Skip main navigation

Single-Cell Transcriptome Analysis of the Circle of Willis in a Mouse Cerebral Aneurysm Model

Originally published 2022;53:2647–2657



The circle of Willis (CoW) is the most common location for aneurysms to form in humans. Although the major cell types of the intracranial vasculature are well known, the heterogeneity and relative contributions of the different cells in healthy and aneurysmal vessels have not been well characterized. Here, we present the first comprehensive analysis of the lineage heterogeneity and altered transcriptomic profiles of vascular cells from healthy and aneurysmal mouse CoW using single-cell RNA sequencing.


Cerebral aneurysms (CAs) were induced in adult male mice using an elastase model. Single-cell RNA sequencing was then performed on CoW samples obtained from animals that either had aneurysms form or rupture 14 days post-induction. Sham-operated animals served as controls.


Unbiased clustering analysis of the transcriptional profiles from >3900 CoW cells identified 19 clusters representing ten cell lineages: vascular smooth muscle cells, endothelial cells fibroblasts, pericytes and immune cells (macrophages, T and B lymphocytes, dendritic cells, mast cells, and neutrophils). The 5 vascular smooth muscle cell subpopulations had distinct transcriptional profiles and were classified as proliferative, stress-induced senescent, quiescent, inflammatory-like, or hyperproliferative. The transcriptional signature of the metabolic pathways of ATP generation was found to be downregulated in 2 major vascular smooth muscle cell clusters when CA was induced. Aneurysm induction led to significant expansion of the total macrophage population, and this expansion was further increased with rupture. Both inflammatory and resolution-phase macrophages were identified, and a massive spike of neutrophils was seen with CA rupture. Additionally, the neutrophil-to-lymphocyte ratio (NLR), which originated from CA induction mirrored what happens in humans.


Our data identify CA disease-relevant transcriptional signatures of vascular cells in the CoW and is searchable via a web-based R/shiny interface.

Cerebral aneurysms (CAs) are common, with a prevalence of 2% to 3% in the general population,1,2 although exceeding 19% in high-risk populations.3 Damage to the endothelial layer is observed both in humans and in animal models of CA followed by vascular remodeling.4,5 Lineage tracing system studies have enabled the unbiased fate mapping and identification of vascular cells during the progression of cardiovascular diseases.6-8 Although the major cell types in the intracranial vasculature are well known,9,10 the heterogeneity and relative contributions of different cells in healthy and aneurysmal vessels are poorly understood. Single-cell RNA sequencing (scRNAseq) provides an opportunity to capture cell-specific changes making it a powerful tool capable of uncovering regulatory relationships between genes and complex cell populations.7,11,12 Here, we present the first comprehensive analysis of the lineage heterogeneity and altered transcriptomic profiles of vascular cells from healthy and aneurysmal mouse circles of Willis (CoW) using scRNAseq. By cluster analysis of the CoW cells, we identified 19 clusters and 10 distinct cell types. We also report the changes of cellular subpopulations, fractions, and transcriptomic profiles in the vessels, expanding established concepts to bridge the gap from basic science to clinical intervention.


For a comprehensive description of the methods see Supplemental Material.

Data Availability

The scRNAseq data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under the accession code GSE193533. Other data supporting the findings of this study are available from the corresponding author upon reasonable request.

A web-based R/shiny app ( was generated.13

Mouse Model of Elastase-Induced CA

Animal research was performed in accordance with Tulane University IACUC. Fifty-eight-week-old male C57/BL6 mice were obtained (Jackson Laboratories). To induce aneurysms, we used a well-established model that involves pharmacological hypertension and stereotactic injection of elastase.14 The mice underwent a unilateral nephrectomy followed by implantation of a deoxycorticosterone acetate pellet in the subcutaneous tissue 1 week later. The mice were given 1% NaCl in their drinking water, and 17.5mU of elastase was injected into the basal cistern.

CoW Dissociation and Single-Cell Preparation

Preparation of the single-cell suspension was performed following a previously described protocol15 with modifications. Cerebral blood vessels from the CoW were harvested, and the connective and arachnoid tissue removed. The CoW pooled from 5 mice in each group were digested with an enzyme solution. The cell suspension was strained, washed, and resuspended in opti-MEM with 10% FBS.

Single-Cell RNA-Sequencing Data Analysis

The cell suspensions underwent scRNAseq using standard 10X Chromium Single Cell Chemistry V3. Raw sequencing data were processed using Cell Ranger to produce gene-level counts for each cell in each sample. The CellRanger mkfastq command was used to generate Fastq files. Data were mapped to a prebuild mouse reference genome. All subsequent analysis was performed using R.

CoW Whole Mount Immunofluorescence and Confocal Microscopy

Staining of surgically isolated CoW arteries from 3 mice per experimental group was performed following a previously described protocol with modifications.16 The vessels were fixed in 4% paraformaldehyde and dehydrated in graded methanol overnight at −20 °C. Tissues were then rehydrated, washed, and blocked with SEA BLOCK Blocking Buffer. They were incubated overnight at 4 °C with primary antibodies, anti-Galectin 3 and anti-αSMA. Secondary antibodies, goat anti-rat Alexa 647 and goat anti-mouse Alexa 488, were used. DAPI was used as the nuclear counterstain. Confocal images were captured using a Nikon TiE-2 Inverted Research Microscope.

To quantify VSMCs, 5 confocal images per animal were used and uploaded into ImageJ. Vessels from 3 mice per group (shams, CA 1 day and CA 1 week) were analyzed according to a previously published method.17

Statistical Analysis

In experiments other than scRNAseq, analysis was performed using GraphPad Prism 7.0 Software. All data were tested for normality and similar variance.


