Transcriptome analysis of the aortic coarctation area

Abstract

Background

Coarctation of the aorta (CoA) is a relatively common congenital heart defect. The underlying causes are not known, but a combination of genetic factors and abnormalities linked to embryonic development is suspected. There are only a few studies of the underlying molecular mechanisms in CoA. The aim of the current study was to expand our understanding of the pathogenesis of CoA by characterizing the transcriptome of the coarctation area.

Methods

Tissue samples from 21 pediatric patients operated for CoA were dissected into separate biopsies consisting of the localized coarctation itself, proximal/distal tissue and ductus. RNA was sequenced to evaluate gene expression in the different biopsies.

Results

We observed an activation of acute phase response in samples from the localized coarctation compared to samples from distal or proximal tissue. However, we observed even bigger differences for patient age and sex than compared to biopsy location. A cluster of genes located at 1q21, including the S100 gene family, displayed contrasting expression depending on patient sex, and appeared to affect the balance between inflammatory and interferon pathways. Biopsies from patients <3 months old were characterized by a significantly higher fibrotic activity compared to samples from older patients. The ductus tissue was characterized by an upregulation of factors associated with proliferation.

Conclusions

The ongoing processes in the coarctation area are influenced by the age and sex of the patient, and possibly by differences in etiology between different patients. The impact of patient attributes must be taken into consideration when performing future studies.

Graphical abstract

Highlights

• •Patient age and sex have a substantial impact on the gene expression of the aortic coarctation area
• •The localized coarctation is characterized by acute phase responses and complement activation
• •Males and females differ in their expression of a cluster of genes on 1q21, which contains the S100 gene family
• •Younger children (<3 months) have a more active fibrosis than older children
• •The ductus tissue is characterized by an upregulation of factors associated with cell division/proliferation

1. Introduction

Coarctation of the aorta (CoA) is defined as a narrowing or obstruction of the aorta, and is a common type of congenital heart disease (CHD), in which the aortic isthmus is narrowed, proximal, distal and/or opposite the insertion of ductus arteriosus (DA) [1], often combined with hypoplasia of the distal aortic arch. The incidence of CoA is reported as 1:2500 births and represents 6–8 % of CHD with a male-to-female dominance [[2], [3], [4]]. Hoffman reported in 1968 the natural history of CHD, where untreated CoA was associated with very high mortality rate (86 %) occurring during the first months of life [5]. Neonatal critical CoA accounts for 60 % of all CoA while the remaining 40 % are asymptomatic and present later in life, for example as hypertension in the upper part of the body [6]. CoA may occur in isolation or in association with other CHD such as a bicuspid aortic valve (BAV), ventricular septal defect (VSD), and varying degrees of left ventricular or aortic arch hypoplasia. BAV is the most common associated CHD and occurs in 60 % of CoA patients [7].

The embryologic mechanism resulting in CoA is not completely understood. There are three leading hypotheses regarding formation of this anomaly. First, tissue from the ductus arteriosus may incorporate into the aortic wall where it connects to the descending aorta. As the ductus arteriosus constricts postnatally, this tissue in the isthmus constricts, leading to CoA [8]. Another theory includes altered hemodynamics, in which abnormal preductal flow or an abnormal angle between the ductus and aorta leads to increased right-to-left ductal flow and decreased flow across the arch and isthmus, resulting in CoA development [9] The third theory involves an abnormal involution of a small segment of the left dorsal aorta, which may later move cranially with the left subclavian artery, forming CoA in the isthmus region [10]. Familial clustering of isolated CoA supports a genetic basis for isolated CoA [11], however, the known genetic etiologies for isolated CoA are heterogeneous and complicated by variable expressivity and penetrance [12]. Genetic causes of isolated CoA include CNVs and sequence variants but the potential monogenic causes are not well understood.

