key: cord-1013998-332stbjl
authors: Liu, Tianyuan; Nogueira, Leandro Balzano; Lleo, Ana; Conesa, Ana
title: Transcriptional differences for COVID-19 Disease Map genes between males and females indicate a different basal immunophenotype relevant to the disease
date: 2020-10-01
journal: bioRxiv
DOI: 10.1101/2020.09.30.321059
sha: 315a17497c771e92afbac989263b1dfb7a6e408c
doc_id: 1013998
cord_uid: 332stbjl

Worldwide COVID-19 epidemiology data indicate clear differences in disease incidence among sex and age groups. Specifically, male patients are at a higher death risk than females. However, whether this difference is the consequence of a pre-existing sex-bias in immune genes or a differential response to the virus has not been studied yet. We created DeCovid, an R shiny app that combines gene expression data of different human tissue from the Genotype-Tissue Expression (GTEx) project and the COVID-19 Disease Map gene collection to explore basal gene expression differences across healthy demographic groups. We used this app to study differential gene expression between men and women for COVID-19 associated genes. We identified that healthy women present higher levels in the expression of interferon genes and the JAK-STAT pathway leading to cell survival.

acknowledges that a vaccine solution for the general population might not be ready before mid or late 2021. At the same time, very recent re-infection cases suggest that immunity may not always persist, and active disease surveillance of past cases may be required. As the second wave of the pandemic steadily progresses in Europe and the US, possibly with new variants of the virus and affecting a more diverse and younger population, novel or improved treatment targets and options are likely to remain as important for the management of the disease.

One of the most intriguing aspects of COVID-19 is the different grade of severity it affects different people. Although the acknowledged risk factors such as pre-existing conditions such as cardiovascular disease (Zheng et al., 2020; D. Wang et al., 2020; Chen et al., 2020) , diabetes (Apicella et al., 2020; Fang et al., 2020; Guo et al., 2020, 19) , obesity (Simonnet et al., 2020) , age (Liu et al., 2020; CDC COVID-19 Response Team, 2020; LitCovid) and sex (higher mortality in men than women (Gebhard et al., 2020; Jin et al., 2020) suggest a relationship between the physical condition and disease progression, the precise pathways of this relationship have not been clearly established yet. Sex-associated differences between males and females that have been proposed to affect COVID-19 incidence include lifestyle (e.g., smoking and drinking) and mental health. The smoking population of males is higher than females, and smoking may be a risk factor for severe COVID-19 (Cai, 2020, 19; Jamal et al., 2016; Grundy et al., 2020) . In contrast, men differ from women during the pandemic in sleeping and signs of depression (C. . A recent study showed that male and female COVID-19 patients differed in their immune response, with the first showing a stronger cytokine response while the second is having a higher T-cell activation pattern (Takahashi et al., 2020) . Interestingly, the pattern of COVID-19 risk factors is not fully shared with other similar recent pandemics such as SARS, MERS, H1N1. For example, SARS patients were more frequent among the healthy young people (Yin and Wunderink, 2018) . Age and complications, but not gender, was the most significant risk factors for mortality in the Arab and South Korean Studies of MERS (Park et al., 2018) .

Several studies on the 2009 H1N1 influenza revealed the younger age, chronic conditions, and female sex as risk factors of the disease (Viasus et al., 2011) . This suggests that the observed COVID-19 risk factors, especially those associated with age and sex groups, might be the result of specific interactions of the SARS-CoV-2 with intrinsic physiological characteristics, including the basal immunophenotype, of these population groups. However, which specific immunological characteristics imprint an existing condition that how this relates to COVID-19 remains to be explored. The COVID-19 Disease Map is a valuable resource to investigate the molecular responses to SARS2-CoV-2 infection and understand the biological pathways leading to the severe manifestations of the disease. This database also creates an opportunity for interrogating existing molecular data on differences associated with COVID-19 risk factors in the general population.

In this paper, we present the DeCovid app, a Shiny app, to explore basal expression level differences in COVID-19 disease map genes between men and women and different age groups. We used data from the GTEx database, which contains RNA-seq profiles for hundreds of demographically diverse healthy individuals in multiple tissues and allows us to interrogate COVID-19 genes globally or individually. We used this resource to study basal expression differences in COVID-19 genes between man and woman and found that a different immunological state is prevalent in both genders.

Interestingly some of these differential pathways have been shown to be relevant for disease severity progression and be characteristic of the sex-biased immune response to the virus, providing ground for hypotheses on the molecular basis of the COVID-19 sex-bias. We anticipate that the DeCovid app could be a useful tool for researchers to explore the molecular etiology of COVID-19 demographic differences.

The DeCovid shiny app combines a selection of human tissue specific GTEx data with the COVID-19 Disease Map database to allow quick exploration of basal gene expression values and differences in the healthy human population for genes described to be important for COVID-19. We included data from blood, lungs, heart, kidney, stomach, and brain for being tissues described to be affected by SARS-CoV-2 infection.

