DISSECTING THE HETEROGENEITY OF HUMAN THORACIC AORTIC ANEURYSMS USING SINGLE-CELL TRANSCRIPTOMICS

Abstract

Thoracic aortic aneurysm (TAA) is a life-threatening condition caused by weakening of the thoracic aorta wall, often developing silently until dissection or rupture occurs. Despite substantial efforts in the past decade, there have been no significant therapeutic advances to prevent or clinically manage diverse forms of TAA and dissection (TAAD) with the only effective treatment being surgical repair. There is an urgent need to understand intra- and inter-aneurysmal heterogeneity underlying TAAD pathogenesis. The human aortic wall consists of many cell types and exhibits significant regional heterogeneity. High throughput single-cell RNA sequencing (scRNA-seq) has emerged as the principal tool to reveal the complexity in human tissues and clinical specimens. Recent scRNA-seq studies of different aortic cell populations both in vivo and in vitro began to dissect this complexity and have provided valuable information. In this review, we summarize these findings and discuss the potential applications of single-cell transcriptomics and related high-content technologies in human TAAD research as well as the challenges associated with it.

Graphical Abstract

REGIONAL AND GENETIC HETEROGENEITY IN HUMAN AORTIC ANEURYSMS

Based on anatomic locations, human aortic aneurysms can be subdivided into 2 groups: TAA and abdominal aortic aneurysm (AAA).¹ TAAs can be further subdivided into aneurysms of aortic root, ascending aorta, aortic arch, and descending thoracic aorta (Figure 1). Despite their common phenotypic manifestations such as reduced aortic elasticity, there are key differences between TAA and AAA. AAA is more common than TAA and is significantly associated with atherosclerosis.² Other AAA risk factors include coronary heart disease, hypertension, smoking, and hyperlipidemia. In contrast, genetic predisposition plays a more dominant role in TAA, which has been associated with mutations in more than 20 genes.³ Strikingly, these mutations do not necessarily increase AAA risk suggesting their region-specific effect on human thoracic aorta. Syndromic TAAs, developing along with extravascular organ abnormalities, account for nearly 20% of cases. These conditions include Marfan syndrome (MFS), Loeys–Dietz syndrome (LDS), vascular Ehlers–Danlos syndrome (vEDS), and Shprintzen–Goldberg syndrome (SGS). LDS and MFS often cause aortic root aneurysm while vEDS causes aortic rupture with or without aneurysm (Figure 1).⁴ vEDS is associated with thoracoabdominal aneurysms with animal models showing descending thoracic aorta rupture.^5,6

Figure 1.

Human aortic tissue applications of scRNA-seq and related technologies to elucidate regional susceptibility to TAAD. The figures were created with BioRender.com; CPC: cardiovascular progenitor cell in the second heart field; NCSC: neural crest stem cell; ISH: in situ hybridization; scATAC-seq: single-cell assay for transposase-accessible chromatin using sequencing; scChIP-seq: single-cell chromatin immunoprecipitation sequencing; scRNA-seq: single-cell RNA sequencing; SMC: vascular smooth muscle cell.

Non-syndromic TAAs are the more common type but they can also show familial inheritance. SMC contractile apparatus genes such as ACTA2, MYH11, PRKG1, and MYLK have been associated with familial TAAs.⁷ However, these genes only explain a small portion of all familial TAA cases suggesting a heterogeneous genetic etiology for familial TAA. A significant portion of sporadic TAAs are associated with bicuspid aortic valve (BAV), the most common cardiovascular malformation affecting 1–2% of all adults. More than 50% of BAV patients develop aortic aneurysms with a higher incidence in ascending aorta and they are at a higher risk of aortic dissection (Figure 1).⁸ Importantly, BAV-associated TAAs have been linked to both hemodynamic changes resulting from aortic valve malformation as well as genetic risk factors including NOTCH1 and ROBO4 mutations.^9–11

The human aorta contains many cell types including vascular smooth muscle cells (SMCs), fibroblasts, endothelial cells (ECs), and immune cells. SMCs play essential roles in maintaining aortic wall integrity. Aberrant SMCs are a hallmark of aortic aneurysms making SMCs the primary focus of TAA research. The disease initiation and propagation have been associated with SMC loss and phenotype switching which causes contractile SMCs to adopt a proliferative dedifferentiated state.^12–14 Importantly, aortic SMCs have heterogeneous embryonic origins depending on their spatial location, and SMC gene expression and functional differences have been attributed to their developmental origins.^15,16 Regional susceptibility to aortic aneurysms is likely caused by lineage-specific defects of SMCs, which retain their differential disease susceptibility even in ex vivo studies.¹⁷