In the current era, CoA is usually diagnosed perinatally by echocardiography in utero or in the immediate postnatal period, when the DA is open and there is antegrade flow through the aortic isthmus. After birth, the oxygen concentration in the blood increases and prostaglandin decreases, which initiates ductal constriction [13]. Medial smooth muscle cells proliferate and migrate towards the intima resulting in complete ductal occlusion and subsequent obliteration [9]. In affected individuals, ductal tissue that has migrated into the aorta contracts, leading to a narrowing of the aortic isthmus in association with ductal closure [14].

If blood flow to the distal aorta is impaired critically by this narrowing, newborns with CoA may present with cardiogenic shock. The acute treatment in neonatal CoA is an infusion of Prostaglandin E1 (Alprostadil), which results in improved antegrade aortic blood flow by relaxing ductal smooth muscle cells in the DA and isthmus. Once the patient is stabilized, the coarctation is surgically removed and an end-to-end or end-to-side anastomosis of the aortic arch is performed. If the aortic arch is hypoplastic as well, patch augmentation of the aortic arch may be necessary, an operation that usually requires cardiopulmonary bypass. In addition, associated CHD may be repaired during the same surgery. Some infants with CoA survive the neonatal period without clinically detected signs or symptoms of CoA. These patients are typically identified following evaluation for a heart murmur or systemic arterial hypertension.

Several studies have described the morphological features of CoA as well as immunohistochemical aspects of the localized coarctation, but there is limited data regarding the molecular properties of the localized coarctation [15]. The aim of the current study was to perform transcriptome analysis to characterize transcriptional activity, signaling pathways and ongoing processes in tissue collected from pediatric patients operated for CoA.

2. Materials and methods

2.1. Patients

The patients were recruited at Skane University Hospital Lund, one of the two tertiary centers for pediatric heart surgery in Sweden. In total 118 patients were referred for operation of coarctation during April 2016 – May 2021 to the Pediatric Heart Center in Lund, Sweden.

In the present study parents to 21 pediatric patients scheduled for surgery of CoA, agreed to save the removed tissue in the national biobank of congenital heart disease in Sweden (www.snab-chd.se). Inclusion criteria in the current study was isolated CoA, with or without associated aortic arch hypoplasia, bicuspid aortic valve (BAV), or other associated congenital heart diseases (CHD) such as ventricular septal defect (VSD), mild aortic stenosis (AS), atrial septal defect (ASD) or patent ductus arteriosus (PDA). Exclusion criteria were CoA as part of critical AS, aortic atresia, Shone complex, borderline or hypoplastic left heart syndrome (HLHS) as well as other more complex CHDs.

The study was conducted in accordance with the Declaration of Helsinki and approved by the regional Swedish Ethical Review Authority ethics committees in Stockholm and Linköping (Dnr 2015/883–31/4; Dnr 2016/389–31). The legal guardians of the children included in this study signed an informed consent for the donation of the removed tissue for research purposes.

2.2. Tissue collection

In the operating theatre, the coarctation area was isolated with vascular clamps and the DA ligated. The coarctation area was resected with sharp scissors and the whole specimen placed in RNAlater. The specimen included a few millimeters of distal aorta, the localized coarctation area, the DA and a proximal part of aorta included isthmus area and sometimes aortic arch tissue. The test tube with the tissue, marked with a unique code, was moved to a refrigerator, and stored at 5 °C for at least 24 h. The specimen was transferred to the tissue bank. RNAlater was removed and the tissue was placed without any additives in a sterile tube and stored in a biobank freezer at −80 °C until requested for transcriptome analysis.

2.3. Tissue dissection

In the laboratory the tissue from the 21 patients was thawed to room temperature prior to dissection. The tissues were divided into a biopsy of the localized coarctation and normal proximal/distal tissue. In cases where the pathologist could determine whether the normal biopsy was from proximal or distal tissue, this was noted. For patients where ductus tissue was available, this biopsy was also harvested.

