key: cord-0264342-ye5g92y1
authors: Le Nguyen, Thanh Thanh; Liu, Duan; Gao, Huanyao; Ye, Zhenqing; Lee, Jeong-Heon; Wei, Lixuan; Yu, Jia; Zhang, Lingxin; Wang, Liewei; Ordog, Tamas; Weinshilboum, Richard M.
title: Glucocorticoids Mediate Transcriptome-wide Alternative Polyadenylation: Mechanisms and Functional Implications
date: 2022-03-30
journal: bioRxiv
DOI: 10.1101/2022.03.30.486468
sha: 9b1229c452137448d786eeba9f3b0c5a1c2d6b17
doc_id: 264342
cord_uid: ye5g92y1

Alternative polyadenylation (APA) is a common genetic regulatory mechanism that generates distinct 3′ ends for RNA transcripts. Changes in APA have been associated with multiple biological processes and disease phenotypes. However, the role of hormones and their drug analogs in APA remains largely unknown. In this study, we investigated transcriptome-wide the impact of glucocorticoids on APA in 30 human B-lymphoblastoid cell lines. We found that glucocorticoids could regulate APA of a subset of genes, possibly by changing the expression of 142 RNA-binding proteins, some with known APA-regulating properties. Interestingly, genes with glucocorticoid-mediated APA were enriched in viral translation-related pathways, while genes with glucocorticoid-mediated expression were enriched in interferon and interleukin pathways, suggesting that glucocorticoid-mediated APA might result in functional consequences distinct from gene expression. Glucocorticoid-mediated APA was also cell-type-specific, suggesting an action of glucocorticoids that may be unique to immune regulation. We also observed evidence for genotype-dependent glucocorticoid-mediated APA, providing potential functional mechanisms for a series of common genetic variants that had previously been associated with immune disorders, but without a clear mechanism. In summary, this study reports a series of novel observations regarding the impact of glucocorticoids on APA, suggesting novel avenues for mechanistic studies of hormonal biology.

Alternative polyadenylation (APA) is a molecular regulatory mechanism that generates RNA isoforms with differential usage of polyadenylation sites (PASs) (1) . When APA takes place in the final exon of a gene, it creates RNA transcripts with differential lengths of the 3'-untranslated region (3'-UTR) of the gene. When APA takes place in introns, it can generates transcript variants with new 3' ends, a process that often involves differential splicing (1) . The functional outcomes of APA include modulation of mRNA stability and translation, mRNA localization, and intriguingly, protein localization independent of mRNA localization (2) (3) (4) (5) (6) . Even though not being as extensively studied as gene transcription, APA is a common phenomenon with at least 70% of mammalian mRNA-coding genes showing APA isoforms (7, 8) , suggesting that it is one of the major regulatory mechanisms of cellular and molecular dynamics with important functional consequences (1) (2) (3) (4) (5) (6) . In fact, dysregulation of APA has been associated with disease phenotypes (9) including cancer (10) (11) (12) (13) , immunological and neurological diseases (14) (15) (16) .

In recent years experimental methods and computational tools for detecting APA have developed rapidly, allowing the detection of transcriptome-wide APA events on a scale that was not possible previously (9, 13) . Of interest, there are now computational algorithms that enable detection of APA events directly from standard poly(A)-enriched RNA-seq data, allowing the generation of novel hypotheses and observations using existing RNA-seq datasets from large databases. For example, using DaPars, an algorithm that detects de novo PAS proximal to the last exon by modeling read counts of RNA-seq, investigators analyzed data from The Cancer Genome Atlas Project (17) and observed a global shortening of 3'UTRs in tumors compared to normal tissue (11, 12) . Similar approach was also applied to analyze RNA-seq datasets generated by the Genotype-Tissue Expression (GTEx) project (18) for identification of global APA across different 4 types of human tissues (19-21). Most interestingly, by integrating genome-wide single-nucleotide polymorphism (SNP) genotyping data with APA, investigators were able to discover quantitative trait loci associated with APA (3'aQTLs), many of which had been associated with disease phenotypes but with functions unexplained by expression QTLs (eQTLs) (20,21). Several other biological processes have been reported to affect global APA including cell proliferation (22), cell differentiation (23), and viral infection (24). However, the effects on APA of hormones and their drug analogs, which have mainly been studied for their effects on gene transcription and expression, remain largely unknown with the exception of a case report for estrogens (25). In the present study, we systematically characterized transcriptome-wide the role of glucocorticoids, a major class of hormones, in the regulation of APA.

