Research ArticleAtherosclerosis

Long noncoding RNA SNHG12 integrates a DNA-PK–mediated DNA damage response and vascular senescence

See allHide authors and affiliations

Science Translational Medicine  19 Feb 2020:
Vol. 12, Issue 531, eaaw1868
DOI: 10.1126/scitranslmed.aaw1868

A senescent lnc to atherosclerosis

DNA damage and senescence are thought to enhance atherosclerotic lesion formation, although how this happens is unclear. Haemmig et al. identified an endothelial-enriched long noncoding RNA (lncRNA) as a DNA-dependent protein kinase (DNA-PK)–dependent regulator of vascular DNA damage and cellular senescence. Small nucleolar host gene-12 (SNHG12) was down-regulated in mouse, pig, and human atherosclerotic arteries and correlated with DNA damage and senescence. Endothelial Snhg12 knockdown exacerbated vascular cellular senescence and lesion formation in mouse models of atherosclerosis. Both nicotinamide riboside administration and intravenous delivery of Snhg12 rescued disease progression in vivo, suggesting potential to combat atherosclerosis by targeting a mechanism of vascular senescence.


Long noncoding RNAs (lncRNAs) are emerging regulators of biological processes in the vessel wall; however, their role in atherosclerosis remains poorly defined. We used RNA sequencing to profile lncRNAs derived specifically from the aortic intima of Ldlr−/− mice on a high-cholesterol diet during lesion progression and regression phases. We found that the evolutionarily conserved lncRNA small nucleolar host gene-12 (SNHG12) is highly expressed in the vascular endothelium and decreases during lesion progression. SNHG12 knockdown accelerated atherosclerotic lesion formation by 2.4-fold in Ldlr−/− mice by increased DNA damage and senescence in the vascular endothelium, independent of effects on lipid profile or vessel wall inflammation. Conversely, intravenous delivery of SNHG12 protected the tunica intima from DNA damage and atherosclerosis. LncRNA pulldown in combination with liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis showed that SNHG12 interacted with DNA-dependent protein kinase (DNA-PK), an important regulator of the DNA damage response. The absence of SNHG12 reduced the DNA-PK interaction with its binding partners Ku70 and Ku80, abrogating DNA damage repair. Moreover, the anti-DNA damage agent nicotinamide riboside (NR), a clinical-grade small-molecule activator of NAD+, fully rescued the increases in lesional DNA damage, senescence, and atherosclerosis mediated by SNHG12 knockdown. SNHG12 expression was also reduced in pig and human atherosclerotic specimens and correlated inversely with DNA damage and senescent markers. These findings reveal a role for this lncRNA in regulating DNA damage repair in the vessel wall and may have implications for chronic vascular disease states and aging.


Atherosclerosis, a chronic arterial disease of medium- to large-sized arteries, is the most frequent cause of death worldwide and is associated with traditional risk factors such as lipoprotein accumulation, immune cell function, and extracellular matrix metabolism (1, 2). Accumulating studies demonstrate that cells of advanced plaques are more prone to senescence, a permanent cellular growth arrest often triggered by DNA damage (35). Reactive oxygen species (ROS)–mediated oxidative stress and DNA damage can contribute to cellular senescence and dysfunction of endothelial cells (ECs) and macrophages (6, 7) and thus to chronic disease such as atherosclerosis, neurodegenerative disorders, and premature aging, among others (8, 9). Lesional DNA damage increases with the progression of atherosclerosis (10). The consequences of unrepaired or extensive DNA damage include growth arrest, cell senescence, and apoptosis, which all rise with plaque severity (11). Advanced plaques contain cells bearing senescence markers such as senescence-associated β-galactosidase (βgal) activity and elevated expression of p16, p21, and p53 (12). Recently, elimination of p16+ senescent cells in advanced lesions from Ldlr−/− mice led to inhibition of lesion growth, prevention of maladaptive plaque remodeling, and a reduction in the secretion of proinflammatory molecules (13). Despite these important findings, major mechanistic gaps remain in the understanding of the underlying molecular signaling events that contribute to the increased DNA damage and senescence in advanced atherosclerotic lesions.

Long noncoding RNAs (lncRNAs) have emerged as powerful regulators of numerous cellular processes through their ability to interact with RNA, DNA, or protein depending, in part, on their cellular localization (14). Although lncRNAs have a low cross-species conservation rate and may have lower copy numbers per cell than mRNAs (15, 16), several studies have identified lncRNAs enriched in a tissue- or cell-specific manner that can exert profound phenotypic effects (17). We detected lncRNAs specifically expressed in the intima of lesions during the progression or regression phases of atherosclerosis to provide a better understanding of their potential stage-specific roles and potentially uncover new insights for DNA damage in advanced lesions.


Identification of dynamically regulated lncRNAs in atherosclerosis

We sought to identify lncRNAs whose expression changes in the aortic intima during the progression or regression phases of atherogenesis. RNA was derived from the aortic intima of Ldlr−/− mice after 0, 2, or 12 weeks of consumption of a high-cholesterol diet (HCD) (progression phase; groups 1 to 3) and at 18 weeks after 6 weeks of resumption of a normal chow diet (regression phase; group 4) (Fig. 1A). RNA sequencing (RNA-seq) profiling captured differentially expressed lncRNAs [log2 fold change, 1.5; false discovery rate (FDR), <0.05] compared to early progressors in group 1 (Fig. 1B). Changes in lesional macrophages in the aortic sinus verified progression and regression of atherosclerosis, and cell type–specific markers established the purity of the endothelial enriched tunica intima RNA (fig. S1, A and B). Because of potential miscalculations by DESeq2 for two transcripts on opposite strands, we also applied an algorithm to exclude all antisense reads. This approach identified 37 lncRNA transcripts from DESeq2 and 19 lncRNAs without overlapping reads in the progression stage (group 3); 14 lncRNAs were commonly dysregulated according to both algorithms (Fig. 1C and fig. S1C). Among these 14 transcripts, we noted a highly conserved and most abundantly expressed lncRNA named small nucleolar host gene-12 (Snhg12), which is expressed from a syntenic location in the mouse, human, and pig genomes. We verified Snhg12 mouse and human sequences by 5′-RACE (rapid amplification of complementary DNA ends)–polymerase chain reaction (PCR) (Fig. 1D). RNA-seq results for Snhg12 were further verified by reverse transcription quantitative PCR (RT-qPCR) and by RNA in situ hybridization, which showed reduced expression of Snhg12 in the aortic intima after 12 weeks of HCD and near normalization during lesion regression (Fig. 1E and fig. S1D). Although SNHG12 contains intronic cryptically encoded small RNAs and overlaps by 9 base pairs (bp) with transfer RNA selenocysteine 1 associated protein 1 (TRNAU1AP), gapmeR-mediated silencing of SNHG12 did not affect the expression of either the small RNAs or TRNAU1AP (fig. S1, E to G). Snhg12 expression in the vascular endothelium exceeded that in the aortic media, peripheral blood mononuclear cells (PBMCs), and bone marrow–derived mononuclear cells (BMDMs) (fig. S1H). In addition, analysis from genotype-tissue expression (GTEx) revealed that cardiovascular or endothelial-enriched tissues expressed SNHG12 (fig. S1I). SNHG12 is a nuclear-expressed lncRNA in human and mouse ECs and primary macrophages (fig. S1, J to M), does not encode short peptides (fig. S1, N and O), and is polyadenylated (fig. S1P).