The sample size for different tissues is different in the GTEx datasets. The largest sample size is the whole blood with 330 samples, while the smallest dataset is the stomach tissue with 57 samples (Figure 1a) . The total number of collected COVID-19

Disease Map is genes 1,523. A Gene Ontology (GO) term enrichment analysis indicates that genes related to cytokines, response to virus, innate immune response, transcriptional regulation, and kinase activity populate this collection (Figure 1b) . The DeCovid app is a user-friendly implementation to allow an analysis of these data for researchers without strong bioinformatics skills (Figure 2) . Extensive documentation and video tutorials are provided to the facility for a quick start. Users should indicate the demographic factor for differential expression analysis. In case age is selected, the age threshold value to classify samples as old or young should be specified. Once a significance p.value and a fold-change threshold are provided, the app computes differential expression and provides results as a list of genes and explore the distribution of gene expression values for specific genes of the COVID-19

Disease Map collection or investigate the enrichment of specific immune responserelated functions among genes with sex or age expression biases.

Screenshot of the DeCovid app showing on the left dialog panels for input parameters (contrast type and significance thresholds), and on the right results panels with list of differentially expressed genes, gene-specific expression plots and enrichment analysis.

Using DeCovid, we analyzed the extent of gene expression differences among demographic groups across different COVID-19 affected tissues (Table 1) . The analysis revealed a wide-range of significant gene expression differences for COVID-19 genes between when comparing sexes and age groups for all evaluated tissues.

Differences where lower when imposing a fold-change threshold of 0.5 for the magnitude of the mean expression level. Whole-blood were the tissue with the highest number of expression differences for COVID-19 genes followed by the brain. From this simple analysis, we concluded that the COVID-19 Disease Map represents a disease signature with significant differences across demographic groups that may be worth exploring for hypothesizing on the molecular basis for differences in disease incidence among these groups. DeCovid identified basal sex-biased expression differences in immune-response genes associated to differential disease incidence between man and women Since COVID-19 fatality rates are consistently higher in men than women in many different countries worldwide (Supplementary Figure 1) , we hypothesized that sex is a pre-existing condition that influences the severity of the disease and that the study of basal differences in expression differences in COVID-19 relevant genes in healthy individuals could reveal insights into the molecular basis of these differences. We used the DeCovid app to explore these differences and concentrated on the blood tissue for capturing systemic immune responses. A total of 67 genes (p.value < 0.05 and fold change > 0.5) were differentially expressed between men and women in blood samples, regardless of the age group (Figure 3a) . We found high expression in females of genes involved in the innate immune response, particularly type I interferons, such as   IFNA17, IFNA2, IFNE, IFIT1, and IFIT3, anti-apoptotic regulators such showed that female patients have a high T-cell level (Takahashi et al., 2020) , which were postulated as a critical factor for the differential fatality incidence between sexes. Therefore, with a high level of baseline expression of type 1 interferons, female patients are more likely to maintain a high T-cell level in the early stage, which might lead to a milder evolution of the disease. Here we speculate that these innate immunity pathways may play a crucial role in the a b early control of respiratory coronavirus infection (Figure 3b) . Moreover, data indicate higher expression in women of JAK, multiple members of the STAT family, and cell survival genes such as MCL1 and BCL2A. It has been shown that the Jak-STAT signaling pathway mediates the synthesis of anti-apoptotic regulators such as MCL1, BCL2A, and BCL2L1 upon induction of high type 1 interferon levels (Sepúlveda et al., 2007) , agreeing with the observed gene expression phenotype in females.

Finally, we detected significant increased tumor necrosis factor TNFSF13B expression in females. This gene stimulates the synthesis of chemokines CCL21 and CXCL12 through NFkB subunit 2 pathway, leading to B-cell homing through the synthesis of CCL13 (López-Giral et al., 2004) , lymphopoiesis through the synthesis of CXCL12 (Silva et al., 2016; Sen, 2006) , and development and survival through the synthesis of TNFSF13B (Caamaño et al., 1998; Gerondakis and Siebenlist, 2010) .

Since B-cells are part of the humoral immunity component of the adaptive immune system and are responsible for mediating the production of antigen-specific immunoglobulins (Ig) directed against invasive pathogens, higher levels of these chemokines in females might lead to a faster response against a potential viral infection.

In males, in turn, the mentioned processes were comparatively downregulated, while the only process upregulated compared to females were those related to the expression of pro-inflammatory cytokines (Figure 3b) , particularly IL17, IL22 and IL3. Also, chemokine CXCL2 levels are higher in men, which supports the hypothesis of a first stage of male response against a viral infection being more related to inflammation and Th17 cell differentiation processes.

Here we present the DeCovid app as a resource to explore sex and age differential expression patterns in the healthy population for genes described to be involved in COVID-19 disease pathways. The GTEx data, used in this work has been recently mined for sex-specific expression differences concluding that these are tissue-specific and generally small, also observed in our analyses (Oliva et al., 2020) . This study focused on the sex-specific genetic effects on gene expression associated to complex genetic traits. In our application we repurpose the valuable GTEx database for the study of COVID-19 relevant pathways while providing a mechanistic model to interpret differential expression results.

