SATB2, coordinated with CUX1, regulates IL‐1β‐induced senescence‐like phenotype in endothelial cells by fine‐tuning the atherosclerosis‐associated p16INK4a expression

Abstract Genome‐wide association studies (GWAS) have validated a strong association of atherosclerosis with the CDKN2A/B locus, a locus harboring three tumor suppressor genes: p14 ARF , p15 INK4b , and p16 INK4a . Post‐GWAS functional analysis reveals that CUX is a transcriptional activator of p16INK4a via its specific binding to a functional SNP (fSNP) rs1537371 on the atherosclerosis‐associated CDKN2A/B locus, regulating endothelial senescence. In this work, we characterize SATB2, another transcription factor that specifically binds to rs1537371. We demonstrate that even though both CUX1 and SATB2 are the homeodomain transcription factors, unlike CUX1, SATB2 is a transcriptional suppressor of p16INK4a and overexpression of SATB2 competes with CUX1 for its binding to rs1537371, which inhibits p16INK4a and p16INK4a‐dependent cellular senescence in human endothelial cells (ECs). Surprisingly, we discovered that SATB2 expression is transcriptionally repressed by CUX1. Therefore, upregulation of CUX1 inhibits SATB2 expression, which enhances the binding of CUX1 to rs1537371 and subsequently fine‐tunes p16INK4a expression. Remarkably, we also demonstrate that IL‐1β, a senescence‐associated secretory phenotype (SASP) gene itself and a biomarker for atherosclerosis, induces cellular senescence also by upregulating CUX1 and/or downregulating SATB2 in human ECs. A model is proposed to reconcile our findings showing how both primary and secondary senescence are activated via the atherosclerosis‐associated p16INK4a expression.


| INTRODUC TI ON
Cellular senescence is defined as an irreversible cell cycle arrest. It is a phenomenon characterized by the cessation of cell division often accompanied by an enlarged and flattened cellular morphology. In association with the arrest, senescent cells also secrete multiple proinflammatory molecules, growth factors, and proteases such as the interleukins IL-6 and IL-1β, intercellular adhesion molecule 1 (ICAM-1), vascular cell adhesion molecule1 (VCAM-1), granulocyte macrophage colony-stimulating factor (GM-CSF), vascular endothelial growth factor (VEGF), and matrix metalloproteinase 8 (MMP8), collectively known as the senescence-associated secretory phenotype (SASP). Based on the initiating trigger, cellular senescence can be classified as replicative or stress-induced senescence (Armanios & Blackburn, 2012;Hayflick & Moorhead, 1961). Both replicative and stress-induced senescence are mediated through p53/p21 and/ or p16 INK4a /RB (retinoblastoma protein) pathways; however, preference for one pathway over the other depends on cell type, species and the stimuli (Gorgoulis et al., 2019). Also, both replicative and stress-induced senescence are now recognized as primary senescence in contrast to secondary senescence that is induced by primary senescent cells either via cell-to-cell direct contact (juxtacrine), or through the secretion of SASP (paracrine) (Admasu et al., 2021).
Increasing evidence demonstrates that the accumulation of senescent cells with age is a main contributor to aging and age-related diseases (Campisi & Robert, 2014;Childs et al., 2015). Induction of p16 INK4a has been reported to lead to cellular senescence in a variety of cells and tissues (Kanavaros et al., 2001;Nielsen et al., 1999;Zindy et al., 1997). Depletion of p16 INK4a -positive cells in either a normal mouse model or a mouse model with accelerated aging using genetic manipulation or synolytic reagents eliminate senescent cells in different organs and tissues, which delays age-related diseases and extends healthy life span (Baker et al., 2011(Baker et al., , 2016Dang et al., 2020;Hickson et al., 2019). These findings suggest that p16 INK4a plays a pivotal role in aging and age-related diseases. In addition, p16 INK4a is also one of the markers for cellular senescence and other senescent markers include SAβ-gal (senescence-associated β-galactosidase), γ-H2AX (H2A histone family member X), telomeric length, the expression of SASP genes and cell cycle arrest. artery disease (CAD), myocardial infarction, aneurysms, peripheral artery disease, and heart failure, in addition to other diseases such as glaucoma, type 2 diabetes, and various forms of cancer (Hannou et al., 2015;Kong et al., 2016). Of note, all these disorders are recognized as age-related diseases in that their incidence markedly increases as a function of age. The same CDKN2A/B locus was also associated with frailty and overall human life span (Fortney et al., 2015;Giuliani et al., 2018). However, very few studies have tried to mechanistically investigate how disease-associated causative fSNPs in the CDKN2A/B locus regulate the expression of the CDKN2A/B genes including p16 INK4a in the context of cellular senescence.
IL-1β is an inflammatory molecule that belongs to the interleukin cytokine 1 family. It is an important mediator of the inflammatory response and is involved in a variety of cellular processes, including cell proliferation, differentiation, and apoptosis (Libby, 2017).
IL-1β is also a SASP gene that is activated in senescent cells (Coppe et al., 2008). Ample evidence has been reported to show the important role that IL-1β plays in atherosclerosis. For example, high levels of plasma IL-1β were detected in patients with atherosclerosis, which are positively correlated with the severity of the disease (Qamar & Rader, 2012) and inhibition of IL-1β-induced signal transduction alleviates atherosclerosis (Libby, 2017;Ridker et al., 2017).
Moreover. IL-1β has different effects on different types of cells involved in atherosclerosis (Libby, 2021). In endothelial cells, IL-1β was reported to induce adhesion molecules such as VCAM-1 and ICAM-1. These molecules could be responsible for recruiting inflammatory monocytes, which promotes their invasion into local intima, a step that occurs at the initiation of atherosclerosis (Bevilacqua et al., 1985). IL-1β also induces cytokines such as IL-6 regulating different immune responses (Loppnow & Libby, 1990).
Of note, all these molecules are now recognized as SASP genes, which suggests that IL-1β is an inducer of cellular senescence.
Previously, using Reel-seq and FREP-MS, two novel techniques developed in our laboratory Zhao et al., 2020), we identified six proteins including CUX1, SATB1, SATB2, HOXA10, NFIC, and MYH9 specifically binding to a fSNP rs1537371 on the CDKN2A/B locus (Jiang et al., 2022). Among these proteins, we demonstrated that CUX1, a homeodomain transcription factor, activates both replicative and stress-induced senescence by upregulating p16 INK4a expression (Jiang et al., 2022). In this report, we show that SATB2, another homeodomain transcription factor, suppresses cellular senescence in human arterial ECs by downregulating p16 INK4a expression via its competitive binding with CUX1 to rs1537371. At the same time, we also demonstrate that CUX1 is a transcriptional repressor of SATB2 since upregulation of CUX1 decreases SATB2 expression. Most important, we discover that IL-1β can activate a senescence-like phenotype by fine-tuning p16 INK4a expression. This is performed by upregulating CUX1 and downregulating SATB2 in human ECs.