Clinically, glucocorticoids represent a class of medication used to treat a variety of inflammatory and autoimmune diseases (26,27). While the role of the glucocorticoids in activating the glucocorticoids receptor (GR) to become a transcription factor has been well-characterized, to our knowledge there have been no studies of their regulation of APA. To characterize transcriptomewide glucocorticoid-dependent APA, we analyzed 90 standard RNA-seq datasets that we had generated from 30 human B-lymphoblastoid cell lines (LCLs) before and after glucocorticoid treatment (28). We discovered a group of immune-specific genes with glucocorticoid-mediated APA that appeared to be functionally distinct from those with glucocorticoid-mediated alterations in expression. We observed that one possible mechanism underlying glucocorticoid-mediated APA involved GR-dependent regulation of RNA-binding proteins (RBP), an important group of APA regulators (29). We also uncovered non-coding genetic risk sequence variants that potentially behaved as alternative polyadenylation quantitative trait loci (3'aQTL), but only after 5 glucocorticoid exposure, some of which had previously been associated with glucocorticoidrelated disease phenotypes during genome/phenome-wide association studies.

Three hundred lymphoblastoid cell lines (LCLs) were obtained from the Coriell Institute and were genotyped with the Illumina HumanHap550K and HumanExon510S-Duo Bead Chips in our laboratory, information that was combined with data generated at the Coriell Institute using the Affymetrix Human SNP Array 6.0. Those genotype data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE23120 (30).

Thirty LCLs with similar levels of GR expression selected from the 300-LCL panel were exposed to cortisol, an endogenous GR ligand, followed by the addition of CORT108297 (C297), a selective GR modulator that could antagonize cortisol activity for 9 hours. RNA-seq for these samples were obtained as described previously (28). The RNA seq data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE185941. RNA-seq fastq files obtained for the A549 lung carcinoma epithelial cell line after 12 hours of dexamethasone treatment were downloaded from the ENCODE portal (series numbers ENCSR632DQP for control and ENCSR154TDP for treatment). A total of four replicates for each treatment were also downloaded. The fastq files were mapped to the human genome GRCh38 (hg38) using STAR (31). Bed files from each RNA-seq dataset were generated using STAR (31). Identification of APA events was performed with DaPars v2.0 (21) using cutoff values 6 for coverage of 10 and a fit value of 10 using the GENECODE v38 human genome gene annotation. Changes of percentage for the distal poly(A) site usage index (PDUI) between drug and vehicle treatments were conducted in R using a two-sided Wilcoxon matched pair signed rank test. Significance was defined as: (1) FDR less than 0.05 across the 30 LCLs and (2) the fact that the change was reversed by the antagonist C297.

Raw counts were generated from the RNA-seq bam files with the Python package "HTseq" (32) and differential expression analysis was conducted with the R package "EdgeR" (33). Only genes that passed CPM of 32 were included in the analysis. Pathway analysis was conducted with EnrichR (34) both for genes with glucocorticoid-mediated APA and genes with glucocorticoidmediated expression. Available bed files for cortisol-regulated RBPs were then downloaded from the Encyclopedia of DNA Elements (ENCODE) Project (39) . The binding sites of these RBPs were overlapped with 3'UTR data containing the cortisol-mediated APA sites for each gene using bedtools (35).

In addition, precise transcriptional regulation of DEGs in the 30 cells was interrogated using previously generated epigenomic datasets including GR-targeted ChIP-seq obtained in the presence or absence of glucocorticoids, HiChIP targeting H3K27ac, a histone mark associated with active enhancers and promoters, generated with or without exposure to glucocoricoids, and a 25chromatin state prediction model (28). To identify RBPs that were directly regulated by GR through GR-mediated enhancers, cortisol-dependent RBPs with loops connecting them to cortisoldependent ChIP-seq peaks were identified using the R packages "GenomicRanges" (36) and bedtools (35). The chromatin states of the cortisol-dependent ChIP-seq peaks were also identified 7 based on the 25-chromatin state prediction model. The sequencing data for these datasets were Treatment time was optimized to achieve the greatest observed effect of glucocorticoid on APA at 3hrs, 6hrs, 9hrs, and 12hrs (data not shown). Since cortisol-induced APA changes were most striking at 9 hours, we repeated the experiments with different dosages of the drugs, specifically, 0nM, 1nM, 10nM, 100nM, and 1000nM after 9 hours of incubation.