2.4. RNA isolation

After tissue dissection, samples were kept in DNA/RNA Shield (Zymo) at room temperature for a maximum of 2 h prior to lysis and RNA isolation. Samples were lysed using ZR BashingBead Lysis Tubes (Zymo) on a TissueLyser (Qiagen) set to 25 Hz for 10 min and RNA was isolated using Quick-RNA Microprep kit (Zymo) according to the manufacturer's protocol. RNA quality was assessed using QIAxpert (Qiagen) and concentration was measured using Qubit™ RNA High Sensitivity kit.

2.5. Library preparation and sequencing

10 ng of RNA was prepared using Solo RNAseq system (Tecan) according to the manufacturer's instructions. Libraries were sequenced on a NextSeq 500 (Illumina) using High Output 150 kit (reads were paired) in batches of 8 resulting in approximately 50 million reads/sample.

2.6. Bioinformatics and statistical analysis

Data was demultiplexed using bcl2fastq. After mapping using STAR [16], PCR duplicates were removed using NuDup in accordance with the library kit manufacturer's instructions. Gene counts were extracted using featureCounts [17]. Differentially expressed genes were calculated using DESeq2 [18]. For comparisons between tissue types (coarctation vs proximal/distal, ductus vs proximal/distal and proximal vs distal), the analysis design controlled for patient. For the comparison between age groups and sex, the analysis design controlled for tissue type. Samples from ductus were excluded from all analyses except for the analysis concerning differential expression in ductus compared to distal/proximal tissue. Detailed information about statistical design, as well as DESeq2 statistical analysis are included as a supplementary file in the gene expression omnibus (GEO) record associated with this study.

Pathway analysis was conducted using Ingenuity Pathway Analysis (IPA) software (Qiagen). For the analysis comparing tissue types, and differences between males and females, p-values <0.05 were used without FDR adjustment. For the age group analysis, and for the analysis of ductus tissue, FDR adjusted p values <0.05 were used. This approach was adopted in order for the number of genes in the pathway analysis to fall within the window defined as optimal by IPA.

IPA was also used to compare profiles of differentially expressed genes between coarctation and proximal/distal tissue in this study to differentially expressed genes between the aortic isthmus of patients with and without CoA in a study by Liu et al. [19].

ORA (over representation analysis) of genes according to cytogenetic location was performed using WebGestalt [20] and visualized using ChromoMap [21].

3. Results

21 patients were included in the study, fifteen patients were male, and six female, age between 2 days and 7.5 years (Table 1). The majority of patients (n = 17; 81 %) underwent CoA repair within the neonatal period to early infancy (i.e. within 3 months of life; Table 1), whereas 4 patients underwent primary CoA repair at or beyond 3 months of age (Table 1). There was a male predominance in both groups, constituting 71 % of the entire cohort. Associated CHD was particularly common in the <3 months of age group; muscular or perimembranous VSD (n = 6), intermediate AV-canal defect (1), valvular (n = 2) or subvalvular (n = 1) AS or aortic arch hypoplasia (n = 4). BAV was present in 12 patients (57 %). Chromosomal aberration was diagnosed in three patients. Likewise, surgical repair using cardiopulmonary bypass, aortic arch reconstruction and associated intracardiac cardiac surgery occurred only in the neonatal group <3 months of age (Table 1). The rate of recurrent CoA (reCoA) in the group of children operated for CoA <3 months of age was 29 % (n = 5), all operated before 14 days of age, whereas no reCoA occurred in patients operated >3 months of age.

Table 1.

Characteristics of the study participants. Abbreviations: ASD = atrial septal defect, AVSD = atrio- ventricular septal defect, BAV = bicuspid aortic valve, CHD = congenital heart defect, CoA = coarctation of aorta, CPB = Cardio-pulmonary bypass, VSD = ventricular septal defect, mVSD = muscular VSD, PAB = pulmonary banding, PDA = patent ductus arteriosus, PGE1 = Prostaglandin E1.

