Background
Most peripheral artery disease (PAD patients) are diagnosed at advanced stages of the disease when the arteries are already seriously compromised [
1]. Beside medical treatment, current therapeutic approaches for symptomatic PAD consist of revascularization with angioplasty and stenting, atherectomy or bypass grafting. However, restenosis is a serious complication occurring in about 2/3 of the patients undergoing revascularization [
2‐
4]. Thus, restenosis after endovascular treatment of atherosclerotic lesions remains a challenging clinical problem [
4] and understanding factors that contribute to its pathophysiology could help predicting occlusion recurrence and/or interfering with it.
Atherosclerotic plaques are characterized by macrophage infiltration and abnormal proliferation of vascular smooth muscle cells (VSMCs) migrated from the medial to the intimal layer. Under the action of pro-inflammatory cytokines released by infiltrating macrophages, intra-plaques VSMCs transdifferentiate from a quiescent-contractile to a secretory-proliferative phenotype. This phenotype is characterized by expression of vascular cells adhesion molecules and inflammatory enzymes such as cyclooxygenase-2 (COX2), and contributes to the progression of atherosclerosis (reviewed in [
5]). Since plaques prone to rupture contain a lower proportion of VSMCs compared to stable plaques, VSMCs survival is a major determinant of plaque stability [
6] and proliferation of VSMCs plays a major role in in-stent restenosis [
7].
The Notch signaling pathway is a major regulator of VSMCs and macrophages functions [
8,
9]. This pathway is mediated by four transmembrane receptors (Notch1-4) and five ligands (Delta like 1, 3, 4, Jagged 1, 2). Ligands expressed on the surface of adjacent cells activate Notch receptors, triggering two proteolytic cuts that generate the active form of the receptor, NotchIC. The latter translocates to the nucleus where it binds transcriptional factor CSL (CBF-1, Suppressor of Hairless and Lag-1) thus promoting the transcription of Notch target genes. Established Notch target genes belong to the Hes (hairy/enhancer of split 1) and Hey (Hes-related proteins) families, involved in the transcription of downstream genes that can either maintain cell in an uncommitted state or induce differentiation [
10,
11]. Different Notch receptors have distinct sets of target genes [
12]. Moreover, activation of Notch receptors by different ligands can trigger different responses [
13‐
17]. Jagged1 (Jag1)—mediated Notch activation is required for the expression of contractility marker smooth muscle (SM) 22 in VSMCs [
18‐
21]. Activation of Notch1 and 3 prevents VSMCs apoptosis [
22] and it counteracts their trans-differentiation from a quiescent/contractile to a proliferative/secretory phenotype promoted by interleukin (IL) 1-β [
23,
24]. Differently from VSMCs, activation of Notch signaling promotes inflammatory response in macrophages [
25] and leads to M1 (pro-inflammatory) rather than M2 (anti-inflammatory) gene expression by activation of NF-kappa B [
26] and by reprogramming mitochondrial metabolism [
27]. Delta-like ligand 4 (Dll4)-Notch3 signaling has been shown to mediate the response of macrophages to pro-inflammatory stimuli [
28,
29]. These data suggest an involvement of the Notch signaling in determining the development and evolution of plaques in PAD patients.
In addition, several microRNAs (miRs) have been identified which play a role in atherosclerosis (reviewed in [
30]). Endothelial cells-secreted miR126-5p promotes VSMCs turnover [
31] and miR424/322 is involved in restenosis in injured rat carotid arteries [
32] and in plaque rupture in mice [
33]. miR21-5p and 146-5p promote VSMCs proliferation [
34] and miR21-5p has been associated to plaque instability [
35,
36]. miR155-5p plays a key role in atherogenic programming of macrophages to sustain and enhance vascular inflammation [
37]. On the contrary, miR125a-5p has a protective effect in atherosclerosis since it decreases the production of inflammatory cytokines in oxLDL (oxidized low density lipoproteins) -stimulated monocyte-derived macrophages [
38] and promotes the anti-inflammatory macrophage phenotype M2 [
37]. Cross-talks between some of these miRs (miR126a-5p, miR21-5p, miR155-5p) and the Notch pathway have been described [
34,
39‐
41].