Fig. 1 Identification of the conserved lncRNA SNHG12 in lesional intima.

(A) RNA derived from aortic intima of Ldlr−/− mice (n = 3; each sample represents RNA pooled from two mice) placed on HCD for 0 weeks (group 1), 2 weeks (group 2), 12 weeks (group 3), or 18 weeks (group 4) after 6 weeks of resumption of a normal chow diet. (B) Workflow of genome-wide RNA-seq profiling for the identification of differentially expressed lncRNAs [log2 fold change (FC), >1.5; FDR, <0.05]. (C) Venn diagram displaying dysregulated lncRNAs detected by DESeq2 with or without overlapping reads. (D) 5′RACE-PCR for Snhg12 in mouse from RNA of the aortic intima and human RNA from HUVECs (n = 3). Visualization of RNA-seq in mouse and pig to verify sequence alignment and splicing junctions. (E) RNA-seq of Snhg12 in groups 1 to 4 was verified by RT-qPCR for the Snhg12-205 isoform. (F) Ldlr−/− mice were intravenously injected with vehicle control–gapmeR or SNHG12-gapmeR (7.5 mg/kg per mouse) twice per week and placed on HCD for 12 weeks (n = 12 per group) and assessed for (G) Snhg12 knockdown (KD) in aortic intima, media, and PBMCs (n = 6 per group). (H) Lesion areas were detected by Oil Red O (ORO) staining aortic sinus sections (n = 10 per group) and (I) thoracoabdominal aorta (n = 12 per group). (J) ApoE−/− mice were intravenously injected with LacZ or Snhg12 RNA twice per week and placed on an HCD for 6 weeks (n = 10 per group). (K) Delivery of RNA to the aortic intima, tunica media, and PBMCs was assessed by RT-qPCR (n = 5 per group). Lesion areas were quantified by Oil Red O staining of aortic sinus (L) and thoracoabdominal aorta (M) (n = 10 per group). All P values by Student’s t test. For all panels, values are means ± SD; *P < 0.05, **P < 0.01, ***P < 0.001.

Impact of loss or gain of function of lncRNA Snhg12 on atherosclerosis

To explore the role of systemically delivered SNHG12-gapmeRs in atherosclerosis, Ldlr−/− mice received vehicle control or SNHG12-gapmeR (7.5 mg/kg, twice weekly, intravenously) for 12 weeks on HCD (Fig. 1F). After 12 weeks on HCD, gapmeR-mediated silencing of Snhg12 reduced its expression in the aortic intima by 40 and 32% in PBMCs, but not in the aortic media (Fig. 1G and fig. S2, A to C). Analysis of atherosclerotic lesion formation by Oil Red O staining revealed a 2.4-fold increase in lesion area in the aortic sinus and 1.7-fold increase in the descending thoracoabdominal aorta compared to negative control (Fig. 1, H and I). To explore the effects of Snhg12 overexpression on atherosclerotic progression, we intravenously delivered a 5′-capped and 2-O-methylated Snhg12 RNA transcript to ApoE−/− mice placed on HCD for 6 weeks (n = 10 per group, 15 μg twice weekly). This intervention achieved a fourfold increase of Snhg12 expression in the aortic intima (Fig. 1, J and K, and fig. S2D) and reduced lesion areas in the aortic sinus and descending aorta by 34 and 40%, respectively (Fig. 1, L and M).

Neither loss nor gain of Snhg12 function changed the accumulation of lesional cells bearing markers of macrophages or vascular smooth muscle cells (VSMCs) or of CD4+ or CD8+ T cells (fig. S2, E and F). SNHG12-gapmeRs also reduced Snhg12 expression in PBMCs but did not alter monocyte polarization as gauged by Ly6C expression monitored by fluorescence-activated cell sorting (FACS) (fig. S2, G and H). Neither loss or gain of function of Snhg12 altered blood cholesterol, triglycerides, high-density lipoproteins (HDLs), or calculated low-density lipoproteins (c-LDLs) (fig. S2, I and J). Activation of the vascular endothelium depends, in large part, on proinflammatory processes mediated primarily through nuclear factor κB (NF-κB)–regulated signaling pathways (2). Snhg12 knockdown in Ldlr−/− mice did not affect nuclear NF-κB p65 detected by immunofluorescence in CD31+ ECs and Mac2+ macrophages of aortic arch sections (fig. S2, K and L). Nor did SNHG12-gapmeRs promote p65 nuclear translocation in human umbilical vein endothelial cells (HUVECs) treated with H2O2 (fig. S2M). These findings indicate that reducing Snhg12 expression in the aortic intima promotes atherosclerotic lesion formation in Ldlr−/− mice, whereas delivery of Snhg12 decreases plaque burden. Furthermore, these effects appear independent of circulating lipid profile or of lesional leukocyte accumulation.

LncRNA SNHG12 interacts with DNA-dependent protein kinase

To identify potential SNHG12-interacting proteins that may inform mechanisms underlying these findings, biotin-labeled, in vitro–transcribed SNHG12 or LacZ was incubated with HUVEC nuclear protein lysate (fig. S3A). We identified peptides that specifically bound to the biotin-labeled SNHG12 transcript by liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. This approach led to the identification of DNA-dependent protein kinase (DNA-PK) (Fig. 2Aand fig. S3B), an important sensor and mediator in the DNA damage response (DDR) and DNA repair process of nonhomologous end joining (NHEJ) (18, 19). DNA-PK was only detectable in the eluate of biotin-labeled SNHG12 compared to LacZ, the antisense transcript of SNHG12, or unlabeled SNHG12, whereas other major regulators of the DDR such as ataxia-telangiectasia mutated kinase (ATM), ataxia-telangiectasia and Rad3-related protein (ATR), and p53 were not found (Fig. 2B and fig. S3C). We validated this interaction in vivo by intravenous injection of biotin-labeled SNHG12 or LacZ in C57Bl/6 mice (fig. S3D). DNA-PK was recovered in the aortic protein lysate of SNHG12-injected mice (n = 4 per group) (Fig. 2C). Biophysical studies determined the equilibrium dissociation constant (KD) for the SNHG12–DNA-PK interaction as 1.318 × 10−7 μM (fig. S3E). To more precisely map the interaction of SNHG12–DNA-PK, we deleted predicted domains of the secondary structure of SNHG12 for subsequent lncRNA pulldown experiments (Fig. 2D). Deletion of domain 4 reduced DNA-PK binding by ~50%, suggesting that this domain participates in binding to DNA-PK (Fig. 2E). Conversely, SNHG12 expression rose sixfold in RNA isolated after DNA-PK immunoprecipitation compared to IgG control; this effect was specific to SNHG12 compared to other transcripts such as HPRT (Fig. 2F). This interaction did not involve transcriptional regulation, as SNHG12 silencing did not affect DNA-PK expression in HUVECs (fig. S3F) or phosphorylation of pDNA-PK, pATM, or pATR (fig. S3G).

