Transcriptional networks specifying homeostatic and inflammatory programs of gene expression in human aortic endothelial cells

  1. Nicholas T Hogan
  2. Michael B Whalen
  3. Lindsey K Stolze
  4. Nizar K Hadeli
  5. Michael T Lam
  6. James R Springstead
  7. Christopher K Glass
  8. Casey E Romanoski  Is a corresponding author
  1. University of California, San Diego, United States
  2. University of Arizona, United States
  3. University of Western Michigan, United States
6 figures, 2 tables and 3 additional files

Figures

Figure 1 with 3 supplements
HAECs display a distinct repertoire of enhancers that nominate combinations of the AP1, ETS, SOX and GATA TF families as major orchestrators of HAEC gene expression.

(a) A heatmap of 16,929 enhancer-like regions were selected by: accessible chromatin (ATAC-seq), coincident with H3K4me2 and H3K27ac deposition (ChIP-seq) in gene-distal positions (≥3 kb from …

https://doi.org/10.7554/eLife.22536.002
Figure 1—figure supplement 1
Analysis of motifs and expression of associated transcription factors.

(a) The Top 19 enriched motifs that occur in active endothelial cell enhancers (Figure 1a) calculated from 200 bp sequences centered on chromatin accessible regions. (b) Ranked TF gene expression …

https://doi.org/10.7554/eLife.22536.003
Figure 1—figure supplement 2
Genetic loci for JUN, JUNB, and JUND.

Chromatin accessibility, histone modifications, and RNA expression of JUN, JUNB, and JUND. SE regions are highlighted in yellow as determined by H3K27ac.

https://doi.org/10.7554/eLife.22536.004
Figure 1—figure supplement 3
Hierarchical clustering of enhancers.

(a) H3K27ac ChIP-seq datasets were downloaded from ENCODE and Roadmap Epigenomics and quantified at 16,929 HAEC enhancers from Figure 1a (x-axis). Enhancers are clustered using hierarchical …

https://doi.org/10.7554/eLife.22536.005
Figure 2 with 2 supplements
Coordinate gene expression across 96 genetically distinct HAEC donors identifies three regulatory programs among ETS, AP1, SOX, and GATA family members.

(a) TF gene expression measured by AffyHU133A array is shown across a population of 97 unique HAEC donors. Array probe set IDs were manually confirmed to cover expressed transcript isoforms of the …

https://doi.org/10.7554/eLife.22536.007
Figure 2—figure supplement 1
The HAEC gene correlation network.

A clustered heatmap of correlation coefficients calculated between the TFs of interest (x-axis) and 7738 transcripts having significant correlation to at least one TF (Pearson correlation p<0.05) on …

https://doi.org/10.7554/eLife.22536.008
Figure 2—figure supplement 2
TF binding at HAEC enhancers.

JUNB, JUN, GATA2, ERG, and ETS1 bind at the center of the enhancer-like regions (from Figure 1a). JUNB, JUN, and ERG binding was measured in HAECs and GATA2 and ETS1 in HUVECs.

https://doi.org/10.7554/eLife.22536.009
Figure 3 with 2 supplements
ERG and JUN co-bind EC enhancers and are enriched at EC-specific genes.

(ac) Promoter-distal regions bound by ERG, JUN, and JUNB, or GATA2 are shown in a, b, and c respectively in a one kilobase window. Each set was centered on the corresponding binding motif, and the …

https://doi.org/10.7554/eLife.22536.010
Figure 3—figure supplement 1
Allele-specific JUN binding at loci with mutated motifs.

JUN binding was measured by ChIP-seq in a HAEC donor with whole-genome sequencing information. Allele-specific JUN binding was calculated at heterozygous loci (y-axis is JUN binding ratio between …

https://doi.org/10.7554/eLife.22536.011
Figure 3—figure supplement 2
JUN and ERG co-occupy loci near EC-specific genes.

ATAC-seq, binding for JUN and ERG, histone modifications H3K4me2 and H3K27ac, and RNA abundance are shown at the genetic loci for CDH5, EGFL7, TEK/TIE2, and PECAM1. Promoters are highlighted in …

https://doi.org/10.7554/eLife.22536.012
Figure 4 with 2 supplements
ERG knockdown elicits a pro-inflammatory gene profile in HAECs.

(a) Genes up-regulated by ERG knockdown (≥4 fold, 5% FDR) were enriched in the indicated functional pathways using Ingenuity Pathway Analysis (IPA). Results are from three experiments using cells …

https://doi.org/10.7554/eLife.22536.013
Figure 4—figure supplement 1
Changes in gene expression with ERG knockdown measured by RNA-seq.

