H19 Induces Abdominal Aortic Aneurysm Development and Progression

All animal protocols were approved by the Administrative Panel on Laboratory Animal Care at Stanford University (http://labanimals.stanford.edu/) or the Karolinska Institutet, and followed the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (NIH Publication No. 85-23, revised 1996). Animals were purchased from The Jackson Laboratory (Bar Harbor, ME). All experiments were performed with 10-week-old male C57BL/6J (porcine pancreatic elastase [PPE]¹⁰) and 10-week-old male apolipoprotein E–deficient (ApoE^−/−) C57BL/6J mice (Angiotensin [Ang] II). A detailed description is provided in the online-only Data Supplement.

One-year-old Yucatan mini-pigs (Ldlr^−/− or Ldlr+/+)¹² were provided by Exemplar Genetics (Sioux Center, IA). The experiments in mini-pigs were approved by the Ethics Board of Northern Stockholm. In a total of 12 genetically mutated mini-pigs, AAAs were induced using the PPE instillation method similar to studies described in mice in our current manuscript, and as previously described by others in landrace pigs.¹³ A detailed description is provided in the online-only Data Supplement.

Either an LNA–anti-H19 GapmeR or scr-GapmeR (LNA negative control GapmeR from Exiqon) was injected intraperitoneally (in 0.2 mL of PBS over 3–6 seconds). The concentration of anti-GapmeR or scrambled GapmeR was 20 mg/kg. Injections were carried out at 0, 7, and 14 days after AAA induction in both murine AAA models. The custom-made LNA–anti-H19 GapmeR sequence was 5’-GACGGAGATGGACGAC-3’. The LNA scrambled GapmeR control sequence was 5’-GTGTAACACGTCTATACGCCCA-3’.

At baseline, and 7, 14 and 28 days after aneurysm induction, B-mode ultrasound imaging was performed to assess the abdominal aortic diameter. For the detailed process, please refer to the Expanded Methods in the online-only Data Supplement.

Total RNA was isolated with a TRIzol-based (Invitrogen) RNA isolation protocol. Gene expression was quantitatively analyzed using Taq-Man quantitative reverse transcription–polymerase chain reaction (qRT-PCR) assays. For the detailed process, please refer to the online-only Data Supplement.

Raw reads from 4 saline and 4 AngII-treated ApoE^−/− mice harvested at 7 days were aligned to the GRCm38 mouse reference genome with STAR¹⁴ using default parameters. The complete dataset has been uploaded to the Gene Expression Omnibus. Detailed procedures are explained in the online-only Data Supplement.

RNA profiling was performed using the Mouse Transcriptome Assay (Affymetrix). The Affymetrix transcriptome analysis console v3.0 was used for differential expression analysis, according to the manufacturer’s instructions. The final analysis included 7-day PPE-treated aortae (n=6) versus controls (sham-operated, saline-treated, n=6). A false detection rate adjusted P value <0.05 was considered significant. The complete dataset has been uploaded into the Gene Expression Omnibus.

Histological staining after harvesting was performed in the same region of the abdominal aorta that was imaged to obtain morphometric data to correlate with ultrasound measurements and gene expression results from qRT-PCR using a standardized protocol as previously described.¹⁵ Detailed procedures are explained in the online-only Data Supplement.

) was performed as described above, utilizing the miRCURY LNA microRNA ISH Optimization Kit (Exiqon; see the online-only Data Supplement) and 5’-Biotin labeled probes for H19 or the HIF1α promoter region. After the ISH procedure, the Duolink assay (Sigma-Aldrich) was initiated with overnight incubation of a mouse antibody against biotin and rabbit antibodies targeting HIF1α or specificity protein 1 (Sp1; Abcam). The procedure was continued the next day according to the manufacturer’s instructions and slightly modified for RNA-protein interactions.¹⁶ The amplified interaction signal was captured and analyzed using a Leica Confocal Microscope.

Human aortic SMC (hAoSMCs) were propagated in growth media (SmGM-2; Lonza) with 5% fetal bovine serum per standard protocols (Lonza; passage no. 4–5). Lipofectamine RNAiMAX (Invitrogen) reagent was used for transfection of small oligonucleotides, and Lipofectamine 3000 (Invitrogen) was used for vector transfection. For detailed process, please refer to the online-only Data Supplement.

The IncuCyte Zoom System (Essen Biosciences) was used for real-time and dynamic assessment of the hAoSMCs status over time¹⁷. For detailed process, please refer to the online-only Data Supplement.