Fig. 2 LncRNA SNHG12 interacts with DNA-PK.

(A) Proteins identified in DNA-PK and negative control LacZ pulldowns of biotinylated SNHG12 (n = 2 independent biological experiments, n = 2 technical replicates each). (B) Immunoblotting for DNA-PK, ATM, ATR, Ku80, Ku70, p53, and control on nuclear protein lysate from HUVECs after lncRNA pulldown (n = 5). (C) Western blot after streptavidin pulldown of nuclear protein lysate derived from the aorta of C57Bl/6 mice after two intravenous injections of biotin-labeled Snhg12 or LacZ (n = 4 per group). (D) Predicted secondary structure of SNHG12 with indicated deletions of domains 1 to 4. (E) LncRNA pulldown of biotinylated SNHG12 domain deletion constructs (n = 4). P value by one-way ANOVA with Fisher’s test. (F) Immunoprecipitation of DNA-PK after RNA isolation and subsequent RT-qPCR for SNHG12 and for negative control HPRT (n = 5). (G) HUVEC nuclear protein lysate was harvested 36 hours after transfection with control-gapmeR or SNHG12-gapmeR (25 nM) in the presence or absence of H2O2 for 1 hour (1 mM), immunoprecipitated with IgG or DNA-PK antibody, and immunoblotted for Ku70, Ku80, and DNA-PK. (H) Quantification of (G) under H2O2 conditions (n = 3). For all panels, values are means ± SD; **P < 0.01. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; n.d., not detectable; WT, wild type; IgG, immunoglobulin G; USF-2, upstream stimulatory factor 2.

To identify the functional consequences of the interaction of SNHG12 with DNA-PK, we assessed the effect of SNHG12 on DNA-PK’s kinase activity (DNA-PKcs). We incubated purified DNA-PK protein with in vitro–transcribed SNHG12 to assess the ability of DNA-PKcs to convert adenosine triphosphate (ATP) to adenosine diphosphate (ADP). In the presence of SNHG12, DNA-PKcs increased by twofold compared to LacZ control and positive control NU7441 (fig. S3, H and I) (20). RNAseA treatment reduced DNA-PKcs activity, suggesting that SNHG12 facilitates DNA-PKcs. DNA-PKcs recruitment to and activation by DNA requires the Ku complex, a heterodimer comprising two subunits of 70 and 80 kDa that binds to DNA double-strand breaks (DSBs) (18). We assessed the ability of DNA-PK to bind Ku70/80 under ROS-induced DNA damage conditions by performing DNA-PK immunoprecipitation and subsequent immunoblotting for Ku70 and Ku80. In HUVECs transfected with SNHG12-gapmeRs, the expression of Ku70 and Ku80 fell significantly (P = 0.0016 and 0.0218, respectively) under H2O2-induced DDR compared to control-gapmeRs after DNA-PK immunoprecipitation (Fig. 2, G and H). Together, these findings indicate that DNA-PK can actively bind lncRNA SNHG12, in turn facilitating the ability of DNA-PKcs to bind Ku70/80, an important mediator of the DDR (fig. S3J).

Knockdown of SNHG12 leads to increased cellular DNA damage

The DDR involves DNA strand break recognition followed by the initiation of a cascade that promotes DNA repair. SNHG12 expression fell dose dependently in response to intrinsic (γ-irradiation) and extrinsic (H2O2) triggers of DSBs (Fig. 3A). Other stress triggers such as turbulent shear stress of the lesser aortic curvature (fig. S4A) and aging (fig. S4B) significantly reduced SNHG12 expression (P = 0.0392 and 0.0420, respectively). We assessed the consequences of SNHG12 on the DDR in ECs treated with γ-irradiation or H2O2. ECs transfected with SNHG12-gapmeRs showed increased nuclear γH2AX foci across all time points, with the most pronounced effect (a 2.5-fold increase) occurring after 12 hours (Fig. 3B). γH2AX protein increased in ECs by 2.2-fold and 1.4-fold under basal and H2O2-induced DNA damage conditions, respectively (Fig. 3C). We recapitulated the effect of Snhg12 on the DDR using two additional antisense oligonucleotides (ASOs) in mouse ECs (fig. S4, C and D) and human arterial ECs (fig. S4, E and F). Conversely, lentiviral overexpression of SNHG12 reduced γH2AX foci formation by 50% as early as 1 hour after γ-irradiation (fig. S4, G and H) and reduced H2O2-induced γH2AX phosphorylation by twofold (Fig. 3D). Although ECs expressed more SNHG12 than leukocytes, we assessed SNHG12-mediated effects on the DDR in human primary macrophages and mouse RAW264.7 cells using camptothecin, a topoisomerase inhibitor that induces DSBs (21). A DNA Comet assay to quantify individual cell DNA damage showed that SNHG12 knockdown in RAW264.7 and human primary macrophages (fig. S4I) prolonged DNA tail length in the presence of camptothecin by 45% (fig. S4J) and increased γH2AX phosphorylation by twofold (fig. S4K).

Fig. 3 DNA damage analysis in vitro and in lesions upon SNHG12 loss- and gain-of-function studies.

(A) SNHG12 expression was analyzed in HUVECs γ-irradiated for 1 min (1.2 Gy−min) and RNA isolated at 0, 1, 2, 4, 8, or 12 hours after irradiation or in the absence or presence of H2O2 (30, 100, 250, 500, or 1000 μM) for 4 hours. (B) Control-gapmeR– or SNHG12-gapmeR–transfected HUVECs were γ-irradiated (1.2 Gy−min) and fixed at 0, 1, 2, 4, 8, or 12 hours after irradiation. Quantification of γH2AX foci with representative images. Scale bar, 20 μm. IR, irradiation. (C) Western blot analysis of protein lysate for γH2AX from HUVECs transfected with gapmeRs or negative control treated for 0 and 30 min with H2O2 (1 mM) (n = 3). (D) HUVECs transduced with control lentivirus or SNHG12 lentivirus were analyzed for γH2AX in the absence or presence of H2O2 (1 mM) for 1 hour. (E) Lesional DNA damage in the vessel wall of the aortic arch in Ldlr−/− mice on HCD injected with control-gapmeR or SNHG12-gapmeR for 12 weeks was quantified by γH2AX with nuclear colocalization of DAPI in lesional CD31+ cells. Representative images are shown. Scale bar, 50 μm (n = 6 mice per group; two to three lesions per arch). (F) Lesional DNA damage was quantified in the aortic arch of ApoE−/− mice after delivery of lacZ or Snhg12 RNA. Representative images are shown. Scale bar, 200 μm (n = 10 mice per group). (G and H) HUVECs were cotransfected with either control-gapmeR, SNHG12-gapmeR, control-siRNA, or siRNA–DNA-PK (25 nM each) and treated with H2O2 (1 mM) for 1 hour before DNA double-strand breaks (DSBs) were assessed by (G) γH2AX Western blot (n = 3) and (H) Comet assay (neutral pH) (n = 3). (I) NHEJ efficiency in HUVECs assessed by FACS for GFP-positive cells as an indicator or repaired DSB in HUVECs overexpressing SNHG12 (cherry reporter) compared to lentiviral control (n = 3). All P values by Student’s t test except for one-way ANOVA with Fisher’s test in (A). For all panels, values are means ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Consistent with these in vitro findings, lesions of Ldlr−/− mice treated with SNHG12-gapmeRs exhibited a twofold increase in γH2AX foci in the vascular endothelium (CD31+ cells) (Fig. 3E). Furthermore, γH2AX rose significantly (P = 0.0157) by 1.4-fold in lesional macrophages (Mac2+ cells) (fig. S4L). Moreover, delivery of Snhg12 to the vessel wall reduced γH2AX foci in ApoE−/− mice after 6 weeks on HCD (Fig. 3F).

