To study the role of the murine ortholog of the human CAD risk interval in atherosclerosis, Chr4^{Δ70kb/Δ70kb} mice were backcrossed into the Ldlr^−/−ApoB^100/100 atherosclerotic mouse strain and studied on those fed the standard laboratory diet and on those fed HFD for 6 and 12 weeks (Figure 1A). The hematopoietic deletion was made by destroying the bone marrow (BM) of Ldlr^−/−ApoB^100/100 mice by irradiation and then transplanting new BM collected from Chr4^{Δ70kb/Δ70kb} or WT mice, after which the mice were fed HFD for 12 weeks (Figure 1B). The size and composition of atherosclerotic plaques were characterized by the histology and immunohistology of aortic roots. Plasma lipids and white blood cells were analyzed at the end of the study. To investigate the expression pattern of human ANRIL and its murine equivalent, Ak148321, in atherosclerotic plaques, scRNA-seq data generated by Wirka et al. (14) from human coronary arteries were used. For mouse atherosclerosis scRNA-seq, the raw sequencing reads generated by Pan et al. (15) from mouse aortas were obtained and reprocessed. Expression of CAD risk area transcripts in mouse spleen and the genes regulating lipid metabolism in the liver were measured with qPCR, using TaqMan-based assays. Mouse BMDMs were extracted, the cells were exposed to oxLDL for 16 h and the expression of pro-inflammatory cytokines was measured. M1/M2 phenotype polarization assay was performed, and foam cell formation and cell proliferation rate were measured in BMDMs. All animal experiments were approved by the National Experimental Animal Board of Finland and carried out following the guidelines of the Finnish Act on Animal Experimentation and Directive 2010/63/EU of the European Parliament. For statistical analyses, the Student's t-test or nonparametric Mann–Whitney test was used for testing the difference of means between the groups. The difference was considered statistically significant when the P-value was ≤ 0.05.

Detailed study methods are available in Supplementary material. All relevant data considered in this article are provided in Section 3 and Supplementary material, and raw data are available upon a reasonable request from the corresponding author.

To study the role of the murine equivalent of the CAD risk locus and ANRIL in atherosclerosis, we assessed the development of atherosclerotic plaques in hypercholesterolemic Ldlr^−/−ApoB^100/100 mice having a systemic deletion orthologous to the human CAD risk interval in Chr9p21.3. These Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice were fed HFD for 6 weeks to induce early atherosclerosis or aged until 6 months of age on standard diet or with 12 HFD for advanced plaques, respectively (Figure 1A). Chr4^{Δ70kb/Δ70kb} increased the development of advanced atherosclerotic plaques in an HFD-dependent manner (P = 0.0001; Figure 2A) in Ldlr^−/−ApoB^100/100 mice after 12 weeks of HFD, whereas there was no effect on atherosclerosis in age-matched mice on standard laboratory diet (Figure 2A). Interestingly, compared with standard diet, HFD increased the atherosclerotic plaque area in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice almost two-fold (P < 0.0001) but did not significantly affect plaque size in Ldlr^−/−ApoB^100/100 control mice (P = 0.144). In addition, Chr4^{Δ70kb/Δ70kb} did not have an effect on early atherosclerosis that was measured after 6 weeks of HFD (Figure 2A). The effect of Chr4^{Δ70kb/Δ70kb} on atherosclerotic plaque morphology was analyzed, but there were no statistically significant differences in the total amount of macrophages, collagen, necrosis, or SMCs (data not shown) or the percentages of these characteristics in relation to the plaque area on the standard laboratory diet (Supplementary Figures 1A–D) or after 6 (Supplementary Figures 2A–E) or 12 weeks of HFD (Supplementary Figures 3A–E). Thus, HFD triggered an increase in advanced plaque area in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice compared with Ldlr^−/−ApoB^100/100, which was caused by a gradual increase in multiple plaque characteristics. Plaque lymphocytes were analyzed after 6 (Supplementary Figure 2B) and 12 weeks of HFD (Supplementary Figure 3B), but their number was low, leaving macrophages as the main inflammatory cell type in the plaques.