A BRIEF HISTORY OF SINGLE-CELL RNA-SEQUENCING

Rapid advances in gene expression profiling in the past decade led to the development of single-cell RNA-sequencing (scRNA-seq), which enables transcriptome-wide gene expression measurements at single-cell resolution. The scalability of the first generation scRNA-seq approaches was limited to less than a hundred cells. However, vast cellular diversity in healthy and diseased tissues necessitates scRNA-seq of thousands of cells to fully capture this complexity and to reach the statistical power to identify cell populations with unique molecular profiles. This motivated the development of high-throughput scRNA-seq approaches utilizing scalable microfluidics. Briefly, these methods rely on co-encapsulating individual cells with mRNA capture beads in droplets or microwells and subsequent parallel incorporation of cell barcodes into cDNA from thousands of cells.^18–21

In recent years, scRNA-seq has emerged as an unbiased way of profiling the molecular signatures of individual cells to identify cell types, states, and the molecular pathways maintaining these states. The initial applications of scRNA-seq on clinical samples have mainly focused on cancer and significantly advanced the knowledge in tumor heterogeneity, providing insights into therapeutic targeting and precision oncology.^22,23 ScRNA-seq has also been a key tool in revealing the developmental trajectories in various developmental systems including the human heart by describing dynamic changes in gene expression programs.^24,25 The application of single-cell transcriptomics in human aneurysm research is at an early stage compared to the other systems where stable references such as cell atlases are established and cross-examined by separate research groups.

INSIGHTS FROM SCRNA-SEQ STUDIES OF ANIMAL MODELS

Due to difficulties in acquiring and processing human aortic aneurysmal tissue, scRNA-seq studies of the aorta were primarily conducted on tissue samples collected from animal models. Using scRNA-seq, Dobnikar et al. discovered a rare SMC population expressing progenitor cell marker Stem cell antigen 1 (Sca1), which plays a critical role in SMC phenotype switching.²⁶ ScRNA-seq profiling of aortic root/ascending tissue from MFS mouse model (Fbn1^C1041G/+) and healthy controls revealed a modulated SMC cluster in the mutant aorta with higher expression of proliferation and ECM synthesis gene signatures.²⁷ Luo et al. induced sporadic AAD in Sting-deficient mice with high fat diet and angiotensin II treatments, and uncovered the role of Sting in SMC death and macrophage activation.²⁸ ScRNA-seq experiments were also conducted on AAA mouse models. Zhao et al. performed scRNA-seq on both elastase-treated and control mice at early and late time points during AAA progression, showing drastic compositional changes in immune cells and SMCs.²⁹ Hadi et al. demonstrated that targeted deletion of netrin-1 is protective against AAA and scRNA-seq revealed significant enrichment of netrin-1 in activated macrophages.³⁰ Yang et al. induced AAA using CaCl₂, which also decreased SMC numbers and increased immune infiltration.³¹ In a seminal scRNA-seq study, SMC-specific disruption of TGFβ signaling in hypercholesterolemic ApoE knockout mice induced trans-differentiation of SMCs into mesenchymal stem cell-like state, implying a key role for SMC reprogramming in aneurysm formation.³² In sum, these animal studies elucidated the molecular characteristics of SMC and immune cell heterogeneity as well as their disease states (Table 1).

Table 1.

Distinctive markers identified in previous human and mouse model scRNA-seq studies of aortic aneurysms.