| Demonstration of allele-imbalanced binding of SATB2 to the atherosclerosis-associated fSNP rs1537371 on the CDKN2A/B locus in human arterial ECs
Previously, we identified that homeodomain transcription factor CUX1, as one of the six proteins in a complex, activates p16 INK4adependent cellular senescence via its specific binding to the atherosclerosis-associated fSNP rs1537371 on the CDKN2A/B locus in human arterial ECs (Jiang et al., 2022). Another one of these six proteins in the complex is SATB2, also known as a homeodomain transcription factor (Britanova et al., 2005). To determine whether SATB2 is also a regulator of cellular senescence, we first demonstrated the specific binding of SATB2 to the fSNP rs1537371 using chromatin immunoprecipitation (ChIP) assay in the SATB2 shRNA knockdown ECs. Knockdown of SATB2 was first confirmed by both qPCR and Western blot as shown in Figure 1a. As a result, a significant decrease in the DNA fragment containing the fSNP rs1537371 was detected in the SATB2 shRNA knockdown ECs compared to the scrambled shRNA control cells (p-value = 0.001) (Figure 1b, right), indicating that there is a specific endogenous binding of SATB2 to the fSNP rs1537371. As a negative control, we performed the same ChIP assay using an isotype control antibody (an anti-IgG antibody), and no obvious difference was observed between the SATB2 shRNA knockdown ECs and the control ECs treated with scrambled shRNA (p-value = 0.5) (Figure 1b, left). Next, we performed a luciferase reporter assay using a reporter construct that contains the risk allele A from the fSNP rs1537371 also in the SATB2 shRNA knockdown cells. This reporter construct was previously used to validate the fSNP rs1537371 by detecting the allele-imbalanced luciferase activity that shows the risk allele A having more luciferase activity than the nonrisk allele C (Jiang et al., 2022). Our data, as shown in Figure 1c (right), indicate that downregulation of SATB2 resulted in a significantly increased luciferase activity (p-value = 0.019) by comparing the SATB2 shRNA knockdown cells with the scrambled shRNA control cells. While this result demonstrates the specific binding of SATB2 to the fSNP rs1537371, it also suggests that SATB2 acts as a suppressor as knockdown of SATB2 increases the luciferase reporter activity. Also, as a negative control, we performed the same luciferase reporter assay with a reporter construct containing an SNP sequence from an irrelevant SNP and as expected, we did not observe any significant difference in luciferase activity between the can be demonstrated. To prove this, we applied allele-imbalanced DNA pulldown-Western blot (AIDP-Wb), a novel method that was recently developed in our laboratory to detect the interaction between a known protein and a confirmed fSNP (Zhao et al., 2020).
Using this method, a differential binding of SATB2 with the risk allele A binding more SATB2 than the nonrisk allele C was observed ( Figure 1d). In addition, the specific binding of SATB2 to the fSNP rs1537371 was further demonstrated by the sequence similarity between the surrounding sequence ATAATTT[A/C]ATT from the fSNP rs1537371 and the MAR (matrix attachment region)-binding motif TTTATTTAATA as SATB2 is a MAR binding protein (Figure 1e) (Alvarez et al., 2000). Together, these data demonstrate that SATB2 is a protein that specifically binds to the fSNP rs1537371 in an alleleimbalanced manner.