The cells were pelleted and harvested for RNA extraction. Total RNA was extracted with the Quick-RNA Miniprep kit (Zymo, Cat# R1055) per the manufacturer's instructions. A total of 2μg of total RNA was used for cDNA synthesis using the SuperScript™ III First-Strand Synthesis System (Thermo Fisher, Cat# 18080-093) and poly(A) oligo(dT)20 (Thermo Fisher, Cat# 18418020) per manufacturer's instruction. Briefly, a mixture of 1μL of oligo(dT)20 (50 μM), 1μL of 10mM dNTP mix (Thermo Fisher, Cat# 18427088), 2μg of total RNA, and distilled water to 13μL was heated to 65°C for 5 minutes and incubated on ice for 1 minute. This step was followed 8 by the addition of 4μL 5X First-Strand Buffer, 1μL 0.1M DTT, 1μL RNAse Inhibitor (Cat # 10777-019, 40 units/ μL), and 1μL of SuperScript™ III reverse transcriptase (200 units/μL). This mixture was then incubated at 25°C for 5 minutes, 50°C for 45 minutes, and 70°C for 15 minutes. For removal of RNA complementary to the cDNA, 1μL of RNase H (Ambion, Cat # AM2293) was incubated with the mixture at 37°C for 20 minutes. The cDNA was then ready for the PCR step.

Primer sequences targeting the predicted APA site in LY6E were designed using Primer3 software.

The following two primers were used to perform this experiment: LY6E_3UTR_1 Forward: excluded. Analysis of quantitative trait loci for delta PDUI was conducted with the R package "Matrix eQTL" (38) using ANOVA model. A total of 515,177,592 associations were conducted, and P-values with significant trends were defined as 5.16 x 10 -8 . Those SNPs were then overlapped with SNPs documented in the GWAS Catalog (https://www.ebi.ac.uk/gwas/) and with UKBiobank data (http://pheweb.sph.umich.edu/) for potentially significant clinical associations. Significant Pvalue cutoffs were defined by the databases queried.

Thirty LCLs of differing genomic backgrounds were treated with vehicle, 100 nM cortisol (a GR agonist) and 100 nM cortisol plus 100 nM of CORT108297 (C297, a GR antagonist). Treatment conditions such as dose and incubation time were optimized before sequencing, and the treatment effect was confirmed for these datasets by expression quantification of prototypic GR-targeted genes such as FKBP5 and TSC22D3. Specifically, the mRNA levels for both of these genes were significantly induced after cortisol treatment and those inductions returned to baseline after C297 was added to the treatment (28). We then analyzed these 90 RNA-seq datasets to identify transcriptome-wide APA events using the computational algorithm DaPars v. After PDUI values were obtained for each gene in each sample, a total of 696 genes that displayed APA had been detected. Of these 696 genes, we found that cortisol regulated APA dynamics for 54 genes (Figure 1A , Supplementary Figure 1B-C) . The addition of C297 reversed this cortisol-mediated repression back to baseline (Figure 1B-C) . Interestingly, we found that cortisol treatment did not change LY6E mRNA levels in our LCLs but that it did mediate 3'UTR APA to reduce the percentage of transcripts with long 3' UTRs ( Figure 1B) . To validate the glucocorticoid-mediated APA observations for LY6E, we quantified the long LY6E 3'UTR transcripts by qRT-PCR using two different sets of primers targeting the LY6E 3'UTR region that covered the proximal PAS. We observed a dose-dependent repression by cortisol, and we replicated that dose-dependent regulation with dexamethasone, a potent synthetic glucocorticoid that is often used in the clinic ( Figure 1D ).