Cytoplasmic and nuclear fractions were isolated using the PARIS kit (Thermo Fisher Scientific), according to the manufacturer’s instructions. Western blotting was carried out as previously described¹⁸ (online-only Data Supplement). Tubulin was used as cytoplasmic loading control, and Lamin B1 as the nuclear loading control.

Transcriptional activity of HIF1α was confirmed by transfecting hAoSMCs with pGL-HIF1A prom luciferase vector (Addgene), with or without recombinant Sp1 protein (Abcam)/AngII treatment and pcDNA-H19 wt/deletion. Luciferase activity, 24 hours after treatment, was determined using the Lightswitch Assay (Switchgear) as previously described.¹⁹

Occupancy of Sp1 in the HIF1α promoter region was evaluated via a chromatin immunoprecipitation assay²⁰. For the detailed process, please refer to the online-only Data Supplement.

Human samples from the Munich Vascular Biobank from patients who underwent surgical repair of their AAA (n=20), as well as abdominal aortic samples from organ donor controls (n=10), were harvested during surgery (or explantation), snap-frozen, and stored at −80°C. Approval for studies on human tissue samples was obtained under informed consent, complying with all guidelines and policies at the Klinikum rechts der Isar, Technical University Munich, and in accordance with the Declaration of Helsinki. Patient samples were matched for age (72.3 on average), sex (only males), cardiovascular risk factors (smoking: all smokers, diabetes mellitus: all nondiabetics, without known medical therapy), dyslipidemia (total cholesterol <200 mg/dL; triglycerides < 185 mg/dL), and medication (no antiplatelet or anticoagulant therapy, all patients on statin therapy).

Data are presented as mean±SEM, unless stated differently. Groups were compared using Student t test. Normality was tested to ensure that parametric testing was appropriate. When comparing multiple groups, data were analyzed using ANOVA with Bonferroni post-test. Sequential measurements (abdominal aortic diameters at consecutive time points) were analyzed by two-way repeated-measures ANOVA. Categorical data were analyzed by using the χ² test. A value of P<0.05 was considered statistically significant.

We performed RNA sequencing in the AngII/ApoE^−/− mouse model to identify dysregulation during aneurysm development. Abdominal aortic tissue from either AngII-treated mice (n=4) or sham controls (n=4) were harvested 7 days after pump implantation. The lncRNA H19 was identified as one of the most substantially upregulated transcripts in AAA tissue (Figure 1b). When cross-checking upregulated transcripts in a second experimental AAA mouse model using PPE infusion, H19 and 2 other lncRNAs (A530020G20Rik, F630028O10Rik) were identified as model-independent lncRNAs with potential implications for experimental AAA progression on analysis of a mouse transcriptome array. Of further importance, all 3 transcripts have human homologues, allowing us to evaluate their relevance and potential changes in AAA samples of human origin (Figure 1a and 1b). Upregulation of all 3 lncRNAs was confirmed by RT-qPCR in AngII-induced AAA samples versus controls (Figure 1c). H19 was chosen for further investigations, as it was the most significantly upregulated IncRNA in mouse and human samples compared to the other lncRNAs and their homologues.

The incidence, growth rate, and size of aneurysmal expansion was measured by ultrasound at baseline (day 0) and after 7, 14, and 28 days post-AAA induction (Figure Ia in the online-only Data Supplement). Of importance, systemic in vivo knockdown of H19, utilizing site-specific antisense oligonucleotides (LNA-GapmeRs; Figure Ib, Ic, and Id in the online-only Data Supplement), significantly limited aneurysm growth in both murine models (Figure 1d and 1e; Figure IIa in the online-only Data Supplement). Mismatched, scrambled oligonucleotides were used as relevant controls. Fluorescent imaging of fresh aortic sections indicated sufficient transvascular distribution of the prelabeled GapmeR (Figure Ic in the online-only Data Supplement). Knockdown of H19 was confirmed with qRT-PCR in various organs in addition to the abdominal aorta (Figure Id in the online-only Data Supplement).

In situ hybridization indicated that H19 is expressed in both medial SMCs and adventitial fibroblasts of the undiseased, nondilated aorta (Figure 1f: NC). On AngII infusion, H19 expression was substantially elevated in medial SMCs in dilated aortic tissue (Figure 1f). This increase was abolished by inhibiting H19 with GapmeRs. These data suggest that H19 participates in AAA development and progression possibly through influencing SMC dynamics.