| SATB2 regulates the expression of p14 ARF , p15 INK4b , p16 INK4a , and ANRIL in human arterial ECs
Given that the atherosclerosis-associated fSNP rs1537371 is located in the CDKN2A/B locus that harbors p14 ARF , p15 INK4b , p16 INK4a , and ANRIL (Hannou et al., 2015), we investigated the expression of these four genes in the same SATB2 shRNA knockdown human ECs as we showed in Figure 1a.  Figure S1a). Thus, both SATB2 shRNA and siRNA knockdowns demonstrate that SATB2 is a transcriptional regulator that controls the expression of p14 ARF , p15 INK4b ,p16 INK4a , and ANRIL in human ECs. Since p14 ARF , p15 INK4b , and p16 INK4a are cell cycle inhibitors regulating cell proliferation, we also performed cell proliferation assay and cell cycle analysis in the SATB2 shRNA knockdown human ECs and observed a defect in cell proliferation ( Figure S1b) and a decrease in cell numbers in the S/ G 2 /M phase in these ECs ( Figure S1c).
These data, together with the data presented in Figure 1a- hypothesis, we first investigated the possible role of SATB2 in replicative senescence by measuring the expression of SATB2 in passage 5 versus passage 10 ECs. As the results can be seen in Figure 2a, a significant decrease in SATB2 expression in the passage 10 ECs was detected, suggesting that SATB2 is a suppressor of cellular senescence. This result is consistent with our previous findings that CUX1 is an activator of cellular senescence and its expression is significantly increased in p10 ECs compared to p5 ECs (Jiang et al., 2022). We next confirmed the upregulated expression To further demonstrate that SATB2 is a suppressor of cellular senescence by inactivating p16 INK4a expression, we first perofmed a functional complementation assay by ectopically overexpressing p16 INK4a in the SATB2 overexpressed human ECs at passage 10.
( Figure S3a). Consistently, we observed a restoration of cellular senescence as detected by SAβ-gal and γ-H2AX staining as well as the expression of the SASP genes IL-6, IL-1β, and ICAM-1 ( Figure S3bd). Next, we performed an immunocytochemical staining on the plaques of pateints with carotid artery disease using antibodies specifically against both SATB2 and p16 INK4a . As we know, atherosclerosis is considered as an age-related disease and cellular senescence has been detected in the human arterial ECs of patients with atherosclerosis (Katsuumi et al., 2018;Minamino et al., 2002). Again, consistently, we observed a reduced SATB2 and an induced p16 INK4a staining in the plaque zones when comparing them to the normalappearing zones even though the overall expression level of SATB2 is low in normal-appearing zones (n = 8) ( Figure S4).
In addition, like p16 INK4a , p15 INK4b is also upregulated in the SATB2 knockdown human ECs as we showed; therefore, we tested the possibility that SATB2 may also suppress cellular senescence via inactivating p15 INK4b . However, our data suggest that p15 INK4b is not responsible for cellular senescence at least in the SATB2 shRNA knockdown human ECs ( Figure S5). We also tested whether ANRIL is involved in regulating cellular senescence in human ECs as recent publications suggested (Muniz et al., 2021;Tan et al., 2019).
However, once more, we did not observe any obvious change in neither SAβ-gal nor γ-H2AX staining in the ANRIL siRNA knockdown human ECs ( Figure S6). F I G U R E 1 SATB2 regulates the expression of CDKN2A/B genes via its specific binding to the fSNP rs1537371. (a) qPCR and Western blot demonstrating downregulation of SATB2 by shRNA in primary human arterial ECs. Relative density of SATB2 in the Western blot was shown. Data for Western blots represent three biologically independent experiments (n = 3). Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (b) ChIP assay demonstrating a decreased binding of SATB2 to rs1537371 in the SATB2 shRNA knockdown ECs. Data for ChIP assay represent three biologically independent experiments (n = 3). Rabbit anti-IgG antibody was used as a control. (c) SATB2-dependent luciferase reporter assay showing an upregulation of luciferase activities under the conditions of shRNA SATB2 knockdown in 293 T cells. RLA: relative luciferase activity; rs1537371-A and Con: luciferase reporter construct pGL3 (Promoter vector, Promega) containing the fSNP rs1537371 with the risk allele A or an irrelevant SNP as a negative control. Data for this assay represent six biologically independent samples (n = 6). (d) AIDP-Wb showing that SATB2 binds to the fSNP rs1537371 in an allele-imbalanced manner with the risk allele A binding more SATB2 than the non-risk allele C. PARP-1, a DNA end-binding protein, was used as an internal loading control. Data for AIDP-Wb represent three biologically independent experiments (n = 3). (e) Sequence analysis showing the sequence similarity between the MAR-binding motif and the fSNP rs1537371 surrounding sequence. (f) qPCR showing a downregulation of p14 ARF and ANRIL, and an upregulation of p15 INK4b and p16 INK4a in the SATB2 shRNA knockdown human ECs. Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (g) Western blot analysis showing a downregulation of p14 ARF , and an upregulation of p15 INK4b , and p16 INK4a in the SATB2 shRNA knockdown human ECs. Relative density of p14 ARF , p15 INK4b , and p16 INK4a in the Western blots was shown. Data for Western blot analysis represent three biologically independent experiments (n = 3). sh: shRNA. α-Tubulin is used as a loading control. *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001.