To test whether the SNHG12-mediated effect on γH2AX foci formation depended on DNA-PK, we silenced DNA-PK with simultaneous SNHG12 knockdown. DNA-PK silencing blocked the SNHG12-gapmeR–mediated increase of γH2AX foci and tail moment after ROS- or camptothecin-induced DNA damage (Fig. 3, G and H). Further, because DNA-PK is implicated in regulating NHEJ (3) and SNHG12 is dependent on DNA-PK, we stably transduced HUVECs with a DNA repair reporter (pDRR). I-SceI cuts the integrated pDRR to generate DSBs; repair through NHEJ leads to expression of green fluorescent protein (GFP) (22). Lentiviral overexpression of SNHG12 promoted NHEJ as indicated by a 2.5-fold increase in GFP-positive cells compared to control (Fig. 3I and fig. S4M). These findings support the role of SNHG12 in the DDR in a DNA-PK–dependent manner and show that this lncRNA can prevent DNA damage accumulation in atherosclerotic lesions.

Phenotypic consequences of accumulating DNA damage

We used genome-wide RNA-seq on SNHG12-gapmeR–transfected ECs exposed to H2O2 to explore the pathways and processes affected by SNHG12 knockdown. We identified a number of up- or down-regulated genes (log2 fold change, >0.5; FDR, <0.05) (Fig. 4A). Gene set enrichment analysis (GSEA) revealed the p53 pathway and ultraviolet (UV) response among the top 10 processes affected (Fig. 4B). p53 was also predicted to be activated (normalized enrichment score = 1.424, FDR q = 0.0089) (Fig. 4C). We confirmed this in vitro, showing that knockdown of SNHG12 significantly (P = 0.0089) induced phosphorylation of p53 in response to ROS without altering total p53 (Fig. 4D). Furthermore, p53 binding affinity to DNA increased upon SNHG12 knockdown (Fig. 4E). Increased p53 activity to maintain genomic integrity is a well-established hallmark in the DDR (23), and the p53-p21 axis importantly regulates stress-induced senescence (24). Because inhibition of DNA-PK accelerates senescence (25), we evaluated the role of SNHG12 in this process. To this end, EC senescence was triggered with a low dose of H2O2 (30 μM) for 1 hour, followed by another 3 days of culturing with normal growth medium. SNHG12 knockdown increased the expression of the senescence markers p16 (by 1.5-fold) and p21 (by 1.5-fold) (Fig. 4F). In lesions with reduced Snhg12 expression after delivery of gapmeRs in vivo, p16, p21, and p27 all rose by up to twofold in the aortic intima of Ldlr−/− mice (Fig. 4G). Furthermore, those lesions harbored more than twofold higher acellular areas than controls (Fig. 4H). Conversely, Snhg12 overexpression in the vessel wall reduced plaque necrosis (fig. S5A). SNHG12 knockdown increased senescence-associated βgal-positive cells in HUVECs (fig. S5, B and C) and human arterial ECs (fig. S5, D and E), whereas overexpression of SNHG12 reduced senescence-associated βgal (Fig 4I). As a consequence of increased DNA damage, ECs may exhibit impaired homeostatic control of functions important to lipoprotein entry and macrophages may display impaired clearance of cellular debris or apoptotic cells. To assess lipoprotein entry, we performed transcytosis assays using confluent monolayers of human coronary artery ECs and total internal reflection fluorescence (TIRF) microscopy (26). SNHG12 knockdown significantly (P < 0.0001) increased red fluorescent labeled 3,3′-dioctadecylindocarbocyanine-LDL (DiI-LDL) transport from the apical to the basolateral membrane by 1.8-fold (Fig. 4J and movies S1 and S2). Conversely, lentiviral overexpression of SNHG12 rescued H2O2-induced EC permeability to LDL (fig. S5F and movies S3 to S6). To address in vivo permeability in arteries, a cone-shaped polyethylene constrictive cuff was placed on the left common carotid artery, as previously described (27), to mimic disturbed flow in lesion prone areas. Flow perturbation was measured 24 hours after surgery by Evans blue extravasation. The downstream Evans blue area of the constrictive cuff was more elongated in longitudinal cross sections of the left common carotid artery after inhibiting Snhg12 expression compared to negative control (fig. S5, G and H). Conversely, administration of Snhg12 RNA by intravenous injection reduced downstream Evans blue extravasation (fig. S5I). Efficiency of Snhg12 knockdown and delivery was assessed in the aortic intima by RT-qPCR analysis (fig. S5J).

Fig. 4 Downstream consequences of SNHG12 on p53 and senescence.

(A) Genome-wide RNA-seq profiling of HUVECs transfected with control-gapmeR or SNHG12-gapmeR treated for 1 hour with H2O2 (1 mM) (n = 3 per group). Volcano plot displaying significantly dysregulated genes (log2 fold change, >1.5; FDR, <0.05). (B) GSEA of the top 10 significantly affected processes and (C) enrichment plot for the p53 pathway of cells in (A) after SNHG12 knockdown. “Tag%” gives an indication of the fraction of genes contributing to the enrichment score. TNF, tumor necrosis factor. (D) p-p53 and total p53 by Western blot in gapmeR-transfected HUVECs with and without H2O2. P values by Student’s t test. (E) Electrophoretic mobility shift assay for p53 using nuclear lysate of control- or SNHG12-gapmeR–transfected HUVECs. (F) p16 and p21 immunoblot in cells treated for 1 hour with H2O2 (30 μM), followed by 3 days of incubation in normal growth medium. (G) RT-qPCR of senescence markers in mouse aortic endothelium–derived RNA (n = 6 per group). (H) Plaque necrosis in the aortic sinus (n = 6 per group; two to three lesions per mouse). Scale bar, 50 μm. (I) βgal staining of lentiviral SNHG12-transduced HUVECs. Representative images are shown. Scale bar, 300 μm (n = 3). (J) Transcytosis of DiI-labeled LDL quantified by TIRF microscopy (movies S1 and S2) (n = 3). (K) In vivo efferocytosis measured by Mac2-associated TUNEL staining in lesions of SNHG12-gapmeR–injected Ldlr−/− mice. Representative images are shown. Scale bar, 100 μm. P value by Student’s t test. For all panels, values are means ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We tested human primary macrophages for their ability to ingest apoptotic cells, a process described in atherosclerotic plaques as efferocytosis (28, 29). SNHG12-gapmeR–transfected macrophages showed a 50% reduced uptake of apoptotic cells (fig. S5K), whereas lentiviral overexpression of SNHG12 increased their ability to phagocytose apoptotic cells by twofold (fig. S5, L and M). Consistently, lesions of SNHG12-gapmeR–injected Ldlr−/− mice had significantly more Mac2-free terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL)–positive cells (P = 0.0021), an indication of impaired efferocytosis in vivo (Fig. 4K) (29). If DNA damage is unrepaired or extensive, cells may undergo senescence or apoptosis (11). Therefore, we analyzed cleaved caspase-3 and TUNEL, markers of apoptosis, in lesions after Snhg12 knockdown in Ldlr−/− mice but found no differences in expression (fig. S6, A to E). Proliferation, however, was strongly affected upon SNHG12 loss- and gain-of-function experiments in HUVECs (fig. S6, F and G). These data indicate that SNHG12 deficiency triggers the DDR and DDR-induced senescence, which exacerbates EC permeability, LDL transcytosis, and macrophage efferocytosis.