The aim of this pilot study was the characterization of Notch signaling in plaque material from atherosclerotic lesions of PAD patients in relation to known markers of inflammations and plaque stability. We focused on the relative quantitation of mRNAs for Notch pathway components (Delta-like ligand 4, Jagged1 and target genes Hes1, Hey1, 2, L) and for markers of inflammation and VSMCs trans-differentiation and survival (CD68, COX2, Bcl2, SM22, miRs21-5p, 125a-5p, 126-5p, 146-5p, 155-5p, 424-5p). The relationship between these variables will be discussed.
Discussion
In this study, we have found that PAD patients could be stratified into three groups based on the type on Notch signaling present in plaque material: a group characterized by high Jag1/Hey2/HeyL and low Dll4 (high-Jag1), a group characterized by high Dll4, low Jag1/Hey2/HeyL (high-Dll4) and a group with intermediate characteristics between the first two. These three groups of patients showed specific gene expressions profiles. A “stable” plaque profile, characterized by high SM22, Bcl2, miR125a-5p, was associated to high-Jag1, an “inflammatory” plaque profile, characterized by high CD68, VCAM1, COX2, miR155-5p, 126-5p, 146-5p, 424-5p, associated to high-Dll4, and a “mixed” plaque profile in the group with intermediate values. These observations reveal a possible role of Notch signaling in the pathophysiology of PAD and suggest that ligand-specific activation of this pathway could determine the plaque characteristics and therefore progression of the disease.
Inflammatory cytokines-induced Dll4/Notch3 signaling leads to macrophages activation [
28] whereas Jag1/Notch1, 3 signaling promotes maturation of VSMCs [
18‐
21]. SM22 is a direct transcriptional target of Notch in VSMCs [
18] but Hey2 has also been involved in the modulation of the differentiation of VSMCs [
21]. Activation of Notch by Jag1 counteracts IL-1β-induced transdifferentiation of VSMCs from a quiescent/contractile to a proliferative/secretory phenotype [
24] by upregulation of Hey1 and HeyL [
23]. On the contrary, Dll4-mediated Notch activation is not able to induce differentiation of mesenchymal cells into VSMCs [
48]. Taking these studies into consideration, the existence of the different Notch signaling profiles in our samples might be explained by a prevalence of macrophages in the high-Dll4 group and of VSMCs in the high-Jag1. Consistent with this hypothesis, we found a positive correlation between Dll4 and CD68 and between Jag1 and SM22 mRNA levels. However, immunohistochemical analysis on plaque material from one high-Dll4 patient and one high-Jag1 patient showed a heterogeneous cells population consisting of spindle-like and round cells, both types expressing variable amount of Dll4, CD68, Jag1, SM22, Notch1 and Notch3. This is consistent with capturing VSMCs at different stages of transdifferentiation into macrophage-like cells. It is becoming widely recognized that intraplaque cells identified as macrophages can derive from VSMCs acquiring macrophage-like morphology and phagocytic activity [
49‐
53]. VSMCs expressing variable amounts of both SM22 and CD68 have been identified in aortic [
51] and coronary [
50] atherosclerotic plaques. From our results, we hypothesize that Dll4 induction in VSMCs and the switch from a Jag1-activated to a Dll4-activated Notch signaling could be a marker or a causative factor of VMSCs transdifferentiation in PAD plaques. We found that COX2 and VCAM1 correlated positively with Dll4 and negatively with Jag1. Jag1/Notch inhibition has been linked to COX2 induction in VSMCs [
24] and Dll4-mediated Notch signaling in macrophages activates NF-k B [
28,
54], which, in turn, induces VCAM1 and COX2 transcription (
http://www.bu.edu/nf-kb/gene-resources/target-genes/). We also found that Bcl2 correlated positively with Jag1 and negatively with Dll4. It has been shown that Notch activation in VSMCs is involved in transcriptional induction of the Bcl2-family of protein [
55] and survival [
22] but there is no evidence linking a specific Notch ligand to the pro-survival activity of this pathway.