Consistent with our in vivo data (Figure 1b and 1c), AngII was able to robustly induce H19 in cultured aortic SMCs (Figure IIIa in the online-only Data Supplement). AngII treatment further induced apoptosis, while decreasing proliferation rates of SMCs (Figure IIIb through IIId in the online-only Data Supplement). These effects were both attenuated by H19 knockdown (Figure IIIb through IIId in the online-only Data Supplement), indicating the regulatory role of H19 in SMC apoptosis under AAA-relevant (AngII) conditions. Of note, this proapoptotic effect could only be observed at a later timepoint (48 h), and not during the early, acute response phase (8 h; Figure III in the online-only Data Supplement).

To further evaluate the cellular effects of H19 modulation in SMCs, we used a kinetic live cell imaging system. First, we either inhibited H19 with LNA-GapmeRs (Figure 2a) or induced H19 with pcDNA-H19–dependent overexpression (Figure 2b) in cultured hAoSMCs. Two different custom anti-H19 GapmeRs were tested for knockdown efficiency in hAoSMCs, and the more potent one was chosen for further studies (and was used for the in vivo experiments; Figure Ib in the online-only Data Supplement). Cells were kinetically monitored for rates of proliferation and apoptosis. Overexpression of H19 (Figure 2b) resulted in enhanced apoptosis (Casp3+, green dots) and decreased proliferation (confluence %, orange mask) of SMCs in a time-dependent manner (Figure 2c through 2e). Knockdown of H19 (Figure 2a) inhibited SMC apoptosis, whereas no significant effect on the proliferation rate could be observed (Figure 2c through 2e; Movies 1 through 4 in the online-only Data Supplement).

Accumulating studies have suggested that H19 may function as a reservoir of miR-675, which is embedded in the first exon of the H19 transcript.^21,22 Thus, we evaluated the role of miR-675 in H19-mediated effects on SMCs. Cultured hAoSMCs were transfected with either H19 wild-type vector (H19 wt) or H19 mutant with deletion of miR-675 (H19 mut, Figure 3a), and monitored over time. Both H19 mut and H19 wt induced apoptosis and reduced proliferation of SMCs (Figure 3b and 3c). Apart from this, no significant difference in miR-675 expression was observed in AngII-induced AAAs by qRT-PCR (Figure 3d). Unlike H19, miR-675 was furthermore not significantly deregulated in tissue samples from human patients with AAA, or in either of the 2 experimental murine models of aneurysm development. Thus, the effects of H19 on SMCs appear independent of miR-675 during AAA development.

Transcription factors have been well recognized as master regulators controlling gene expression, chromatin stability and cell homeostasis.^23–25 To explore how H19 regulates SMC apoptosis, we proposed that H19 might affect molecular processes via alteration of transcription factors. Based on an in silico analysis of regulatory RNA elements,²⁶ we identified several transcription factors that were predicted to interact with H19, but had not been experimentally validated. The analysis concluded that cAMP responsive element binding protein 1 (CREB1), transcription factor CP2 (TFCP2), GTF2I repeat domain containing 1 (GTF2IRD1), SMAD family member 4 (SMAD4), HIF1α, zinc finger and BTB domain containing 14 (ZBTB14), Krüppel-like factor 11 (KLF11), and transcription factor AP-2 alpha (TFAP2α) are possibly involved in the molecular function of H19. All of these transcription factors had a predicted binding free energy of <−30 Kcal/mol (Figure 3e).

First, to determine whether these transcription factors are regulated on the RNA level as part of a feed-forward loop, a custom-designed RNA array was used to investigate changes in transcription factor expression in SMCs on overexpression and knockdown of H19. Results revealed that HIF1α was the only differentially regulated factor (Figure 3f). In addition, HIF1α protein levels decreased after H19 knockdown and concomitantly increased after vector-based H19 induction (Figure IVa in the online-only Data Supplement).