| SATB2, repressed by CUX1, competes with CUX1 for binding to the fSNP rs1537371 in human ECs
We previously showed that SATB2 and CUX1 were pulled out in the same protein complex using a 31 bp DNA fragment containing the fSNP rs1537371 as a "bait." (Jiang et al., 2022). Like SATB2, CUX1 also binds to rs1537371 with the risk allele binding more than the non-risk allele; however, CUX1 is an activator of p16 INK4a and p16 INK4a -dependent endothelial senescence (Jiang et al., 2022).
Based on these data, we hypothesize that SATB2 might compete with CUX1 for its binding to the fSNP rs1537371 and then suppressing p16 INK4a and p16 INK4a -dependent endothelial senescence. To test this hypothesis, we first performed a ChIP assay using an anti-CUX1 antibody on the DNA fragment containing the fSNP rs1537371 in human ECs, in which SATB2 is ectopically overexpressed using pLVX-SATB2 vector. Western blot analysis confirmed the overexpression of SATB2 in human ECs, whereas there is no change of CUX1 expression ( Figure 3a). As shown in Figure 3b (right), a decreased binding of CUX1 to the rs1537371-containing DNA fragment was detected in the SATB2 overexpressed ECs compared to the control ECs transfected with pLVX-puro vector (Clontech, Cat#: 632164) (pvalue = 0.0001). As a negative control, the same ChIP assay was performed using an anti-IgG isotype antibody, and no obvious difference was observed (p-value = 0.75) (Figure 3b, left). These data clearly demonstrate that SATB2 competes with CUX1 for its binding to the fSNP rs1537371 and overexpression of SATB2 decreases the binding of CUX1 to the fSNP rs1537371 without changing the level of CUX1 expression in human ECs.
To further demonstrate that SATB2 competes with CUX1 for its binding to the fSNP rs1537371, we performed the same ChIP assay; however, we used an anti-SATB2 antibody in the CUX1 overexpressed human ECs. To our surprise, in the CUX1 overexpressed human ECs, a decreased expression of SABT2 was observed ( Figure 3c), which suggests that CUX1 is a suppressor of SATB2.
Consistent with this expression profile, a significant reduction in the binding of SATB2 to the fSNP rs1537371-containing DNA fragment was detected in the CUX1-overexpressed ECs compared to the control ECs transfected with pLVX-puro vector (Figure 3d, right). To further confirm that CUX1 is a negative regulator of SATB2, we performed a CUX1 shRNA knockdown in human ECs. Not surprising, a significant upregulation of SATB2 was detected at both the protein and mRNA levels as shown in Figure 3e,f. Consistent with this expression profile and, also as expected, we identified a significant increase in the binding of SATB2 to the DNA fragment containing fSNP rs1537371 in the ChIP assay using the anti-SATB2 antibody ( Figure 3g).
To further demonstrate that CUX1 is a transcription factor suppressing SATB2 expression, we searched the genomic sequence on the promotor region of SATB2 and identified a CUX1 core binding motif ATCAAT (Vadnais et al., 2013) as a putative CUX1-binding site (PCBS), which is located ~4.8 kb upstream of the transcription start site of SATB2 (Figure 3h). To demonstrate that this PCBS is the binding site for CUX1, we first performed a DNA pulldown Western blot analysis. As shown in Figure 3i, a strong binding of CUX1 to the PCBS was observed; however, very little binding of CUX1 to the PCBS was detected with the CUX1 core binding motif mutated to ATCCAT. Next, we performed a luciferase reporter assay by cloning the PCBS (ATCAAT) into the promoter vector pGL3 (Promega) under the condition that CUX1 was downregulated. As it can seen in Figure 3j, a significant reduction in luciferase reporter activity was detected in the CUX1 knockdown cells compared to the scrambled shRNA control. While these data demonstrate that CUX1 is a transcription factor for SATB2, it also indicates that, besides the PCBS, CUX1 might need other element(s) on the genomic sequence in order to function as a transcriptional suppressor in repressing SATB2 expression. To further demonstrate that CUX1 binding to the PCBS, we also performed a ChIP assay using anti-CUX1 antibody to pull down the DNA fragment that contains the PCBS. As the results shown in Figure 3k, a significant increase in the binding of CUX1 to the PCBS was detected in comparison with the control using an isotype antibody. Taken together, our data suggest that while SATB2 suppresses p16 INK4a expression by competing with CUX1 for its binding to the fSNP rs1537371 in human ECs, SATB2 itself is also transcriptionally inhibited by CUX1.

