£1018 8.59£1020 3.49£1016 1.65£1030 4.65£1028 6.33£1027 3.25£1023 cg17515347 1:159,047,163 AIM2 1.11£1010 2.03£108 7.51£106 1.10£103 8.25£1020 9.06£1016 1.01£1017 1.57£1014 8.19£1030 6.39£1027 1.40£1025 7.59£1021 cg04736673 10:61,647,141 CCDC6 1.99£104 1.90 1.45£103 4.52£101 7.20£108 4.97£103 3.29£106 1.59£102 7.20£1012 4.71£107 1.86£1010 1.69£106 cg02003183 14:103,415,882 CDC42BPB 6.08£1012 1.03£1010 2.29£104 7.79 2.24£109 1.84£105 5.97£106 7.48£101 8.41£1021 2.27£1017 3.59£1011 1.03£107 cg15355235 18:35,001,518 CELF4 9.54£103 1.53£102 2.28£102 2.10£102 1.47£103 1.38£101 1.11£102 2.30£102 2.52£105 6.81 5.12£104 2.79£101 cg10947500 6:118,985,774 CEP85L 9.99£107 1.77£103 3.20£105 9.67£101 9.32£108 3.80£103 2.37£108 3.15£104 1.30£1014 3.63£1010 1.52£1014 3.17£1010 cg19225688 2:218,990,043 CXCR2 1.04£109 1.09£107 3.93£104 2.70 1.37£1015 6.56£1011 2.53£1012 2.96£108 1.01£1024 5.62£1021 1.77£1017 2.97£1013 cg02872426 6:110,736,772 DDO 4.73£109 1.55£106 1.29£102 3.80£102 2.14£105 1.27£101 3.29£106 1.67£102 1.41£1014 3.33£1010 1.68£108 1.54£104 cg08309069 6:31,240,651 HLA-C 4.28£109 1.77£106 1.11£103 6.07£101 2.25£102 6.96£103 5.22£102 5.00£103 1.28£109 2.14£105 1.71£105 8.44 cg05030953 6:31,241,000 HLA-C 6.79£107 2.90£103 2.09£103 2.83£101 2.13£101 6.13£104 2.86£101 1.00£103 6.73£106 2.76£101 5.12£104 2.84£101 cg13452062 1:79,088,559 IFI44L 1.55£106 1.01£103 8.40£104 9.77£101 5.31£1011 9.10£106 1.01£107 6.60£103 1.75£1016 3.73£1012 1.11£1011 3.64£107 cg26931608 1:170,036,455 KIFAP3 4.10£1011 7.52£108 8.97£104 8.18£101 6.54£108 5.50£103 4.93£105 9.11 1.06£1019 1.37£1016 5.17£1011 6.48£106 cg07796016 1:152,779,584 LCE1C 6.14£103 2.58£102 2.62£103 1.93£101 4.46£104 5.01£101 1.21£103 2.36£101 4.13£106 4.62£101 3.21£106 5.38£101 cg13571460 9:124,989,337 LHX6 1.69£103 1.26£101 3.40£104 4.64 5.91£102 2.41£103 4.53£102 5.00£103 1.53£104 9.81£101 2.52£105 5.53 cg07381806 19:2094,327 MOBKL2A 8.19£1011 3.00£108 3.93£104 3.28 4.78£108 7.62£103 4.91£106 9.82£101 1.31£1018 8.42£1014 2.11£1010 1.37£106 cg17178900 1:205,818,956 PM20D1 3.60£103 4.94£102 1.13£102 4.40£102 3.76£103 4.95£102 1.84£103 1.44£101 1.41£105 1.26£101 7.50£106 2.09£101 cg14893161 1:205,819,251 PM20D1 4.97£103 3.34£102 2.99£102 1.60£102 1.43£103 1.42£101 4.66£104 6.54£101 7.95£106 2.32£101 5.46£106 3.02£101 cg10188795 10:52,158,244 SGMS1 1.15£1010 1.92£108 1.30£104 1.57£101 4.31£1020 2.23£1017 1.45£1016 7.55£1012 7.58£1030 7.09£1027 1.32£1021 5.88£1017 cg24795173 10:108,751,940 SORCS1 2.64£104 1.32 2.10£103 2.72£101 1.23£102 1.34£102 1.17£102 2.10£102 5.33£106 3.53£101 8.79£106 1.74£101 cg14859874 1:154,238,265 UBAP2L 1.07£105 8.28£101 4.90£104 1.99 1.72£102 9.34£103 3.80£102 6.00£103 3.78£107 5.68£102 3.30£106 5.06£101 cg12682382 8:74,787,918 UBE2W 3.27£105 1.95£101 2.37£103 2.26£101 1.42£108 2.70£104 1.52£106 4.15£102 1.82£1013 2.23£109 7.83£1012 5.86£107 cg26063719 10:17,273,187 VIM 7.68£1012 7.09£109 3.93£104 2.77 6.58£1018 2.68£1014 6.57£1014 1.35£1010 2.66£1029 1.40£1027 3.69£1019 1.69£1015 cg06760111 3:21,551,522 ZNF385D 4.45£109 1.68£106 1.40£102 3.50£102 9.63£107 3.32£102 3.90£104 8.87£101 4.95£1016 1.21£1012 1.96£107 1.06£103 Notes: CpG ID corresponds to the unique CpG site identifier in the HumanMethylationEPIC array (Illumina). Chromosomal location denoted according human reference assembly GRCh37/hg19. All depicted P values are Benjamini-Hochberg adjusted p values. M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 7 Fig. 2. Heatmap representing the entire cohort clustered by methylation beta values of the 44 CpGs defining the EPICOVID signature. Cluster analysis was performed using the Ward.D2 clustering method and assuming Euclidean distances. Figure S3. Finally, although our study was specifically designed to curve and calculated the Area