When we conducted pathway analysis of 1362 cortisol-responsive genes from our RNA-seq results (FDR < 0.05), we found that "RNA-binding protein" (RBP) was the most enriched pathway (adjusted p-value = 7.39E-04), raising the possibility that cortisol might regulate downstream RBPs with APA-regulating properties (Figure 2A) . Therefore, we hypothesized that glucocorticoid-activated GR might regulate the expression of RBPs with known APA-regulating properties, which in turn might drive the changes observed for APA. From the pathway analysis, we observed that 142 proteins with RNA-binding properties were regulated by cortisol and that their cortisol-mediated expression was reversed by C297 ( Figure 2B) . To identify which of those 142 DEGs were directly regulated by GR, we consulted GR-targeted ChIP-seq and H3K27ac

HiChIP datasets that we had previously generated in LCLs both with and without cortisol treatment, together with a 25-chromatin-state prediction model (28). We found that 27 (18%) of the 142 glucocorticoid-regulated RBPs had H3K27ac HiChIP loops that connected them with one or more cortisol-induced GR peaks, suggesting direct regulation through cortisol-mediated enhancers ( Figure 2C) .

We then explored the function of the 142 cortisol-regulated RBPs by consulting enhanced crosslinking and immunoprecipitation sequencing (eCLIP-seq) datasets for more than 70 RBPs and their annotated functions that had been generated by the ENCODE project (39) . We found that 12 of those cortisol-regulated RBPs had information for transcriptome-wide RNA-binding sites readily available in two human cell lines (Figure 2D ; see Supplementary (42) (43) (44) (45) (46) (47) , but their roles-if any-in the regulation of APAs have yet to be determined.

When we integrated the eCLIP-seq data for APA-regulating RBPs PABPN1 and EWSR1, as described above, with our APA analysis, we found that 68% of 54 genes with GR-mediated APA events were bound by either EWSR1 or PABPN1 at 3'UTRs containing the alternative PAS ( Figure 2E) . For example, based on eCLIP-seq data, EWSR1 bound to multiple sites in the LY6E

3'-UTR surrounding the predicted proximal PAS (Figure 2F ). EWSR1 was downregulated by cortisol via a cortisol-induced GR enhancer that loops across 100kb to the EWSR1 promoter ( Figure 2G) , possibly causing the observed repression of the distal PAS usage. Taken together, 12 these observations supported our hypothesis that glucocorticoids can mediate APA, at least in part, through the transcriptional regulation of RBPs with APA-regulating properties.

GR is known to act as an RBP itself (48) . However, a previous study reported that, in general, GR binds to the 5'UTR rather than the 3'UTR of mRNAs and that GR-RNA binding is independent of AU-rich elements, an RNA-binding motif that is often found in 3'UTRs (48) . This observation suggests that the RNA-binding property of GR might have limited effect on the glucocorticoiddependent APA that we observed.

As described above, we identified 54 glucocorticoid-mediated APA genes, a much smaller number than the 1,362 glucocorticoid-mediated DEGs in these same LCLs. Furthermore, we noticed that only 5 of those 54 APA genes overlapped with the 1,362 DEGs ( Figure 3A) . Therefore, we asked what the function(s) of the glucocorticoid-mediated APA genes might be and how they might differ from those of the glucocorticoid-mediated DEGs. It is well known that one of the mechanisms underlying the immunosuppressant actions of glucocorticoids is medicated through the inhibition of interleukin and cytokine signaling pathways (49) . For example, glucocorticoids induce the expression of glucocorticoid-induced leucine zipper (GILZ, or TSC22D3), an inhibitor of NF-kB (50) and they inhibit the expression of pro-inflammatory cytokines and chemokines (49) .

We observed similar effects of cortisol in our LCLs, as demonstrated by pathway analysis of cortisol mediated DEGs (Figure 3B, Supplementary Table 3) . However, we found that genes which displayed glucocorticoid-mediated APA were primarily enriched in viral translation-related pathways, with p-values more significant than those for pathways enriched for cortisol-mediated DEGs despite a smaller number of this class of genes (Figure 3C-D) . That observation was The functional consequences for viral biology of the shortening of LY6E 3'UTR after cortisol treatment, however, remains to be determined.