| SATB2 regulates p16 INK4a -dependent cellular senescence by competing with CUX1 for binding to the fSNP rs1537371
To demonstrate that SATB2 and CUX1 coordinately regulate cellular These data, together with the data presented above, demonstrate that SATB2 is a transcriptional target of CUX1 and inhibits p16 INK4a -dependent cellular senescence by competing with CUX1 for binding to the fSNP rs1537371 in human ECs. Therefore, decreased expression of SATB2 in the CUX1 upregulated cells results in a coordinative binding of SATB2 with CUX1 to the fSNP rs1537371 for fine-tuning the expression of p16 INK4a and for cellular senescence.

| IL-1β induces a senescence-like phenotype in human ECs by activating p16 INK4a expression via upregulating CUX1 and/or downregulating SATB2 in human ECs
To determine whether IL-1β plays a role in the induction of cellular senescence in human ECs as a causal condition of atherosclerosis, human ECs treated with IL-1β at a concentration of 15 ng/ml for 24 h were stained by both SAβ-gal and γ-H2AX. Although we did not observe any obvious change in SAβ-gal staining at this condition, to our surprise, a dramatic induction of γ-H2AX staining was detected ( Figure 5a). Consistently, an increased expression of the SASP genes including IL-6, ICAM-1 and IL-1β itself ( Figure 5b) and a defective cell proliferation (Figure 5c), as two more senescence markers, were detected in the IL-1β-treated human ECs by qPCR and BrdU incorporation, respectively. Together, these data suggest that IL-1β is an inducer that can activate a senescence-like phenotype in human ECs.
To understand how IL-1β induces a senescence-like phenotype in human ECs., we first checked the expression of p16 INK4a , another marker for cellular senescence, in the IL-1β-treated human ECs To demonstrate that the CUX1/SATB2/ p16 INK4a pathway is responsible for IL-1β-induced senescence-like phenotype, we also performed functional complementation assays by either downregulating CUX1 or over-expressing SATB2 in the IL-1β treated human ECs ( Figure S7a,b). As expected, IL-1β treatment induced the expression of p16 INK4a and both downregulation of CUX1 and upregulation of SATB2 in the IL-1β-treated human ECs reduced the expression of p16 INK4a to a level that is equivalent to that in the control cells ( Figure S7a,b, middle and right lane). Consistent with the p16 INK4a expression profile, we observed a restoration of cellular senescence as evidenced by a reduced γ-H2AX staining and a decreased SASP gene expression ( Figure S7c,d, second from the left, second from the right F I G U R E 3 SATB2, repressed by CUX1, competes with CUX1 for its binding to the fSNP rs1537371 in human ECs. (a) Western blot analysis showing no obvious change of CUX1 expression in the SATB2-overexpressed human ECs. Data for the Western blot analysis represent three biologically independent experiments (n = 3). (b) ChIP assay using an anti-CUX1 antibody demonstrating that the binding of CUX1 to the fSNP rs1537371 is significantly reduced (p-value <0.0001) in the SATB2-overexpressed human ECs (right). As a control, anti-IgG isotype antibody was used, and no obvious change was observed (p-value = 0.75) (left). Data for ChIP assay represent three biologically independent samples (n = 3), each performed in duplicate. (c) Western blot analysis showing a downregulation of SATB2 expression in the CUX1-overexpressed human ECs. Data for the Western blot analysis represent three biologically independent experiments (n = 3). (d) ChIP assay using an anti-SATB2 antibody demonstrating that the binding of SATB2 to the fSNP rs1537371 is significantly reduced (p-value = 0.0008) in the CUX1 overexpressed human ECs (right). As a control, anti-IgG isotype antibody was used (left). Data for ChIP assay represent three biologically independent samples (n = 3), each performed in duplicate. (e) Western blots and (f) qPCR showing an upregulation of SATB2 expression in the CUX1 shRNA knockdown human ECs. Data for the Western blot analysis represent three biologically independent experiments (n = 3). Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (g) ChIP assay using an anti-SATB2 antibody demonstrating that the binding of SATB2 to the fSNP rs1537371 is significantly increased (p-value = 0.006) in the CUX1 shRNA knockdown human ECs (right). As a control, anti-IgG isotype antibody was used, and no change was observed (p-value =0.95) (left). Data for ChIP analysis represent three biologically independent samples (n = 3), each performed in duplicate. (h) Diagram showing the sequence of a putative CUX1 binding site (PCBS) and its relative location (4.8 kb upstream of the SATB2 transcription start site) within the SATB2 promoter region. Underlined nucleotides represent the CUX1 core binding motif ATCAAT. (i) DNA pulldown Western blot analysis demonstrating a specific binding of CUX1 to the 30-bp PCBS containing the CUX1 core binding motif ATCAAT. As a control, PCBS containing a mutated CUX1 core binding motif ATCCAT was used. PARP-1 was used as an internal loading control. (j) CUX1-dependent luciferase reporter assay showing a downregulation of luciferase activity in the CUX1 shRNA knockdown 293 T cells. RLA: relative luciferase activity; PCBS (ATCAAT): luciferase reporter construct pGL3 (Promoter vector, Promega) containing PCBS with the CUX1 core binding motif ATCAAT. Control (ATCCAT): a control luciferase reporter construct pGL3 containing a mutated CUX1 core binding motif ATCCAT that was used as a negative control. Data for this assay represent six biologically independent samples (n = 6). (k) ChIP assay in ECs using an anti-CUX1 antibody showing an increased binding of CUX1 to the putative CUX1 binding site (PCBS) on the SATB2 promoter. For negative control anti-IgG antibody was used. pLVX: overexpression vector control; pLVX SATB2 and pLVX CUX1: overexpression of SATB2 and CUX1; and sh: shRNA. Relative density of CUX1 and SATB2 in all the Western blots was shown. *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001. For control, an anti-IgG isotype antibody was used.