Disease Model/Patient	Aortic Region	Marker	Cell Type(s)	Mechanism/Phenotype	Ref.
TGFβR2iSMC-Apoe mice on hypercholesterolemic diet	Thoracic	Klf4	MSC	Klf4 is a key regulator controlling SMC reprogramming into MSCs.	32
AngII-induced Ntn1−/− ApoE−/− mice	Abdominal	Netrin-1	Macrophage	Macrophage-derived netrin-1 promotes AAA formation.	30
AngII induced, acute lung injury mice	Abdominal	Ripk3	Macrophage	Acute lung injury-induced HMGB1 causes mitochondrial dysfunction in macrophages through RIPK3, and stimulates AAA.	77
CaCl2-induced AAA mice	Abdominal	Pf4, Lyz2, Mrc1	Anti-inflammatory macrophage	Three types of macrophages with distinct expression profiles were identified in CaCl2 -induced AAA mice.	31
		Lyz2, Il1b, H2-Ab1	Inflammatory macrophage
		Lyz2, Stmn1, Mki67	Proliferative macrophage
Stinggt/gt mice with/ without high-fat diet+AngII	Thoracic	Sting, Casp1, Htra2, Ripk3, Cxcr3	SMC	Sting is enriched in SMC and macrophage of challenged WT mice and STING signaling increases aortic degeneration in TAAD.	28
		Sting, Mmp2, Mmp9, Casp1, Cxcr3, Il1b	Macrophage
Fbn1C1041G/+ mice	Thoracic	Tnfrsf11b, Fn1, Mgp, Nupr1, Eln	Modulated SMC	A modulated SMC population with increased expression of ECM synthesis, adhesion, and cell proliferation markers was detected in MFS mouse model.	27
Myh11-CreERT2/ Confetti mice	Thoracic	Sca1	Phenotype switched SMC	Sca1 elevation is an indicator of SMC undergoing phenotype switching.	26
AngII induced, Mef2c-Cre +/0 ROSA26RmT/mG mice	Thoracic	H2afz, Cks2, Cenpa, Cdk1, Mki67	SHF-derived Fibroblast	A unique SHF-derived fibroblast population was identified in AngII induced TAA mice.	33
β-aminopropionitrile-induced TAAD mice	Thoracic	Il1rn, Trem1	Pro-inflammatory macrophage	A novel pro-inflammatory macrophage population was found in β-aminopropionitrile-induced TAAD mice.	78
Human MFS TAA	Thoracic	TNFRSF11B, MMP2, FN1, CYTL1, COL8A1	Modulated SMC	The modulated SMC population identified in Fbn1C1041G mice was confirmed in a human MFS root tissue.	27
Human AAA	Abdominal	JMJD3	Macrophage/ monocyte	Increased JMJD3 levels in macrophages aggravates AAA progression.	36
Human MFS TAA	Thoracic	SMTN, MYH11, CNN1	Contractile SMC	Multiple SMC populations were identified in control and MFS ascending aorta, and proliferative SMCs were enriched in MFS tissue.	35
		JUN, JUNB, FOS, FOSB, EGR1	Stressed SMC
		CCND1, CCND2	Proliferative SMC
		TNFRSF11B, LTBP2, GSN	Fibromyocyte
Human TAA	Thoracic	ACTC1, ACTA2, MYL9, CARMN	Contractile SMC	Multiple SMC, fibroblast and immune cell populations were identified in control and TAA tissue.	34
		FOS, ATF3, JUN, HSPB8	Stressed SMC
		MGP, TPM4, MYH10	Proliferative SMC
		COL8A1, FBLN5, CDH13, NOV	Fibromyocyte
		CCL3L1, CCL4L2, CCL4, TNF	M1 macrophage like
		PDK4, STAB1, TXNIP, MAF	M2 macrophage like

ECM: extracellular matrix; TAAD: thoracic aortic aneurysm and dissection; SHF: second heart field; SMC: vascular smooth muscle cell; MFS: Marfan syndrome; MSC: mesenchymal stem cell-like; AngII: angiotensin-II.

Importantly, most of our knowledge about lineage-specific SMCs, their regional distribution and differential disease susceptibility comes from mouse model studies. In a recent study, Sawada et al. performed scRNA-seq of second heart-field derived SMCs and fibroblasts from both healthy and angiotensin II-induced ascending aortic aneurysm tissue, and identified a novel fibroblast population in angiotensin II infused mice, potentially contributing to the aneurysm development.³³ Combined with improved precision in lineage-tracing methods, scRNA-seq of mouse models could provide further insights into region-specific SMCs and other aortic cell types including fibroblasts. Of note, the degree of commonality between mouse and human aortic cell heterogeneity and disease states needs to be further validated with region-specific scRNA-seq studies of healthy and aneurysmal human aortic tissue.

APPLICATIONS OF SINGLE-CELL TRANSCRIPTOMIC APPROACHES ON HUMAN THORACIC AORTA

ScRNA-seq studies of tissue samples require the preparation of single-cell suspensions through cell dissociation, which alters gene expression profiles of individual cells and causes loss of spatial information. The human aorta is a large elastic artery with unusually strong mechanical properties. Extracellular matrix proteins such as elastin and collagen account for a significant portion of aortic dry weight, making it a difficult tissue to reliably dissociate. Despite this difficulty, there have been a number of recent scRNA-seq studies of both human TAA and AAA (Table 1). These human aorta studies implicate a more complex picture for aortic SMC diversity with multiple disease-associated states and provide valuable data resources.