Single-Cell RNA Sequencing Revealed 19 Clusters Representing 10-Cell Lineages

A total of 4544, 5241, and 3975 of qualified cells were obtained from sham, formed and ruptured, respectively (Figure S1) and a total of 19 clusters were identified (Figure 1B through 1F). Cell type-specific canonical markers were used to distinguish and identify 10 distinct cell lineages (Figure 1E and 1F): (1) VSMCs; (2) monocytes and macrophages (Mo/MΦ); (3) neutrophils; (4) fibroblasts; (5) endothelial cells; (6) T cells; (7) pericytes; (8) B lymphocytes; (9) dendritic cells (DC); and (10) mast cells (Figure 1E and 1F). Although the expression of natural killer (NK) cell marker genes was also detected in Cluster 9, no NK cells were singled out, and this cluster showed the highest percentage of expression of the T-cell markers Cd3d and Cd3g. Interestingly, a small population of erythrocyte cells were singled out in clusters 8 and 16 that were identified as being VSMC and Mo/MΦ, respectively (Figure S2). We also unbiasedly identified the top 10 marker genes (sorted by P) for each cluster, relative to all other clusters (Figure 2A). Distinct expression patterns across all clusters are listed in Table S1.

Figure 1.

Figure 1. Identification of cell clusters present in the mouse circle of Willis (CoW) by single-cell RNA sequencing (scRNAseq). A, Schematic diagram indicating the procedure for scRNAseq. B, UMAP plot of aggregate CoW cells with colors denoting different clusters. C, UMAP plot of cell clusters in CoW cells across the indicated conditions: Sham, formed CA 14 d after induction and ruptured CA 14 d after induction. D, Circos plot displaying the linkage between the cells from the different treatment groups to individual clusters. The clusters marked by arrows have asymmetrical cellular contributions (yellow, blue, and red indicate vascular smooth muscle cell [VSMC] clusters, neutrophils, and fibroblasts, respectively). E, Clusters and major cell types of correspondence. F, Dot plot of selected marker genes for each cluster and lineage in aggregate CoW. DC indicates dendritic cells; DOCA, deoxycorticosterone acetate; EC, endothelial cells; and MC, mast cells.

Figure 2.

Figure 2. Categorization of mouse circle of Willis (CoW) cell clusters and populations. A, Heatmap of the top 10 highly expressed genes for each cluster across conditions (Sham, Formed, and Ruptured) in aggregate CoW cells. B, Fraction of each cell type in CoW cells across conditions; (C) cell population percentages across conditions. D, Cell percentages of separate cell types in relation to the total number of cells across conditions: vascular smooth muscle cell (VSMC), fibroblasts, EC, immune cells (Mo/MΦ, neutrophils, T, B, MC).

After determining cell populations and clusters, we created an online platform for a simpler and interactive visualization of expression patterns ( This tool allows any user to choose gene(s) of interest, provided it is expressed in our dataset, and view expression values in a heatmap, violin, or t-SNE plot.

Decreased VSMCs and Expansion of Monocyte/Macrophages in the Formation Process of CA

Among the cell types, VSMCs and Mo/MΦs were clustered into multiple subpopulations: 5 VSMC and 6 Mo/MΦ (Figure 1E). At the single-cell level, comparison of healthy CoW vessels with those that had CAs that either formed or ruptured demonstrated a progressive loss of VSMCs and expansion of immune cell populations (Figure 1C and 1D [yellow arrow] and Figure 2B through 2D). Consistent with pathological changes seen in CAs, the VSMC populations decreased18,19 relative to sham (Figure 2D). Macrophages, however, increased when CAs formed and ruptured when compared with sham (Figure 2D). Endothelial cells and fibroblasts represented <15% of the total number of cells. CA induction also influenced the proliferation of fibroblasts (sham 3% versus formed 11%), which contributed to the growth/formation of aneurysmal tissue (Figure 1C and 1D [red arrow] and Figure 2B through 2D). Among the immune cells, neutrophils and DCs also showed expansion during CA progression and rupture (Figure 1C and 1D, blue arrow).

Qualitative confocal microscopy demonstrated that 1 week after aneurysm induction, the macrophage marker, Galectin-3, was highly expressed (Figure 3C). Galectin-3 is a β-galactoside-binding lectin that serves as a substrate for metalloproteinases amplifying the inflammatory response.20 Semiquantitative analysis of vessels confirmed a significant loss of VSMCs after 1 week following induction (Figure 3D). These results corroborate the scRNAseq findings and reiterate that loss of mural cells acts as a driving force for aneurysmal growth and leads to an increased inflammatory reaction, severe wall degeneration, and eventual rupture.18

Figure 3.

Figure 3. Whole mount immunolabeling of circle of Willis vessels. A, Sham, (B) 1 d after elastase cerebral aneurysm (CA) induction, (C) 7 d after elastase CA induction and (D) quantification of VSMC as a percentage of total number of cells measured. Vascular smooth muscle cells (αSMA+) are stained in green, macrophages (Galectin-3+) are stained in pink, and nuclei staining by DAPI are indicated in blue. For the quantification, data are presented as mean±SEM. One-way ANOVA followed by Tukey post hoc analysis were performed. Scale bar from representative images=50 μm.

VSMC Transcriptomes Reflect Diversity of Phenotypes in Response to CA Induction

Five distinct VSMC clusters were identified based on expression of specific markers (Figure 1B and 1C [yellow arrow] and Figure 1E). Clusters 0, 1, and 5 accounted for 90% of the VSMCs in the sham group, and clusters 8 and 11 represented 10% (Figure 4A).

