Endothelial NOTCH1 is suppressed by circulating lipids and antagonizes inflammation during atherosclerosis

Abstract

Although much progress has been made in identifying the mechanisms that trigger endothelial activation and inflammatory cell recruitment during atherosclerosis, less is known about the intrinsic pathways that counteract these events. Here we identified NOTCH1 as an antagonist of endothelial cell (EC) activation. NOTCH1 was constitutively expressed by adult arterial endothelium, but levels were significantly reduced by high-fat diet. Furthermore, treatment of human aortic ECs (HAECs) with inflammatory lipids (oxidized 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine [Ox-PAPC]) and proinflammatory cytokines (TNF and IL1β) decreased Notch1 expression and signaling in vitro through a mechanism that requires STAT3 activation. Reduction of NOTCH1 in HAECs by siRNA, in the absence of inflammatory lipids or cytokines, increased inflammatory molecules and binding of monocytes. Conversely, some of the effects mediated by Ox-PAPC were reversed by increased NOTCH1 signaling, suggesting a link between lipid-mediated inflammation and Notch1. Interestingly, reduction of NOTCH1 by Ox-PAPC in HAECs was associated with a genetic variant previously correlated to high-density lipoprotein in a human genome-wide association study. Finally, endothelial Notch1 heterozygous mice showed higher diet-induced atherosclerosis. Based on these findings, we propose that reduction of endothelial NOTCH1 is a predisposing factor in the onset of vascular inflammation and initiation of atherosclerosis.

Figure 1.

NOTCH1 is constitutively expressed by adult arterial endothelium. (A) Relative levels of Notch receptors measured by qRT-PCR in HAECs (n = 11; three donors). (B and C) Western blot performed on cultured HAECs (n = 3 donors; VEGFR2 positive) and HASMCs (n = 3 donors; SM22 and Calponin positive). Relative protein level was measured by densitometry and normalized to loading control γ-Tubulin (γTUB). Mean values detected in HAECs are used as the reference point (value = 1), and data are presented as mean ± SEM. ****, P < 0.0001; ***, P < 0.001 by paired (A) and unpaired (C) Student’s t test. (D) Immunohistochemistry performed on human coronary arteries from normal individuals. NOTCH1 was detected in the endothelium (left and middle; arrows), which was also positive for CD31 (right). Dotted box panels (middle; NOTCH1 staining) are higher magnifications of the dotted areas presented in the adjacent left panels. Bars, 50 µm. (E) Quantification of area covered by NOTCH1 in human coronary arteries from vessels without atherosclerosis (n = 14 sections) on a linear region of 500 µm in length. (F) NOTCH1 transcript levels from 147 HAECs isolated from independent individual heart transplant donors were measured by microarray analysis and represented as Log2 expression values. Data are represented as mean ± SEM.

Figure 2.

Notch signaling is repressed by proatherogenic stimuli in vivo. (A and B) C57BL/6J wild-type mice were fed a standard diet (chow) or HFD for 4–13 d or 4 d followed by a 3-d chow diet. After an overnight fast (A) circulating cholesterol levels were measured (B). The data are represented as fold change compared with the mean chow (B). (C) ECs were isolated from the descending aorta, and qRT-PCR was performed to determine Notch1 and Hey1 levels under HFD compared with chow. (B and C) Chow n = 29 and HFD n = 6–10. (D) Correlations between circulating cholesterol levels and endothelial mRNA levels of Notch1 and Hey1 at early time points (n = 47). ****, P < 0.0001; ***, P < 0.001; *, P < 0.05 relative to Chow; ^###, P < 0.001; ^#, P < 0.05 relative to 4HFD, by unpaired Student’s t test. Data are represented as mean ± SEM.

Figure 3.