Consistent with a role for Notch in VSMCs trans-differentiation into macrophages-like cells, we found that cholesterol loading of RASMCs led to reduction of contractility and induction of inflammatory markers in association with reduced levels of Jag1 and Hey2 and increased levels of Dll4 mRNAs. Further studies should determine the molecular mechanisms underlying the cholesterol-induced Jag1–Dll4 switch and whether this switch of ligands, alone or in synergism with other factors (i.e. endothelium secreted-miR126, circulating glucose levels, hemodynamics), plays a role in the transdifferentiation of VSMCs into macrophages-like cells.
We detected high levels of pro-inflammatory miR126-5p, 146-5p, 155-5p and 424-5p and low level of anti-inflammatory miR125a-5p in high-Dll4 patients. In addition, our analysis showed a negative correlation between Jag1 and miR155-5p suggesting that Jag1, but not Dll4, could be the ligand involved in the Notch-mediated downregulation of miR155-5p [
41]. We found that Dll4 positively correlated with miR424-5p, consistent with a study showing that hypoxic conditions, also present in plaques, lead to increased miR424-5p which stabilizes hypoxia-inducible factor-2 α (HIF-2α) [
56], required for Dll4 expression [
57]. In our study, Dll4 was also negatively correlated with miR125a: interestingly, miR125a downregulates Lunatic Fringe, a glycosylation enzyme which determines the selective response of Notch receptors to different ligands [
58].
Our study suggests the existence of a Jag1-activated Notch signaling, associated to quiescence/contractility of VSMCs intra-plaque and low levels of inflammation, and a Dll4-activated Notch signaling associated to markers of inflammation, which could be involved in the progression of the disease. Consistently, Delta-like ligands have been shown to activate small mother against decapentaplegic (SMAD) signaling [
59] which has been linked to restenosis of superficial femoral artery [
60]. The existence of an “inflamed” or “stable” plaque signature has been reported by a microarray-based study of 101 PAD specimens and the Notch target gene Hey2 was included in the “stable” signature [
61]. Microarray analysis on laser-microdissected samples of atheroma containing a prevalence of macrophages or VSMCs led the authors to link the “inflamed” phenotype to the predominance of macrophagic component and “stable” phenotype to a prevalence of VSMCs [
61]. There are several other examples of specific phenotypes associated to a specific combination of Notch ligands and receptors. Jag1 and Dll4 have opposite effects on sprouting angiogenesis [
13] and differential effects of Jag1 and Dll1,4 have been described on T cells proliferation [
15] and the regulation of T cell effector functions in autoimmunity [
14]. Furthermore, there is evidence of differential regulation of osteoclastogenesis by Notch2/Dll1 and Notch1/Jagged1 axes [
16].
Clinical follow-up showed that patients with a “stable” plaque profile (Jag1-activated Notch) were all asymptomatic at 6- and 12-months, whereas patients with an “inflamed” (Dll4-activated Notch) or “mixed” (with high Dll4/COX2 or high COX2) profile presented symptoms to the treated or to the other leg (Additional file
1: Table S4). If confirmed in a larger set of patients, our findings, linking a specific Notch signaling to plaque phenotype, could have clinical relevance. To this end it is important to mention that soluble Dll4-mediated Notch signaling blockade interferes with atherosclerosis progression in an animal model of metabolic syndrome [
54] and that antibodies antagonizing Dll4-Notch signaling, are already under clinical investigation in the oncology setting [
62]. We also found that the expression of several intra-plaque miRNAs is associated to Jag1 or Dll4 levels. Circulating miRNAs are being investigated as possible biomarkers for early detection and progression of PAD [
63]. Since miR126-5p is secreted by endothelial cells [
31], the high levels of miRNA126-5p observed in the “inflammatory” plaques could be mirrored in the serum levels of this miRNA. Further studies are needed to investigate this possibility and to assess the clinical utility of the circulating level of miR126-5p.
Authors’ contributions
Conception and design of the research: LT, RF, PR, PF, AC; Acquisition of data: GA, CF, SD, MP, MBM, CC, FC, PF, AO, AP; Analysis and interpretation of data: PR, GA, CF, AP, LM; Statistical analysis: AP, PR, GA, CF; Drafting of the manuscript: PR, GA, CF, LT, RF; Critical revision of the manuscript: MDM, AO, AP, LM, AP, RF, PR, AC. All authors read and approved the final manuscript.