Under the Curve (AUC=92.1%) to fur- identify epigenetic biomarkers associated with patients that required ther assess and visualize the model performance (Figure S5). As in hospitalized oxygen therapy vs those asymptomatic or paucisympto- the discovery cohort, supervised hierarchical clustering of the matic who were non-hospitalized (Table S2), we also show the differ- COVID-19 signature in the validation cohort also distinguished two entially methylated CpG sites in all possible comparisons between branches that were significantly enriched with respect to paucisymp- G1/G2/G3 for the discovery, validation and entire cohort (Table S5) to tomatology compared with respiratory failure onset (Fisher’s foster future research efforts in the field. exact test, P = 8.8e16) (Figure S6). The use of the EPICOVID sig- Identifying single DNA methylation loci associated with the onset nature in the supervised hierarchical clustering for the entire of COVID-19 disease could be very useful, but the development an cohort of COVID-19 cases (n = 407) confirmed the existence of overall epigenomic signature based on our EWAS data would also be two branches that classified patients as those exhibiting mild extremely helpful for early stratification of COVID-19 patients symptoms or a worse clinical course (Fisher’s exact test, according to their clinical risk. To achieve this, we selected the 44 P = 2.9e50) (Fig. 2), further strengthening the value of the dis- methylation sites that, adjusted by gender and age, were correlated covered DNA methylation signature to indicate the risk of disease with COVID-19 severity (Table S2) to train in our discovery cohort severity. Importantly, the EPICOVID signature was not only asso- (n = 207), using a metamodel generated by combinating six different ciated with clinical parameters of COVID-19 severity, but also machine learning algorithms (see Methods). Using this approach, we with its corresponding laboratory findings [46,47] such as the lev- obtained an epigenomic signature, hereafter referred to as the EPICO- els of the proinflammatory cytokine interleukin-6, c-reactive pro- VID signature, that predicted COVID-19 severity with 90.18% accu- tein, ferritin, fibrinogen and D-dimer, and the total lymphocyte racy (mean Kappa = 0.804). Supervised hierarchical clustering using count (for all cases Fisher’s exact test, P < 0.01) (Table S6). the COVID-19 signature also clearly distinguished two branches that To further demonstrate the specificity of the EPICOVID signature were significantly enriched with respect to each condition: those for COVID-19, we determined if the identified epigenomic profile who were asymptomatic/paucisymptomatic and those with respira- was overrepresented in DNA methylation datasets available in the tory failure (Fisher’s exact test, P = 7.29e18) (Figure S4), providing public functional genomics data repository GEO for other different additional evidence of the relevance of this epigenomic signature to respiratory inflammatory pathologies and diseases that involve disease severity. hyperactivation of the inflammatory cascade. We observed that the Once we had found that EPICOVID signature was a predictor of EPICOVID signature was not enriched in other respiratory inflamma- COVID-19 disease severity in the discovery cohort, we examined tory diseases such as tuberculosis [48], chronic obstructive pulmo- whether the signature retained its predictive value in our validation nary disease (COPD) [49], asthma [50] and other respiratory allergies cohort of COVID-19 patients (n = 200). We observed that the EPICO- [51] (For all cases, hypergeometric test P > 0.05). In addition, the EPI- VID-positive signature was also associated with worse COVID-19 COVID signature was also not enriched