11-week male C57B6/J mice were implanted with osmotic minipumps delivering of 2.8 mg/kg/d of AngII (Sigma, A9525) for 2 weeks. To confirm the effect of the pharmacological treatment, blood pressure signals from aortic arch before (CTRL); under AngII up to 2 weeks (AngII); and 2 and 3 weeks after the interruption of AngII (AngII+2W; AngII+3w, respectively) were recorded in selected, conscious, and unrestrained animals, with surgically implanted miniaturized telemetry devices (DSI, USA) as described previously (24). Mice were sacrificed after anesthesia at the same time points and age-matched littermate served as control (Figure 1A). All the investigations conformed to the Guide for the Care and Use of Laboratory Animals (NIH Publications no. 8023, revised 2011) and were approved by the Institutional Animal Care and Research Advisory Committee of the Université Catholique de Louvain.

Morphometric and histologic measurements were obtained from hearts arrested in diastole in KCl solution, washed, subsequently fixed with 4% formaldehyde and paraffin embedded. To assess cardiac myocyte transverse area, tissue was costained with wheat germ agglutinin (WGA; for plasma membrane staining, rhodamine-conjugated) and isolectin B4 (GS-IB4; for endothelial staining, biotin-conjugated and revealed with fluorescein-conjugated streptavidin). Cell area from 150 to 200 cells per slide was determined using AxioVision software 4.8.2.0.

Vessels (carotid arteries and aortas in totality) were carefully dissected, washed in cold PBS, and divided in several pieces for further experiments. One piece was subsequently fixed with 4% formaldehyde and paraffin embedded. To assess endothelial activation, carotid sections were stained with ICAM-1 primary antibody (R&D #AF796, 1/1000). On the aortic tissue, to evaluate inflammatory cells infiltration and proliferation, sections were stained with anti-CD45 primary antibody (BD Biosciences #55053P, 1/50) or rabbit anti-Ki67 primary antibody (CST #12202, 1/200), respectively. Smooth muscle actin and H3K27me3 stainings were performed using anti-αSMA primary antibody (CST #19245, 1/400) and anti-tri-methyl-histone H3 antibody (CST #9733, 1/200). Primary antibodies were revealed with EnVision-HRP systems (Agilent) and DAB (Agilent). Finally, nuclei were counterstained with H&E (Agilent). Slides were digitized with a slide scanner (SCN400 Leica) and blindly analyzed with QuPath software (University of Edinburgh) (25).

One-third of aorta was immediately frozen in liquid nitrogen in dry Eppendorf. Because of low yield of RNA extracted in preliminary experiments, and to ensure sufficient amount and quality of RNA for sequencing, tissues from two animals were pooled in 1 ml of tri-reagent (TR118, MRC) for subsequent homogenization with the use of a Precellys Evolution homogenizer (Bertin Instrument, France). Total RNA was extracted with the PureLink™ RNA Micro Scale Kit (Invitrogen) according to the manufacturer's instructions, including a DNase step. In total, RNA from 8 samples, corresponding to 16 mice, were quantified by Qubit RNA BR assay kit (Thermo Fisher Scientific, Q10211) on a Qubit 4 Fluorometer (Thermo Fisher Scientific). RNA integrity was evaluated on the Agilent 2100 Bioanalyzer using the RNA 6000 nanokit (Agilent, 5067-1511). All the samples had RNA integrity number values between 6.8 and 7.7.

Libraries were prepared starting from 150 ng of total RNA using the KAPA RNA HyperPrep Kit with RiboErase (HMR) (KAPA Biosystems, KK8560) following the manufacturer's recommendations (KR1351—version 1.16). Libraries were equimolarly pooled and sequenced on a single lane on an Illumina NovaSeq 6000 platform. All the libraries were paired end (2 × 100 bp reads) sequenced and a minimum of 35 million paired end reads were generated per sample.