Figure 4.

Figure 4. Comparison of vascular smooth muscle cell (VSMC) subpopulations in circle of Willis across conditions. A, Percentages of the VSMC subpopulations across conditions. B and C, Heatmap showing relative gene expression levels (columns) of the indicated metabolic pathways in VSMC subpopulations and across conditions.

Each cluster had a different profile and classification was determined by analyzing the differential expression in 1 cluster versus all the others (Table S1). The identity of each cluster was based on biological process (BP) and molecular function obtained from Gene Ontology and KEGG pathway information (Figure S4A). Cluster 0 was identified as being proliferative with differentially expressed genes (DEGs) significantly enriched in pathways including ECM (extracellular matrix)-receptor interaction and focal adhesion which are strongly correlated with CA progression.21 Gene ontology term analysis for BP indicated significant enrichment in cell adhesion, angiogenesis and positive regulation of cell migration. Cluster 1 was identified as stress-induced senescent as DEGs were associated with TNF-α signaling and oxidative phosphorylation (OXPHOS). BP terms indicated significant enrichment in Notch signaling pathway, cell adhesion, and response to hypoxia. Furthermore, this cluster highly expressed genes that contribute to VSMC dedifferentiation (Klf4, Atf3, Jun, and Fos). Cluster 5 overexpressed genes related to blood vessel integrity and suppression of cell cycle progression (Rasl11a, Mt1, Gper1, Tsc22d1, Errfi1, and Rrad) and was identified as quiescent. Pathway analysis revealed changes in cGMP-PKG and p53 signaling, while BP terms included ATP metabolic process, ATP synthesis coupled proton transport, and actin cytoskeleton organization indicating these cells are gearing towards normal vascular development and homeostasis. Cluster 8 showed higher expression of pro-inflammatory factors (Il1b, Cxcl2, and Ccrl2) and was termed inflammatory-like VSMCs. The top 3 BP terms identified were angiogenesis, inflammatory response, and neutrophil chemotaxis. We also observed a well-defined cluster of erythrocytes in the ruptured sample that did not overlap with VSMCs (Figure S3, green arrow). Last, we explored how ATP metabolic processes related to the phenotypic transformations and found that cluster 11 exhibited the highest metabolic rate with increased expression of genes involved in the tricarboxylic acid cycle (TCA) and OXPHOS pathways (Figure 4B). It also presented increased expression of Fblx22 and Mob2, which are involved in the maintenance of muscle fiber structure and cell cycle progression, respectively.22,23 The BP terms identified for this cluster included oxidation-reduction process, Notch signaling pathway and ATP biosynthetic process. Thus, cluster 11 was classified as synthetic hyperproliferative.

Next, we evaluated mitochondrial function as a metabolic feature that controls VSMC phenotype. We found an important downregulation of genes related to NADH complex assembly and ATP metabolic processes in the formed and ruptured samples from clusters 0 and 1 (Figure 4B and 4C). These 2 clusters also exhibited increased expression of genes involved in cholesterol biosynthesis when CA was induced and subsequently ruptured (Figure S4B). Suppression of mtDNA transcription results in gradual loss of the genes necessary for OXPHOS, ATP production, and energy-dependent functions, such as cell contractility.24 Since clusters 0 and 1 accounted for most of the VSMC population, we suggest that mitochondrial dysfunction plays a central role in the phenotypic modulation of such cells.

Macrophage Diversity in CA Progression

Six Mo/MΦ clusters were singled out (Figure 1E and 1F) and KEGG pathway analysis found DEGs significantly enriched in pathways related to inflammatory response and processes involved in the development or functioning of the immune system. In the sham group, clusters 2 and 3 corresponded to 43.6% and 34.4% of Mo/MΦ cells, respectively. The remaining clusters accounted for 22% of the cells (Figure 5A). Similar to one of the VSMC clusters, Mo/MΦ cluster 16 co-localized with a small population of erythrocyte cells (Figure S5). Of note, although cluster 12 weakly expressed the canonical markers, Cd68, C1qa, and C1qb, it was identified as macrophage. This classification was based on the uniquely expressed transcripts Ly6c2, Chil3, and Plac8 (Figure 2A), which are indicative of infiltrating macrophages.25 Cluster 2 showed high expression of complement and pro-inflammatory cytokine genes, which are involved in recruitment of new inflammatory cells. This cluster also showed enrichment of Apoe, Wfdc17, Pf4, which act as counter-regulators of the pro-inflammatory response indicating differentiation of monocytes into macrophages. Cluster 3 had a similar pattern but was also enriched for antigen presentation/MHC-II (Cd74, H2-Aa, H2-Ab1, H2-Eb1) so cells from this cluster were identified as resolution-phase macrophages. Cluster 10 revealed an overrepresentation of genes related to ribosomal and mitochondrial proteins (Table S1). This cluster was also the most metabolically active compared with all other Mo/MΦ populations. It unambiguously demonstrated high levels of fatty acid β-oxidation and OXPHOS (Figure 6C and 6D) and modest expression of chemokines and cytokines (Figure 5B and 5C) indicating these cells are M2-like macrophages. Last, cluster 18 showed a proliferative signature with high expression of genes associated with proliferation, microtubule activity, apoptosis inhibition, and transcription regulation. Importantly, we found an overlap in gene expression in the Mo/MΦ clusters and DC cluster 15 with DCs expressing significantly higher levels of Ifitm1.

Figure 5.