Patient age	<3 months (n = 17)	>3 months (n = 4)
Demographic characteristics
Sex	12 Male, 5 Female	3 Male, 1 Female
Weight at repair	3.3 (2.2–6.2) kg	15.2 (6.2–23) kg
Age at repair	10 (2–47) days	2.7 (0.33–7.5) years
Preterm birth (<37 weeks gestational age)	1	0
Extracardiac malformations or syndromes	3
Heredity	0	0
Preoperative PGE1 infusion	14	0
Cardiac characteristics
Associated CHD	8	1
Associated CHD excl. sm. mVSD	5	1
BAV	10	2
VSD	6	0
Incomplete AVSD	1
Arcushypoplasia	4	0
Valvular Aortic stenosis	2	0
Subvalvular Aortic stenosis	1	0
Large ASD	1	0
PDA	14 (on PGE1)	1
Operative and postoperative characteristics
CPB	5	0
Extended end-to-end or end-to-side anastomosis	12	4
Arch reconstruction	5	0
Re-Coarctation	5	0
Simultaneous other cardiac surgery	5 (PAB, VSD, aortic valve commissurotomy, Rastelli)	1 (PDA)
Death	1	0

3.1. Clustering based on biopsy location, age and sex

One of the major findings in this study was that there was a greater difference in gene expression according to age and sex than according to localization of the specific aortic biopsy (Fig. 1, Supplementary Fig. 1). Using PCA, we observed no outliers and a tendency of samples to cluster by patient (Fig. 1B). Supervised clustering (PLS-DA) of data revealed differences between samples based on biopsy location (Fig. 1C), patient sex (Fig. 1D) and patient age (Fig. 1E), with clearer clusters when separating the data based on sex and age compared to biopsy location.

Fig. 1.

(A) Method overview: resected tissue from 21 patients with coarctation of the aorta was included in the study. The tissue was dissected into separate biopsies consisting of the localized coarctation itself (n = 21), proximal/distal tissue (n = 25) and ductus (n = 4). RNA was extracted and sequenced to evaluate gene expression in the different biopsies. Statistical analysis was performed based on biopsy location as well as patient sex and age at surgery. Differentially expressed genes were subjected to pathway analysis to gain insight into the processes taking place in the tissue. (B) PCA analysis of data after normalization (rlog) showed no outliers and a tendency of samples to cluster by patient. Supervised clustering (PLS-DA) of data revealed differences between samples based on biopsy location (C), patient sex (D) and patient age (E), with bigger differences observed for patient age and sex compared to biopsy location. Patient number refers to the original patient ID in the tissue biobank for the 21 patients included in this study.

In addition to performing statistical analysis based on patient age and sex, we also performed analysis to identify differentially expressed genes between patients with and without bicuspid aortic valve (BAV). The results from this comparison are provided together with the other statistical analyses performed in the supplemental data in the GEO record associated with this study.

3.2. Acute phase responses in the localized coarctation compared to proximal/distal part of aorta

Statistical analysis was performed to identify differentially expressed genes in biopsies taken from the localized coarctation compared to biopsies from proximal/distal tissue (Fig. 2). Genes with differential expression were subjected to pathway analysis to identify functional clusters and signaling pathways that characterize the coarctation. We found an increase in complement activation and acute phase response via enhanced NFkB, EGR2/3 and JUN/FOS signaling, whereas STAT3 signaling was decreased (Fig. 2). Upstream, these pathways are predicted to be mediated by p38 mitogen-activated protein kinase (MAPK38) signaling. Neuropilin-1 (NRP-1), a receptor for Vascular endothelial growth factor (VEGF), which has been shown to activate MAPK38, was found to be upregulated in the coarctation area and is likely connected to the induction of the observed pathways, although other growth receptors may also be involved [22]. Downstream of the activated transcription factors in the coarctation area, we could see an upregulation of complement components including C1 and C3, as well as an upregulation of cytokines involved in the acute phase response such as IL6 and IL23. Several other stress- and/or inflammation-associated genes reported to contain binding sites for NF-κB and CREB were also augmented, among others the inflammatory factors IκBζ (NFKBIZ) and CEBPD, the immediate early stress response gene ATF3 and the regulator of inflammation DUSP1. Of note, in addition to being both induced by and regulating inflammation, DUSP1 is also associated with suppression of epithelial cell activation in areas of the aorta exposed to high or pulsatile shear stress [23,24].