in other inflammatory or viral clinical course with a specificity of 88.18%, a sensitivity of 77.78%, and infectious diseases such as rheumatoid arthritis [52], multiple sclero- positive and negative predictive values (PPV and NPV) of 84.34% and sis [53], systemic lupus erythematosus [54], Sjo€ gren’s syndrome [55], 82.91%, respectively. The accuracy was 83.5% and the mean Kappa inflammatory bowel disease [56], hepatitis C infection [57] and 0.6643. We also plotted the Receiver Operating Characteristic (ROC) amongst HIV-infected individuals [58] (For all cases, hypergeometric 8 M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 test P > 0.05). Thus, these results further support the specificity of the immune responses [35]. In this regard, viral infections of the lung EPICOVID signature for COVID-19 cases. associated with abnormal pulmonary inflammation, pneumonitis, We validated the DNA methylation data derived from the micro- fibrosis and respiratory tract sequelae have related AIM2 hyperactiv- array platform with a different method, pyrosequencing (Methods). ity [63,64]. The EWAS-derived candidate HLA-C also probably con- The pyrosequencing results of the methylation levels of 21 CpG sites tributes to the severity of COVID-19. HLA-C is found on the cell (from the 44 CpG sites that define the EPICOVID signature) compar- surface and its main function is to display intracellular peptide frag- ing asymptomatic/paucisymptomatic and severe COVID-19 cases (a ments to CD8+ cytotoxic T cells, triggering an immune response [36], total of 39 patients) matched the DNA methylation microarray data. a mechanism used for virus-infected host cells to induce immune Illustrative examples are shown in Figure S7. The use of more user- surveillance and eliminate infected cells. However, as also occurred friendly PCR approaches, such as the described pyrosequencing tech- with AIM2, the overactivation of HLA-C can cause autoimmune reac- nology, could facilitate the analyses at the common hospital labora- tions and the excessive release of cytokines leading to an exacerbated tory level. In this regard, it is worth noting that, if instead of the immune response. In this regard, HLA-C genetic variants are known comprehensive EPICOVID signature we selected the top five CpG sites to be associated with COVID-19 clinical outcome [37-40]. Importantly associated with severity according to P-value (Table S2), its single dif- for the natural history of COVID-19 infection, GWAS have also shown ferential methylation status was still associated with COVID-19 that HLA-C constitutes a susceptibility locus for human pneumonia severity (69%76% accuracy range). Thus, a more restricted signature [65,66]. Similar links between the other candidate epigenetic targets derived from the heatmap and clustering analyses might be useful in identified here and immune imbalance associated with lung inflam- future prospective multicentre studies. mation can be found, such as the case of IFI44L, which is overex- Finally, having demonstrated the EPICOVID signature to be associ- pressed in the bronchoalveolar lavage of severe COVID-19 patients ated with patients who developed severe COVID-19 symptomatol- [67] and SARS-CoV-2-infected bronchial epithelial cells [68]. Thus, ogy, we set out to measure the frequency of the epigenomic profile in the genes identified by EWAS are bona fide players in COVID-19 path- the general population. To achieve this goal, we determined the DNA ophysiology that merit further functional and translational attention. methylation status of whole blood samples of 338 healthy individuals The development of the EPICOVID signature to predict the risk of interrogated with the same epigenomic microarray platform [59-62]. COVID-19 severity