Inflammatory cytokines and Ox-PAPC repress NOTCH1 expression and signaling in HAECs. (A and B) HAECs were treated with recombinant 10 ng/ml IL1β, 10 ng/ml TNF, or 50 µg/ml Ox-PAPC for 4 h. Transcript levels of Notch signaling molecules were measured by qRT-PCR (n = 6–15; three donors). (C and D) HAECs were treated with 10 ng/ml TNF or 50 µg/ml Ox-PAPC for 4 h in the presence of STAT3 inhibitor (10 µM Stattic) or vehicle control (n = 8–14; four to five donors); mRNAs level of NOTCH1 and HES1 were measured by qRT-PCR. (E) Microarray analysis of NOTCH1 levels in 147 HAECs isolated from individual donors untreated (black dots) or treated with 40 µg/ml Ox-PAPC (red dots) for 4 h are shown and represented as Log2 expression values. (F–K) HAECs were treated with 50 µg/ml Ox-PAPC for the indicated times. (F and G) NOTCH1 protein was detected by Western blot, and relative amount was measured by densitometry and normalized by γ-TUBULIN (γTUB; n = 3 donors). (H and I) NOTCH1 mRNA level at 4 and 6 h after treatment with Ox-PAPC in three independent donors (#1–3). (J and K) Transcript levels of NOTCH1, JAG1, HES1, and HEYL were measured over time after Ox-PAPC treatment by qRT-PCR (n = 4; two donors). Bottom (Donor#1) and top (Donor#2) arrows indicate the differences in the regulation of target genes after 4 h of treatment. Data are represented as mean ± SEM. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05 by unpaired Student’s t test. In C and D, ^####, P < 0.0001; ^###, P < 0.001; ^##, P < 0.01; ^#, P < 0.05 to control-treated cells by unpaired Student’s t test. (L and M) Degrees of NOTCH1 repression after treatment with Ox-PAPC were mapped to SNPs across the genome using data from 147 donors. The Manhattan plot shows the significance of association at each SNP marker across the genome. The red arrow shows the peak SNP of association (L). Boxplots of the change in NOTCH1 expression in each donor are shown based on the genotype of the peak association SNP (M).

Figure 4.

Knockdown of NOTCH1 shares transcriptional targets with Ox-PAPC. (A–C) The transcriptional profile of HAECs treated with siRNA NOTCH1 or 50 µg/ml Ox-PAPC (6 h) was analyzed by gene microarray. (A) Venn diagram indicating overlapping and independent transcripts changed by Ox-PAPC or knockdown of NOTCH1. (B and C) Validation by qRT-PCR analyses (n = 3 each condition). (D) IL8, CXCL1, TDAG51, SELE, and CHST1 were measured by qRT-PCR in additional HAEC donors treated with siRNA control or targeting NOTCH1 (n = 9–11). (E) IL8 and CXCL1 protein levels were measured by ELISA in the culture supernatant of HAECs treated with siRNA control or targeting NOTCH1 (n = 6; three donors). (F–H) HAECs from 147 heart transplant donors were treated (red dots) or not (black dots) with 40 µg/ml Ox-PAPC (4 h). mRNA levels of IL8 (F), CXCL1 (G), and TDAG51 (H) were measured by gene microarray analysis and represented as Log2 expression values. The percentage of donors with increased level of the interest transcript under Ox-PAPC treatment is indicated on each graph in red. (I–K) HAECs were treated for the indicated time with 50 µg/ml Ox-PAPC. Levels of IL8, CXCL1, and TDGA51 were measured by qRT-PCR (n = 4; two donors). ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05 by unpaired Student’s t test. Data are represented as mean ± SEM.

Figure 5.

Maintenance of NOTCH1 signaling rescues the induction of TDAG51 and IL8 by Ox-PAPC. (A) Schema of ZEDN1 (rat NICD bound to the plasma membrane) construct that was cloned in a CMV-GFP lentiviral vector. S1–3, proteolytic cleavage sites; LNR, Lin12/Notch repeats; HD, heterodimerization domain; RAM, RBP-jk–associated molecule; ANK, ankyrin repeats; TAD, transactivation domain; PEST, proline, glutamate, serine, threonine rich. (B) Transcript of rat Notch1 was detected by RT-PCR in HAECs transduced with lenti-ZEDN1. (C) Detection of endogenous NOTCH1-p120 (S1; black arrow) and ZEDN1 (red arrow) protein was detected in HAECs transduced with lenti-ZEDN1 by Western blot and normalized to γ-TUBULIN. (D) After transduction with lenti-ZEDN1 or GFP, NOTCH1, JAG1, HES1, IL8, CXCL1, TDAG51, and HEYL mRNA expression was measured by qRT-PCR. (E and F) HAECs were transduced with lentivirus encoding ZEDN1 or virus control (vGFP) and treated with 50 µg/ml Ox-PAPC (6 h) or control media. Notch signaling molecules (E) and downstream proatherogenic target (F) mRNA levels were determined by qRT-PCR. (D–F) n = 5–6; four donors. Data are represented as mean ± SEM. ***, P < 0.001; **, P < 0.01; *, P < 0.05 by paired Student’s t test.

Figure 6.