Pedroza et al. performed single-cell transcriptomics on a root sample from a Marfan Syndrome patient, which confirmed the SMC phenotype switch observed in Fbn1^C1041G mice.²⁷ Li et al. conducted a more comprehensive study on multiple sporadic ascending aorta aneurysm samples as well as region-matched controls from heart transplant donors and recipients.³⁴ Although the immune cells were surprisingly ubiquitous in this dataset, unsupervised clustering uncovered major cell types in both control and aneurysmal ascending aortas highlighting five SMC, two fibroblast, and two endothelial cell clusters. Furthermore, a recent study conducted on multiple aneurysmal samples from Marfan Syndrome patients identified several SMC clusters and speculated that impaired TGFβ signaling contributes to the disease pathology.³⁵ An scRNA-seq study of both human and mouse AAA samples revealed elevated levels of proinflammatory monocytes and macrophages in the aneurysmal samples and emphasized the role of JMJD3 in macrophage inflammation and aortic dilation.³⁶ With the increasing number of scRNA-seq studies of human aorta, a meta-analysis study could reveal additional molecular signatures of region-specific aneurysm development.

Considerations for Human Aortic Tissue Applications of Single-Cell Transcriptomics

Defining cell types using cell clustering analyses and subsequent differential expression analyses have emerged as the principal application of scRNA-seq. However, precautions should be taken in the biological interpretation of the clustering data. The workflow for scRNA-seq analysis leading up to the cell clustering involves quality control steps to avoid cell clusters that are technical artifacts. ScRNA-seq data can have high levels of noise due to several factors including batch effects, insufficient sequencing depth and mRNA capture efficiency, which confound the data analysis. As these are inherent limitations to almost all scRNA-seq data analysis so several computational tools with a range of performances have been developed to distinguish true biological differences and technical artifacts.^37–39

ScRNA-seq of human aortic tissue also comes with unique challenges. Cell dissociation is a particularly concerning step in the studies of the human aorta, an exceptionally difficult tissue to dissociate. Suboptimal cell dissociation (e.g. incomplete dissociation, long duration) can significantly alter compositional profiles of different cell types and native gene expression levels. A prime example for dissociation-based expression difference is high mitochondrial content. Removal of apoptotic cells does not adequately address this problem, as some remaining cells that pass the applied threshold could still form discrete clusters. Furthermore, use of identical dissociation protocols for control and fibrotic human aneurysmal tissue may not have the same dissociation efficiency, potentially misrepresenting cell composition differences in healthy vs. aneurysmal samples.

Given all the technical confounders, scRNA-seq data does not necessarily represent the ground truth about the biological differences. Instead, researchers should use scRNA-seq data to generate predictions that can be validated by orthogonal validation tools. Orthogonal validations in the context of scRNA-seq refer to the independent experimental methods to evaluate the validity of scRNA-seq findings such as cell clustering and differential gene expression results. Successful scRNA-seq experiments yield a significant amount of data hindering the validations of all data-driven predictions. Therefore, the experimental validations can be conducted to substantiate the key findings of the study.

In situ hybridization (ISH) assays can provide a visual confirmation of scRNA-seq results. Multiplexed ISH methods have been developed enabling simultaneous detection of multiple RNA targets. Particularly, single molecule fluorescence in-situ hybridization is a powerful method to validate the predictions of scRNA-seq data.⁴⁰ It allows the imaging-based detection and quantification of individual RNA molecules within cells while preserving spatial information. Traditional experimental approaches have also been efficiently used to validate key predictions of scRNA-seq datasets. Immunohistochemical analysis provides visual information about co-expression and mutual exclusivity of unique cluster markers at the protein level. When gene expression changes are not exclusive to a rare cell type and are observed in the broader dataset, bulk approaches such as quantitative real-time PCR and western blots can be utilized. Although it requires the preparation of single-cell suspensions, flow cytometric analyses of different cell types can support the validity of predictions about the cell composition differences between healthy and aneurysmal samples. In sum, scRNA-seq is a powerful hypothesis generating tool that generates many predictions that need to be substantiated by experimental validation. These validations can also provide additional information about spatial organization of cells and protein marker levels. In the context of human aortic research, orthogonal validations are essential to confirm key cell type gene expression differences and true tissue-resident cell types and states.