To determine whether the glucocorticoid-mediated APA effects that we observed in LCLs might apply to other types of cells, we analyzed published RNA-seq datasets from the ENCODE portal generated for A549 lung carcinoma epithelial cells before and after 100nM dexamethasone treatment for 12 hours, a treatment condition comparable to that used in our LCL study (56) . Using similar analytical approaches to those which we applied for LCLs, we identified 2448 DEGs after dexamethasone treatment of A549 cells (FDR < 0.05), with 243 genes overlapping with DEGs that we had identified in LCLs (see Figure 4A) . Interestingly, the most significant pathway enriched for glucocorticoid-regulated DEGs in A549 cells was the DNA-binding pathway (adjusted p-value = 3.022e-8) (Figure 4B) , in contrast to the RNA-binding pathways observed in LCLs, as described above (Figure 2A) . Using the DaPars v2.0 algorithm, we were able to detect 460 genes with APA events at baseline in A549 cells, genes which overlapped with 56% of the genes with APA that we had detected in LCLs (see Figure 4C ). However, we did not detect any change in APA events 14 after the dexamethasone treatment of A549 cells (FDR < 0.05) (Figure 4D) . Consistent with these observations, much fewer RNA-binding-related genes were differentially expressed in A549 cells than that in LCLs (21 versus 142 genes) after glucocorticoids treatment ( Figure 4A) . Importantly, two genes with known APA-regulating properties and explaining 68% of glucocorticoiddependent APA observed in the LCLs, PABPN1 and EWSR1 (Figure 2E) , were not differentially expressed in A549 cells after glucocorticoid treatment ( Figure 4A ). This result might explain why no glucocorticoid-mediated APA events were detected in A549 cells. These observations suggest that GR-mediated APA regulation may be highly cell-type specific and that it may occur more often in association with immune-related functions of GR.

Taking advantage of the genome-wide genotype data that we had already generated for our 30 Figure 5A) .

To identify possible glucocorticoid-mediated PGx-3'aQTL associations, we associated SNP genotypes with changes in percentage of distal APA usage between glucocorticoid treatment and vehicle exposure across our 30 LCLs (see Methods). That step identified a total of 66 cortisoldependent candidate PGx-3'aQTLs (P < 5.16 × 10 -8 ), all of which no longer exhibited SNP-APA change associations after the addition of the GR antagonist, C297 (P ≥ 0.05), demonstrating a cortisol-dependent effect ( Figure 5B, Supplementary Table 4 ). To investigate whether these glucocorticoid-dependent PGx-3'aQTLs could potentially explain the function of SNPs previously associated with clinical phenotypes but with unclear functional mechanisms, we overlapped PGx-3'aQTL SNPs with those already reported to be genome-wide significant signals in the GWAS Catalog (58) and/or the UK Biobank PheWAS Catalog (59) . We found that five of our PGx-3'aQTLs had been associated with disease phenotypes (Figure 5C) , four of which were related to known effects of glucocorticoid usage such as bacterial infections (60), inflammatory bowel disease (61, 62) , and cardiovascular disease (63, 64) . Furthermore, the function of the genes modulated by the PGx-3'aQTL appeared to be related to these associated phenotypes. For example, rs770150 was associated with bacterial enteritis based on PheWAS (P = 4.7 × 10 -05 ), and it was found in our study to be a PGx-3'aQTL for DDX3X. Specifically, cortisol decreased PDUI in subjects with the AA genotypes but not subjects with other genotypes at this locus, and this repression was reversed by C297 ( Figure 5D ). DDX3X encodes a RBP that is known to play a critical role in antimicrobial innate immunity (65, 66) . Another example involved rs2483280, a SNP previously associated by GWAS with electrocardiographic QRS duration (2.0 × 10 -11 ). That 16 SNP was associated with repressed PDUI for CCR7 in a genotype-and drug-dependent manner ( Figure 5E ). CCR7 encodes a chemokine receptor that has been shown to play a role in cardiac function (67, 68) . While these associations with disease through possible glucocorticoid-dependent mechanisms are intriguing, these SNPs did not disrupt GR binding sites and appeared to act through trans mechanisms that remain to be determined and validated in future studies with larger sample sizes.

Our study has provided a series of observations with regard to the impact of glucocorticoids on alternative polyadenylation which, to our knowledge, had not been reported previously. In the clinic, glucocorticoids are used to treat a variety of inflammatory and autoimmune diseases (26,27). Traditional understanding of glucocorticoid mechanisms in immune regulation has been based on their activation of the glucocorticoid receptor (GR), which is then translocated to the nucleus where it binds to DNA to initiate gene transcription (17, 18) . We demonstrated that glucocorticoids could also regulate global APA for immune-related genes in human LCLs (Figure   1 ), genes that were functionally distinct from those influenced by glucocorticoids on expression level (Figure 3) . Even though the number of glucocorticoid-dependent APA genes was smaller than that of glucocorticoid-dependent DEGs, glucocorticoid-dependent APA genes were highly enrich in the "viral infection-related" pathways ( Figure 3C) , indicating a more specific function of glucocorticoid-dependent APA. Characterization of glucocorticoid-dependent APA in viral infection might help to better understand the mechanism of glucocorticoids' therapeutic effects in diseases such as COVID-19 (69, 70) .