upon SARS-CoV-2 infection could also be of value All of them were  61-year-old adults, as in our discovery and valida- in the clinical setting. Given the pandemic extension of the disease, tion cohorts. Most importantly, all these samples were collected all the actions that might improve patient care by optimizing the effi- before the emergence of COVID-19, so these donors were never ciency of the medical resources and costs are welcome additions to exposed to the SARS-CoV-2 virus. Using the described cohort, we our prognostic portfolio. In this regard, early categorization and strat- observed that only 13.3% (45 of 338) of individuals exhibited the EPI- ification of COVID-19 patients according to their potential clinical COVID+ signature, whereas 86.7% (293 of 338) did not carry the DNA severity could also help reduce associated morbidity and mortality. methylation profile derived from our study that was associated with The epigenetic biomarker-based COVID-19 risk assessment signature a severe COVID-19 clinical course. Without any definitive epidemio- developed in our study, combined with other genetic, cellular, sero- logical data indicating which percentage of the described worldwide logical and clinical parameters, could identify patients who require adult population infected by the SARS-CoV-2 virus will eventually close monitoring and early active treatments to prevent the progres- require hospitalization with oxygen therapy, the epigenomic signa- sion of the disease as far as possible. Our findings might also have ture could be helpful to provide such estimation. These findings rein- implications for COVID-19 vaccination. In these pressing times, force the aforementioned association between the EPICOVID+ where the newly developed vaccines against COVID-19 are not yet signature and the severity of the disease, and its potential role as a universally available, the identification of particular sets of individu- useful biomarker in the design of early prevention and therapeutic als who become critically ill when infected, compared with those strategies to reduce the morbidity and mortality of the disorder. who experience no or mild symptoms could be of great benefit to public health triage. This is an exciting area in which comprehensive 4. Discussion and rigorous studies, dealing with more biological questions, are warranted in the immediate future. For example, vaccine adjuvants To our knowledge, this is the first study to report an epigenome- activate inflammasomes, amongst them the one assembled by AIM2 wide association with COVID-19 severity. Using a large collection of [69], and it is possible that epigenetic variants of this gene and others individuals who were categorized as positive for the SARS-CoV-2 influence the efficacy of COVID-19 vaccination. virus and achieved very different outcomes, from asymptomatic or In conclusion, our results indicate that the epigenomic back- mild symptomatology to ward hospitalization with respiratory sup- ground of the individuals infected by SARS-CoV-2 affects the degree port that could even require invasive mechanical ventilation, we of COVID-19 severity. Together with other laboratory and clinical interrogated close to 850,000 DNA methylation sites of the human characteristics, it would be helpful to be able to predict who will genome and identified epigenetic loci that were associated with the experience a pure asymptomatic onset or a severe form that compro- worse clinical course. The epigenomic landscape of the severe COVID- mises respiratory function in patients with no associated comorbid- 19 patients obtained reflects an enrichment of genes involved in ities. For other COVID-19 populations, such as older patients and interferon-related pathways, involving