Generally, elevated plasma total cholesterol and lipoprotein levels due to HFD are known to increase the number of circulating leukocytes (16, 17). Interestingly, the total number of blood leukocytes in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice was increased (9.68 ± 2.26 × 10⁹ cells/L) compared with Ldlr^−/−ApoB^100/100 mice (7.53 ± 1.17 × 10⁹ cells/L; P = 0.038; Figure 2B) after 12 weeks of HFD, but Chr4^{Δ70kb/Δ70kb} did not affect the total number of blood leukocytes on the standard laboratory diet (Figure 2C) or after 6 weeks of HFD (Figure 2D). There was no difference in white blood cell differential count between the genotypes on a standard laboratory diet or after 6 or 12 weeks of HFD (Table 1). Body weight, plasma total cholesterol, LDL, HDL, and triglyceride levels were equal in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 mice on the standard laboratory diet and after 6 and 12 weeks of HFD (Table 1). Fasting blood glucose levels were higher in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice (7.35 ± 0.58 mmol/L) compared with Ldlr^−/−ApoB^100/100 mice (6.59 ± 0.62 mmol/L) on the standard laboratory diet (P = 0.043; Table 1), but not in 6- or 12-week HFD mice.

As the Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice showed increased atherosclerosis and an elevated number of circulating leukocytes but no difference in plasma cholesterol levels after prolonged HFD, we investigated the effects of hematopoietic deficiency of the risk interval on atherosclerosis (Figure 1B). Therefore, WT or Chr4^{Δ70kb/Δ70kb} BM was transplanted into Ldlr^−/−ApoB^100/100 mice, and the engraftment of the transplant was determined from white blood cells (Figure 3A). We found that Chr4^{Δ70kb/Δ70kb} in BM-derived cells significantly increases the size of aortic root atherosclerotic plaques after 12 weeks of HFD in Ldlr^−/−ApoB^100/100 mice (P = 0.009; Figure 3B) compared with Ldlr^−/−ApoB^100/100 mice that received the WT BM. Characterization of plaque morphology revealed no statistically significant differences in total area (data not shown) and proportion of macrophages (Figure 3C), lymphocytes (Figure 3D), collagen (Figure 3E), or necrosis (Figure 3F) in relation to the plaque area between the BM transplant groups. Hematopoietic Chr4^{Δ70kb/Δ70kb} increased the area of SMCs with respect to the total plaque area compared with Ldlr^−/−ApoB^100/100 mice with WT BM (P = 0.004; Figure 3G). However, the number of SMCs was low, averaging < 1% of the plaque area in both groups. Parallel to total knockout, Chr4^{Δ70kb/Δ70kb} BM-transplanted mice also showed a tendency to increase blood leukocyte number (12.09 ± 3.65 × 10⁹ cells/L) compared with WT BM recipients (9.42 ± 2.82 × 10⁹ cells/L), but the difference between the groups did not reach statistical significance (P = 0.087; Table 2). The white blood cell differential counts were similar between the groups.

After 12 weeks of HFD, Chr4^{Δ70kb/Δ70kb} BM recipient mice showed significantly increased plasma total cholesterol (26.91 ± 3.77 mmol/L, P = 0.012) and triglyceride (3.31 ± 1.03 mmol/L, P = 0.001) levels compared with WT BM recipients in an Ldlr^−/−ApoB^100/100 background (cholesterol 22.35 ± 3.03 mmol/L, triglycerides 1.60 ± 0.57; Table 2). There were no differences in plasma LDL or HDL levels between the groups. In addition, the fasting blood glucose level was increased in Chr4^{Δ70kb/Δ70kb} BM recipient mice (8.25 ± 0.93 mmol/L) compared with WT BM recipients in the Ldlr^−/−ApoB^100/100 background (7.44 ± 1.09 mmol/L), but the difference was not statistically significant (P = 0.054). As the BM-specific deletion of the murine CAD risk locus increased the plasma total cholesterol and triglyceride levels, we measured the expression of master regulators of liver fatty acid and cholesterol synthesis. Interestingly, the hematopoietic Chr4^{Δ70kb/Δ70kb} significantly increased the hepatic expression of Fasn in Ldlr^−/−ApoB^100/100 mouse livers (P = 0.048; Supplementary Figure 4A) but did not affect the expression of Srebf1 (Supplementary Figure 4B) and Srebf2 (Supplementary Figure 4C) compared with WT BM recipients after 12 weeks of HFD. Nevertheless, the hepatic steatosis and the accumulation of inflammatory cells in the liver were equal between the Chr4^{Δ70kb/Δ70kb} BM and WT BM transplanted Ldlr^−/−ApoB^100/100 mice after 12 weeks of HFD (Supplementary Figure 4D).