Glucocorticoids were known to play important functional roles in a variety of multiple physiological processes and diseases including autoimmunity (49) , osteoporosis (71) , mood disorders (72, 73) , cancer (74) , and metabolism (75) . However, the therapeutic usage of glucocorticoids remains largely dependent on their function in immunosuppression. Of interest, glucocorticoid-dependent APA, which possibly occurred through glucocorticoid-mediated transcriptional regulation of RBPs with APA-regulating properties (Figure 2) , appears to be celltype specific (Figure 4) . Although comparable numbers of APA genes were identified in both LCLs and A549 lung cancer cells (Figure 4C) , none of the glucocorticoid-dependent APA genes were identified in the A549 cell line (Figure 4D) , raising the possibility that glucocorticoiddependent APA could be more often immune-related. This possibility requires further investigation with more cell lines and tissue types, which, once confirmed, might expand our understanding of the mechanism(s) of glucocorticoids' immunosuppressive properties. However, the present study did not determine the functional consequences of these APA changes in terms of molecular and cellular dynamic changes induced by glucocorticoids. While mRNA stability is one of the known functional outcomes of APA (1), we found that glucocorticoid-mediated changes in APA were only associated with changes in gene expression in a small minority of DEGs in LCLs.

We also observed trends that glucocorticoids could regulate APA in a genotype-dependent manner, a phenomenon that we referred to as "PGx-3'aQTL" (Figure 5 ). PGx-3'aQTLs could represent a novel type of context-dependent SNPs for which pharmacological or physiological reagents can "unmask" the function of silent non-coding SNPs with previously unknown functional roles. They could also potentially explain an aspect of gene × environment interaction in disease risk and variation in drug response that otherwise could not be explained by eQTLs or context-dependent eQTLs. Unlike diseases associated with glucocorticoid-mediated PGx-eQTLs which spanned a broad spectrum of glucocorticoid-related diseases as identified in our previous study (28), diseases or pathological states associated with glucocorticoid-mediated PGx-3'aQTLs appeared to be 18 limited to or focused on immune-related phenotypes-at least in LCLs. This observation is consistent with the fact that glucocorticoids regulate APA in a cell-type specific manner and appeared-in LCLs-to do so primarily for genes with immune-related functions. While these observations are intriguing, they need to be validated with larger sample size.

In conclusion, our study has revealed a series of novel aspects of genomic regulation by glucocorticoids, particularly their impact on RNA polyadenylation with its possible role in disease risk, aspects that should be explored for other hormones and nuclear receptors beyond GR. Li 

Alternative polyadenylation of mRNA precursors

Mechanisms and consequences of alternative polyadenylation

Alternative cleavage and polyadenylation: extent, regulation and function

Alternative 3' UTRs act as scaffolds to regulate membrane protein localization

The end of the message: multiple protein-RNA interactions define the mRNA polyadenylation site

Evolution and Biological Roles of Alternative 3'UTRs

A quantitative atlas of polyadenylation in five mammals

Analysis of alternative cleavage and polyadenylation by 3' region extraction and deep sequencing

Alternative cleavage and polyadenylation in health and disease

A germline variant in the TP53 polyadenylation signal confers cancer susceptibility

Dynamic analyses of alternative polyadenylation from RNA-seq reveal a 3'-UTR landscape across seven tumour types

Comprehensive Characterization of Alternative Polyadenylation in Human Cancer

Alternative polyadenylation: methods, mechanism, function, and role in cancer

A rare polyadenylation signal mutation of the FOXP3 gene (AAUAAA-->AAUGAA) leads to the IPEX syndrome

The human GIMAP5 gene has a common polyadenylation polymorphism increasing risk to systemic lupus erythematosus

Three functional variants of IFN regulatory factor 5 (IRF5) define risk and protective haplotypes for human lupus

The Cancer Genome Atlas Pan-Cancer analysis project

The Genotype-Tissue Expression (GTEx) project

A large-scale binding and functional map of human RNA-binding proteins

The poly(A)-binding protein nuclear 1 suppresses alternative cleavage and polyadenylation sites