Figure 5. Comparison of Mo/MΦ subpopulations in CoW across conditions. A, Percentages of the Mo/MΦ subpopulations across conditions. B, The inflammation-associated genes belonging to CCL and CXCL expressed by each Mo/MΦ subpopulations with color denoting experimental conditions. C, Heatmap showing gene expression levels of the cytokines and chemokines (columns) for each cell cluster in aggregate CoW cells. D, Cytokine-cytokine receptor interaction and (E) chemokine signaling pathway GO enrichment analysis of Mo/MΦ in the different treatment. KEGG indicates Kyoto Encyclopedia of Genes and Genomes.

CA-induction prompted significant expansion of the total macrophage population compared with the sham group, and the expansion was further witnessed in the ruptured sample (Figure 2B). Intriguingly, clusters 2, 3, 16 and 18 exhibited co-expression of inflammatory genes and the anti-inflammatory gene, Il10 (Figure 5B), which implies a potential compensation to inhibit an uncontrolled pro-inflammatory response to reduce tissue damage.

Immune Response Contributions to Formation and Rupture of CA

Neutrophils are associated with significant increases in rupture risk of CAs. We found an accumulation of neutrophils with induction of CA as indicated by the increased percentages of these cells in formed (2.9 %) and ruptured (25.7%) samples compared with sham (0.6%; Figure 1C [blue arrow] and Figure 2C). Pathway analysis identified TNF-α signaling and NF-κB as being the most significantly enriched (Table S2; Figure S7A). They were characterized by enhanced collagenolytic activity (Figure S7B). Furthermore, we characterized 2 lymphocyte clusters, cluster 9 (T cells) and cluster 14 (B cells; Table S2). The ribosome pathway was the most significantly enriched in both lymphocyte clusters indicating increased translational capacity during cell activation.

Dynamic changes were also observed in neutrophil-to-lymphocyte ratios (NLR). A higher NLR was associated with an unfavorable outcome (sham=0.12, formed=1.19, and ruptured=4.08).

Pathway Enrichment Profile in VSMC and Mo/MΦ

Heatmap analysis compared expression of metabolic genes from glycolysis, tricarboxylic acid cycle, and fatty acid β-oxidation pathways. The overview of the heatmaps indicated that all 3 pathways were highly upregulated in VSMC clusters (0,1,5,11), fibroblasts (6), and pericytes (13; Figure S6). To gain mechanistic insight into how pathways influence VSMC phenotypic plasticity, we used Normalized Enrichment Score (NES) to determine the relative importance of individual pathways in different treatment groups.26 Scores from all 3 metabolic pathways were significantly higher in the sample from animals that formed CA compared with sham. Conversely, NES from ruptured CA were significantly lower when compared with sham (Figure 6A through 6C). This suggests that VSMCs use both glycolysis and tricarboxylic acid cycle for energy production during proliferation and dedifferentiation, which goes against the idea of a metabolic switch, where one or the other is active. Further work is needed to resolve this discrepancy.

Figure 6.

Figure 6. Gene ontology (GO) enrichment analysis of vascular smooth muscle cells. NES indicate the distribution of GO categories across a list of genes ranked by hypergeometrical score (HGS). Higher enrichment scores indicate a shift of genes belonging to certain GO representing more upregulation. A, Glycolysis/gluconeogenesis, (B) tricarboxylic acid cycle, (C) fatty acid metabolism, (D) regulation of actin cytoskeleton, (E) cell cycle, and (F) mTOR signaling pathway. mTOR indicates mammalian target of rapamycin.

mTOR (mammalian target of rapamycin) exhibits a dual role as both an upstream activator of Akt and the downstream effector of the PI3K/Akt/mTOR pathway influencing VSMC growth, differentiation, and metabolism by regulating protein synthesis and mRNA translation.27 We calculated the statistical significance of the NES for gene sets from mTOR signaling, regulation of actin cytoskeleton, and cell cycle in VSMCs. Results highlighted a significant activation of all 3 pathways in formed compared with sham. NES from ruptured CA were significantly lower when compared with formed (Figure 6D through 6F). These results suggest that increased VSMC loss outweighs their proliferation. This leads to a loss of structural integrity in the vessel and subsequent rupture.

To assess cellular sources of inflammation, we looked at a heatmap of the dynamic expression of the core genes involved in cytokine and chemokine pathways (Figure 5C). We found genes from these pathways were upregulated in clusters 2, 3, 12, 16, 18 indicating macrophages are the major source of vascular inflammation. Enrichment analysis using NES showed significant activation of cytokines and chemokines when CAs formed and even more so when they ruptured (Figure 5D and 5E).


The CoW acts to provide collateral blood flow between the anterior and posterior circulations of the brain and is the most common location for aneurysms to form.28 We identified 19 clusters and 10 distinct cell lineages within the CoW and delineated the transcriptional profiles of VSMC and Mo/MΦ in the sham, formed and ruptured tissue samples using scRNAseq. Our results showed that metabolic heterogeneity exists between and within cell types of the CoW and that VSMC are the most metabolically active cells (Figures 4B, 6A through 6C; Figure S6). Moreover, assessment of the dynamic immune response provided evidence that Mo/MΦ are the major contributors to the inflammatory state (Figure 5B through 5E). Our data were very similar to those gained from transcriptomic profiles of vascular cells from healthy and elastase-induced aneurysmal mouse infrarenal abdominal aortas using scRNAseq.29

We isolated 5 distinct VSMC subpopulations within the CoW based on previous studies on heterogeneity using lineage tracing and scRNAseq approaches.11,30 Clusters 0 and 11 were classified as synthetic and increased their proportional numbers in formed CA compared with sham (Figure 4A). Cluster 1, stress-induced senescent, pins down a crosstalk between Notch signaling and hypoxia that leads to vascular differentiation.31 Senescent VSMCs had increased expression of the transcription factor, HIF1-α (hypoxia-inducible factor), in formed and ruptured samples (data not shown) which enhances Notch-dependent activation, providing a mechanism by which hypoxia can regulate the differentiation status of a cell.32 Cluster 5, quiescent, maintained the cell number proportions in sham compared with formed. This cluster also overexpressed genes that are necessary for mature vessels to maintain physiological homeostasis/differentiated status. Of note, a small population of RBCs co-localized with VSMCs in cluster 8. This may indicate contamination as we did not perfuse the animals with PBS, nor depleted RBCs from samples due to the limited number of total cells and to avoid extra stress-related RNA disturbances. The cells from the ruptured cluster 8 had a specific population of RBCs as a result of SAH which did not express VSMC markers (Figure S3, green arrow).