All the sequencing data were analyzed using the Automated Reproducible MOdular workflow for preprocessing and differential analysis of RNA-seq data (ARMOR) pipeline v.1.2.0 (26). In this pipeline, reads underwent a quality check using FastQC v0.11.7 (27). Quantification and quality control results were summarized in a MultiQC report (28) before being mapped using Salmon (29) to the transcriptome index which was built using all Ensembl cDNA sequences obtained in the Mus_musculus.GRCm38.cdna.all.fa (release 101) file (30). Then, the estimated transcript abundances from Salmon were imported into R using the tximeta (1.7.14) package (31, 32) and analyzed for differential gene expression with edgeR (3.31.4) package, in which p-value was adjusted using Benjamini-Hochberg method (33). Accordingly, each experimental group (AngII, AngII+2W, and AngII+3W) was compared against the control group (CTRL), thereby producing 3 lists of differentially expressed genes. Differential gene expression results from edgeR were used to conduct Over-Representation Analysis (ORA) and Gene Set Enrichment Analysis (GSEA) with the WebGestaltR (v.0.4.3) package (34). These analyses were conducted on the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome database. RNA-seq full data are available in the NCBI Gene Expression Omnibus (GEO) database under the study accession code GSE175588.

Total RNA was extracted from one-third of the total aorta in a replication cohort of mice, with the same protocol and extraction procedure, i.e., freeze drying, further homogenization in Trizol with Precellys Evolution homogenizer, and extraction with PureLink™ RNA Micro Scale Kit. Extracted RNA was reverse-transcribed and analyzed by quantitative polymerase chain reaction (qPCR) with GAPDH as housekeeping gene. The primer sequences used for qPCR are presented in Table I in Supplemental Material.

Human aortic vascular smooth muscle cells (HAVSMC) were purchased at ScienCell Research Laboratories. Cells were grown in full Smooth Muscle Cell medium (ScienCell #1101), in T75 flask coated with polylysine (2 μg/cm²); medium was renewed every 24 h as cells reached 80% confluence. After serum starvation (0.1% FBS for 18 h) cells were treated with AngII 1μM for 72 h (AngII), or 48 h and then 24 h in control, serum-deprived medium (MemAngII); and compared with the cells maintained 72 h in control, serum-deprived medium (CTRL). Cells between passages 4 and 8 were used for experiments. Total RNA was extracted with Maxwell Kit (Promega, #AS1340). Extracted RNA was reverse-transcribed and analyzed by quantitative polymerase chain reaction (qPCR) with GAPDH as housekeeping gene. The primer sequences used for qPCR are presented in Table I in Supplemental Material. For protein extraction, cells were scrapped in RIPA buffer containing proteinases and phosphates inhibitors. Denatured proteins (in Laemmli buffer) were separated by SDS-PAGE and transferred on PVDF membrane. Membranes were then blocked 1 h in 5% non-fat dry milk in TBS-Tween and incubated overnight at 4°C in 1% milk Tween-TBS with primary antibodies. Antibodies were αSMA (CST #19245, 1/10000) and HSP90 (BD Biosciences #610419, 1/2500). Membranes were visualized by enhanced chemiluminescence on CL-Xposure film (Thermo Fisher Scientific).

Statistical tests were performed using GraphPadPrism (GraphPad Software Incorporation, San Diego, California, USA). Results are reported as mean and standard error of the mean. Statistical analysis was performed using parametric or non-parametric tests where appropriate after verifying normality of values distribution. P < 0.05 is considered as significant with ^* meaning P < 0.05.

As expected (4, 35), we observed a significant increase in total heart/tibial length (TH/TL), left ventricular weight/tibial length (LVW/TL), and left ventricular weight/body weight (LVW/BW) ratios in the AngII group (Figures 1D–F). This hypertrophic phenotype was sustained in time with ratios significantly increased in the AngII+2W and AngII+3W groups (Figures 1D–F). This was reflected by concordant increases in cardiac myocyte transverse area in the AngII group (Figure 1G), which remained significantly elevated in the AngII+2W and AngII+3W groups.