innate and adaptive immu- those with previous pathologies, the extrapolation of these data nity, from interleukin and chemokine activity and viral response net- would require further research. The DNA methylation landscape dis- works to the major histocompatibility and inflammasome complexes. covered relates to the immune function of the host, suggesting that The most prominent cases are those revealed by the DNA methyl- when the infection has not been cleared by the immune system, it ation susceptibility sites discovered in the AIM2 and HLA-C genes. elicits an inflammatory response in the lung and other organs medi- The AIM2 gene is expressed in the cellular cytosol and, upon detec- ated by inflammasome, major histocompatibility complex, cytokine, tion of host-DNA released in the context of viral infections, initiates interleukin and interferon-induced genes that are associated with the assembly of the inflammasome complex, subsequent activation the clinical severity of COVID-19. of the caspase-1 and a marked innate immune response involving release of cytokines IL-1b and IL-18. AIM2 cannot distinguish Contributors between viral DNA and self-DNA, so, whereas this lack of specificity provides a good defence against a wide spectrum of pathogens, the MCdM, VD and ME designed the study, contributed to the analy- activation of AIM2 also affects host cells that produce exacerbated sis, and wrote the first draft of the manuscript. LPS and AP did further M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 9 data analysis. DAE and CA performed the DNA methylation microar- [16] Moran S, Martínez-Cardu  s A, Sayols S, et al. Epigenetic profiling to classify cancer ray assay. In-depth clinical and pathological characterization and of unknown primary: a multicentre, retrospective analysis. Lancet Oncol 2016;17:1386–95. recruitment of patients were carried out by MR, SAA, JT, JVR, VVS, [17] Capper D, Stichel D, Sahm F, et al. Practical implementation of DNA methylation ARP, JVG, JPH, SA, CC, AR, LR, BD, DD, MJA, LLC, AMP, JPT, IFC, PV, JT, and copy-number-based CNS tumor diagnostics: the Heidelberg experience. Acta RC, AMN, PSP, FV and AP. All authors helped draft the manuscript or Neuropathol 2018;136:181–210. [18] Fernandez AF, Rosales C, Lopez-Nieva P, et al. The dynamic DNA methylomes of revise it critically for significant intellectual content, and made sub- double-stranded DNA viruses associated with human cancer. Genome Res stantial contributions to the concept and design of the study and 2009;19:438–51. acquisition, analysis and interpretation of data. [19] Pei Y, Wong JH, Robertson ES. Herpesvirus Epigenetic Reprogramming and Onco- genesis. Annu Rev Virol 2020;7:309–31. [20] Moron-Lopez S, Urrea V, Dalmau J, et al. The genome-wide methylation profile of Data sharing CD4+ T cells from HIV-infected individuals identifies distinct patterns associated with disease progression. Clin Infect Dis 2020 published online Jul 26. doi: 10.1093/cid/ciaa1047. The complete DNA methylation raw data of the 407 COVID-19 [21] Oriol-Tordera B, Berdasco M, Llano A, et al. Methylation regulation of antiviral cases is available on the GEO repository under accession number host factors, interferon stimulated genes (ISGs) and T-cell responses associated GSE168739 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? with natural HIV control. PLoS Pathog 2020;16:e1008678. [22] Grifoni A., Weiskopf D., Ramirez S.I., et al. Targets of T cell responses to SARS-CoV- acc=GSE168739. 