To confirm that HIF1α mediates the effects of H19, we used siRNA-induced inhibition of HIF1α. Both mRNA (Figure 4b in the online-only Data Supplement) and protein (Figure 4c in the online-only Data Supplement) levels of HIF1α were significantly blocked. Interestingly, H19 overexpression as well as AngII-treatment dramatically increased the expression of HIF1α and consequently apoptosis in SMCs (Figure 3g, upper panel). This effect could be abolished by inhibition of H19 or HIF1α silencing (Figure 3g, bottom panel and right panel). H19 levels in SMCs remained unchanged on HiF1α inhibition (Figure 4d in the online-only Data Supplement). Importantly, the proapoptotic protein Bcl-2 associated X (BAX) displayed the same pattern as HIF1α, whereas the antiapoptotic mediator B-cell CCL/lymphoma 2 (BCL2) showed the opposite trend (Figure 4e in the online-only Data Supplement), suggesting that AngII/H19-induced SMC apoptosis is dependent on changes in HIF1α expression.

To further dissect the mechanism through which HIF1α triggers SMC apoptosis, we either performed HIF1α overexpression or induced its expression via the AAA-related factor AngII. In both cases, BAX expression and the SMC apoptotic ratio substantially increased. This could be attenuated by p53 knockdown (Figure 3h and 3i). It was recently reported that BAX is regulated by p53²⁷ and has multiple p53 binding sites in its promoter region (Figure 4f in the online-only Data Supplement). Previous studies have shown that p53 is controlled by the pro- to-oncogene transformed mouse 3T3 cell double minute 2 (Mdm2)²⁸, a ubiquitin E3 ligase, via a feedback mechanism in lung cancer cells and mouse embryonic fibroblasts.¹⁷ p53 is an established factor that triggers aneurysm expansion.^29,30 We were able to confirm that HIF1α directly interacts with Mdm2, preventing Mdm2-mediated reduction of p53 (Figure 3j and 3k). This suggests that HIF1α induces SMC apoptosis via regulating Mdm2-p53 activity during aortic dilatation.

To confirm that HIF1α contributes to H19-induced SMC apoptosis and is relevant in AAA, we evaluated the expression of HIF1α and apoptosis-related genes in both mouse and human AAA samples with qRT-PCR. As expected, H19, HIF1α, and BAX were all significantly increased, whereas the antiapoptotic protein BCL2 was markedly decreased in the murine AngII and PPE models, as well as in human aortic aneurysms (Figures 4a and 5a; Figure IIb in the online-only Data Supplement). Utilizing immunohistochemical analysis, we detected HIF1α upregulation in medial SMCs, correlating with cell apoptosis in both experimental AAA models (Figure 4c: SCR; Figure IIc through IIe in the online-only Data Supplement: SCR). Depletion of H19 sufficiently suppressed the expression of HIF1α and significantly limited SMC apoptosis (Figure 4b and 4c: H19 KD; Figure IId and IIe in the online-only Data Supplement: H19 KD). Again, colocalization of upregulated H19 and HIF1α, as well as a correlation with cellular apoptosis, was observed in SMC layers of human AAA specimens (Figure 5b, Figure V in the online-only Data Supplement). H19 and HIF1α were consistently induced in both murine AAA models, whereas specific changes for known HIF1α downstream targets were inconsistent in the different AAA models (Figure Vc in the online-only Data Supplement), indicating that these changes are likely to be model-related rather than HIF1α-dependent.

Finally, AAA was induced in wild-type (WT) or Ldlr^−/− Yucatan mini-pigs through PPE perfusion, similar to the murine model. The incidence, growth rate, and absolute size of aneurysmal expansion were measured by ultrasound at baseline, and at 7 and 28 days after AAA induction (Figure 5c). Compared with the sham-operated group, the expression of H19 was significantly upregulated along with enhanced expression of HIF1α in PPE-induced AAAs (Figure 5d and 5e).

Cultured hAoSMCs subjected to either H19 modulation or treatment with AngII underwent ISH, immunofluorescent staining, and proximity ligation assay. The analysis indicated that both AngII treatment and H19 overexpression resulted in increased expression of HIF1α and colocalization with H19, whereas knockdown of H19 reduced HIF1α (Figure 6a).

Colocalization only indicates closeness of different components, and fluorescence microscopy is insufficient to identify the physical apposition of 2 molecules.³¹ Bioinformatic analysis²⁶ revealed the secondary structure of a HIF1α binding site on H19, with a low binding free energy of −33.9 Kcal/mol (Figure 6b). To assess the physical interaction between H19 and HIF1α, we used ISH and proximity ligation assay, in which amplified signals can be detected once 2 molecules reach close proximity (<40 nm), considered proof of their interaction,³² Indeed, AngII treatment as well as H19 overexpression substantially enhanced the interaction between H19 and HIF1α, predominantly in the cytoplasm (Figure 6c).