Similar analyses were performed on the vascular phenotype. Histological analysis of carotid sections in the AngII group revealed an increased endothelial expression of ICAM-1, reflective of endothelial activation. ICAM-1 expression remained elevated in the AngII+2W and AngII+3W group (Figure 2A). Consistently, CD45 labeling was increased in the aortic tissue of the AngII group, reflective of inflammation that was also sustained in time in the AngII+2W and AngII+3W groups (Figure 2B). In line with the known proliferative effect of AngII in vascular tissue (36, 37), Ki67 labeling of aortic tissue was increased in the AngII group and, again, maintained in the AngII+3W group (Figure 2C).

AngII stimulates tissue nicotinamide adenine dinucleotide phosphate (NADPH) oxidases in cardiac and vascular cells to produce superoxide anions and, upon dismutation with extracellular SOD, the secondary oxidizing product H₂O₂ (38). Accordingly, plasma hydroperoxides were elevated in the AngII group, and also in the AngII+2W group (Supplementary Figure 1), reflecting persistent systemic oxidant stress at least up to 2 weeks after Ang II removal in our model.

To gain further mechanistic insight into the sustained effect of AngII on this vascular phenotype, we used an unbiased approach through whole aortic tissue RNA-sequencing. We first assessed the differential expression of genes and underlying signaling pathways in the AngII group compared with control, untreated mice. Next, we performed a similar comparison between the AngII+2W and AngII+3W groups (i.e., “memory” conditions) vs. control, untreated mice. We also compared the resulting list of genes and pathways to identify transcripts similarly modulated in AngII and “memory” groups, assuming that these sustained up and/or downregulated transcripts may be related to the observed “memory” phenotype.

Volcano plots in Figures 3A–C represent differential expression of gene data sets comparing each condition to CTRL, considering FDR < 0.05 as a cutoff. As shown in Figures 3A,C, we found 808 genes significantly differentially expressed in AngII and 13 genes in AngII+3W memory condition. Unlike the AngII condition (Figure 3A) in which 55% of genes were upregulated, we found a vast majority of significantly underexpressed genes in AngII+3W memory condition (Figure 3C). As no significant differentially expressed gene was found with this FDR cutoff in the AngII+2W group (Figure 3B), we focused on the AngII+3W memory group compared with AngII for further analysis.

Gene set enrichment analysis of the 2 data sets is summarized in Figures 3D–F. A total of 62 KEGG pathways were significantly differentially modulated in AngII group, and 23 in AngII+3W group, using FDR < 0.1 as a cut-off. The Venn diagram in Figure 3D shows that 10 of them were commonly modulated pathways between the 2 groups. Among these 10 pathways, we next searched for those modulated in the same direction, e.g., up- or downregulated. Four of the 10 commonly modulated pathways were similarly downregulated, and identified as “dilated cardiomyopathy” (mmu05414); “hypertrophic cardiomyopathy” (mmu05410); “longevity regulating pathway” (mmu04213) and, notably, “vascular smooth muscle contraction” (mmu04270) (Figure 3E; Supplementary Figure 2). Conversely, other pathways known to be regulated by AngII, such as, “Renin angiotensin system” (mmu044614), “NF-KappaB signaling” (mmmu04064) and “cGMP signaling” (mmu04022) were, as expected, up (for the first two) or downregulated (for the latter) in the AngII group, but returned to normal level of enrichment in the AngII+3W “memory” group (Figure 3F; Supplementary Figure 2).

We next compared the gene lists to identify transcripts commonly regulated in both AngII and AngII+3W conditions. A total of 13 genes were commonly differentially expressed in both conditions, i.e., 13 genes similarly and significantly modulated both in the “memory” group vs. control, untreated mice and in the AngII group vs. control, untreated mice. Figure 4A represents a heat map with these 13 common transcripts, illustrating changes in the expression level in each condition (AngII and AngII+3W) compared with control, untreated condition (see also Supplementary Figure 3).

Among these, our attention was drawn to Acta2. Acta2 is a gene coding for alpha-smooth muscle actin (αSMA), a protein of the cytoskeleton mainly expressed in smooth muscle cells that is involved in vascular contractility and blood pressure homeostasis. Mutations in this gene cause a variety of vascular diseases, such as thoracic dilated aortic disease, coronary artery disease, stroke, and Moyamoya disease (39), and also multisytemic smooth muscle dysfunction syndrome (40). Downregulation of αSMA expression induced by AngII has also been described in vascular SMC in vitro (41, 42).

