Expression and clinical significance of SARS-CoV-2 human targets in lung tissues: normal, primary tumor and metastasis

The recent COVID-19 outbreak in China led to a worldwide pandemic associated with a severe acute respiratory illness. A higher incidence of COVID-19 infection was demonstrated in cancer patients, including patients with lung cancer. This study was conducted to get insights into the reasons for this enhanced frequency of COVID-19 infection. Methods: Using different bioinformatic tools, the expression and methylation pattern of ACE2 and TMPRSS2 gene were analyzed in healthy and malignant tissues with a focus on lung adenocarcinoma (LUAD) and correlated to clinical parameters and smoking history. Results: ACE2 and TMPRSS2 were heterogeneously expressed across 36 healthy tissues with the highest expression in digestive, urinary and reproductive organs, while their expression was significantly lower in 36 cancer tissues. In LUAD, ACE2, but not TMPRSS2 was overexpressed, which inversely correlated to the promoter methylation. An age-dependent upregulation of ACE2 expression was found in LUAD compared to normal lung tissues. In a healthy lung, TMPRSS2 expression was dependent on sex and smoking history and downregulated in LUAD of smokers. Cancer progression was associated with decreased TMPRSS2, but unaltered ACE2 expression, while ACE2 expression in lung metastases of different cancers was higher than in metastasis of other sites. TMPRSS2, but not ACE2 expression, was associated with LUAD patients’ survival. Conclusions: Comprehensive molecular analyses revealed a heterogeneous, distinct expression and methylation profile of ACE2 and TMPRSS2 in healthy lung vs LUAD tissues across sex, age and smoking history, which is associated with clinical parameters and might have implications for COVID-19 disease.

Infection with SARS-CoV, MERS-CoV and COVID-19 caused a severe acute respiratory illness with a mild to severe patients' outcome. In patients with severe disease manifestation, diffuse alveolar damage with severe capillary congestion was detected, suggesting a vascular dysfunction and inflammation [14,15].
There exist several reports on an increased susceptibility and incidence of cancer patients to COVID-19 infection compared to the overall population [19][20][21][22][23][24], in particular in patients older than 60 years with lung carcinoma as the most common [25]. Upon infection, cancer patients often display more severe disease courses with multiple organ dysfunctions [22], higher mortality and intensive care unit admission [24]. Furthermore, smokers have a higher incidence and severity of COVID-19 infection than non-smokers [26]. This might be associated with structural alterations, e.g., fibrosis leading to changes in lung architecture and intra-and peri-tumoral microenvironment and inflammation due to tobacco-related lung damage and lung cancer [26].
Profiling of ACE2 and TMPRSS2 expression and their distribution across different cell types in lung tissues and in cells derived from subsegmental bronchial branches demonstrated high expression levels of TMPRSS2 in all the tissues, whereas ACE2 was heterogeneously expressed in the transient secretory cell type of the subsegmental bronchial branches suggesting that bronchial branches might be more vulnerable for COVID-19 infection [27].
ACE2 expression significantly differed between lung cancer subtypes. A decreased ACE2 expression was found in all non-small cell lung cancer (NSCLC) tissues when compared to healthy control, which negatively correlated with the intensity of neoangiogenesis and sensitivity to cytostatics of tumor cells [28,29]. Concerning NSCLC subentities, ACE2 expression was significantly upregulated in lung adenocarcinoma (LUAD) and remained unchanged in lung squamous cell carcinoma [30].
In order to search for factors to predict susceptibility or severity of COVID-19 infection in   cancer patients, the expression patterns of ACE2 and TMPRSS2 were compared across 36   different human healthy and cancer tissues with a focus on healthy lung and LUAD tissues   and associated to the methylation pattern, sex, age, clinical parameters and smoking history using different bioinformatics tools.