Considerations for Aortic Tissue Collection and Study Design

Studies of the human aorta also have unique biological challenges. Single-cell transcriptomics is a sensitive experiment, and surgical differences and the duration between the tissue collection and processing can significantly affect RNA integrity and gene expression patterns. Thoracic aorta samples can be collected during various surgical procedures including coronary artery bypass graft surgery (CABG), heart transplant, aneurysmal tissue repair, and aortic valve replacement, which have different operational practices (Figure 1). Ascending aorta samples can be collected during CABG and aortic valve replacement (Figure 1); however, the amount of tissue will be limited. Although CABG offers a strategy to collect non-aneurysmal control samples, the CABG tissue often has low RNA quality. Heart transplant can provide large amounts of non-aneurysmal thoracic aorta tissue except the root; however, the tissue quality can be low due to prolonged ischemic times. Aortic aneurysm repair enables the collection of large amounts of tissue from different thoracic aorta regions. AAA repair is often conducted using minimally invasive endovascular aneurysm repair limiting tissue availability from the abdominal aorta.

Human TAA single-cell transcriptomics studies should also take basic biological variables such as patients’ age, sex and race into consideration for meaningful data interpretation. Age is an important risk factor for both sporadic TAA, and age-related changes have been reported in thoracic aorta.⁴¹ Furthermore, aortic aneurysms are more prevalent in men, and TAA-associated autosomal mutations are more likely to cause aneurysms in male patients.⁴² Although the cost of single-cell transcriptomics is a limiting factor, orthogonal validations can be performed in different age and sex groups depending on the sample availability. Lastly, animal studies revealed that major cell types in different regions of aorta arise from different embryonic sources (Figure 1).^43,44 Intriguingly, human ascending aorta SMCs are likely derived from multiple lineages further adding to aortic SMC complexity. Different aortic regions are also differentially susceptible to TAAD, rupture, atherosclerosis, and calcification implying lineage-specific effects of aneurysm risk factors (Figure 1). Given the regional cellular heterogeneity and disease susceptibility of the human aorta, comparison of region-matched samples is essential to assess the molecular defects in the aneurysmal tissue (Figure 1).

Future Directions for Human Thoracic Aorta Applications of Single-Cell Genomics

High-throughput scRNA-seq approaches require preparation of single-cell suspensions resulting in the loss of valuable spatial information. This led to the emergence of spatial transcriptomics which allows gene expression measurements across tissue space. The aorta has a layered structure and TAAD is likely caused by the weakening of the medial layer populated by SMCs. Therefore, revealing spatially resolved gene expression changes in each layer of the aorta is essential to fully understand the molecular and cellular mechanisms underlying TAAD and rupture (Figure 1).

The current spatial transcriptomics technologies can be broadly sub-grouped into two categories: sequencing-based approaches and ISH-based approaches (Figure 1). Each of these methods has its pros and cons in terms of gene throughput, spatial resolution, sensitivity, and its applicability. Sequencing-based approaches are more suitable for a discovery-based experiment as they enable the profiling of the entire transcriptome.^45,46 They also provide sequencing information allowing the extraction of additional features such as splicing isoforms. The sequencing-based approaches have limited resolution but recent advances in sequencing-based spatial transcriptomics aim to overcome this disadvantage.⁴⁷ ISH-based approaches utilize sequential rounds of imaging probe hybridization.^48,49 Following successive hybridization rounds, serial images are demultiplexed and linked to each transcript. These approaches require a priori knowledge of the model system as they are limited to the profiling of a gene panel. However, their throughput has been significantly improved with recent advances in the technology.⁵⁰ Importantly, ISH-based approaches offer near single-molecule sensitivity and high spatial resolution revealing subcellular localization of transcripts. There are also in situ sequencing based methods which have features of both approaches mentioned above. These methods provide sequencing information by performing the sequencing directly within the tissue.^51,52 In situ sequencing based methods also have lower gene throughput but a highly multiplexed technology has been developed.⁵³ Of note, these spatial transcriptomics approaches have been unexplored in human aortic research. The robustness of these approaches can show tissue-type dependence and is not characterized on surgically-resected human aortic tissue. Given the unique features of human aorta, significant optimization (e.g. permeabilization) will likely be required to adopt and apply these technologies on human aorta samples for reliable data generation. In sum, spatial transcriptomics is a rapidly developing field with a number of approaches that could be applied to human aortic aneurysm research. In addition to providing the spatial maps of aneurysmal tissues, these approaches could also be utilized to infer cellular interactions by revealing the ligand-receptor pairs in neighboring aortic cells or layers.