As the CAD risk locus and its transcripts may potentially promote atherogenesis via inflammatory cell-specific regulation in Ldlr^−/−ApoB^100/100 mice and macrophages are the most prominent inflammatory cell type in mouse atherosclerotic plaques, we extracted BM-derived macrophages (BMDMs) from Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 mice and aimed to clarify the role of the CAD risk locus in macrophage polarization, foam cell formation, and proliferation. Pro-inflammatory, IFN-γ-induced, M1 macrophages are classically considered proatherogenic, whereas IL4-induced M2 macrophages are anti-inflammatory and provide atheroprotective effects. Interestingly, the secretion of several pro-inflammatory cytokines and chemokines, including IL6 (P = 0.040), MCP1 (P = 0.033), and RANTES, was significantly increased at the protein level in response to oxLDL in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs compared with Ldlr^−/−ApoB^100/100 BMDMs (Figure 4A). CXCL1/KC (P = 0.053), CXCL2/MIP2a (P = 0.08), and CCL9/MIP1y (P = 0.051) also showed increased protein-level expression, but without statistical significance. The enhanced pro-inflammatory phenotype was also supported by the macrophage polarization assay as the mRNA expression of M1 marker Tnf was already higher at the basal level (P = 0.005) and also in response to both IFN-γ (P = 0.003) and IL4 (P = 0.016) in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100BMDMs compared with Ldlr^−/−ApoB^100/100BMDMs (Figure 4B). Moreover, the expression of another M1 marker, Il6, was increased at the mRNA level in response to IFN-γ (P = 0.004) and was not altered on the basal level or after IL4 treatment in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs compared with Ldlr^−/−ApoB^100/100 BMDMs (Figure 4C). The expression of macrophage M2 phenotype markers, ArgI and Fizz1, was also measured on the basal level and in response to IL4 and IFN-γ in BMDMs. Surprisingly, ArgI expression was significantly higher at the basal level (P = 0.035) and in response to IFN-γ (P = 0.029) in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs compared with Ldlr^−/−ApoB^100/100 BMDMs, but expression levels were equal between genotypes in response to IL4 (Figure 4D). Fizz1 expression in BMDMs was equal between genotypes at the basal level and in response to IL4 and IFN-γ (Figure 4E). However, the enhanced pro-inflammatory activity in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs did not have an effect on macrophage cholesterol accumulation and foam cell formation as the BMDMs were treated with oxLDL and the cellular neutral lipid content was measured by Oil Red O, but no differences were observed between the genotypes (Figure 4F). Finally, Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs showed significantly lower proliferation both at the basal level (P = 0.001; Figure 4G) and after IFN-γ (P = 0.008; Figure 4H) and oxLDL (P = 0.005; Figure 4I) but not after IL4 (Figure 4J) treatment compared with Ldlr^−/−ApoB^100/100 BMDMs.

Because both global and hematopoietic deletions in the murine CAD risk orthologous locus led to increased atherosclerosis, global Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice showed increased blood leukocyte count, and Chr4^{Δ70kb/Δ70kb} BMDMs showed more pro-inflammatory phenotype in vitro, we decided to investigate if human ANRIL and its murine equivalent (Figure 5A) are expressed in inflammatory cell populations in atherosclerotic plaques, and how their expression levels are in comparison with other vascular cell populations. We analyzed public single-cell RNA sequencing data from Pan et al. (15) and found that murine ANRIL equivalent Ak148321 is expressed most in macrophages and SMCs of mouse atherosclerotic aortas (Figure 5B). We were also able to focus on the major macrophage and SMC subtypes, and the results suggest that a proliferating macrophage subpopulation has the highest expression of Ak148321. Out of the plaque SMCs, the transitioning SMCs had the highest expression of Ak148321. For human atherosclerosis, we analyzed the scRNA sequencing data of human coronary plaques from Wirka et al. (14) and found that ANRIL expression was highest in T-cells and B-cells, followed by macrophages (Figure 5C).