Dataset selection
Datasets were collected from gene expression omnibus (GEO), The Cancer Genome Atlas (TCGA), and individual publications. The analysis of GEO was performed by GENT2 [31].
For datasets not been deposited into GEO, the selection was performed by a literature search and by referencing other commonly used databases.

Analysis of gene expression patterns
Expression data were downloaded from GENT2 system, a publicly accessible online cancer microarray database, which provides a landscape of gene expression profile across 72 normal and tumor tissues. The values of the ACE2 and TMPRSS2 mRNA expression were before analysis in normal and cancer tissues (log2-transformed) and then subjected to statistical comparison.

Meta-analysis
The Lung Cancer Explorer (LCE), an open-access web resource, is housing expression data and clinical data from more than 6700 patients in 56 studies [32]. Meta-analysis effectively combines the statistical strength from multiple data sets, which allows higher precision than using any of the single studies. Forest plots of ACE2 and TMPRSS2 were employed to summarize tumornormal standardized mean difference for tumor vs normal meta-analysis. The Cancer Genome Atlas (TCGA). ACE2 and TMPRSS2 transcriptomes were compared between lung tissues with others in various cancer metastasis.

Co-expression analysis
The cBioPortal online platform [42, 43] was used to query the TCGA lung adenocarcinoma (LUAD). In total, 566 tumor samples (from the TCGA database) with messenger RNA (mRNA) next-generation sequencing data were empoyed.

Statistics
All statistical tests provided, such as a two-sample t-test, log-rank test, and meta-survival analysis, were implemented using R with Bioconductor plugins [31,32]. Co-expressed genes were predicted by cBioPortal online analysis with the Pearson correlation coefficient.
Student's t test was used to compare ACE2 and TMPRSS2 gene expression levels between males and females, younger (< 60 years) and older (> 60 years) individuals and between smokers and non-smokers in healthy and tumor tissues. P-values of less than 0.05 were considered statistically significant.

Heterogeneous expression profile of COVID receptor and COVID infection associated gene in healthy tissues and tumors
A tissue-wide gene expression profile of ACE2 and TMPRSS2 demonstrated a heterogeneous expression of both genes among 36 human healthy tissues using the GLP570 data. ACE2 was highly expressed in gallbladder (12.48 ± 0.54), testis (11.06 ± 0.35), kidney (10.44 ± 0.08), colon (8.9 ± 0.06) and small intestine (8.18 ± 0.84) ( Figure 1A cancer tissues demonstrated significantly lower expression of ACE2 in cancer tissues when compared to normal tissues ( Figure 2A; Table 1). However, detailed analysis revealed that ACE2 was overexpressed in some tumor entities, including lung adenocarcinoma (LUAD) with a 0.301-fold expression difference between tumor and normal tissues (p < 0.001; logFC -0.299). A heterogeneous gene expression pattern was also found for TMPRSS2 with a significant downregulation of TMPRSS2 in lung cancer (p< 0.001; logFC -0.962) ( Figure 2B; Table 1).
Next, meta-analyses based on the lung cancer explorer were performed for ACE2 and  Figure 3B). Interestingly, the ACE2 ( Figure 4A) and TMPRSS2 ( Figure 4C) expression in healthy lung tissues as well as in LUAD was inversely correlated to the methylation pattern of the ACE2 ( Figure 4B) or TMPRSS2 promoter ( Figure   4D), respectively.

Dependence of ACE2 expression in normal lung and LUAD tissues on sex, age and smoking history
The dependence of ACE2 expression on sex, age and smoking history was determined in healthy individuals as well as in LUAD patients. Using the TCGA data set, ACE2 expression was higher in male than in female normal lung tissues (p = 0.17), which was confirmed by Oklahoma and coauthors (p = 0.23) [44]. In contrast, there existed no sex-dependent ACE2 expression in LUAD patients (Supplementary Figure 1A).
Since an age-dependent incidence of COVID-19 infection was reported [45], the ACE2 expression was assessed in both healthy lung tissues and LUAD tissues from individuals of < 60 years and older. In healthy lung tissues, the ACE2 expression was slightly, but not statistically significantly lower in individuals < 60 years (p = 0.32) ( Figure 5A), whereas a statistical upregulation of ACE2 expression (p = 0.0049) was detected in LUAD patients over 60 years ( Figure 5B), while there is no alteration (p = 0.88) in ACE2 expression between normal and LUAD of < 60 years ( Figure 5C); there was a significant higher expression in LUAD patients > 60 years compared to healthy control samples ( Figure 5D).
Besides, a link between ACE2 expression and smoking history was investigated in healthy and LUAD tissues. The ACE2 expression was altered in normal lung tissues of smokers vs.