Gene expression levels of various genes according to RNA-seq data in three different donors, each in control (siSCR) and ERG knockdown (siERG) conditions.

https://doi.org/10.7554/eLife.22536.014
Figure 4—figure supplement 2
Changes in gene expression with ERG knockdown measured by RT qPCR.

Gene expression measured by RT- qPCR in no transfection controls, with scrambled siRNA, individual siERG oligos, and pooled siERG oligos for genes that either inflammatory or endothelial-specific …

https://doi.org/10.7554/eLife.22536.015
Figure 5 with 4 supplements
Inflammatory signals activate enhancer-like elements with distinct motifs and suggest that CEBP and IRF members mediate responsiveness to TNFα and IL1β.

(a) A schematic for de novo enhancer activation by combinations of signal-dependent and collaborative TFs. (b) Enrichments of the NRF2, NFκB, IRF, C/EBP, AP1, and ETS motifs (y-axis) are shown for …

https://doi.org/10.7554/eLife.22536.017
Figure 5—source data 1

Motif enrichments in 100 bp sequences defined by promoter-distal (≥3 kb) loci gaining ATAC-seq and H3K27ac upon 4 hr oxPAPC, TNFα, and IL1β treatments.

https://doi.org/10.7554/eLife.22536.018
Figure 5—figure supplement 1
The transcriptional response to oxPAPC, TNFα, and IL1β.

(a) Enriched pathways for genes regulated after 4-hr treatment with 40 µg/ml oxPAPC, 10 ng/ml IL1β, and 2 ng/ml TNFα are shown. (b) The proportion of molecule types regulated per cell treatment. (c–g

https://doi.org/10.7554/eLife.22536.019
Figure 5—figure supplement 2
Signal-responsive transcription factor NRF2 binds endothelial enhancers.

(a) HMOX1, TXNRD1, NQO1, and GCLM loci are co-bound by ERG, JUN and NRF2 and acquire H3K27ac upon oxPAPC treatment. (b) NRF2 binding was measured by ChIP-seq after oxPAPC treatment and coincides …

https://doi.org/10.7554/eLife.22536.020
Figure 5—figure supplement 3
Signal-responsive transcription factor NFκB binds endothelial enhancers.

(a) The heme oxygenase 1 (HMOX1) and thioredoxin reductase 1 (TXNRD1) loci are co-bound by ERG, JUN and NRF2 and accumulate H3K27ac upon 4-hr oxPAPC treatment. (b) NRF2 binding was measured by …

https://doi.org/10.7554/eLife.22536.021
Figure 5—figure supplement 4
Binding of p65 at mutated motifs at each allele p65 (NFκB) binding was measured by ChIP-seq in a HAEC donor with whole-genome sequencing information.

Allele-specific p65 binding was calculated at heterozygous loci (y-axis is p65 binding ratio between alleles). Allele-specific motif mutations were calculated as instances where an allele mutated …

https://doi.org/10.7554/eLife.22536.022
Figure 6 with 1 supplement
CEBPD and IRF1 knockdown effect on gene expression response to TNFα.

(a) Binding of p65, CEBPD, and IRF1 was measured by ChIP-seq and is shown at de novo enhancers elicited by IL1β (e) and TNFα (f). Factor binding was measured by CHIP-seq upon 4-hr cytokine …

https://doi.org/10.7554/eLife.22536.023
Figure 6—source data 1

Transcripts up-regulated by more than twofold by CEBPD knockdown in untreated HAECs compared to scrambled control.

Differential expression performed in DEseq.

https://doi.org/10.7554/eLife.22536.024
Figure 6—figure supplement 1
ERG and p65 co-bind loci with a distinctive ETS/NFκB motif.

(a) ERG mRNA was measured in HAECs by RNA-seq in three independent experiments in media lacking (no treatment) or containing 2 ng/µl TNFα or 10 ng/µl IL1β for 4 hr. (b) ATAC-seq, RNA-seq, and …

https://doi.org/10.7554/eLife.22536.025

Tables

Table 1

Overlap of HAEC enhancers with GWAS loci reported for coronary artery disease (CAD) or hypertension (HT). Associated SNPs were downloaded from the NHGRI-EBI Catalog of published genome-wide …

