Regulation of the ACE2 locus in human airways cells

The angiotensin-converting enzyme 2 (ACE2) receptor is the gateway for SARS-CoV-2 to airway epithelium1,2 and the strong inflammatory response after viral infection is a hallmark in COVID-19 patients. Deciphering the regulation of the ACE2 gene is paramount for understanding the cell tropism of SARS-CoV-2 infection. Here we identify candidate regulatory elements in the ACE2 locus in human primary airway cells and lung tissue. Activating histone and promoter marks and Pol II loading characterize the intronic dACE2 and define novel candidate enhancers distal to the genuine ACE2 promoter and within additional introns. dACE2, and to a lesser extent ACE2, RNA levels increased in primary bronchial cells treated with interferons and this induction was mitigated by Janus kinase (JAK) inhibitors that are used therapeutically in COVID-19 patients. Our analyses provide insight into regulatory elements governing the ACE2 locus and highlight that JAK inhibitors are suitable tools to suppress interferon-activated genetic programs in bronchial cells.


Introduction
Recent studies 3,4 have identified a novel short form of ACE2, called dACE2, that originates from an intronic promoter activated by interferons. Onabajo et al. 4 used ENCODE data for chromatin modification marks (H3K4me3, H3K4me1 and H3K27ac) as well as DNase I hypersensitive (DHS) sites in cell lines to label putative regulatory elements at the newly identified exon (ex1c) located within intron 9 of the ACE2 gene. 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. ; https://doi.org/10.1101/2020. 10.04.325415 doi: bioRxiv preprint However, no regulatory elements were detected in the vicinity of the 5' end of the fulllength transcript encoding biologically active ACE2, and in sequences distal to the genuine promoter. Since these data sets were obtained from a wide range of cell lines and not from human primary airway cells, the principal target of SARS-CoV-2, they might not present a comprehensive picture of the regulatory regions controlling expression of the dACE2 and the full-length ACE2 transcripts in bronchial tissue.

Results
To comprehensively identify the genetic elements controlling the extended ACE2 locus, with an emphasis on its interferon response, we focused on human primary Small Airway Epithelial Cells (SAEC), which express a wide range of cytokine receptors and key mechanistic components of the executing JAK/STAT signal transduction pathway (Supplementary Table 1). We stimulated SAECs with interferon type I (IFNa and IFNb), type II (IFNg) and type III (IFNl) as well as with growth hormone (GH), Interleukin 6 (IL6) and IL7, followed by RNA-seq transcriptome analyses (Supplementary Tables 2-8).
Increased ACE2 expression was obtained with the interferons but not with GH, IL6 and IL7 (Supplementary Figure 1). However, the induction was less than that seen for classical interferon stimulated genes (ISG), such as STAT1. In agreement with earlier studies 3,4 , we detected the novel dACE2 N-terminal exon (ex1c) within intron 9 of the ACE2 gene (Supplementary Figure 1). To obtain more definitive information on the interferon response of the dACE2 and ACE2 promoters, we used RNA-seq and determined the respective read counts over the three alternative first exons ( Figure 1a and Supplementary Figure 1d). While the increase of RNA-seq reads induced by IFNa/b was highest (~25-fold) over ex1c, a lesser, yet significant, ~2-10-fold increase was detected over ex1a and ex1b, supporting the notion that expression of the full-length ACE2 transcript is also under interferon control. As an independent assay we used qRT-PCR and determined that IFN a/b stimulation led to a 8 to 15-fold increase of dACE2 and an approximately ~3-fold increase of ACE2 RNA (Figure 1b). However, the degree of induction of either form was lower than that seen for bona fide ISGs (Supplementary Figure 1). Previous studies in normal human bronchial epithelium (NHBE) did not reveal an interferon response of the native ACE2 promoter 3,4 suggesting differences between 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. ; https://doi.org/10.1101/2020.10.04.325415 doi: bioRxiv preprint 3 cell types or growth conditions. The mouse Ace2 gene is induced by cytokines through a STAT5-based enhancer in the second intron 5 and a DHS site is located in the equivalent location in the human ACE2 gene in SAEC and lung tissue. This suggests the presence of additional regulatory elements controlling expression of the full-length ACE2 mRNA.
To identify candidate regulatory elements controlling the extended ACE2 locus, including ACE2 and dACE2, in primary airway cells, we dug deeper and conducted ChIPseq for the chromatin marks H3K27ac (activate loci), H3K4me1 (enhancers), H3K4me3 there is little evidence of H3K4me3 and H3K27ac marks. However, it is well known that there is no direct relationship between gene activity and the presence of these marks. 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020.

Discussion
In summary, we have assessed the extended ACE2 locus for regulatory elements   and Pol II loading. The DHS data were obtained from ENCODE 6,7 . Yellow shade, candidate enhancers and blue shade, predicted promoter. The P/E region within intron 9 probably constitutes a combined promoter/enhancer. d. Putative STAT5 enhancer in the ACE2 gene was identified using ChIP-seq data from IFNb treated K562 cells 8 . biological replicates (n = 3). One-way ANOVA with followed by Dunnett's multiple comparisons test was used to evaluate the statistical significance of differences. e-h. 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. ; https://doi.org/10.1101/2020.10.04.325415 doi: bioRxiv preprint 7 H3K27ac marks, and Pol II loading at the ACE2, STAT1 and ISG15 loci in SAECs in the absence and presence of IFNb and the JAK inhibitor, Ruxolitinib.
105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. ; https://doi.org/10.1101/2020.10.04.325415 doi: bioRxiv preprint Figure 1 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. ; https://doi.org/10.1101/2020.10.04.325415 doi: bioRxiv preprint Figure 2 105 and is also made available for use under a CC0 license.
(which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted November 6, 2020. Total RNA-seq read quality control was done using Trimmomatic 12 (version 0.36) and STAR RNA-seq 13 (version STAR 2.5.4a) using 50bp paired-end mode was used to align the reads (hg19). HTSeq 14 was to retrieve the raw counts and subsequently, R (https://www.R-project.org/), Bioconductor 15 and DESeq2 16 were used. Additionally, the RUVSeq 17 package was applied to remove confounding factors. The data were prefiltered keeping only those genes, which have at least ten reads in total. Genes were categorized as significantly differentially expressed with an adjusted p-value (pAdj) below 0.05 and a fold change > 2 for up-regulated genes and a fold change of < -2 for downregulated ones. The visualization was done using dplyr (https://CRAN.Rproject.org/package=dplyr) and ggplot2 18 . Sequence read numbers were calculated using Samtools 19 software with sorted bam files.

Chromatin immunoprecipitation sequencing (ChIP-seq) and data analysis.
Chromatin was fixed with formaldehyde (1% final concentration) for 15 min at room temperature, and then quenched with glycine (0.125 M final concentration). Samples were processed as previously described 20 Table 3. List of all genes with normalized read counts in each replicate at Control and IFNb treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 4. List of all genes with normalized read counts in each replicate at Control and IFNg treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 5. List of all genes with normalized read counts in each replicate at Control and IFNl3 treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 6. List of all genes with normalized read counts in each replicate at Control and IL6 treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 7. List of all genes with normalized read counts in each replicate at Control and IL7 treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 8. List of all genes with normalized read counts in each replicate at Control and GH treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 9. List of all genes with normalized read counts in each replicate at IFNb and Baricitinib with IFNb, treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis. Table 10. List of all genes with normalized read counts in each replicate at IFNb and Ruxolitinib with IFNb, treated SAEC, log2 (fold change), p-value and adjusted p-value as well as upregulated gene list and GSEA analysis.