| DISCUSS ION
Previously, we demonstrated that CUX1 activates p16 INK4adependent cellular senescence in human ECs via binding to the atherosclerosis-associated fSNP rs1537371, with the risk allele A binding more CUX1 than the non-risk allele C (Jiang et al., 2022), which reveals a mechanism underlying the contribution of the atherosclerosis-associated fSNP rs1537371 to the susceptibility of this disease. In this report, we demonstrate the same specific binding of SATB2 to the fSNP rs1537371, with also the risk allele A binding more SATB2 than the non-risk allele C. However, unlike CUX1, SATB2 is a suppressor of p16 INK4a and binding of SATB2 to rs1537371 results in downregulation of p16 INK4a expression and inhibition of cellular senescence. Therefore, SATB2 can compete with CUX1 in regulating p16 INK4a expression and endothelial senescence.
Interestingly, we also find that CUX1 is a transcriptional suppressor of SATB2, and its expression can decrease the binding of SATB2 to rs1537371 by reducing the expression of SATB2. Thus, SATB2, coordinated with CUX1, fine-tunes the expression of p16 INK4a , which ensures a precise and tight regulation of cellular senescence.
Both experimental and clinical evidence support that IL-1β is an important contributor to atherosclerosis and its complications (Libby, 2017). However, the underlying mechanism is not yet wellunderstood. In this report, we find that IL-1β can induce a p16 INK4adependent senescence-like phenotype in human ECs as evidenced by increased γ-H2AX staining, upregulation of both p16 INK4a and SASP gene expression, as well as arrested cell cycle. We further demonstrate that IL-1β induces a senescence-like phenotype in human ECs also via activation of CUX1 and/or suppression of SATB2. However, currently, we do not know the signal transduction pathway that IL-1β activates CUX1 and/or suppreses SATB2. We also do not know how IL-1β is activated at first place in the development of atherosclerosis.
Previously, we also demonstrated that induction of cellular senescence by telomeric shortening, DNA damage induced by bleomycin, and oxidative stress as mimiced by H 2 O 2 at least in human ECs uses the same CUX1/p16 INK4a pathway as IL-1β does, resulting in upregulation of SASP genes including IL-1β itself (Jiang et al., 2022). Cellular senescence induced by the abovementioned stresses is now recognized as primary senescence in contrast to secondary senescence that is induced by primary senescent cells either via cell-to-cell direct contact (juxtacrine), or through the secretion of SASP (paracrine) (Admasu et al., 2021). While it still remains to be determined if secreted IL-1β during primary senescence can paracrinely activate secondary senescence both in vitro and in vivo, our findings suggest that secondary senescecne induced by IL-1β in human ECs and/or VSMCs might be one of the machnisms that contribute to the accumulation of senescent cells, which could lead to atherosclerosis and its complications.
Collectively, based on our current work together with our previous findings (Jiang et al., 2022), we believe that both stressesinduced primary senescence and IL-1β-induced secondary senescence at least in human ECs could be through activation of CUX1 expression as shown in the model described in Figure 6.
This model also reveals that CUX1 is a transcriptional suppressor of SATB2 and upregulation of CUX1 can result in downregulation of SATB2. Therefore, together, CUX1 and SATB2 fine-tune the expression of p16 INK4a via their competitive, but also coordinated binding to the atherosclerosis-associated fSNP rs1537371, which precisely and tightly regulates cellular senescence in human ECs. Considering that endothelial senescence is the initial step of developing atherosclerosis (Davignon & Ganz, 2004) and IL-1β is an important mediator and biomarker of atherosclerosis and a SASP gene itself, we believe that understanding the mechanism and the signal transduction pathway of endothelial senescence as we present in this model will give us new insights into the pathogenesis of atherosclerosis.
Of note, we present this model based on our post-GWAS functional analysis on the atherosclerosis-associated fSNP rs1537371 (Hannou et al., 2015). However, as we revealed in Wu et al. are the signaling pathways that activate these regulatory elements?
Do these cis-REs coordinate with each other regulating p16 INK4adependent cellular senescence? Once these questions and many other questions are answered, we will have a better idea about how disease-associated endothelial senescence is regulated, which, we believe, will provide us a better strategy to develop therapeutics for these diseases.