https://doi.org/10.7554/eLife.22536.006
GWAS SNPHAEC enhancer
SNP in enhancerLD to lead SNP from studyp-Value of leadTraitReported gene of leadPubMed IDPosition
(chr, start bp, end bp)
Nearest geneType
 rs12091564Lead2.0E-07CADHFE2216261371, 145395579, 145395699LOC101928979common
 rs72701850LD, rs12091564, r2 = 0.953462.0E-07CADHFE2216261371, 145396840, 145397006LOC101928979common
 rs72701850LD, rs10218795, r2 = 0.953462.0E-07CADHFE2216261371, 145396840, 145397006LOC101928979common
 rs56348932LD, rs17114036, r2 = 0.9168234.0E-19CADPLPP321378990, 242623251, 56988477, 56988661PLPP3EC-specific
 rs56348932LD, rs9970807, r2 = 0.9428682.0E-09CADPLPP3263433871, 56988477, 56988661PLPP3EC-specific
 rs56348932LD, rs17114046, r2 = 0.9428683.0E-07CADPLPP321846871, 213789881, 56988477, 56988661PLPP3EC-specific
 rs10047079LD, rs2229238, r2 = 0.8668487.0E-07CADILR6223190201, 154468114, 154468189SHEEC-specific
 rs55916033LD, rs10496288, r2 = 12.0E-09HTintergenic216261372, 83278987, 83279062LOC1720EC-specific
 rs55916033LD, rs10496289, r2 = 12.0E-09HTintergenic216261372, 83278987, 83279062LOC1720EC-specific
 rs72836880LD, rs10496288, r2 = 12.0E-09HTintergenic216261372, 83308909, 83309314LOC1720EC-specific
 rs72836880LD, rs10496289, r2 = 12.0E-09HTintergenic216261372, 83308909, 83309314LOC1720EC-specific
 rs112798061LD, rs10496289, r2 = 12.0E-09HTintergenic216261372, 83308909, 83309314LOC1720EC-specific
 rs3748861LD, rs13420028, r2 = 0.9162661.0E-10HTGPR39216261372, 133196310, 133196505GPR39mix
 rs3748861LD, rs10188442, r2 = 0.9162661.0E-10HTGPR39216261372, 133196310, 133196505GPR39mix
 rs144505847LD, rs6725887, r2 = 11.0E-09CADWDR1221378990, 242623252, 203672243, 203672412ICA1LEC-specific
 rs144505847LD, rs7582720, r2 = 13.0E-08CADWDR12242623252, 203672243, 203672412ICA1LEC-specific
 rs56155140LD, rs17087335, r2 = 0.9791125.0E-08CADNOA1, REST263433874, 57824385, 57824541NOA1mix
 rs5869162LD, rs6452524, r2 = 0.9246982.0E-07HTXRCC4216261375, 82393827, 82393921XRCC4EC-specific
 rs5869162LD, rs6887846, r2 = 0.9246982.0E-07HTXRCC4216261375, 82393827, 82393921XRCC4EC-specific
 rs6475604LD, rs7865618, r2 = 0.9405972.0E-27CADMTAP216061359, 22052677, 22052823CDKN2BEC-specific
 rs17293632LD, rs72743461, r2 = 11.0E-07CADSMAD32634338715, 67442510, 67442670SMAD3common
 rs17293632LD, rs56062135, r2 = 0.9884895.0E-09CADSMAD32634338715, 67442510, 67442670SMAD3common
 rs17227883LD, rs17228212, r2 = 0.9814382.0E-07CADSMAD31763444915, 67442769, 67443128SMAD3common
 rs1563966LD, rs1231206, r2 = 0.8441519.0E-10CADintergenic2137899017, 2095878, 2096222LOC101927839mix
 rs1563966LD, rs216172, r2 = 0.9093151.0E-09CADSMG6, SRR21378990, 2634338717, 2095878, 2096222LOC101927839mix
 rs7408563LD, rs7246657, r2 = 0.9005127.0E-06CADZNF3832387019519, 37808501, 37809067HKR1common
Table 2

Molecular trait changes observed upon HAEC exposure to pro-inflammatory stimuli. Differential expression was determined in DESeq with duplicate RNA-seq experiments.

https://doi.org/10.7554/eLife.22536.016
StimulusRegulated genes
(>2 fold, 5%FDR)
de novo enhancers increased H3K27ac and accessibility upon stimulation
 oxPAPC 4 hr (40 µg/ml) versus control322 (242 up, 80 down)839
 TNFα 4 hr (2 ng/ml) versus control840 (611 up, 229 down)266
 IL1β 4 hr (10 ng/ml) versus control1174 (807 up, 367 down)3199

Additional files

Supplementary file 1

Public H3K27ac ChIP-seq datasets used to define endothelial-specific enhancer-like regions.

The GEO accession number and cell or tissue type is listed for each dataset in the analysis.

https://doi.org/10.7554/eLife.22536.026
Supplementary file 2

Knockdown of ERG with siRNA modulates transcription of numerous other HAEC transcription factors.

Above are the transcription factors shown to be modulated in two separate HAEC donors (by poly-A RNA-sequencing, greater than twofold change in transcript level, adjusted p<0.05).

https://doi.org/10.7554/eLife.22536.027
Supplementary file 3

List of primers and siRNA oligos used in study

https://doi.org/10.7554/eLife.22536.028

Download links