2 coronavirus in humans with COVID-19 disease and unexposed individuals. Cell2020; 181: 1489501.e15. [23] Kreer C, Zehner M, Weber T, et al. Longitudinal isolation of potent near-germline Declaration of Competing Interest SARS-CoV-2-neutralizing antibodies from COVID-19 patients. Cell 2020;182:843– 54 e12. Dr. Esteller reports grants from Ferrer International, personal fees [24] Meckiff BJ, Ramírez-Sua stegui C, Fajardo V, et al. Imbalance of regulatory and cyto- toxic SARS-CoV-2-reactive CD4+ T cells in COVID-19. Cell 2020;183:1340–53 e16. from Quimatryx, outside the submitted work. The other authors [25] Loo Yau H, Ettayebi I, De Carvalho DD. The cancer epigenome: exploiting its vul- declare no conflicts of interest. nerabilities for immunotherapy. Trends Cell Biol 2019;29:31–43.  [26] Villanueva L, Alvarez-Errico D, Esteller M. The contribution of epigenetics to can- cer immunotherapy. Trends Immunol 2020;41:676–91. Acknowledgements [27] Duruisseaux M, Martínez-Cardu  s A, Calleja-Cervantes ME, et al. Epigenetic pre- diction of response to anti-PD-1 treatment in non-small-cell lung cancer: a multi- We thank the Health Department and Centres de Recerca de Cata- centre, retrospective analysis. Lancet Respir Med 2018;6:771–81. [28] Jung H, Kim HS, Kim JY, et al. DNA methylation loss promotes immune evasion of lunya (CERCA) Programme of the Generalitat de Catalunya, the Josep tumours with high mutation and copy number load. Nat Commun 2019;10:4278. Carreras Leukaemia Foundation and the Cellex Foundation for institu- [29] Heyn H, Sayols S, Moutinho C, et al. Linkage of DNA methylation quantitative trait tional support. loci to human cancer risk. Cell Rep 2014;7:331–8. [30] Ongen H, Andersen CL, Bramsen JB, et al. Putative cis-regulatory drivers in colo- rectal cancer. Nature 2014;512:87–90. Supplementary materials [31] Moran S, Arribas C, Esteller M. Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences. Epige- nomics 2016;8:389–99. Supplementary material associated with this article can be found, [32] Teschendorff AE, Menon U, Gentry-Maharaj A, et al. An epigenetic signature in in the online version, at doi:10.1016/j.ebiom.2021.103339. peripheral blood predicts active ovarian cancer. PLoS ONE 2009;4:e8274. [33] Iorio F, Knijnenburg TA, Vis DJ, et al. A landscape of pharmacogenomic interac- tions in cancer. Cell 2016;166:740–54. References [34] van der Made CI, Simons A, Schuurs-Hoeijmakers J, et al. Presence of genetic var- iants among young men with severe COVID-19. JAMA 2020;324:1–11. [1] Zhu N, Zhang D, Wang W, et al. China novel coronavirus investigating and [35] Kumari P, Russo AJ, Shivcharan S, Rathinam VA. AIM2 in health and disease: research team. A novel coronavirus from patients with pneumonia in China. N inflammasome and beyond. Immunol Rev 2020;297:83–95. Engl J Med 2020 2019;382:727–33. [36] Dendrou CA, Petersen J, Rossjohn J, Fugger L. HLA variation and disease. Nat Rev [2] Coronaviridae study group of the international committee on taxonomy of Immunol 2018;18:325–39. viruses. The species severe acute respiratory syndrome-related coronavirus: clas- [37] Sakuraba A, Haider H, Sato T. Population difference in allele frequency of HLA- sifying 2019-nCoV and naming it SARS-CoV-2. Nat Microbiol 2020;5:536–44. C*05 and its correlation with COVID-19 mortality. Viruses 2020;12:1333. [3] Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavi- [38] Lorente L, Martín MM, Franco A, et al. HLA genetic polymorphisms and prognosis rus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 of patients with COVID-19. Med Intensiva 2020 published online Sep 6.. doi: cases from the Chinese center for disease control and prevention. JAMA 10.1016/j.medin.2020.08.004. 