In vivo rescue of DNA damage by nicotinamide riboside attenuates Snhg12-deficient progression of atherosclerosis in Ldlr−/− mice

Recent work has demonstrated that ROS-induced DNA damage in tissues relates to impaired NAD+ (nicotinamide adenine dinucleotide) metabolism (30). Furthermore, administration of the NAD+ precursor nicotinamide riboside (NR) limited DNA damage and prolonged the life span in mice by ameliorating the DDR and senescence (30). To assess whether Snhg12-deficient lesion progression involves increased DNA damage, SNHG12-gapmeR was delivered intravenously as described in Fig. 1F and an HCD was supplemented with NR (400 mg/kg per day) (Fig. 5A). Analysis of atherosclerotic lesion formation after 12 weeks on HCD and NR showed no significant increase in lesion areas between SNHG12-gapmeR and control-gapmeR groups (P = 0.1729; Fig. 5B). Moreover, NR administration abrogated the SNHG12-gapmeR–mediated effect on increased γH2AX foci in lesional ECs in the aortic arch (Fig. 5C). Consistent with less DNA damage, the induction of senescence markers p16, p21, and p27 decreased in the presence of NR compared to Ldlr−/− mice treated with SNHG12-gapmeRs without NR (Fig. 5, D and E). In addition, NR treatment eliminated differences in plaque necrosis (Fig. 5F) but reduced TUNEL-positive cells compared to groups without NR (Fig. 5G). Together, the rescue effects mediated by NR support the findings that the lncRNA Snhg12 regulates the DDR both in vitro and in vivo.

Fig. 5 NR rescued progression of atherosclerotic lesions in Ldlr −/− mice induced by Snhg12 silencing.

(A) Ldlr−/− mice were intravenously injected with vehicle control or SNHG12-gapmeR (7.5 mg/kg per mouse) twice per week and placed on an HCD containing NR (400 mg/kg per day) for 12 weeks (n = 12 per group). (B) Lesion areas were quantified using Oil Red O area on mouse aortic sinus sections (n = 10 per group). Representative images are shown. Scale bars, 200 μm. Fold change was calculated to no NR–treated, control-gapmeR–injected mice. (C) Lesional DNA damage in the vascular endothelium of the aortic arch was quantified in Ldlr−/− mice by γH2AX alongside nuclear colocalization with DAPI in CD31+ cells. Representative images are shown. Scale bar, 100 μm. RT-qPCR of (D) Snhg12 and (E) senescence markers p16, p21, and p27 from RNA isolated from the aortic intima of mice on HCD plus NR (n = 6 per group). (F) Acellular areas in the aortic sinus for each of the indicated groups (n = 6 per group; two to three lesions per mice). Representative images shown are shown. (G) TUNEL staining was quantified in the aortic sinus for each of the indicated groups (n = 7 per group). All P values by Student’s t test. For all panels, values are means ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Mitochondrial stress as a consequence of accumulating DNA damage

Mitochondrial dysfunction is a hallmark of senescence, and reparative responses during oxidative insults require effective energy metabolism. Oxidative stress–induced DNA damage activates NAD+ consumption pathways (31). We used Seahorse analyses to examine the effect of SNHG12 knockdown on key mitochondrial activities such as mitochondrial respiration and glycolysis in ECs. Loss of SNHG12 decreased the extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) of live cells (fig. S7, A and B). These alterations yielded a reduction in total ATP production, likely due to cumulative DNA damage. NR itself did not affect SNHG12 expression (fig. S7C). However, knockdown of SNHG12 decreased the NAD+/NADH ratio, and NR partially rescued this reduction (fig. S7D). Collectively, these findings support the premise that deficiency of SNHG12 increased DNA damage, leading to higher NAD+ consumption, increased mitochondrial stress, and less ATP production.

SNHG12 is a highly conserved lncRNA that inversely correlates with DNA damage and senescence markers in human, mouse, and pig atherosclerotic specimens

SNHG12 is conserved across mouse, human, and pig (Fig. 1D). To assess the translational relevance of SNHG12, we isolated RNA and protein from human nondiseased control carotid arteries and atherosclerotic carotid arteries. Human atherosclerotic arteries (n = 23) displayed significantly reduced expression of SNHG12 (by 58%, P = 0.002) compared to control arteries (n = 8) (Fig. 6A). Emerging studies demonstrate that DNA damage increases with progression of atherosclerosis (10). Furthermore, in vivo elimination of p16+ cells in lesions markedly reduces progression of atherosclerosis, suggesting the importance of senescence for the development of atherosclerosis (13, 32). Consistent with these studies, we found that the expression of γH2AX, p16, and p21 increased markedly (10-, 23-, and 6.5-fold, respectively) in human atherosclerotic carotid arteries compared to control arteries (Fig. 6B). From an independent study, we analyzed specimen samples and RNA-seq data from Yorkshire pigs that were placed for up to 60 weeks on an HCD and developed atherosclerosis. To this end, we separated carotid cross sections into mild, moderate, and severe atherosclerosis groups for progression of atherosclerosis based on histopathological markers (Oil Red O staining, intima/media ratio, CD45, plaque–internal elastic lamina, elastin) (Fig. 6C). RT-qPCR expression analyses of these groups revealed that SNHG12 also decreased ~50% with progression of disease as shown for mice and humans (Fig. 6D). Conversely, the gene expression of senescence markers p21 and p16 increased in intermediate and severe atherosclerosis compared to mild lesions (Fig. 6E). In summary, these results demonstrate evolutionary conservation of lncRNA SNHG12, its reduced expression during progression of atherosclerosis, and a consistent and inverse correlation with DNA damage and senescence across mouse, human, and pig atherosclerotic lesions (fig. S8).

Fig. 6 SNHG12 expression inversely correlates with DNA damage and senescence markers in human and pig atherosclerotic specimens.

(A) SNHG12 expression in RNA isolated from nondiseased control human carotid arteries (n = 8) or atherosclerotic carotid arteries (n = 23). P value by Student’s t test. (B) Total γH2AX, p16, and p21 protein assessed from the same samples in (A), normalized by GAPDH. P value by Student’s t test. Plots in (A) and (B) indicate fold change relative to control arteries. (C) Oil Red O staining of fresh-frozen carotid artery cross sections from Yorkshire pigs fed an HCD for up to 60 weeks. (D) RT-qPCR analysis of SNHG12 expression in RNA from specimens in (C), normalized by GAPDH. P value by one-way ANOVA with Fisher’s test. (E) RNA-seq transcriptomic analysis of p16 and p21 expression. P value by one-way ANOVA with Fisher’s test. For all panels, values are means ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.