Fig. 2.

Activated pathways in coarctation compared to proximal/distal tissue. Differentially expressed genes in biopsies taken from the localized coarctation compared to proximal/distal tissue were assembled into pathways using IPA and literature search [25,26]. EGFR/PI3k signaling was predicted to be activated, leading to the activation of NFkB and EGR2/3, which in turn activates Jun/Fos. Together, these transcription factors lead to complement activation - seen as upregulation of components of the classical complement pathway, and acute phase signaling – seen as upregulation of IL-6 and IL-23. A number of interferon-inducible factors were also upregulated (see supplementary/raw data), indicating an involvement of interferon signaling. STAT3 downstream molecules IL-17, IL-1a and CCL20 were downregulated, likely through a mechanism involving SOCS3. BTG2 levels were upregulated. Paired analysis of localized coarctation (n = 21) and proximal/distal biopsies (n = 25) from 21 patients. 17 patients had one of each biopsy type, 4 patients had one coarctation and two proximal/distal biopsies. Differentially expressed genes (DESeq2): p < 0.05 without FDR adjustment. IPA analysis “Acute Phase Response Signaling”: 17 molecules, z-score 1.897, p-value 4.73 × 10⁻⁸, “Complement system”: 8 molecules, z-score 2.646, p-value 2.56 × 10⁻⁷.

To evaluate whether the differences we observed between the localized coarctation and surrounding tissue can be generalized to reflect the differences between aorta with coarctation and healthy aorta, we compared our results to results from a transcriptomics study comparing patients with and without CoA published by Liu et al. [19]. Data from transcriptomics of tissue at the aortic isthmus of 12 patients with CoA and 10 healthy controls, kindly provided by Liu et al. [19], was subjected to the same pathway analysis using IPA as the data in our current study, and the resulting differentially regulated pathways were compared (Supplementary Fig. 2). This comparison revealed that upregulation of genes associated with acute responses such as the complement system and inflammation could be observed in both our data and the data provided by Liu et al. in association with coarctation. The comparison also identified that upregulation of pathways associated with cytoskeletal remodeling and phagocytosis could only be observed in data from the Liu et al. study, but not in our data.

In addition to comparing the localized coarctation to the proximal/distal tissue, proximal tissue was compared to distal tissue for the biopsies where this information was available. As only 4 patients had this information in addition to having both proximal and distal biopsies, this analysis was not included in the main manuscript. The results from statistical analysis on differentially expressed genes between proximal and distal tissue are provided together with the other statistical analysis performed in the supplemental data in the GEO record associated with this study.

Expression profiles of the differentially expressed genes showed considerable variation between individuals, suggesting that there may be differences in pathogenesis mechanisms between the different patients (Supplementary Fig. 3).

3.3. Inflammation and sex

Statistical analysis was performed to identify differentially expressed genes in biopsies taken from males compared to females. Differentially expressed genes in the localized coarctation vs proximal/distal tissue in males and females were subjected to pathway analysis using IPA. In addition to genes identified by comparison analysis using DESeq2, genes on the Y chromosome only expressed by males were included in the analysis. Females had a higher immune activation overall and had a more pronounced elevation of inflammatory factors in the localized coarctation compared to the surrounding tissue (Supplementary Fig. 4).