We attempted to elucidate mechanistic insights into the dysregulated metabolism of VSMC in CA. We show that the polarization of VSMCs toward a synthetic state, that is, when CAs formed, is dependent on metabolic rewiring where cells acquire sufficient nutrients such as glucose, amino acids, lipids and nucleotides that are necessary to support cell growth, and manage the redox.33 Interestingly, the tricarboxylic acid cycle is significantly increased and not mutually exclusive with glycolysis as routes for energy production, suggesting it is equally required for proliferation and survival. Conversely, we saw a significant reduction in metabolic activity following rupture, which may have contributed to a derangement of oxidative metabolism, failure of oxygen utilization, and secondary injury (Figure 6). This suggests that there is a fundamental link between CA induction and mitochondrial dysfunction, which may cause further cellular dysregulation and result in chronic organelle damage and the activation of apoptotic pathways.

We confirm that VSMCs are pleiotropic, expressing unique permutations and combinations of both contractile and synthetic genes that are continuously modulated in response to elastin degradation and hypertension.32,34 The phenotypic switch following injury is accompanied by modulation of transforming growth factor (TGF)-β, mTOR, and genes associated with actin skeleton, oxygen and metabolic homeostasis, cell growth and apoptosis (Figures 6; Figure S8).

During CA progression, 1 dramatic change in the adventitia is Mo/MΦ infiltration in the aneurysmal tissue.35–39 Mo/MΦ clusters represented the second largest population within the vascular wall and were involved in the amplification of the local inflammatory response. Despite the well-recognized polarized states of M1 and M2 macrophages, we were not able to distinguish this polarization in vivo. In fact, the 6 subpopulations exhibited dual expression of pro-inflammatory genes and anti-inflammatory genes, albeit at varying levels. This discrepancy is likely due to the prototypical states being mainly achieved in controlled in vitro cultures.7

The inflammatory Mo/MΦs from cluster 2 highly expressed pro-inflammatory factors while the subpopulation from cluster 3 appeared to be involved in the resolution of the CA-induction associated acute injury. Similar to in vitro findings, Mo/MΦ from cluster 3 were specifically enriched for the biochemical machinery necessary for antigen processing and presentation, and secretion of T- and B-cell chemoattractants.40 Interestingly, Mo/MΦ from cluster 10 expressed high levels of genes for fatty acid β-oxidation and OXPHOS but low levels of inflammatory makers (Figure 5C and 5D; Figure S6). The key metabolic signature of alternatively activated macrophages is the consumption of fatty acids and increase in the mitochondrial respiratory capacity.41,42 Thus, cluster 10 was identified as M2-like. Cluster 18 was identified as proliferative, consistent with their known self-renewing properties. During CA formation, these cells continue to proliferate but their relative proportion decreases after rupture suggesting that this local proliferation could be an alternative mechanism of inflammation that allows these macrophages to accumulate in sufficient numbers to perform critical functions, such as wound repair, in the absence of potentially damaging cell recruitment.43

Previously, our group has identified lymphocytes as a contributing factor to CA formation and rupture.44 We identified the presence of T and B lymphocytes, DCs, and pericytes in CA vessels at the single-cell level, indicating that antigen-dependent activation of naïve T cells and adaptive immune response should be involved in the elastase-induced CA progression. However, we did not focus on the differences of lymphocytes subpopulations and their contributions to aneurysms as our goal was to characterize the 2 main cell types involved in CA progression. We identified neutrophils and found an important pathogenic role for these cells in CA rupture. The enormous spike in a proportional number of neutrophils in the ruptured sample confirms its pivotal role in the maintenance and exacerbation of inflammatory responses, facilitating the degenerative changes in lesions.45 Additionally, the NLR ratio was substantially different in the tissue sample from unruptured compared with the ruptured, mirroring what happens in humans with CA. Patients with ruptured CAs that had elevated baseline NLR levels have been associated with poor postoperative functional outcomes.46 Our findings support the potential prognostic value of NLR in patients with CA.

Our study uncovered the hallmarks of CoW cell population heterogeneity in an unbiased manner; however, certain limitations must be acknowledged. While scRNAseq permits simultaneous characterization of every cell within the CoW, this data provides a limited view of the true functional changes in CA pathogenesis that are undoubtedly affected by cellular processes downstream of gene transcription. We included the entirety of the CoW to ensure adequate cell yield for scRNAseq, limiting the specificity of our findings since ideally, we would have isolated only the aneurysm dome. Another issue is the lack of a time-course profile, as 2 weeks after elastase injection effectively marks the study’s end point of complete aneurysm formation or rupture. Other metabolomic tools will be able to further dissect and distinguish the relationship between CoW metabolic dysfunction and CA formation, potentially determining the severity of brain injury and predicting the pathological progression and outcomes after SAH.

