Thymidine phosphorylase facilitates retinoic acid inducible gene-I induced endothelial dysfunction

Abstract

Activation of nucleic acid sensors in endothelial cells (ECs) has been shown to drive inflammation across pathologies including cancer, atherosclerosis and obesity. We previously showed that enhancing cytosolic DNA sensing by inhibiting three prime exonuclease 1 (TREX1) in ECs led to EC dysfunction and impaired angiogenesis. Here we show that activation of a cytosolic RNA sensor, Retinoic acid Induced Gene 1 (RIG-I) diminishes EC survival, angiogenesis and triggers tissue specific gene expression programs. We discovered a RIG-I dependent 7 gene signature that affects angiogenesis, inflammation and coagulation. Among these, we identified the thymidine phosphorylase TYMP as a key mediator of RIG-I induced EC dysfunction via its regulation of a subset of interferon stimulated genes. Our RIG-I induced gene signature was also conserved in the context of human diseases - in lung cancer vasculature and herpesvirus infection of lung endothelial cells. Pharmacological or genetic inhibition of TYMP rescues RIG-I induced EC death, migration arrest and restores sprouting angiogenesis. Interestingly, using RNAseq we identified a gene expression program that was RIG-I induced but TYMP dependent. Analysis of this dataset indicated that IRF1 and IRF8 dependent transcription is diminished in RIG-I activated cells when TYMP is inhibited. Functional RNAi screen of our TYMP dependent EC genes, we found that a group of 5 genes - Flot1, Ccl5, Vars2, Samd9l and Ube2l6 are critical for endothelial cell death mediated by RIG-I activation. Our observations identify mechanisms by which RIG-I drives EC dysfunction and define pathways that can be pharmacologically targeted to ameliorate RIG-I induced vascular inflammation.

Fig. 1. RIG-I activation causes EC dysfunction.

HUVECs or HMVECS were treated with either control agonists or RIG-I agonists (1 ug/ml) (A-E). Viability was measured using Cell Titer glo (A) and Cell Death was measured using Caspase 3&7 glo (B–C) assays. D Migration was measured using a scratch assay. E Sprouting angiogenesis was measured using a 3D tube formation assay in a fibrin gel. Scale bar = 100 μm. Quantification of sprout area stained using U.europaeus lectin. Dots represent individual beads. **P < 0.01, ***P < 0.005 using ANOVA or two-tailed Student’s T-test and Mann–Whitney U-test for (E). Bars represent mean ± SEM of independent replicates.

Fig. 2. RIG-I drives a specific angiogenic gene signature in ECs.

A–C HUVECs or B–D HMVECs were treated with RIG agonists or control agonist. 24 h later RNA was extracted and RNA sequencing was performed. Data was analyzed using RaNAseq pipeline as described in Prieto & Barrios, Bioinformatics 2020. All genes with adjusted P-values <0.05 are plotted and critical IFN responsive genes are colored and labeled. C–D Gene set enrichment analysis using Enrichr. E Heatmap depicting gene expression changes in indicated tissues from WT vs RIG-I −/− mice (n = 2) from an angiogenesis qRT-PCR array. F Schematic depicting a 7 gene RIG-I dependent angiogenic signature. G Heatmap depicting enrichment of endothelial RIG-I signature from F in public datasets. Left panel depicts correlation coefficients of RIG-I with the 7 genes across four major cell types from the lung-tumor microenvironment interactome, a dataset of scRNA-seq from patients with squamous cell carcinoma or adenocarcinoma of the lung. Right panel depicts normalized expression scores of RIG-I and the 7 gene signature in human pulmonary ECs during infection with human herpesvirus 8 (HHV-8) plotted from EndoDB dataset E-GEOD-6489.

Fig. 3. TYMP inhibition partially rescues RIG-I activation induced phenotypes.

A TYMP induction was measured using qRT-PCR. B HUVECs were treated with a control or RIG-I agonist in combination with Tipiracil. B Cell Death was assessed using a Caspase Glo assay (C) Annexin V luminescence assay and (D) Necrosis assay. E–F HUVECs were transfected with either a control siRNA or TYMP siRNA and treated with a control agonist or RIG-I agonist at the indicated doses. E mRNA levels showing efficient knockdown of TYMP 24 h post transfection. F Cell death was measured using Caspase-Glo assay. G–H Migration was assessed using a scratch assay in a 6-well TC plate. Scale bar = 100 uM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001 using ANOVA or two-tailed Student’s T-test. Bars represent mean ± SEM of independent replicates. One of two independent experiments.

Fig. 4. TYMP dependent genes that impact RIG-I induced EC dysfunction.

A, B HUVECs were treated with control agonist or RIG-I agonist in combination with Tipiracil or vehicle (PBS). RNAseq was performed 24 h after treatment. A Volcano plot depicting differentially expressed genes (B) Gene set enrichment analysis (Enrichr) is depicted. C, D 19 genes were identified as differentially expressed (ie up with RIG-I but down with RIG-I + TPI or vice-versa). E Consensus transcription factor analysis from ENCODE and ChEA (Enrichr) (F) HUVECs were transfected with 4 pooled siRNAs against each of the 19 genes. 24 h later cells were treated with control or RIG-I agonist and cell death was measured. Loss of significance in bar graph indicate the siRNAs that prevent RIG-I induced EC death. One of two independent experiments.