While females had more pronounced inflammation, males had a higher activation of pathways associated with interferon production, including a number of IRFs (interferon regulatory factors) (Fig. 3, Supplemental data). This difference in the balance between interferon and inflammatory signaling appeared to involve a difference between the S100 molecules expressed by each sex, with males having increased expression of S100A8/A9 and S100A12 while females had higher expression of S100A1 and S100A7. The distinct S100 expression profiles were also linked to altered expression of various GPRs (G-coupled protein receptors) and appeared to originate from signaling events related to the increased estrogen receptor signaling observed in females (Fig. 3, Supplemental data).

Fig. 3.

Estradiol signaling was significantly lower in males compared to females according to IPA analysis (80 molecules, activation z-score − 2.3, p = 3.3 × 10⁻³). Estradiol, as well as other factors reported in the literature to lead to nitric oxide production were lower in males than females. S100 signaling appeared to be differentially regulated in males compared to females (31 molecules, z-score − 2.8, p = 3.53 × 10⁻³), with males having increased expression of S100A8/A9 and S100A12 leading to IRF activation and interferon production while females had higher expression of S100A1 and S100A7, leading to higher inflammation. 33 biopsies from male patients (n = 15) and 13 biopsies from female patients (n = 6) were compared. The design for the comparison controlled for biopsy location. Differentially expressed genes p < 0.05 without FDR adjustment were used for the pathway analysis.

ORA (overrepresentation analysis) was performed to examine whether the differentially expressed genes between males and females clustered according to cytogenetic location. In addition to observing that genes significantly upregulated in females compared to males were over-represented on the X chromosome, the analysis identified cytogenetic bands chr1q21.2 and chr1q21.3 as significantly over-represented among genes whose expression differs according to sex (Fig. 4, Supplementary Fig. 6). These bands contained 17 genes whose expression differed between males and females, including a gene cluster containing the S100 molecules described above.

Fig. 4.

Chromosomal location of genes differentially expressed in males vs females. Overrepresentation analysis (ORA) of differentially expressed genes in males (n = 15) vs females (n = 6) identified chr1q21.2 and chr1q21.3 as significantly over-represented. This region contains 17 genes with differential expression between males and females (p < 0.05, without FDR adjustment), including the S100 genes: PDE4DIP, FCGR1A, H3C13, H3C15, H2AC21, CIART, ADAMTSL4, ANXA9, THEM5, LCE2B, RPLP0P4, S100A9, S100A12, S100A8, S100A7, S100A1, RAB13.

Similar to our observations on comparison of biopsy location as described above, there was considerable intra-patient variation in gene expression between individuals (Supplementary Fig. 5).

3.4. Fibrosis and age

Statistical analyses were performed to identify differentially expressed genes in biopsies from patients <3 months and > 3 months of age (Fig. 5, Supplementary Fig. 7). Pathway analysis of the significantly differentially expressed genes revealed a dramatically greater expression of factors associated with fibrosis, fibroblast proliferation and extracellular matrix accumulation in biopsies in patients <3 months compared to >3 months old, where biopsies clearly clustered in their two respective groups regarding expression of genes implicated in fibrosis (Supplementary Fig. 8).

Fig. 5.

Factors associated with aorta maturation and extracellular matrix accumulation are upregulated in patients < 3 months old compared to patients > 3 months old. Pathway analysis revealed that in biopsies from patients <3 months old (37 biopsies from 17 patients) there was increased activation of TGF-β signaling, Wnt signaling and the Hippo and Hedgehog pathways, leading to production of extracellular matrix proteins and aorta maturation. In contrast the JAK/STAT pathway was predicted to be more active in patients >3 months (9 biopsies from 4 patients). Differentially expressed genes p < 0.05 with FDR adjustment were used for the pathway analysis.

In the biopsies from patients <3 months, there was a marked increase in production of a plethora of type 1 collagens, as well as other matrix components such as elastin (ELN), aggrecan (ACAN), fibrinogen type 2 (FBN2), Fibronectin 1 (FN1) and Lysyl oxidase (LOX). Our data indicates that this upregulation is mediated by several different pathways.