2020;323:1239–42. [39] Pisanti S, Deelen J, Gallina AM, et al. Correlation of the two most frequent HLA [4] Berlin DA, Gulick RM, Martinez FJ. Severe Covid-19. N Engl J Med 2020 published haplotypes in the Italian population to the differential regional incidence of online May 15.. doi: 10.1056/NEJMcp2009575. Covid-19. J Transl Med 2020;18:352. [5] Marini JJ, Gattinoni L. Management of COVID-19 respiratory distress. JAMA [40] Moura RR, Agrelli A, Santos-Silva CA, et al. Immunoinformatic approach to assess 2020;323:2329–30. SARS-CoV-2 protein S epitopes recognised by the most frequent MHC-I alleles in [6] Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mortality of adult inpa- the Brazilian population. J Clin Pathol 2020 published online Aug 5. doi: 10.1136/ tients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet jclinpath-2020-206946. 2020;395:1054–62. [41] Benson KK, Hu W, Weller AH, et al. Natural human genetic variation determines [7] Li X, Xu S, Yu M, et al. Risk factors for severity and mortality in adult COVID-19 basal and inducible expression of PM20D1, an obesity-associated gene. Proc Natl inpatients in Wuhan. J Allergy Clin Immunol 2020;146:110–8. Acad Sci U S A 2019;116:23232–42. [8] Covid-19 GWAS Group Severe, Ellinghaus D, Degenhardt F, et al. Genomewide [42] Satake W, Nakabayashi Y, Mizuta I, et al. Genome-wide association study identi- association study of severe Covid-19 with respiratory failure. N Engl J Med fies common variants at four loci as genetic risk factors for Parkinson's disease. 2020;383:1522–34. Nat Genet 2009;41:1303–7. [9] Pairo-Castineira E, Clohisey S, Klaric L, et al. Genetic mechanisms of critical illness [43] Sanchez-Mut JV, Heyn H, Silva BA, et al. PM20D1 is a quantitative trait locus asso- in Covid-19. Nature 2020 published online Dec 11. doi: 10.1038/s41586-020- ciated with Alzheimer's disease. Nat Med 2018;24:598–603. 03065-y. [44] Sanchez-Mut JV, Glauser L, Monk D, Gra €ff J. Comprehensive analysis of PM20D1 [10] Zhang Q, Bastard P, Liu Z, et al. Inborn errors of type I IFN immunity in patients QTL in Alzheimer's disease. Clin Epigenetics 2020;12:20. with life-threatening COVID-19. Science 2020;370:eabd4570. [45] Singh H, Choudhari R, Nema V, Khan AA. ACE2 and TMPRSS2 polymorphisms in [11] Bastard P, Rosen LB, Zhang Q, et al. Autoantibodies against type I IFNs in patients various diseases with special reference to its impact on COVID-19 disease. Microb with life-threatening COVID-19. Science 2020;370:eabd4585. Pathog 2020;150:104621. [12] Feinberg AP. The key role of epigenetics in human disease prevention and mitiga- [46] Ji P, Zhu J, Zhong Z, et al. Association of elevated inflammatory markers and tion. N Engl J Med 2018;378:1323–34. severe COVID-19: a meta-analysis. Medicine (Baltimore) 2020;99:e23315. [13] Berdasco M, Esteller M. Clinical epigenetics: seizing opportunities for translation. [47] Chaudhary R, Garg J, Houghton DE, et al. Thrombo-inflammatory biomarkers in Nat Rev Genet 2019;20:109–27. COVID-19: systematic review and meta-analysis of 17,052 patients. Mayo Clin [14] Esteller M, Garcia-Foncillas J, Andion E, et al. Inactivation of the DNA-repair gene Proc Innov Qual Outcomes 2021 Feb 8. doi: 10.1016/j.mayocpiqo.2021.01.009. MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med [48] Chen YC, Hsiao CC, Chen TW, et al. Whole genome DNA methylation analysis of 2000;343:1350–4. active pulmonary tuberculosis disease identifies novel epigenotypes: pARP9/ [15] Hegi ME, Diserens AC, Gorlia T, et al. MGMT gene silencing and benefit from miR-505/RASGRP4/GNG12 gene methylation and clinical phenotypes. Int J Mol temozolomide in glioblastoma. N Engl J Med 2005;352:997–1003. Sci 2020;21:3180. 