| Cell culture and reagents
Primary human arterial ECs (Cat#: CC-2535) were purchased from Lonza. Cells were cultured at 37°C in 5% CO 2 in basal medium EGM-2 supplemented with 10% fetal bovine serum. All cells are free of mycoplasma.

| Primers and antibodies
All primers used in this study were purchased from IDT and are listed in Table S1. All antibodies used are listed in Table S2 with the corresponding supplier information.
Briefly, scrambled shRNA control human ECs and the SATB2 shRNA knockdown human ECs were cross-linked with 1% formaldehyde for 10 min. Sonication was carried out at 30% amplitude with 20 s F I G U R E 5 IL-1β induces a senescence-like phenotype in human ECs. (a) An increased γ-H2AX staining in human ECs treated with 15 ng/ ml IL-1β for 24 h. quantitative plots for γ-H2AX foci/cells (%) after γ-H2AX staining are shown on the right side of the panel. Data for γ-H2AX staining represent three biologically independent experiments (n = 3). (b) qPCR analysis demonstrating a significant increase in the expression of the SASP genes IL-6, IL-1β, and ICAM-1 in human ECs treated with IL-1β. Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (c) BrdU incorporation assay showing a defect in cell proliferation in human ECs treated with IL-1β. Data for BrdU incorporation assay represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (d) Western blot analysis showing an upregulation of p16 INK4a and CUX1 and a downregulation of SATB2 in the human ECs treated with 15 ng/ml IL-1β for 24 h. Relative density of CUX1, SATB2 and p16 INK4a in the Western blot was shown. Data for the Western blot analysis represent three biologically independent experiments (n = 3). (e) qPCR analysis showing an upregulation of p16 INK4a and CUX1 and a downregulation of SATB2 in the human ECs treated with 15 ng/ml IL-1β for 24 h. Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. (f) Time courses of CUX1, SATB2 and p16 INK4a expression in the human ECs treated with 15 ng/ml IL-1β for 12, 24, 48 and 72 h, respectively, as shown by qPCR analysis. All the qPCR data were normalized to the untreated control. Data for qPCR analysis represent a combination of three biologically independent samples (n = 3), each performed in duplicate. *p-value <0.05; **p-value <0.01; ***p-value <0.001 F I G U R E 6 Model highlighting the CUX1/SATB2/ p16 INK4a signal transduction pathway leading to cellular senescence in human ECs via the atherosclerosis-associated fSNP rs1537371 on the CADassociated CDKN2A/B locus. In human ECs, both stress-induced primary sensencece and IL-1β-induced secondary senescence are activated via upregulation of CUX1 expression. However, it is not known how CUX1 is upregulated. Upregulation of CUX1, which suppresses SATB2 expression, enhances the binding of CUX1 to the atherosclerosis-associated fSNP rs1537371(A/C) by reducing the ability of SATB2 to compete for the binding. Therefore, coordinated binding of CUX1 and SATB2 to the fSNP rs1537371 fine-tunes p16 INK4a expression, which guarantees a precise and tight regulation of p16 INK4a -dependent endothelial senescence. on and 50 s off for 5 min. Ten micrograms anti-SATB2 antibody coupled to Dynabeads™ Protein A/G (Thermo Fisher Scientific, Cat#:10001D and 10003D) was incubated with sonicated nuclei at 4°C overnight. DNA was pulled down and purified using PCR purification kit (Macherwy-nagel Cat#: 740609.5) after reversal of the crosslink. The purified DNA was then used for real-time PCR analysis of the sequences around the fSNP rs1537371. Rabbit IgG was used as an isotype control. The data represent the combination of three independent samples (n = 3).