This study provides evidence for the dynamic regulation of lncRNA SNHG12 during stress-induced DNA damage. The reduction in expression of this lncRNA in atherosclerotic arteries of mice, pigs, and humans highlights its evolutionary conserved expression in chronic vascular disease states. Loss- or gain-of-function studies of SNHG12 in atherosclerosis regulated the DDR and vascular senescence without altered lipoprotein profile, lesional accumulation of leukocyte subsets, or intimal NF-κB activation, indicating that modulation of pathways beyond lipid risk factors or inflammation contributes substantially to atherosclerotic lesion progression. Mechanistically, SNHG12 knockdown increased markers of DSBs such as γH2AX, in part, by inhibiting the DNA-PK interaction with Ku70 and Ku80, heterodimeric proteins that bind DSBs and that facilitate the NHEJ pathway. These findings identify SNHG12 as a regulator of DNA-PK and the DDR in vitro and in vivo. Although SNHG12 can increase DNA-PKcs activity, SNHG12 is probably not essential for the V(D)J recombination function of DNA-PK, as other studies have shown that minimal DNA-PKcs protein is suffice to mediate V(D)J recombination, but not the DDR evoked by ionizing radiation (33).

Sustained DNA damage may lead to cellular senescence and aggravate the pathogenesis of chronic disease states such as atherosclerosis. Although any definitive role for DNA-PK in atherosclerotic lesion progression is poorly defined, DNA-PK activity increases with progression of atherosclerosis, potentially as a means to repair DNA damage observed in the vessel wall (34). Furthermore, markers of DNA DSBs, oxidative DNA damage, and DNA reparative enzymes increase with advanced atherosclerotic lesions across mice, rabbits, and humans, highlighting the evolutionary conservation of this pathway, an indication of its potential importance (3437). In addition to atherosclerosis, the maintenance of genomic integrity resists malignant transformation of a cell provoked by genotoxic stress and carcinogenic insults such as irradiation (38). To date, SNHG12 has been described in several forms of cancer as, for example, in prostate (39), gastric (40), and breast cancer (41). Up-regulation of SNHG12 in some cancer cell types increased cellular proliferation and resistance to cell death insults in vitro (42). However, a definitive in vivo role of Snhg12 in murine tumors is lacking. The findings from this study may have translational value not only for atherosclerosis but also for cancer, as this study demonstrates the regulatory role of the lncRNA SNHG12 in maintenance of cellular genomic stability via its interaction with DNA-PK.

Although Snhg12 knockdown elevated markers for DSBs and senescence primarily in the vascular endothelium and macrophages of lesions, we cannot rule out the possibility of similar effects in other cell types such as VSMCs. This possibility arises as delivery of ASO Snhg12 by intravenous tail vein injection may not penetrate the VSMC-enriched aortic media sufficiently to reduce Snhg12 expression. The recent recognition that VSMC-derived DNA damage has minimal effects on atherogenesis, but alters fibrous cap areas in advanced lesions, also suggests a modest contribution of DNA damage derived from this cell type (43).

As a consequence of increased DNA damage, cells may exhibit impaired homeostatic control of functions important to lipoprotein entry or clearance of cellular debris or apoptotic cells (12). Consistent with this notion, knockdown of SNHG12 in ECs increased permeability by LDL transcytosis and impaired macrophage efferocytosis, or the ability to engulf apoptotic cells, a process implicated in the clearance of cells from the core of plaques. Oxidative stress induces genomic instability that can lead to DNA damage checkpoint arrest and, if not appropriately repaired, to senescence or apoptosis (11). Most lesional DNA defects occur through the generation ROS via the NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase (44, 45). NR is a clinical-grade small-molecule activator of NAD+ that serves as a precursor for NADPH, which, in turn, activates a number of canonical pathways that reduce oxidative stress and hence DNA damage (46). Here, we demonstrate that NR completely rescued the effects of increased DNA damage, vascular senescence, and atherosclerosis in SNHG12-gapmeR–injected Ldlr−/− mice. Because clinical trials are investigating the use of NR formulations for diverse chronic conditions including peripheral artery disease (NCT03743636), heart failure (NCT03423342), and cognitive function (NCT03562468), the findings in this study provide new mechanistic insights into the role of NR in vascular senescence applicable to these conditions. Furthermore, the findings that NR can rescue accelerated atherosclerosis may inform new strategies to ameliorate a range of vascular occlusive disease states.

Our study builds upon the emerging roles of lncRNAs in the progression of atherosclerosis. For example, the lncRNAs LeXis and MeXis control lipid metabolism via liver X receptor (LXR) pathways (47, 48). Deficiency of lncRNA Malat1 accelerates atherosclerosis and triggers robust immune system dysregulation mediated by bone marrow–derived cells (49). Collectively, these studies highlight that lncRNAs exert profound regulation of key signaling pathways relevant to homeostasis in the vessel wall.

Limitations of this study include the nonselective delivery approach to target or deliver Snhg12 RNA and the lack of a cell-specific genetically altered Snhg12 mouse model. We not only identified Snhg12 expression as enriched in the endothelial intima but also showed that Snhg12 is also expressed, to a lesser extent, in other cell types such as macrophages. Nonetheless, gain- and loss-of-function studies in vitro and in vivo both revealed concordant findings in macrophages, highlighting that the mechanism by which Snhg12 regulates the DDR and cellular senescence is not limited to one cell type. Last, because lncRNAs may interact with DNA, RNA, and protein, we cannot exclude the possibility that SNHG12 may interact with other substrates in defined cellular systems.

In summary, we have identified lncRNA SNHG12 in atherosclerotic lesions as a homeostatic regulator of genomic stability by interaction with DNA-PK, a key mediator of the DDR. Knockdown of the lncRNA SNHG12 impairs DNA damage repair, leading to lesional DNA damage, vascular senescence, and accelerated atherosclerosis independent of effects on lipid-lowering or lesional inflammation. Intravenous administration of the lncRNA SNHG12 reduced lesional DNA damage and plaque burden. Strategies aimed at restoring SNHG12 expression or facilitating SNHG12–DNA-PK interactions may provide a translational approach to limit DNA damage and vascular senescence.


Study design

The main goal of this study was to identify dynamically regulated lncRNAs during the progression of atherosclerosis. Among them, we identified the lncRNA Snhg12 as a highly abundant and evolutionary conserved lncRNA from mouse to pig to human. Loss-of-function studies were performed with modified ASOs (gampeRs) to assess the role of SNHG12 in the progression of atherosclerosis. In vivo experiments for loss-of-function studies were performed in atherosclerotic-prone Ldlr−/− mice on an HCD by tail vein administration of gapmeRs twice per week over the course of 12 weeks in the presence or absence of NR. For gain-of-function experiments, Snhg12 RNA was delivered over the course of 6 weeks via the same route in ApoE−/− mice on an HCD starting at 12 weeks of age. Experimental groups included at least 10 mice per group to robustly identify alterations in disease progression, and mice were randomly assigned to each group. All measurements were blinded. GampeR-mediated knockdown and lentiviral overexpression in human ECs and human primary macrophages were performed to verify the evolutionary conserved function of SNHG12. In vitro experiments included a minimum of n = 3 independently performed replicates. Expression data from human and pig carotid cross sections were used to assess the atherosclerosis-specific expression profile of SNHG12. No data points were excluded as outliers.