Although scRNA-seq is the most popular single-cell genomics technique, measurements from individual cells are not limited to the transcriptome. In addition to the transcriptome, single-cell omics assays can profile other molecular features including chromatin accessibility, histone modifications, transcription factor-DNA interactions, and even the proteome at individual cell level (Figure 1).^54–56 Integrative single-cell omics approaches have also been developed to provide multiple molecular measurements from the same cell to uncover complex biological processes.⁵⁷ These methods include cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq), which allows simultaneous measurements of surface protein expression and mRNA profiles from the same cell. Combining CITE-seq with T cell receptor sequencing, Fernandez et al. revealed the T cell repertoire in human atherosclerosis.⁵⁸ Swanson et al. developed TEA-seq, which allows simultaneous profiling of gene expression, protein epitopes as well as chromatin accessibility (scATAC-seq) and applied it to interrogate peripheral blood mononuclear cell diversity.⁵⁹ There are still technical limitations in terms of the sensitivity and throughput of these multiomics approaches and computational challenges for integrative analysis lie ahead. With further advancements in integrative single-cell omics technologies and data analysis tools, multimodal single-cell platforms will be instrumental in uncovering cell-type specific gene regulation, and causal factors that define aneurysmal cell states.

APPLICATIONS OF SCRNA-SEQ IN HIPSC MODELS OF HUMAN THORACIC AORTIC ANEURYSMS

Human induced pluripotent stem cells (hiPSCs) provide a powerful platform to model rare genetic diseases. For many genetic forms of TAA, it is difficult to obtain aneurysmal tissue due to the low prevalence of the disease (e.g. Marfan Syndrome) and changing practices in surgical treatment (AAA). These factors also limit the availability of primary aortic SMCs. In addition, primary cells have limited expandability, changing their identity after a few passages. Combined with CRISPR-Cas9 gene editing, hiPSCs offer many avenues to investigate the molecular defects caused by patient-specific mutations associated with TAA. Genome edited cells with their isogenic controls enable an unbiased analysis of the molecular differences and allow researchers to examine the differences that are independent of genetic background. Importantly, hiPSCs have unlimited expandability increasing the spectrum of experimental and functional experiments that can be conducted. They can also be differentiated into regionally-specific aortic cell populations with distinct embryonic origins enabling the investigation of lineage-specific pathogenicity. Lastly, end-stage aneurysmal tissue is not always ideal to discern early molecular events underlying aneurysm initiation, largely due to epigenetic and compensatory changes accumulating over many years that mask the molecular causes of the disease.

HiPSC disease modeling has proven to be useful in cardiovascular research. The modeling of Marfan Syndrome (MFS) captured many molecular features of MFS-associated aneurysm and identified KLF4 as a key regulator of the disease phenotype.⁶⁰ LDS modeling revealed lineage-specific defects caused by a SMAD3 mutation in contractile gene expression and contractile function.⁶¹ In a recent study, we generated mutant hiPSCs carrying an LDS-causing TGFBR1 point mutation using donor hiPSCs, and corrected the mutation in patient derived hiPSCs using CRISPR-Cas9 gene editing.⁶² Following their lineage-specific SMC differentiation through CPC and NCSC lineages, we investigated the molecular and cellular defects caused by the mutation. We performed large scale scRNA-seq on mutant SMCs and validated the key scRNA-seq findings (e.g. increased cell proliferation in TGFBR1 mutant cells) using conventional molecular techniques and functional assays. ScRNA-seq also identified multiple cell clusters with unique markers even in the control samples suggesting the molecular heterogeneity of SMCs generated using standard differentiation protocols. Similarly, hiPSC derived endothelial cells (ECs) have been instrumental in modelling genetic causes of vascular dysfunction. ScRNA-seq of the hiPSC derived ECs revealed significant heterogeneity demonstrating the need to optimize vascular cell differentiation protocols.⁶³ Lastly, scRNA-seq studies of hiPSC derived cardiomyocytes also uncovered significant cellular heterogeneity as well as cellular immaturity further highlighting a disadvantage of hiPSC models.^64,65

Challenges with hiPSC Models of Human Thoracic Aneurysms

There are a number of limitations in the use of hiPSC derived cells in human TAAD research. Several monolayer-based SMC differentiation protocols have been developed to differentiate hiPSCs to lineage-specific SMC populations; however, the efficiency of these protocols varies considerably and the purity and maturity of the differentiated cells are often unknown.⁴⁴ In monolayer culture protocols, hiPSCs are first differentiated into intermediate progenitors of paraxial mesoderm, lateral plate mesoderm, or neural crest origin using different combinations, amounts, and treatment durations of key morphogens such as TGFβ, BMP (Bone morphogenetic protein), SHH (Sonic hedgehog), and WNT. Subsequent SMC induction protocols also use different media components with induction durations ranging between less than a week to more than a month. Although the overall rationale for each SMC differentiation protocol may be supported by in vivo evidence, their induction efficiency and lineage-specificity were either not reported or limited to the measurements of a small number of SMC markers.⁴⁴ SMCs can also be derived from embryoid bodies but with unknown lineage-specificity.⁴⁴ It should be noted that unwanted cell heterogeneity and cell immaturity are well-known constraints of many hiPSC differentiation protocols and are not exclusive to aortic SMC protocols. Overall, the absence of gold standard aortic SMC differentiation protocols represents both a challenge and an application for single-cell transcriptomics in hiPSC disease modeling.