10 M. Castro de Moura et al. / EBioMedicine 66 (2021) 103339 [49] Chen YC, Tsai YH, Wang CC, et al. Epigenome-wide association study on asthma [60] Novakovic B, Lewis S, Halliday J, et al. Assisted reproductive technologies are and chronic obstructive pulmonary disease overlap reveals aberrant DNA methyl- associated with limited epigenetic variation at birth that largely resolves by ations related to clinical phenotypes. Sci Rep 2021;11:5022. adulthood. Nat Commun 2019;10:3922. [50] Yang IV, Richards A, Davidson EJ, et al. The nasal methylome: a key to under- [61] Soriano-Ta rraga C, Lazcano U, Giralt-Steinhauer E, et al. Identification of 20 novel standing allergic asthma. Am J Respir Crit Care Med 2017;195:829–31. loci associated with ischaemic stroke. Epigenome-wide association study. Epige- [51] Langie SA, Szarc Vel Szic K, Declerck K, et al. Whole-genome saliva and blood DNA netics 2020;15:988–97. methylation profiling in individuals with a respiratory allergy. PLoS ONE 2016;11 [62] Gallego-Fabrega C, Cullell N, Soriano-Ta rraga C, et al. DNA methylation of MMPs e0151109. and TIMPs in atherothrombosis process in carotid plaques and blood tissues. [52] Liu Y, Aryee MJ, Padyukov L, et al. Epigenome-wide association data implicate Oncotarget 2020;11:905–12. DNA methylation as an intermediary of genetic risk in rheumatoid arthritis. Nat [63] Yoo JK, Kim TS, Hufford MM, Braciale TJ. Viral infection of the lung: host response Biotechnol 2013;31:142–7. and sequelae. J Allergy Clin Immunol 2013;132:1263–76. [53] Kular L, Liu Y, Ruhrmann S, et al. DNA methylation as a mediator of HLA-DRB1*15:01 [64] Ma R, Ortiz Serrano TP, Davis J, Prigge AD, Ridge KM. The cGAS-STING pathway: and a protective variant in multiple sclerosis. Nat Commun 2018;9:2397. the role of self-DNA sensing in inflammatory lung disease. FASEB J [54] Joseph S, George NI, Green-Knox B, et al. Epigenome-wide association study of 2020;34:13156–70. peripheral blood mononuclear cells in systemic lupus erythematosus: identifying [65] Tian C, Hromatka BS, Kiefer AK, et al. Genome-wide association and HLA region DNA methylation signatures associated with interferon-related genes based on fine-mapping studies identify susceptibility loci for multiple common infections. ethnicity and SLEDAI. J Autoimmun 2019;96:147–57. Nat Commun 2017;8:599. [55] Brække Norheim K, Imgenberg-Kreuz J, Jonsdottir K, et al. Epigenome-wide DNA [66] Khadzhieva MB, Kuzovlev AN, Salnikova LE. Pneumonia: host susceptibility and methylation patterns associated with fatigue in primary Sjo € gren's syndrome. shared genetics with pulmonary function and other traits. Clin Exp Immunol Rheumatology (Oxford) 2016;55:1074–82. 2019;198:367–80. [56] Ventham NT, Kennedy NA, Adams AT, et al. Integrative epigenome-wide analysis [67] Shaath H, Vishnubalaji R, Elkord E, Alajez NM. Single-cell transcriptome analysis demonstrates that DNA methylation may mediate genetic risk in inflammatory highlights a role for neutrophils and inflammatory macrophages in the pathogen- bowel disease. Nat Commun 2016;7:13507. esis of severe COVID-19. Cells 2020;9:2374. [57] Zhang X, Hu Y, Justice AC, Li B, et al. DNA methylation signatures of illicit drug [68] Vishnubalaji R, Shaath H, Alajez NM. Protein coding and long noncoding RNA injection and hepatitis C are associated with HIV frailty. Nat Commun (lncRNA) transcriptional landscape in SARS-CoV-2 infected bronchial epithelial 2017;8:2243. cells highlight a role for interferon and inflammatory response. Genes (Basel) [58] Oriol-Tordera B, Berdasco M, Llano A, et al. Methylation regulation of antiviral 2020;11:760. host factors, interferon stimulated genes (ISGs) and T-cell responses associated [69] Reinke S, Thakur A, Gartlan C, Bezbradica JS, Milicic A. Inflammasome-mediated with natural HIV control. PLoS Pathog 2020;16:e1008678. immunogenicity of clinical and experimental vaccine adjuvants. Vaccines (Basel) [59] Salas LA, Koestler DC, Butler RA, et al. An optimized library for reference-based 2020;8:554. deconvolution of whole-blood biospecimens assayed using the Illumina Human- MethylationEPIC BeadArray. Genome Biol 2018;19:64.