HIF1α is required for embryonic cardiovascular development, and thus is considered essential for cell survival during hypoxia. During this process, induced HIF1α translocates into the nucleus and activates transcription of many genes involved in the attenuation of the hypoxic damage.³³ Cobalt chloride (CoCl₂), a well-established chemical inducer of HIF1α,³⁴ successfully increased accumulation of HIF1α and triggered its translocation into the nucleus (Figure 6d). Interestingly, H19 attenuated the translocation and detained HIF1α within the cytoplasm, suggesting that this direct interaction of H19 binding cytoplasmic HIF1α can stimulate SMC apoptosis and enhance AAA progression.

To further dissect the molecular mechanism of HIF1α regulation by H19, we assessed the transcriptional activity of HIF1α. It has previously been shown that HIF1α transcription is dependent on the transcription factor Sp1,^35,36 with multiple binding sites located in the HIF1α promoter (Figure 7a). Thus, we investigated whether H19—apart from its regulatory role in the cytoplasm—can also directly influence HIF1α transcription by interacting with Sp1-mediated effects in the promoter region of the HIF1α gene. We designed a biotin-labeled ISH probe, targeting the HIF1α promoter region, and used the proximity ligation assay-based interaction detection method as described above. Under control conditions (using a scrambled probe), only weak Sp1 binding to the HIF1α promoter was observed (Figure 7b). The interaction became drastically enhanced on AngII treatment (Figure 7b), as well as on overexpression of H19 (Figure 7b). A diminished effect was seen when performing a proof-of-concept study with knockdown of H19 (Figure 7b). The average signal intensity is summarized in Figure 7c. A similar interaction pattern between Sp1 and H19 was confirmed in the nucleus (Figure VIa through VIc in the online-only Data Supplement). In addition, no significant changes were seen in the expression level of Sp1 in cells on modulation of H19 or AngII treatment, as well as in our murine AAA models (Figure VId in the online-only Data Supplement).

To assess the transcriptional activity, we then used a luciferase reporter assay containing the HIF1α wt or Sp1-binding site mutants in the promoter region. Cultured hAoSMCs were transfected with a pGL-HIF1α promoter luciferase vector, with or without the recombinant Sp1 protein/AngII treatment and pcDNA-H19 wt/deletion. Sp1 treatment clearly led to an induction of luciferase activity, representing the transcriptional activity of HIF1α, which served as a positive control (Figure 7d). Overexpression of H19 or AngII treatment displayed similar effects (Figure 7d). The combined treatment of Sp1 and H19 overexpression substantially increased transcriptional activity, suggesting that H19 recruits and facilitates Sp1 binding to the HIF1α promoter region (Figure 7d). These effects were abolished in the presence of the Sp1 inhibitor mithramycine A, further supporting the transcriptional role of Sp1 in this scenario (Figure 7d). In addition, luciferase assays of HIF1α promoter mutants (M1-M5) indicated 2 Sp1 binding sites (−66 and −79) in the HIF1α promoter region as being essential for these effects (Figure 7e). Moreover, analyzing the sequence of H19 and HIF1α promoter (+500 to −2000 bp) using the LongTarget tool,³⁷ a H19-HIF1α binding pattern was predicted as shown in Figure 7f. Deletion of this region in H19 (H19 del) retained proapoptotic capacity (Figure VIe in the online-only Data Supplement) but failed to enhance HIF1α transcriptional activity as H19 wt did (Figure 7g). Taken together these data suggest that H19 enhances HIF1α transcription via recruitment of Sp1 to its promoter region.

Finally, a chromatin immune precipitation assay was able to confirm the occupancy of Sp1 and H19 in the HIF1α promoter. SMCs were subjected to chromatin immune precipitation with Sp1 antibodies, followed by qRT-PCR measurements with specific probes targeting the HIF1α promoter or −10kb upstream (control), or H19 respectively. The results indicated that the occupancy of Sp1 in the HIF1α promoter region was significantly increased compared to that of the −10kb upstream/control (Figure 7h) in response to treatment with AngII or H19 overexpression, but not on H19 knockdown. Similar trends were observed for H19 levels (Figure 7i). In addition, H19 was significantly increased with Sp1 RNA-chromatin immune precipitation on stimulation with AngII or H19 overexpression, further validating the direct interaction between Sp1 and H19 (Figure VIf in the online-only Data Supplement).