There are a number of gene editing methods utilizing different nucleases including CRISPR associated Cas9 nuclease, zinc finger nucleases, and transcription activator-like effector nucleases. The CRISPR-Cas9 system is the most applicable tool avoiding complex protein engineering and is commonly used in hiPSC disease modeling. CRISPR-Cas9 gene editing is often employed to introduce disease-causing mutations in healthy donor hiPSCs, and to correct the same mutation in patient-derived hiPSCs. However, potential off-target effects of CRISPR-Cas9 and the inherent susceptibility of hiPSCs to genomic instability are of concern, and the disease phenotype should be validated in different genetic backgrounds. This also applies to single-cell transcriptomics studies where the key findings should be replicated in hiPSCs derived from different individuals such as donor vs. donor mutation knock-in cells and patient vs. patient-corrected cells. Although scRNA-seq of hiPSC models is straightforward due to the relative ease of cell dissociation compared to the human aorta, orthogonal validations of key findings in multiple independent clones and genetic backgrounds are necessary for data interpretation.

Another challenge with the hiPSC approach is that two-dimensional monolayer culture conditions represent a reductive environment and will not capture key contributors of disease pathogenesis such as hemodynamic stress, aortic geometry, and complex ECM and cell-to-cell interactions. To address these concerns, three-dimensional (3D) culture models including tissue-engineered blood vessels have been developed and hiPSC derived SMCs were successfully used to generate implantable blood vessels.^66,67 We anticipate that these 3D models will be increasingly utilized in TAA disease modeling studies, and exploited for delineating the molecular dynamics of aneurysm initiation and propagation.

Future Directions for Single-Cell Transcriptomics in hiPSC Models

Given its high gene throughput and ability to reveal cell heterogeneity, scRNA-seq can be utilized to optimize hiPSC-SMC differentiation protocols (Figure 2). ScRNA-seq on cultured cells is relatively straightforward as the duration of cell suspension preparation is reduced to minutes. In lineage-specific hiPSC-SMC differentiation protocol, time series scRNA-seq could identify cell types and states at both intermediate progenitor cell and differentiated SMC states and provide a whole-transcriptome view of gene expression profiles. Furthermore, single-cell trajectory analysis would uncover different cell transitions between heterogeneous cell populations providing strategies to enrich target intermediate progenitor and SMC populations. Combined with other single-cell omics approaches such as scATAC-seq, these datasets would identify the potential transcription factors involved in lineage-specific SMC priming. Mapping of hiPSC derived SMC to human reference SMCs would be ideal to examine the efficiency of each differentiation protocol. Although there is currently no human reference atlas for regionally-specific aortic SMCs, we anticipate the generation of an aortic cell atlas in the near future with the increasing number of scRNA-seq studies.

Figure 2.

An overview of scRNA-seq applications in hiPSC models of human aortic aneurysm. ECM: extracellular matrix; hiPSC: human induced pluripotent stem cell; scRNA-seq: single-cell RNA sequencing; SMC: vascular smooth muscle cell; PBMC: peripheral blood mononuclear cell.

The use of genome editing has expanded the impact of hiPSC disease modeling. Researchers can now introduce clinically identified TAA mutations in hiPSCs, and eventually examine the molecular and cellular defects in mutation-carrying cell lines compared to their isogenic controls. HiPSC disease modeling for a limited number of mutations has been conducted to study the molecular basis of aneurysm formation, and scRNA-seq profiling of hiPSC derived SMCs has elucidated the genome-wide transcriptional defects in mutant cell clusters.⁶² However, the genetic risk factors that predispose individuals to aortic aneurysm show significant heterogeneity and require additional studies.