Position-specific binding of FUS to nascent RNA regulates mRNA length

Genetic and physical interactions between factors involved in both cell cycle progression and pre-mRNA splicing in Saccharomyces cerevisiae

ZRANB2 and SYF2-mediated splicing programs converging on ECT2 are involved in breast cancer cell resistance to doxorubicin

Mak5p, which is required for the maintenance of the M1 dsRNA virus, is encoded by the yeast ORF YBR142w and is involved in the biogenesis of the 60S subunit of the ribosome

The G-patch protein NF-kappaBrepressing factor mediates the recruitment of the exonuclease XRN2 and activation of the RNA helicase DHX15 in human ribosome biogenesis

Aberrant splicing and defective mRNA production induced by somatic spliceosome mutations in myelodysplasia

NSUN2-Mediated m5C Methylation and METTL3/METTL14-Mediated m6A Methylation Cooperatively Enhance p21 Translation

The human glucocorticoid receptor as an RNA-binding protein: global analysis of glucocorticoid receptor-associated transcripts and identification of a target RNA motif

Immune regulation by glucocorticoids

Role of GILZ in immune regulation, glucocorticoid actions and rheumatoid arthritis

Posttranscriptional regulation of gene expression in innate immunity

Emerging Role of LY6E in Virus-Host Interactions. Viruses

LY6E Restricts Entry of Human Coronaviruses, Including Currently Pandemic SARS-CoV-2

LY6E impairs coronavirus fusion and confers immune control of viral disease

Interferon-inducible LY6E Protein Promotes HIV-1 Infection

The Encyclopedia of DNA elements (ENCODE): data portal update

Alternative polyadenylation mediates genetic regulation of gene expression

The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog)

Exploring and visualizing large-scale genetic associations by using PheWeb

Infection Risk and Safety of Corticosteroid Use

Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease

Association analyses identify 38 susceptibility loci for inflammatory bowel disease and highlight shared genetic risk across populations

Glucocorticoid use and risk of atrial fibrillation or flutter: a population-based, casecontrol study

Taking glucocorticoids by prescription is associated with subsequent cardiovascular disease

The RNA helicase DDX3X is an essential mediator of innate antimicrobial immunity

The DEAD-box helicase DDX3X is a critical component of the TANK-binding kinase 1-dependent innate immune response

CCR7-CCL19/CCL21 Axis is Essential for Effective Arteriogenesis in a Murine Model of Hindlimb Ischemia

Pericardial Adipose Tissue Regulates Granulopoiesis, Fibrosis, and Cardiac Function After Myocardial Infarction

Association Between Administration of Systemic Corticosteroids and Mortality Among Critically Ill Patients With COVID-19: A Meta-analysis

Dexamethasone in Hospitalized Patients with Covid-19

Glucocorticoid-Induced Osteoporosis

Prospective biomarkers of major depressive disorder: a systematic review and meta-analysis

The HPA axis in bipolar disorder: Systematic review and meta-analysis

Glucocorticoids promote breast cancer metastasis

Glucocorticoids, metabolism and metabolic diseases

Acknowledgements. We thank the ENCODE Consortium for their generation of the epigenomic datasets used in this study.

OneOme, LLC. Other authors declare no conflict of interests

and (C) that are regulated directly by GR through GR-mediated enhancers that loop to genes to transcriptionally regulate their expression. (D) Twelve cortisol-mediated RBPs that have genome-wide RNA-binding patterns available on ENCORE through eCLIP assay. The Y axis represents z-scores for gene expression after exposure to vehicle, CORT or CORT/C297. (E) The number of genes with cortisol-mediated APA that have RBPs with known APA-regulating properties that bind at their 3'-UTRs containing the predicted APA site based on data from (D). (F) An IGV plot showing the LY6E 3'UTR, a glucocorticoid-mediated APA example from Figure 1 , in which an RBP with known APA-regulatory properties, EWSR1, binds to multiple regions surrounding the predicted APA site. (G) EWSR1 expression is repressed by cortisol via a distal enhancer, as shown in the IGV plot. Loops that directly interact between the GR-binding site and the EWSR1 promoter are colored black. Chromatin state abbreviations: Enh_str_loop: Enhancer with strong looping properties; Tss_tes_loop: POLR2 initiation & stop with enhancer-looping property.