Decrease in endothelial NOTCH1 expression increases monocyte binding in vitro and in vivo. (A) NOTCH1 protein level was evaluated in HAECs transfected with control siRNA or siRNA targeting NOTCH1. γ-TUBULIN (γTUB) was used as a loading control. (B and C) The confluent monolayers of HAECs were then cocultured with CFSE-labeled THP-1 monocytes. CFSE-labeled THP-1 (green, B) bound to the monolayer was counted in 10 fields per donor and condition (n = 6; three donors). Σ indicates data from the three donors grouped. ****, P < 0.0001 by unpaired Student’s t test. Mean ± SEM is shown. (D) CD45^POS leukocytes were detected by immunohistochemistry on aortic sections from N1^ECWT and N1^EC+/− mice. Black arrows, CD45^POS cells at the surface of the endothelium; open arrows, CD45^POS cells underneath the endothelium. (E) Co-immunostaining of CXCL1/GRO-α and αSMA was performed on aortic sections from N1^ECWT and N1^EC+/− mice. Lower dotted box panels are higher magnifications of the dotted boxes in the panels above. (D and E) Six to nine animals per genotype were examined; representative sections are shown. (F–J) 5-wk-old Notch1-floxed or Cdh5-Cre^ERT2–negative (N1^ECWT; n = 4) or –positive (N1^ECKO; n = 5) littermates were injected with tamoxifen to induce Notch1 deletion in the endothelium. Descending aortae were harvested 2 wk later, stained for β-Catenin to define cell junctions (green), and used for en face confocal imaging. Recombined, Notch1 knockout, ECs were positive for Tomato reporter (red, G–J). Tomato-negative leukocytes were detected at the endothelium surface of N1^ECKO animals (arrows; G–J). (J) 3D reconstitution of I. Bars: (B) 50 µm; (D and E) 25 µm; (F–J) 20 µm. (E–J) Nuclei were stained with Dapi.

Figure 7.

Decrease in endothelial Notch1 increases basal inflammation in L-sIDOL animals. (A) Mice with heterozygous deletion of Notch1 in the endothelium N1^EC+/− were crossed with transgenic L-sIDOL mice (N1^EC+/−/L-sIDOL). (B and C) CD45^POS leukocytes were detected by immunohistochemistry on aortic sections from N1^EC+/−, N1^ECWT/L-sIDOL, and N1^EC+/−/L-sIDOL animals. Black arrows, CD45^POS cells at the surface of the endothelium; open arrows, CD45^POS cells underneath the endothelium. Quantification of accumulated CD45^POS cells was assessed per 280-µm endothelium length in aortic cross section from adult animals. ****, P < 0.0001 by unpaired Student’s t test. Mean ± SEM is shown. (D) Co-immunostaining of CXCL1/GRO-α and αSMA was performed on aortic sections from N1^EC+/−, N1^ECWT/L-sIDOL, and N1^EC+/−/L-sIDOL animals. Dotted box panels are higher magnifications of the cells indicated by arrowheads in each panel. Nuclei are stained with Dapi. Cross sections from 9–10 animals per genotype were examined; representative sections are shown. Bars, 25 µm.

Figure 8.

Hemizygous loss of endothelial Notch1 increases diet-induced atherosclerosis in L-sIDOL mice. 6-wk-old mice with heterozygous deletion of Notch1 in the endothelium were crossed with transgenic L-sIDOL mice (N1^EC+/−/L-sIDOL) and fed for 28 wk a standard diet (chow) or HFD. (A) Plasma cholesterol was measured after 28 wk on chow or HFD. (B–E) En face aorta atherosclerosis was assessed by Sudan IV staining to detect the subintimal accumulation of lipids. Mean ± SEM are shown (A–C). In the N1^EC+/− animals, lesions were observed in the descending aorta (b, ellipsis) and femoral arteries (a and a′, white arrows; E); in the N1^EC+/+/L-sIDOL and N1^EC+/−/L-sIDOL animals, lesions were also found in the aortic arch (c and c′, ellipses; E) and the arch branches (open arrows; E). (A) ****, P < 0.0001; **, P < 0.01 by unpaired Student’s t test. (C) ^#, P = 0.011 relative to N1^EC+/−; *, P = 0.038 by Mann–Whitney test relative to N1^EC+/+/L-sIDOL. Animals fed chow, n = 5 per genotype; HFD, N1^EC+/− and N1^EC+/+/L-sIDOL n = 6 and N1^EC+/−/L-sIDOL n = 8.