Overall, our findings provide novel insights into the function and regulation of CA onset and progression and raise important questions for future studies, including defining the mechanisms regulating metabolic phenotypes of VSMCs and whether these changes are adaptive or deleterious in the context of CA. The availability of our ready to use web-based shiny app will facilitate further exploration of these data by a full spectrum of researchers and can generate information for future projects.

Article Information


The authors would like to thank Dr Jay K. Kolls and Kejing Song of the Sequencing Core at CTRII Tulane University for their assistance in performing the single-cell RNA sequencing.

Supplemental Material

Supplemental Methods

Tables S1–S2

Figures S1–S8

Nonstandard Abbreviations and Acronyms


biological process


cerebral aneurysm


Circle of Willis


extracellular matrix


hypoxia-inducible factor


normalized enrichment score


neutrophil-to-lymphocyte ratio


oxidative phosphorylation


single-cell RNA sequencing


tricarboxylic acid cycle


vascular smooth muscle cell

Disclosures None.


*A.N. Martinez and G.G. Tortelote contributed equally.

Supplemental Material is available at

For Sources of Funding and Disclosures, see page 2656.

Correspondence to: Aaron S. Dumont, MD, Department of Neurosurgery, Tulane Center for Clinical Neurosciences, Tulane University School of Medicine 131 South Robertson St, Ste 1300, New Orleans, LA 70112. Email