Dependence of TMPRSS2 expression in normal lung and LUAD tissues on sex, age and smoking history
Next to ACE2, the TMPRSS2 expression levels were analyzed in non-tumorous lung and LUAD tissues and compared to sex, age and smoking history. The  Figure 6H)). Interestingly, a combination of higher age (> 60 years) and smoking history was correlated with higher TMPRSS2 expression in LUAD patients (p = 0.06) (Supplementary Figure 2B).

Clinical relevance of ACE2 and TMPRSS2 expression in LUAD
In order to address whether the higher ACE2 expression in LUAD tissues might be of clinical relevance, the ACE2 expression was determined in different stages of LUAD using the UALCAN database. ACE2 was not differentially expressed ( Figure 7A) or methylated ( Figure   7B) in stage I tumors (n = 277) compared to stage III (n = 85) and stage IV (n = 28) tumors.
Furthermore, no significant differences in the ACE2 expression/methylation levels of lymph node metastasis (N1-N3) ( Figure 7C, Figure 7D) and primary tumor LUAD without metastasis (NO) was detected. In contrast, ACE2 expression was higher in lung metastasis of other tumor entities, e.g., breast, prostate and liver carcinoma, than in metastasis of these malignancies in other locations, e.g., lymph node, brain, and liver (Table 2 and Supplementary Figure 3). Thus, the ACE2 expression levels were more pronounced in lung metastasis independent of the tumor type. In contrast to ACE2, TMPRSS2 was downregulated during disease progression from stage I to stage IV ( Figure 7E), which was directly associated with enhanced promoter methylation of TMPRSS2 ( Figure 7F). Likewise, the expression ( Figure 7G) of TMPRSS2 in LUAD was more reduced coupled with increased methylation ( Figure 7H) in metastasis compared to primary tumors.

Correlation of ACE2 and TMPRSS2 expression with patients' survival
To determine whether the level of ACE2 expression was associated with the overall survival (OS) of LUAD patients, meta-analyses were performed from 19 studies. No correlation of ACE2 expression with the OS of LUAD patients (HR: 0.95; p = 0.26) was found ( Figure 8A).
In contrast, a meta-analysis of TMPRSS2 expression from 21 studies demonstrated a significantly reduced OS with an HR of 0.77, p < 0.01 ( Figure 8B) suggesting that TMPRSS2 might be a more suitable biomarker for OS than ACE2.

Discussion
Although it has been suggested that patients with cancer have a higher likelihood of being infected by COVID-19 [22], the current data available are insufficient to conclude due to the low number and heterogeneity of samples. In addition, most COVID-19 infected patients have mild disease [25]. expression levels in normal lung epithelium between males and females, younger and older individuals indicates that the infection might be associated with gender and age.
Using TCGA data, our study demonstrated a significantly decreased expression of ACE2 and TMPRSS2 in many cancer types compared to normal adjacent tissues ( Figure 2; Table 1).

Conclusions
Despite this bioinformatics study provides novel information on the role of ACE2 and  Availability of data and material All data generated or analyzed during this study are included in this published article.

 Competing interests
There exists no financial and non-financial competing interests.

 Acknowledgements
We would like to thank Nicole Ott for excellent secretarial help.    In the forest plot, the name of each study is followed by the number of total tumor samples.