To further examine the expression pattern of the CAD risk locus murine equivalent in an inflammatory cell context, we measured the expression of three Ak148321 exons and four expected circular isoforms in the spleens of Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 mice on the standard laboratory diet and in response to prolonged HFD (Figure 6A). HFD did not significantly affect the expression of Ak148321 exons 3, 6, or 9 in the spleens of Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 or Ldlr^−/−ApoB^100/100 mice (Figures 6B–D), However, exon 9 expression was significantly higher in the spleens of Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice compared with Ldlr^−/−ApoB^100/100 mice on standard laboratory diet (P = 0.042). Next, we measured the expression of four expected Ak148321 circular isoforms in response to HFD. Interestingly, the expression of Circular 1 was increased in Ldlr^−/−ApoB^100/100 mouse spleen by HFD (P = 0.024; Figure 6E), but HFD had no effect on the expression of Circular 2 (Figure 6F). Circular 3 and Circular 4 remained undetected in Ldlr^−/−ApoB^100/100 mouse spleen, and none of the circular Ak148321 isoforms were detected in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mouse spleen (Figures 6G, H). Finally, to mimic the atherosclerotic plaque environment, the expression of three Ak148321 exons was measured in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 BMDMs in response to oxLDL. Ldlr^−/−ApoB^100/100 BMDMs showed a tendency to a higher expression of Ak148321 exons 3 (Figure 6I) and 6 (Figure 6J) in response to oxLDL, but the difference was not statistically significant. As expected by the knockout, Ak148321 exon 6 was not detected in Chr4^{Δ70kb/Δ70kb} BMDMs (Figure 6J). There were no differences in the expression pattern of Ak148321 exon 9 in BMDMs in response to oxLDL or by the Chr4^{Δ70kb/Δ70kb} (Figure 6K).

Two cell cycle regulator genes, CDKN2A and CDKN2B, neighboring the CAD risk interval have been suggested to play a role in the CAD risk associated with the 9p21.3 risk locus. When analyzing single-cell RNA sequencing data from mouse (15) and human (14), we found that, in mouse aortic atherosclerotic plaques, Cdkn2a and Cdkn2b were expressed mostly in SMCs (Supplementary Figures 5A, B), while in human coronary plaques, they were mostly expressed in T cells (Supplementary Figures 5C, D). To evaluate the impact of Cdkn2a and Cdkn2b in atherosclerosis and the regulation of macrophage phenotype in the Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mouse model, we measured the expression of Cdkn2a and Cdkn2b in Chr4^{Δ70kb/Δ70kb} and Ldlr^−/−ApoB^100/100 BMDMs and the spleens of Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 mice after oxLDL treatment or HFD, respectively. The expression of Cdkn2a was increased in nontreated Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 BMDMs (P = 0.029), but not in natLDL or oxLDL-treated Chr4^{Δ70kb/Δ70kb} BMDMs compared with Ldlr^−/−ApoB^100/100 BMDMs (Supplementary Figure 5E). Cdkn2b expression did not significantly differ between Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 and Ldlr^−/−ApoB^100/100 BMDMs at the basal level or in response to native or oxLDL treatment (Supplementary Figure 5F). Finally, the splenic expression of Cdkn2a was upregulated by the HFD in Ldlr^−/−ApoB^100/100 mice (P = 0.024), while in Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 spleens, the expression was not modulated by diet (Supplementary Figure 5G). There were no differences in the expression of Cdkn2a or Cdkn2b in the spleen of Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice compared with Ldlr^−/−ApoB^100/100 mice on the standard laboratory diet or after 6 or 12 weeks of HFD (Supplementary Figures 5G, H).