  • 1. Rinkel GJ. Natural history, epidemiology and screening of unruptured intracranial aneurysms.J Neuroradiol. 2008; 35:99–103. doi: 10.1016/j.neurad.2007.11.004CrossrefMedlineGoogle Scholar
  • 2. Komotar RJ, Zacharia BE, Mocco J, Connolly ES. Controversies in the surgical treatment of ruptured intracranial aneurysms: the First Annual J. Lawrence Pool Memorial Research Symposium--controversies in the management of cerebral aneurysms.Neurosurgery. 2008; 62:396–407. doi: 10.1227/01.neu.0000316006.26635.b0CrossrefGoogle Scholar
  • 3. Brown RD, Huston J, Hornung R, Foroud T, Kallmes DF, Kleindorfer D, Meissner I, Woo D, Sauerbeck L, Broderick J. Screening for brain aneurysm in the Familial Intracranial Aneurysm study: frequency and predictors of lesion detection.J Neurosurg. 2008; 108:1132–1138. doi: 10.3171/JNS/2008/108/6/1132CrossrefMedlineGoogle Scholar
  • 4. Chalouhi N, Ali MS, Jabbour PM, Tjoumakaris SI, Gonzalez LF, Rosenwasser RH, Koch WJ, Dumont AS. Biology of intracranial aneurysms: role of inflammation.J Cereb Blood Flow Metab. 2012; 32:1659–1676. doi: 10.1038/jcbfm.2012.84CrossrefMedlineGoogle Scholar
  • 5. Hashimoto T, Meng H, Young WL. Intracranial aneurysms: links among inflammation, hemodynamics and vascular remodeling.Neurol Res. 2006; 28:372–380. doi: 10.1179/016164106X14973CrossrefMedlineGoogle Scholar
  • 6. Biddy BA, Kong W, Kamimoto K, Guo C, Waye SE, Sun T, Morris SA. Single-cell mapping of lineage and identity in direct reprogramming.Nature. 2018; 564:219–224. doi: 10.1038/s41586-018-0744-4CrossrefMedlineGoogle Scholar
  • 7. Cochain C, Vafadarnejad E, Arampatzi P, Pelisek J, Winkels H, Ley K, Wolf D, Saliba AE, Zernecke A. Single-cell RNA-Seq reveals the transcriptional landscape and heterogeneity of aortic macrophages in murine atherosclerosis.Circ Res. 2018; 122:1661–1674. doi: 10.1161/CIRCRESAHA.117.312509LinkGoogle Scholar
  • 8. Feil S, Fehrenbacher B, Lukowski R, Essmann F, Schulze-Osthoff K, Schaller M, Feil R. Transdifferentiation of vascular smooth muscle cells to macrophage-like cells during atherogenesis.Circ Res. 2014; 115:662–667. doi: 10.1161/CIRCRESAHA.115.304634LinkGoogle Scholar
  • 9. Canham PB, Talman EA, Finlay HM, Dixon JG. Medial collagen organization in human arteries of the heart and brain by polarized light microscopy.Connect Tissue Res. 1991; 26:121–134. doi: 10.3109/03008209109152168CrossrefMedlineGoogle Scholar
  • 10. Walmsley JG, Campling MR, Chertkow HM. Interrelationships among wall structure, smooth muscle orientation, and contraction in human major cerebral arteries.Stroke. 1983; 14:781–790. doi: 10.1161/01.str.14.5.781LinkGoogle Scholar
  • 11. Dobnikar L, Taylor AL, Chappell J, Oldach P, Harman JL, Oerton E, Dzierzak E, Bennett MR, Spivakov M, Jørgensen HF. Publisher correction: disease-relevant transcriptional signatures identified in individual smooth muscle cells from healthy mouse vessels.Nat Commun. 2018; 9:5401. doi: 10.1038/s41467-018-07887-3CrossrefMedlineGoogle Scholar
  • 12. Li Y, Ren P, Dawson A, Vasquez HG, Ageedi W, Zhang C, Luo W, Chen R, Li Y, Kim S, et al. Single-cell transcriptome analysis reveals dynamic cell populations and differential gene expression patterns in control and aneurysmal human aortic tissue.Circulation. 2020; 142:1374–1388. doi: 10.1161/CIRCULATIONAHA.120.046528LinkGoogle Scholar
  • 13. Ouyang JF, Kamaraj US, Cao EY, Rackham OJL. ShinyCell: simple and sharable visualisation of single-cell gene expression data.Bioinformatics. 2021. doi: 10.1093/bioinformatics/btab209Google Scholar
  • 14. Nuki Y, Tsou TL, Kurihara C, Kanematsu M, Kanematsu Y, Hashimoto T. Elastase-induced intracranial aneurysms in hypertensive mice.Hypertension (Dallas, Tex: 1979). 2009; 54:1337–1344. doi: 10.1161/hypertensionaha.109.138297AbstractGoogle Scholar
  • 15. Martinez AN, Pascale CL, Amenta PS, Israilevich R, Dumont AS. Cell culture model to study cerebral aneurysm biology.Acta Neurochir Suppl. 2020; 127:29–34. doi: 10.1007/978-3-030-04615-6_5Google Scholar
  • 16. Kumar M, Tanwar P. Organ culture and whole mount immunofluorescence staining of mouse Wolffian ducts.J Vis Exp. 2017;119:55134. doi: 10.3791/55134Google Scholar
  • 17. Shihan MH, Novo SG, Le Marchand SJ, Wang Y, Duncan MK. A simple method for quantitating confocal fluorescent images.Biochem Biophys Rep. 2021; 25:100916. doi: 10.1016/j.bbrep.2021.100916Google Scholar
  • 18. Oka M, Shimo S, Ohno N, Imai H, Abekura Y, Koseki H, Miyata H, Shimizu K, Kushamae M, Ono I, et al. Dedifferentiation of smooth muscle cells in intracranial aneurysms and its potential contribution to the pathogenesis.Sci Rep. 2020; 10:8330. doi: 10.1038/s41598-020-65361-xGoogle Scholar
  • 19. Starke RM, Chalouhi N, Ding D, Raper DM, McKisic MS, Owens GK, Hasan DM, Medel R, Dumont AS. Vascular smooth muscle cells in cerebral aneurysm pathogenesis.Transl Stroke Res. 2014; 5:338–346. doi: 10.1007/s12975-013-0290-1CrossrefMedlineGoogle Scholar
  • 20. Di Gregoli K, Somerville M, Bianco R, Thomas AC, Frankow A, Newby AC, George SJ, Jackson CL, Johnson JL. Galectin-3 identifies a subset of macrophages with a potential beneficial role in atherosclerosis.Arterioscler Thromb Vasc Biol. 2020; 40:1491–1509. doi: 10.1161/ATVBAHA.120.314252LinkGoogle Scholar
  • 21. Liu P, Song Y, Zhou Y, Liu Y, Qiu T, An Q, Song J, Li P, Shi Y, Li S, et al. Cyclic mechanical stretch induced smooth muscle cell changes in cerebral aneurysm progress by reducing collagen type IV and collagen type VI levels.Cell Physiol Biochem. 2018; 45:1051–1060. doi: 10.1159/000487347Google Scholar
  • 22. Hughes DC, Baehr LM, Driscoll JR, Lynch SA, Waddell DS, Bodine SC. Identification and characterization of Fbxl22, a novel skeletal muscle atrophy-promoting E3 ubiquitin ligase.Am J Physiol Cell Physiol. 2020; 319:C700–C719. doi: 10.1152/ajpcell.00253.2020CrossrefGoogle Scholar