We used DeCovid to investigate the gene expression differences between men and woman that could explain differences in disease severity among these demographic groups, with men being more likely to have a fatal outcome than women. We found that two major immune response pathways were differentially expressed between men and women in the healthy population. Women showed higher expression of interferon genes while men have higher levels of pro-inflammatory Th17 cytokine genes. These two pathways have been proposed to be critical factors for fatal response to the virus infection, with patients that present an interferon-mediated response having a better prognosis than those responding with a massive cytokine activation (Blanco-Melo et al., 2020, 19; Hadjadj et al., 2020) . In agreement, animal models of SARS-CoV-1 and MERS-CoV infection showed that failure to elicit an early type 1 interferon response correlates with the disease severity (Channappanavar et al., 2016) . Perhaps more importantly, these models demonstrate that timing is key, as type 1 interferons are protective at the early stages of a viral disease. On the other hand, clinical studies showed a differences in the immunophenotype between female and male COVID-19

patients with women having a more robust T-cells activation while male patients presenting a higher plasma level of pro-inflammatory cytokines (Takahashi et al., 2020) . In this study we showed that these immune system differences, which are critical to COVID-19 progression, represent a sex-associated pre-existing condition that is likely to create a differential predisposition to disease outcome between men and women and may explain the observed sex-bias in COVID-19 mortality. Future studies should address whether these gene expression patterns translate into functional differences in patients infected with SARS-CoV-2, whether they have prognosis value and of if they present any relationship with hormonal levels, which have also been linked to COVID-19 severity (Strope et al., 2020; Channappanavar et al., 2017; Scully et al., 2020) . All together our results show that DeCovid is a useful resource to investigate the status of COVID-19 genes across demographic groups and human tissues and postulate on the differential pathways to disease operating in each case.

We used RNA-seq data from the Genotype-Tissue Expression project (GTEx) that contains data from 44 tissue sources obtained from deceased but considered healthy individuals from both sexes and a wide range of ages (Lonsdale et al., 2013) . Gene read counts matrix and the annotation files from the GTEx portal (https://www.gtexportal.org/home/datasets). In other to provide a homogenous set of human tissue reference samples, GTEx data were analyzed for possible biases, and samples belong to individuals with a "ventilator case" label as a cause of death were discarded as they clearly segregated from remaining samples on a Principal Component Analysis plot (Supplementary Figure 1) . The COVID-19 Disease Map genes were downloaded from https://github.com/wikipathways/cord-19 and available on March 17th, 2020. The list is mined from 9996 PMC papers in COVID-19 Open Research Dataset (CORD-19) using machine learning approaches, so they are positively related to the COVID-19 disease process (Lu .

The DeCovid software is a Shiny app written in R with a user-friendly interface. The app can be installed through a Docker image or directly download from GitHub (See supplementary materials for installation). DeCovid already has integrated all essential data from GTEx, and no additional downloads are necessary. Differential gene expression is calculated using edgeR (Robinson et al., 2010) and multiple testing correction is applied following the Benjamini and Hochberg(BH)method (Benjamini and Hochberg, 1995) . Results are provided either as lists of differentially expressed genes, heatmaps of sex and age mean expression values, and gene-specific boxplots

showing the distribution of expression values across demographic groups. GO enrichment analysis of significant gene sets is provided through the cluster profile R package and uses the list of COVID-19 disease map as a reference set (Yu et al., 2012) . 

COVID-19 in people with diabetes: understanding the reasons for worse outcomes

COVID-19 Disease Map, building a computational repository of SARS-CoV-2 virus-host interaction mechanisms

MERS transmission and risk factors: a systematic review

edgeR: a Bioconductor package for differential expression analysis of digital gene expression data

Considering how biological sex impacts immune responses and COVID-19 outcomes

Control of B lymphocyte apoptosis by the transcription factor NF-kappaB

BCL-2 expression is mainly regulated by JAK/STAT3 pathway in human CD34 + hematopoietic cells

The TLR4-TRIF Pathway Protects against H5N1 Influenza Virus Infection

Transcription factors of the alternative NF-κB pathway are required for germinal center B-cell development

High Prevalence of Obesity in Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) Requiring Invasive Mechanical Ventilation

SeXX Matters in Infectious Disease Pathogenesis

Are sex discordant outcomes in COVID-19 related to sex hormones?

Sex differences in immune responses that underlie COVID-19 disease outcomes

Factors associated with severe disease in hospitalized adults with pandemic (H1N1) 2009 in Spain

Immediate Psychological Responses and Associated Factors during the Initial Stage of the 2019 Coronavirus Disease (COVID-19) Epidemic among the General Population in China

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan

MERS, SARS and other coronaviruses as causes of pneumonia