Several mechanisms for how ANRIL and its splicing isoforms affect atherogenesis have been proposed. In human studies, circular ANRIL isoforms have been linked to possible protection against CAD, while linear isoforms are considered pro-atherogenic (6). In the present study, we demonstrated for the first time two circular RNA isoforms being expressed from the CAD risk orthologous locus in mice. Considering the multiple isoforms of ANRIL and its murine equivalent, Ak148321, both disruption of the full-length linear transcript and deficiency of the potentially protective circular transcripts are considered to mediate the proatherogenic effect of Chr4^{Δ70kb/Δ70kb} in mice. Moreover, the increased expression of one of the circular isoforms in the spleen by HFD further suggests the activation of the circular isoforms under metabolic stress. Holdt et al. reported that circANRIL plays a role in vascular cell proliferation and apoptosis, showing an association between higher expression of circANRIL, stronger induction of apoptosis, and reduced proliferation in human SMCs, as well as demonstrating how circANRIL acts as a binding component of proteins regulating ribosomal RNA maturation (6).

A previous study using ApoE^−/− mice (9) has concluded that the deletion of the orthologous CAD risk locus in mice increases the size of advanced atherosclerotic plaques, with no other effects on plaque vulnerability than increased plaque calcification. That study proposed that the deletion of the risk locus accelerates vascular SMC proliferation, the expression of the calcification-regulating genes, and the sensitivity to ossification. In line with these reports, we indicated that both total and hematopoietic deletion of the risk locus ortholog in mouse models increases the development of advanced but not early atherosclerotic plaques. Interestingly, we did not detect any major changes in plaque morphology or stability in early or advanced atherosclerosis. However, the increased proportion of SMCs found in atherosclerotic plaques of hematopoietic Chr4^{Δ70kb/Δ70kb} mice is in line with the previous study on ApoE^−/− mice. Pro-inflammatory activation is known to promote SMC migration and proliferation in the vascular wall (23). Nevertheless, only a small number of SMCs were present in both early and advanced murine atherosclerotic plaques in this study and thus possibly have only a minor contribution to plaque development.

Despite their structural differences, the single-cell sequencing analysis from atherosclerotic plaques demonstrated the expression of both murine Ak148321 and human ANRIL being highest in macrophages and T- and B-cells. In this study, the number of plaque lymphocytes was remarkably low in both systemic and hematopoietic Chr4^{Δ70kb/Δ70kb} mouse plaques, leaving macrophages as the dominant inflammatory cell type. The Chr4^{Δ70kb/Δ70kb} BMDMs showed a more pro-inflammatory phenotype in vitro, with elevated expression of Il6 and other pro-inflammatory cytokines and chemokines after exposure to oxLDL and IFN-γ. The vascular pro-inflammatory activity regulated by ANRIL proposed in the present study is also supported by others, as macrophages from MI patients carrying the homozygous risk haplotype have been reported to express increased levels of pro-inflammatory chemokines MCP-1 and −2 compared with patients with the nonrisk haplotype (24). ANRIL has also been reported to regulate the inflammatory response in human endothelial cells by direct binding with Yin Yang 1 (YY1) and thereby regulating the NF-κB pathway and other inflammatory genes, such as IL6 and IL8, downstream of the TNF pathway (7). In our study, the proliferation rate of macrophages was reduced by Chr4^{Δ70kb/Δ70kb} at the basal level and in response to oxLDL and IFN-γ, which may indicate increased apoptosis in macrophages and, potentially, increased secondary inflammation, as also proposed by Holdt et al. (6). To conclude, the biology of the risk locus transcripts in the regulation of macrophage function and atherogenesis seem s to be at least partially similar between human and mouse models, and it may explain the higher susceptibility to atherosclerosis in hypercholesterolemic Chr4^{Δ70kb/Δ70kb} mice and human carriers of the risk SNPs.