Animal studies

All protocols concerning animal use were approved by the Institutional Animal Care and Use Committee at Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA and conducted in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. Studies were performed in Ldlr−/− mice (The Jackson Laboratory), ApoE−/− mice (12 weeks old; The Jackson Laboratory), or C57Bl/6 mice (Charles River).


An efferocytosis assay was performed as described in (50). Briefly, Jurkat cells (5 × 106 cells/ml) were labeled with 5 μM calcein-AM (Invitrogen). After 2-hour incubation, cells were washed and irradiated with UV (150 mJ/cm2) with an open lid, followed by another 2 hours of incubation before apoptotic cells were added in a 1:1 ratio to gapmeR-transfected primary macrophages. After several rounds of gentle washing, macrophages were counted positive for internalized green apoptotic bodies if they contained >3-μm clusters of green dots. Quantification was performed from four images with a total of 400 macrophages. For quantification of in vivo efferocytosis, macrophages were stained using 1:100 rat anti-Mac2 (Cedarlane) as described below for immunofluorescence. TUNEL was performed on the basis of the manufacturer’s protocol (In Situ Cell Death Detection Kit, TMR red, Roche). The ratio of macrophage-free TUNEL over macrophage-associated TUNEL signaling was calculated as described (29).

En face RNA isolation

RNA from the lesser and greater curvature (LC and GC) was isolated from C57Bl/6 mice using nitrocellulose slides (ONCYTE NOVA). To this end, aortas were isolated and cut for areas of LC or GC and placed with the tunica intima side toward the nitrocellulose side for 15 min at room temperature (n = 3 mice were pooled to represent n = 1) (51, 52). After removing the aorta from the slide, RNA lysis buffer was directly added to the slide, followed by RNA isolation (RNeasy Plus Micro Kit, QIAGEN).

Evans blue extravasation

In vivo permeability in arteries with Evans blue extravasation was performed as described (27). Briefly, a cone-shaped polyethylene was placed in the left common carotid artery and secured by a circumferential suture. Evans blue extravasation was determined following 24 hours after surgery. C57Bl/6 mice were injected intravenously either with SNHG12-gapmeRs or control-gapmeRs (7.5 mg/kg) or with in vitro–transcribed SNHG12 RNA (15 μg per injection) [as described below in microscale thermophoresis (MST) assay] on two constitutive days followed by cuff surgery on day 3. Downstream Evans blue area of the constrictive cuff elongation in longitudinal cross sections of the left common carotid artery was compared to corresponding controls.

Human atherosclerotic specimens

Frozen sections were prepared from human normal carotid arteries and carotid atherosclerotic lesions that were obtained from the Division of Cardiovascular Medicine, Brigham and Women’s Hospital in accordance with the Institutional Review Board–approved protocol for use of discarded human tissues (protocol #2010-P-001930/2).

Immunohistology and characterization of atherosclerotic lesions

To quantify atherosclerosis in Ldlr−/− mice on HCD (Research Diets Inc.), aortic roots and aortic arch were embedded in optimal cutting temperature (OCT) compound and frozen at −80°C. Serial cryostat sections (6 μm) were prepared using a tissue processor (Leica CM3050). Lesion characterizations, including Oil Red O staining of the thoracic–abdominal aorta and aortic root and staining for macrophages (1:900; anti-Mac3, BD Pharmingen), T cells (1:90; anti-CD4, BD Pharmingen; 1:100; anti-CD8, Chemicon, CBL1318), and VSMCs (1:500; SM-α-actin, Sigma), were performed as previously described (53, 54). The staining area was measured using Image-Pro Plus software (Media Cybernetics), and CD4+ and CD8+ cells were counted manually.

Intimal RNA isolation from aorta tissue

Isolation of intimal RNA from aorta was performed as previously described in (54, 55). Briefly, aortas were carefully flushed with phosphate-buffered saline (PBS), followed by intima peeling using TRIzol reagent (Invitrogen). TRIzol was flushed for 10 s, followed by a 10-s pause, flushed another 10 s, collected in an Eppendorf tube (~300 to 400 μl in total), and snap-frozen in liquid nitrogen.

LDL transcytosis assay

LDL transcytosis assay was performed as previously described (56). Briefly, TIRF microscopy uses an evanescent wave to illuminate just the proximal ~100 nm of the cell, thereby facilitating selective imaging of the basal membrane of a live EC with minimal confounding from the overlying cytoplasm and apical surface. Confluent human coronary artery endothelial cells (HCAEC) monolayers are exposed to a fluorophore-tagged ligand added to the apical cell surface, whereas the basal membrane of the cell is imaged by TIRF. Cytoplasmic vesicles undergoing exocytosis with the basal membrane were directly visualized and quantified. TIRF microscopy was performed on a Leica DMi8 microscope [with 63×/1.47 (O) objectives; 405-, 488-, 561-, and 637-nm laser lines; 450/50, 525/50, 600/50, 610/75, and 700/75 emission filters] and run with Quorum acquisition software. Microscope settings were kept constant between conditions. Briefly, cells at 100% confluency were placed in a live cell imaging chamber and treated with DiI-LDL (20 μg/ml) in cold HEPES-buffered RPMI (HPMI) media for 10 min at 4°C to allow apical membrane binding. After membrane binding, cells were washed twice with cold PBS+ to remove unbound ligand and room temperature HPMI was added. Cells were incubated on the live cell imaging stage at 37°C for 2 min before initial image acquisition. Confluent regions of the monolayer were selected by viewing the number of nuclei in the 4′,6-diamidino-2-phenylindole (DAPI) field of view after staining with NucBlue Live ReadyProbes Reagent (Thermo Fisher Scientific), and TIRF microscopy of the basal membrane was performed to visualize exocytosis. For each coverslip, 10 to 15 videos of 150 frames (100-ms exposure) were captured. Image analysis was performed using a custom MATLAB single particle–tracking algorithm (56).