TAA-associated genetic mutations can be broadly categorized into four groups (Figure 2). The first group of genes encodes components of TGFβ signaling. Mutations in canonical TGFβ signaling genes (TGFBR1, TGFBR2, TGFB2, TGFB3, SMAD2 and SMAD3) have been associated with different LDS subtypes.⁴ Mutations in SKI gene also disrupt TGFβ signaling and cause Shprintzen–Goldberg syndrome which has significant phenotypic overlap with LDS.⁶⁸ Extracellular matrix genes constitute the second group. MFS is caused by mutations in FBN1 gene, which encodes a large ECM glycoprotein.⁶⁹ COL3A1 mutations have been identified in patients with vascular EDS, a syndrome characterized by a fragile aortic wall that is easily ruptured.⁷⁰ Rarely, other subtypes of EDS caused by mutations in COL1A1, COL5A1 and PLOD1 can exhibit vEDS-like phenotypic manifestations.⁷¹ TAA is also observed in families with mutations in LOX gene, which encodes a protein promoting elastin-collagen crosslinking.⁷² The third group consists of genes (ACTA2, MYH11, MYLK, PRKG1) involved in SMC contractile apparatus.⁷ Patients carrying mutations in these genes can have familial TAA with no overt connective tissue abnormalities as observed in LDS, MFS and vEDS patients. The last group consists of genes linked to other heritable TAA cases including BAV-associated genes such as NOTCH1 and ROBO4.^11,73 Furthermore, mutations in SLC2A10 gene causes arterial tortuosity syndrome which can also lead to aortic aneurysm.⁷⁴ Of note, a significant portion of the families with heritable TAA still does not have an easily identifiable genetic cause, so the list of TAA-associated mutations will likely continue to grow. Collectively, the mutations in these aforementioned genes cause TAAD or rupture in a region-specific manner with different aggressiveness, systemic features and disease onsets. ScRNA-seq profiling of hiPSC derived SMCs carrying clinically identified pathogenic TAA variants would provide an important resource to dissect the lineage-specific effects of each mutation and to tailor potential patient-specific treatment strategies (Figure 2).

HiPSCs derived from patients with genetic vulnerabilities also have increasing use in studying the effects of different perturbations including potential therapeutic agents. In addition to its use in identifying global transcriptional defects caused by aneurysm-causing mutations, high-content readout of scRNA-seq may provide opportunities to address single-cell level molecular recovery in response to drug treatment (Figure 2). As discussed above, hiPSC differentiation protocols often yields a non-uniform cell population with unresolved cell maturity and lineage specificity, and bulk approaches are unable to measure transcriptional responses to any perturbation in each cell type/state. ScRNA-seq has been previously utilized to predict cell-type specific drug susceptibility and resistance of heterogenous patient tumors.^75,76 In a similar manner, scRNA-seq of hiPSC derived cells could both reveal targetable pathways and provide a high-content measurement of drug response in a patient-specific manner. Importantly, this approach could also be extended to understand the cell-specific transcriptional response in patient-derived primary cells which usually consist of a mixture of cells from different lineages.

CONCLUDING REMARKS

In this review, we discussed the current and potential applications of single-cell transcriptomics and complementary technologies in human thoracic aortic aneurysm research, and highlighted some considerations to substantiate data-driven predictions and draw biologically meaningful conclusions from such data. Given the diversity in SMC developmental origins and genetic etiology of TAAD, high-content readout of single-cell transcriptomics can be leveraged to resolve inter-aneurysmal heterogeneity and to optimize hiPSC-based TAAD modeling. With further advances in single-cell genomics technologies and disease modeling approaches, we believe that single-cell platforms will be instrumental in uncovering actionable pathways underlying TAAD formation and tailoring personalized treatment strategies.

HIGHLIGHTS.

• Single-cell RNA sequencing and complementary approaches including spatial transcriptomics should be increasingly employed to dissect the regional and genetic heterogeneity of human thoracic aortic aneurysms (TAA).

• The application of single-cell transcriptomics on human TAA samples is at an early stage and orthogonally validated reference atlases need to be established for tissue-resident cell types.

• hiPSC disease modeling will be instrumental to study TAA and single-cell transcriptomics in this context offer new avenues into understanding the molecular causes of TAA formation leading to patient- and cell-type specific therapeutics.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

SMC

vascular smooth muscle cell

scRNA-seq

single-cell RNA-sequencing

TAAD

Thoracic aortic aneurysm and dissection

CPC

cardiovascular progenitor cell in the second heart field

NCSC

neural crest stem cell

ISH

In situ hybridization

MFS

Marfan syndrome

LDS

Loeys–Dietz syndrome

vEDS

vascular Ehlers–Danlos syndrome

TGFβ

transforming growth factor beta

ECM

extracellular matrix