To better delineate the putative role of Acta2 in the context of our pathway enrichment analysis, we drew a virtual protein–protein interaction network using the Cytoscape software (fed from the String database). We only included the list of genes that were differentially expressed in AngII+3W memory condition (compared with CTRL), with enlarged FDR cut-off of < 0.1 for a more comprehensive picture (Figure 4C). Note that this interactome did not consider the direction of modulation (e.g., if a transcript was either up or downregulated, to the extent that protein abundance is modulated similarly). Interestingly, we observed that Acta2 was located at a central hub of this protein–protein interaction network.

Based on the earlier observations, we focused on Acta2 as a potential driver of our phenotype.

We confirmed our RNAseq data in a replication cohort of identically treated mice, in which we found a downregulation of Acta2 mRNA expression by RT-qPCR in AngII, and both AngII+ 2W and AngII+3W groups, compared with CTRL (Figure 4B); among the 13 genes identified in the RNAseq data, downregulation of Myl9, Kcnc4, and Lancl3 mRNA expression were also confirmed (Figure 4B). αSMA protein levels were also evaluated through quantitative immunostaining on aortic tissue. We confirmed a significant decrease in αSMA protein expression in the AngII group, which was significantly maintained in both “memory” conditions, AngII+2W and AngII+3W group (Figure 5A). Remarkably, the cytoskeleton in smooth muscle cells of the arterial media in these two groups was profoundly altered, with important structural defects and disorganization (see representative pictures in Figure 5).

Among factors controling αSMA expression are transcription factors Myocardin (Myocd) and Serum Response Factor (SRF). Altogether they form a complex which binds CArG (CCA/T_(rich)GG) sequence motif upstream Acta2 (and other contractile genes), SRF serving as doking platform for Myocd activity, leading to active contractile transcription machinery. In our RNAseq data, we observed that Myocd and SRF transcripts were significantly downregulated under AngII (−2,07 log₂FC, FDR = 0,02), with the same trend in the memory condition for Myocd (−1,42 log₂FC, FDR= 0,13). We decided to evaluate Myocd expression by RT qPCR in our replication group of mice. We observed that this transcription factor was robustly repressed under AngII infusion and that this repression was sustained in time despite the end of the pharmacological stimulation (Figure 5B).

To gain further understanding of the sustained downregulation of Acta2 and Myocd, we examined putative epigenetic regulatory mechanisms. Indeed, our pathway analysis using ORA or GSEA, identified significant enrichement of several epigenetic pathways under AngII such as “HDAC's deacetylase histone” (R-MMU-3214815), “HATs acetylate histone” (R-MMU-3214847), and also “PRC2 methylates histone and DNA”(R-MMU-21230) (see Table I in Supplemental Material, Reactome T1 ORA, Reactome T1 GSEA). Interstingly, PRC2 is a protein complex that keeps transcriptionally silent genes in a repressed state by trimethylating histone H3 on lysine 27. We then evaluated the status of this epigenetic mark in our model, using immunostaining on aortic sections. Notably, this revealed a significant increase in H3K27me3, only in the 2 memory conditions, AngII+2W and AngII+3W (Figure 5C).

To verify these observations in a homotypic cell system of human origin, we developed an in vitro model of AngII memory on cultured human aortic VSMCs. Cells were exposed to continuous AngII at 1μM for 72 h (AngII); or to the same serum-deprived control medium for 72 h (CTRL), or 48 h of AngII, followed by 24 h of control medium (MemAngII). First, we observed the same downregulation of ACTA2 under AngII stimulation, as depicted by mRNA expression and protein levels (Figures 6A,C). Notably, this downregulation was also maintained in the “memory” condition. This was paralleled with a downregulation of mRNA expression of the transcription factor MYOCD, upon continuous AngII, as well as in the “memory” condition (Figure 6B).