Liquid chromatography–mass spectrometry

LC-MS/MS was performed as previously described (57). Briefly, lncRNA pulldown of SNHG12 or LacZ purified samples was reduced with 10 mM dithiothreitol for 30 min at 56°C in the presence of 0.1% RapiGest SF (Waters). Cysteines were alkylated with 22.5 mM iodoacetamide for 20 min at room temperature in the dark. Samples were digested overnight at 37°C with trypsin. RapiGest was then cleaved according to the manufacturer’s instructions, and peptides were purified by reversed-phase and strong cation exchange chromatography. Peptides were loaded onto a precolumn (4-cm POROS 10R2, Applied Biosystems), resolved on a self-packed analytical column (12-cm Monitor C18, Column Engineering) after gradient elution (NanoAcquity UPLC system, Waters; 5 to 35% B in 90 min; A = 0.2 M acetic acid in water, B = 0.2 M acetic acid in acetonitrile), and introduced to the MS (TripleTOF 5600, AB Sciex) by electrospray ionization (spray voltage = 2.2 kV). The mass spectrometer was programmed to perform data-dependent MS/MS [unit resolution: mass/charge ratio (m/z), 100 to 2000] on the 20 most abundant precursors in each MS1 scan (m/z, 300 to 2000; accumulation time, 0.5 s; threshold, 70 counts; charge state, 2+ to 5+) using a rolling collision energy. After MS/MS, each precursor was excluded for 25 s. Raw data were converted to .mgf using AB Sciex MS Data Converter; precursor and product ions were recalibrated using a linear equation derived from fitting experimentally observed masses obtained in an initial low mass tolerance database search. Recalibrated data were matched to peptide sequences in a forward/reversed human NCBI Refseq release 83 database using Mascot version 2.4.1. Search parameters included trypsin specificity with up to two missed cleavages, fixed carbamidomethylation (C, +57 Da), and variable oxidation (M, +16 Da). Precursor and product mass tolerances were 12 ppm and 25 mmu (millimass units), respectively. Protein hits with FDR < 0.1 from SNHG12-specific pulldown were compared to negative control (LacZ) (n = 2 independent biological experiments, N = 2 technical replicates each).

NHEJ repair by FACS

NHEJ ability was assessed as previously described (22). Lentivirus was produced in 293T cells as described above under “Lentivirus production and transduction” (in the Supplementary Materials) for pDRR (double-strand break repair reporter) (pLCN DSB Repair Reporter, Addgene) and pCBASceI (Addgene). HUVECs were transduced in a 10-cm dish with pDRR without geneticin selection, because >85% cells were GFP positive after 4 to 5 days. Cells were transfected with GapmeRs and small interfering RNA (siRNA) or transduced with lentivirus for SNHG12, and 24 to 48 hours after, cells were transduced with lentivirus for pCBASceI. FACS analysis for GFP-positive cells was performed 48 to 72 hours after pCBASceI transduction.

Pig atherosclerotic samples

Fifteen male hypercholesterolemic Yorkshire swine were placed on an HCD for up to 60 weeks. Detailed sectioning of 3-mm coronary artery segments was performed so that the gene sequencing samples were derived from the exact same portions of the coronary artery plaques used for the histology and immunohistochemistry analyses. Histology and immunohistochemistry analyses included hematoxylin and eosin, van Gieson elastin staining, smooth muscle cell α-actin, Oil Red O staining, picrosirius red staining, and CD31 and CD45 cells, as described in (58).

Statistical analysis

Data throughout the paper are expressed as means ± SD. Data were tested for normality before using unpaired two-tailed Student’s t test. For more than two groups, one-way analysis of variance (ANOVA) was used. P < 0.05 was considered statistically significant. ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Differentially expressed genes were visualized using GraphPad software (V.7.0a or 8.2.0).


Materials and Methods

Fig. S1. Identification and characterization of lncRNA SNHG12 in mouse and human cells.

Fig. S2. SNHG12 does not affect lipid metabolism or inflammation.

Fig. S3. Identification of DNA-PK as an SNHG12 interactor.

Fig. S4. SNHG12 silencing impairs DDR in ECs and macrophages.

Fig. S5. Phenotypic effects of SNHG12 on senescence, EC permeability, and efferocytosis.

Fig. S6. SNHG12 has no regulatory role in apoptosis.

Fig. S7. Accumulating DNA damage and its effect on mitochondrial stress.

Fig. S8. Proposed mechanism of SNHG12 regulation of atherosclerosis through a DNA-PK–mediated DDR in the vascular endothelium.

Table S1. Primer list.

Data file S1. Sequences for cloning.

Data file S2. Raw data from figures.

Movie S1. Transcytosis assay of control gapmeR (25 nM)–transfected HCAECs.

Movie S2. Transcytosis assay of SNHG12 gapmeR (25 nM)–transfected HCAECs.

Movie S3. Transcytosis assay of control lentivirus–transduced HCAECs.

Movie S4. Transcytosis assay of SNHG12 lentivirus–transduced HCAECs.

Movie S5. Transcytosis assay of control lentivirus–transduced HCAECs in the presence of ROS.

Movie S6. Transcytosis assay of SNHG12 lentivirus–transduced HCAECs in the presence of ROS.

References (5963)


Acknowledgments: We would like to thank J. Hutchinson and L. Pantano (Harvard Chan Bioinformatics Core, Harvard T.H. Chan School of Public Health, Boston, MA) for assistance with RNA-seq analysis. We thank M. L. Oscane (Neurobiology Imaging Facility, Harvard Medical School, Boston, MA) for assistance with imaging (NINDS P30 Core Center Grant #NS072030). We thank F. W. Luscinskas and G. A. Newton for technical advice. We also thank A. Lay-Hong and A. P. Gad for their assistance with immunofluorescence imaging. Funding: This work was supported by the NIH (HL115141, HL117994, HL134849, and GM115605 to M.W.F.; HL134892-01A1 and HL080472 to P.L.), the Arthur K. Watson Charitable Trust (to M.W.F.), the Dr. Ralph and Marian Falk Medical Research Trust (to M.W.F.), the RRM Charitable Fund (to P.L.), the École Polytechnique Fédérale de Lausanne (to J.A.), the Swiss National Science Foundation (P2BEP3_162063 to S.H. and 310030B-160318 to J.A.), the American Heart Association (18POST34030395 to S.H., 18CSA34080399 to P.L., and 18SFRN33900144 to M.W.F.), the National Natural Science Foundation of China (81570334 and 81770358 to T.Y.), the Xiangya Eminent Doctor Project (#013 to T.Y.), the Heart and Stroke Foundation of Canada (G-16-00013521 to W.L.L.), and a Canada Research Chair (to W.L.L.). Author contributions: M.W.F. and S.H. conceived the hypothesis. S.H., D.Y., X.S., D.D., S.G., Y.D., D.F., R.M., L.C., N.M., Y.T., J.F.L., and E.S. performed the experiments. S.H., A.K.W., V.S., G.S., J.A.M., P.H.S., W.L.L., T.Y., J.A., P.L., and M.W.F. designed and interpreted the results. S.H. and M.W.F. wrote the manuscript. Competing interests: J.A.M. serves on the SAB of 908 Devices. M.W.F. and Brigham and Women’s Hospital filed a patent application for use of atherosclerosis-associated lncRNAs entitled “Targeting lncRNA in cardiovascular disease” (PCT 62/757,832, PCT 62/905,479, PCT/US2019/061006). P.L. has a financial interest in Xbiotech, a company developing therapeutic human antibodies. P.L.’s interests were reviewed and are managed by Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict of interest policies. The other authors declare that they have competing interests. Data and materials availability: All data can be found in the main text or the Supplementary Materials. RNA-seq data are available through the Gene Expression Omnibus (GSE138219). MS data for lncRNA pulldown are available through MassIVE under accession MSV000084424 (doi:10.25345/C51T1G).

Stay Connected to Science Translational Medicine

Navigate This Article