Interestingly, the hematopoietic Chr4^{Δ70kb/Δ70kb} in hypercholesterolemic mice elevated plasma total cholesterol and triglyceride levels after 12 weeks of HFD, and that potentially at least partially accelerated the development of the atherosclerotic lesion. However, as this elevation was not observed in the systemic Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mice that have a deletion in all tissues, including the master lipid metabolism regulating the organs such as the liver, it is likely causing the slightly different metabolic phenotype compared with hematopoietic Chr4^{Δ70kb/Δ70kb} mice. Moreover, lipid metabolism is essential for the inflammatory response, and both enhanced pro-inflammatory stimuli and low-grade inflammation are known to promote liver de novo lipogenesis. Therefore, the elevated plasma lipid levels in hematopoietic Chr4^{Δ70kb/Δ70kb} mice are most likely a secondary effect of increased pro-inflammatory activity (25). In addition, hematopoietic Chr4^{Δ70kb/Δ70kb} surprisingly increased the expression of the master regulator of fatty acid synthesis, Fasn, which possibly reflected elevated plasma triglyceride and cholesterol levels but did not significantly increase the expression of Srebf s in the liver of Ldlr^−/−ApoB^100/100 mice. These mice also showed other metabolic changes, like increased percentual body weight gain, but the body weight did not differ between the groups in the end after 12 weeks of HFD. Plasma fasting glucose levels were also elevated in hematopoietic Chr4^{Δ70kb/Δ70kb} mice, but the difference did not reach statistical significance. It is notable that the systemic Chr4^{Δ70kb/Δ70kb} mice also had higher plasma fasting glucose levels compared with wild-type mice in Ldlr^−/−ApoB^100/100 background on a standard laboratory diet. Nonetheless, they did not show any difference in the development of atherosclerosis on the standard diet, and this possible metabolic change was not observed between genotypes after 6 or 12 weeks of HFD, possibly due to massive lipid overload and obesity in both controls and Chr4^{Δ70kb/Δ70kb} mice. These features together indicate a metabolic phenotype by hematopoietic Chr4^{Δ70kb/Δ70kb} in hypercholesterolemic mice, but further studies are required to clarify the mechanism and whether it is associated with the activation of the immune response.

Since there are two major tumor suppressor genes, CDKN2A and CDKN2B, neighboring the CAD risk interval, the atherogenic effects of the risk locus mediated by ANRIL regulating the expression of these genes were observed. The monocyte-specific Cdkn2a deficiency has been previously reported to lead to accelerated atherosclerosis, induced pro-inflammatory gene expression, and increased proliferation of macrophages in mice (26). In our study, the pro-inflammatory and pro-atherogenic stimuli were not mediated via reduced expression of Cdkn2a or Cdkn2b as the expressions of Cdkn2a and Cdkn2b were not changed by Chr4^{Δ70kb/Δ70kb} in mouse spleen or BMDMs in response to oxLDL. Instead, the basal expression of Cdkn2a was increased, and consistently, the proliferation rate decreased in Chr4^{Δ70kb/Δ70kb} BMDMs. Parallel to our study, Holdt et al. reported that circANRIL acts independently of CDKN2A/B, cis-regulation of the locus, and miRNA sponging, and they also demonstrated that linANRIL is unaffected by circANRIL expression (6).

The pro-atherogenic and pro-inflammatory effects of Chr4^{Δ70kb/Δ70kb} observed in this study can be derived either or both from the lack of the potentially protective murine circular Ak148321 isoforms or from the disruption of the full-length linear isoform. However, the study was not able to specify the role of different murine or human isoforms in atherogenesis and macrophage polarization. In addition, the role of different ANRIL and Ak148321 isoforms in NF-κB-signaling in inflammatory cells and the question of how other cell types in the arterial wall are affected by ANRIL and Ak148321 isoforms in response to pro-atherogenic stimuli require further studies.

To the best of our knowledge, our study demonstrates for the first time the hematopoietic cell-specific effect of the murine ortholog of the human 9p21.3 CAD risk locus and its murine circular isoforms on atherosclerosis in hypercholesterolemic mice. Even though in humans, the pro-atherogenic effects of 9p21.3 are driven by the risk SNPs rather than a full deletion in the locus, the Chr4^{Δ70kb/Δ70kb}Ldlr^−/−ApoB^100/100 mouse model represents similar pro-atherogenic and pro-inflammatory consequences as seen in humans. The pro-atherogenic and pro-inflammatory phenotype of hematopoietic knockout, along with the single-cell sequencing data from both human and mouse models, tracks down the mechanism for the regulation of inflammatory cell functions.