| Luciferase reporter assay
Luciferase reporter assay was performed in 293 T cells using pGL3 luciferase reporter vector (Promega, cat#: E1761). Insert target sequences are listed in Table S1. Luciferase reporter construct DNA was transfected into 293 T cells by FuGENE HD transfection reagent (Promega, Cat#: E2311) together with the control vector. Luciferase activity was measured by the Dual-Glo® Luciferase Reporter Assay System (Promega, Cat#: E2920). All experiments were performed according to the manufacturer's protocol. The data represent six independent biological replicates (n = 6).
In brief, a 31-bp biotinylated SNP sequence centered with either the risk or non-risk allele was generated by annealing two biotinylated primers (IDT). Approximately 1 μg DNA was then attached to 40 μl of Dynabead™M-280 Streptavidin. DNA-beads were mixed with ~100 μg nuclear extract isolated from ECs at RT for 1 h with rotation. Nuclear extracts were isolated using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, Cat#: 78835).
After washing off the unbound proteins, the DNA bound proteins were eluted by sample buffer and resolved on an SDS-PAGE gel for Western blot analysis using an antibody directed against SATB2. For an internal control, the same blot was probed using an antibody directed against PARP-1. The data represent three independent biological replicates (n = 3).  Table S1.

| qPCR analysis
The data represent the combination of three independent samples (n = 3).

| Western blot analysis
Whole cell lysates prepared using RIPA buffer (Sigma,Cat#: 89900) were used for Western blot analysis. Proteins were resolved on SDS-PAGE gels, transferred to PVDF membranes and then detected using gene-specific antibodies. All antibodies were purchased and used as listed in Table S2. For a loading control, α-Tubulin was used.
The data represent three independent biological replicates (n = 3).
Images were analyzed using ImageJ software (version 1.52 K, NIH).
The data represent three independent biological replicates (n = 3).
Visualization was carried out using an RVL-100-G microscope (Echo Laboratories, San Diego, CA, USA). Images were analyzed using ImageJ software (version 1.52 K, NIH). The data represent three independent biological replicates (n = 3).

| RNAi knockdown
For siRNA transient knockdown in human ECs, siRNAs for human SATB2, CUX1 and p16 INK4a were purchased from Horizon Discovery and knockdown was performed according to the manufacturer's protocol. For SATB2 and CUX1 shRNA knockdown in human ECs, lentiviruses were generated using the pLKO.1-puro vector (Addgene, Plasmid #8453). Assays for cellular senescence including SAβ-Gal and γ-H2AX staining as well as SASP gene expression were performed 48 h after siRNA transfection or shRNA lentiviral infection.
The targeted sequences are listed in Table S1.

| Immunocytochemical staining for p16 INK4a and SATB2
Human atherosclerotic plaques were obtained from patients undergoing carotid endarterectomy. Part of the carotid artery that shows calcified hard tissue was used as a plaque zone and part that is far from the calcified zone was used as normal-appearing zone. Both plaque and normal-appearing zones were separated and fixed with 4% buffered formalin for 2 h and stored in 30% sucrose solution containing 0.05% sodium azide overnight. Sections were made of 10 micrometers thickness, permeabilized with 0.1% triton X-100 for 4 h, and blocked overnight in PBS containing 2% BSA. Sections were further incubated for 24 h with primary antibodies against p16 INK4a (Invitrogen, cat#: MA5-17142; 1:500 dilution) and SATB2 (Novus Biologicals Cat#: NBP1-03328; 1:1000 dilution). After washing with PBS, sections were incubated for 1 h at room temperature with fluorochrome conjugated secondary antibodies (Alexa fluor-488 goat anti-mouse and Alexa fluor-647 goat anti-rabbit). Tissue sections were stained and mounted with VECTASHIELD DAPI. Images were taken using confocal laser microscopy and analyzed using image J.
The data represent two independent experiments with eight plaque zones (n = 8) and eight normal-appearing zones (n = 8).

| Statistical analysis
For normally distributed data, all data were represented as standard error mean (SEM). p-values were calculated using Student's ttest with two tails. The non-normally distributed data related to the quantification of SATB2 and p16 INK4a immunocytochemical staining in Figure 3 were represented as Mean ± SEM interquartile range, and p-values were calculated with the non-parametric Mann-Whitney test for pairwise comparisons.

AUTH O R CO NTR I B UTI O N S
Gang Li designed the study, analyzed the data, and drafted and re-

ACK N OWLED G M ENTS
We thank Drs. Xiaojun Tan, Aditi Uday Gurkar, Jie Liu, and Shihui Liu for scientific discussions concerning this work. We also want to give special thanks to Dr. Toren Finkel for his proposal on this study and for revising the manuscript.

FU N D I N G I N FO R M ATI O N
This work was supported partly by grants from NIH NIA R01AG056279 (GL) and R01AG065229 (GL).

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data and reagents are available upon reasonable request.