  • 23. Gomez V, Gundogdu R, Gomez M, Hoa L, Panchal N, O’Driscoll M, Hergovich A. Regulation of DNA damage responses and cell cycle progression by hMOB2.Cell Signal. 2015; 27:326–339. doi: 10.1016/j.cellsig.2014.11.016Google Scholar
  • 24. Liu YF, Zhu JJ, Yu Tian X, Liu H, Zhang T, Zhang YP, Xie SA, Zheng M, Kong W, Yao WJ, et al. Hypermethylation of mitochondrial DNA in vascular smooth muscle cells impairs cell contractility.Cell Death Dis. 2020; 11:35. doi: 10.1038/s41419-020-2240-7Google Scholar
  • 25. Zimmerman KA, Bentley MR, Lever JM, Li Z, Crossman DK, Song CJ, Liu S, Crowley MR, George JF, Mrug M, et al. Single-cell RNA sequencing identifies candidate renal resident macrophage gene expression signatures across species.J Am Soc Nephrol. 2019; 30:767–781. doi: 10.1681/ASN.2018090931CrossrefMedlineGoogle Scholar
  • 26. Patil I. Visualizations with statistical details: The ‘ggstatsplot’ approach.J Open Source Softw. 2021; 6:3167. doi: 10.21105/joss.03167Google Scholar
  • 27. Saxton RA, Sabatini DM. mTOR signaling in growth, metabolism, and disease.Cell. 2017; 169:361–371. doi: 10.1016/j.cell.2017.03.035CrossrefMedlineGoogle Scholar
  • 28. Vrselja Z, Brkic H, Mrdenovic S, Radic R, Curic G. Function of circle of Willis.J Cereb Blood Flow Metab. 2014; 34:578–584. doi: 10.1038/jcbfm.2014.7CrossrefMedlineGoogle Scholar
  • 29. Zhao G, Lu H, Chang Z, Zhao Y, Zhu T, Chang L, Guo Y, Garcia-Barrio MT, Chen YE, Zhang J. Single cell RNA sequencing reveals the cellular heterogeneity of aneurysmal infrarenal abdominal aorta.Cardiovasc Res. 2021;117:1402–1416. doi: 10.1093/cvr/cvaa214Google Scholar
  • 30. Liu M, Gomez D. Smooth muscle cell phenotypic diversity.Arterioscler Thromb Vasc Biol. 2019; 39:1715–1723. doi: 10.1161/ATVBAHA.119.312131LinkGoogle Scholar
  • 31. Gustafsson MV, Zheng X, Pereira T, Gradin K, Jin S, Lundkvist J, Ruas JL, Poellinger L, Lendahl U, Bondesson M. Hypoxia requires notch signaling to maintain the undifferentiated cell state.Dev Cell. 2005; 9:617–628. doi: 10.1016/j.devcel.2005.09.010CrossrefMedlineGoogle Scholar
  • 32. Morrow D, Guha S, Sweeney C, Birney Y, Walshe T, O’Brien C, Walls D, Redmond EM, Cahill PA. Notch and vascular smooth muscle cell phenotype.Circ Res. 2008; 103:1370–1382. doi: 10.1161/CIRCRESAHA.108.187534LinkGoogle Scholar
  • 33. Zhu J, Thompson CB. Metabolic regulation of cell growth and proliferation.Nat Rev Mol Cell Biol. 2019; 20:436–450. doi: 10.1038/s41580-019-0123-5CrossrefMedlineGoogle Scholar
  • 34. Yoshida T, Owens GK. Molecular determinants of vascular smooth muscle cell diversity.Circ Res. 2005; 96:280–291. doi: 10.1161/01.RES.0000155951.62152.2eLinkGoogle Scholar
  • 35. Ollikainen E, Tulamo R, Kaitainen S, Honkanen P, Lehti S, Liimatainen T, Hernesniemi J, Niemelä M, Kovanen PT, Frösen J. Macrophage infiltration in the saccular intracranial aneurysm wall as a response to locally lysed erythrocytes that promote degeneration.J Neuropathol Exp Neurol. 2018; 77:890–903. doi: 10.1093/jnen/nly068Google Scholar
  • 36. Kanematsu Y, Kanematsu M, Kurihara C, Tada Y, Tsou TL, van Rooijen N, Lawton MT, Young WL, Liang EI, Nuki Y, et al. Critical roles of macrophages in the formation of intracranial aneurysm.Stroke. 2011; 42:173–178. doi: 10.1161/STROKEAHA.110.590976LinkGoogle Scholar
  • 37. Yamashiro S, Uchikawa H, Yoshikawa M, Kuriwaki K, Hitoshi Y, Yoshida A, Komohara Y, Mukasa A. Histological analysis of infiltrating macrophages in the cerebral aneurysm walls.J Clin Neurosci. 2019; 67:204–209. doi: 10.1016/j.jocn.2019.05.027CrossrefGoogle Scholar
  • 38. Shao L, Qin X, Liu J, Jian Z, Xiong X, Liu R. Macrophage polarization in cerebral aneurysm: perspectives and potential targets.J Immunol Res. 2017; 2017:8160589. doi: 10.1155/2017/8160589Google Scholar
  • 39. He L, Vanlandewijck M, Mäe MA, Andrae J, Ando K, Del Gaudio F, Nahar K, Lebouvier T, Laviña B, Gouveia L, et al. Single-cell RNA sequencing of mouse brain and lung vascular and vessel-associated cell types.Sci Data. 2018; 5:180160. doi: 10.1038/sdata.2018.160CrossrefMedlineGoogle Scholar
  • 40. Stables MJ, Shah S, Camon EB, Lovering RC, Newson J, Bystrom J, Farrow S, Gilroy DW. Transcriptomic analyses of murine resolution-phase macrophages.Blood. 2011; 118:e192–e208. doi: 10.1182/blood-2011-04-345330Google Scholar
  • 41. Thapa B, Lee K. Metabolic influence on macrophage polarization and pathogenesis.BMB Rep. 2019; 52:360–372. doi: 10.5483/BMBRep.2019.52.6.140CrossrefGoogle Scholar
  • 42. Tavakoli S, Zamora D, Ullevig S, Asmis R. Bioenergetic profiles diverge during macrophage polarization: implications for the interpretation of 18F-FDG PET imaging of atherosclerosis.J Nucl Med. 2013; 54:1661–1667. doi: 10.2967/jnumed.112.119099CrossrefMedlineGoogle Scholar
  • 43. Jenkins SJ, Ruckerl D, Cook PC, Jones LH, Finkelman FD, van Rooijen N, MacDonald AS, Allen JE. Local macrophage proliferation, rather than recruitment from the blood, is a signature of TH2 inflammation.Science (New York, NY). 2011; 332:1284–1288. doi: 10.1126/science.1204351CrossrefMedlineGoogle Scholar
  • 44. Sawyer DM, Pace LA, Pascale CL, Kutchin AC, O’Neill BE, Starke RM, Dumont AS. Lymphocytes influence intracranial aneurysm formation and rupture: role of extracellular matrix remodeling and phenotypic modulation of vascular smooth muscle cells.J Neuroinflammation. 2016; 13:185. doi: 10.1186/s12974-016-0654-zGoogle Scholar
  • 45. Kushamae M, Miyata H, Shirai M, Shimizu K, Oka M, Koseki H, Abekura Y, Ono I, Nozaki K, Mizutani T, et al. Involvement of neutrophils in machineries underlying the rupture of intracranial aneurysms in rats.Sci Rep. 2020; 10:20004. doi: 10.1038/s41598-020-74594-9Google Scholar
  • 46. Cho A, Czech T, Wang WT, Dodier P, Reinprecht A, Bavinzski G. Peri-interventional behavior of the neutrophil to lymphocyte ratio in patients with intracranial aneurysms.World Neurosurg. 2020; 141:e223–e230. doi: 10.1016/j.wneu.2020.05.084CrossrefGoogle Scholar


eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.