Both thrombospondin 1 (THBS1) and its receptor components integrin subunit β1 (ITGB1) and αV (ITGAV) were significantly overexpressed in patients <3 months, indicating an activation of the Hippo pathway and implicating involvement of YAP signaling.

In addition to the Hippo pathway, THBS1 has also been shown to lead to extracellular matrix accumulation by upregulating TGF-β (TGFB). Expression of TGF-β itself, as well as its receptor TGFBR1 and genes downstream from SMAD3 signaling including PGDA, CNN2 (CTGF) MMP2, MMP14 as well as various extracellular matrix components were significantly upregulated in biopsies from patients <3 months old.

We observed evidence of activation of the canonical Wnt/β-catenin pathway in biopsies from <3 month old patients in the form of the significant upregulation of several molecules reported to be part of this pathway – the Wnt molecule WNT2B as well as the Frizzled receptor FZD2 and the transcription factor TCF3. Part of the β-catenin destruction complex, AXIN1, but not other members of the complex (APC/GSK3β), was found to be significantly upregulated in biopsies from patients <3 months old. The Frizzled receptor FZD3, which has been implicated in non-canonical/β-catenin-independent signaling, was significantly downregulated in patients <3 months old.

In addition to the Wnt/β-catenin pathway, Notch3 signaling has also been shown to upregulate TCF3. Notch3 expression, as well as the expression of another transcription factor reported to be downstream of Notch3 – HES1 – were significantly upregulated in biopsies from patients <3 months. Downstream from Notch3, we could observe the activity of the Ephrin signaling pathway. We could see the significant upregulation of EFNB2, its receptor EPHB3 and SDC2, NCK2 and RGS3 directly downstream from these molecules in biopsies from patients <3 months old.

Expression of SMO and GLI were upregulated in biopsies from patients <3 months, implicating the activity of the Hedgehog pathway as well. The JAK/STAT pathway, however, appeared to be less active in biopsies from <3 months old patients compared to older patients. JAK expression, as well as levels of IL-6, IL-16 and IL-18 were significantly higher in the older patients, as well as several interferons/interferon-inducible genes (IFNK, IRF1, GVINP1, IFI44, IFI44L and IFIH1).

3.5. Analysis of ductus arteriosus tissue

For the patients where ductus tissue was able to be obtained (n = 4), genes differentially expressed between the ductus and proximal/distal tissue were calculated (Fig. 6, Supplementary Fig. 9). An upregulation in genes involved in mitosis in ductus compared to adjacent aortic tissue was observed, specifically components of the centromere (CENPA) and outer kinetochore (KLN1, ZWINT, NDC80, NUF2, SPC24, SPC25, and SKA3).

Fig. 6.

Factors associated with mitosis are upregulated in ductus compared to adjacent aortic tissue. Pathway analysis of differentially expressed genes in ductus biopsies (n = 4) compared to biopsies taken from aortic tissue proximal/distal to the coarctation revealed an upregulation of genes coding for components of the kinetochore in ductus tissue. The comparison analysis was paired according to patient. Differentially expressed genes p < 0.05 with FDR adjustment were analyzed. IPA “Kinetochore signaling pathway” 32 molecules, z-score 3.78, p = 5.38 × 10⁻³³.

4. Discussion

4.1. Clustering based on biopsy location, age and sex

To our knowledge, this is the first study of CoA patients that has performed transcriptomics of the resected tissue and the coarctation itself. Our study is limited by a small sample size (21 patients), however it provides valuable insight into the signaling pathways that characterize the coarctation. Our study also clearly illustrates the immense importance of controlling for patient age and sex when performing CoA studies.

While we observed distinct differences between biopsies taken from the localized coarctation compared to distal/proximal tissue, it was apparent that there were greater differences in gene expression in the samples attributable to patient characteristics than differences attributable to biopsy location. Patient sex (male vs female) and patient age (<3 months vs >3 months old) had a considerable impact on gene expression pattern, stipulating that future studies should take these factors into account to avoid results being contaminated by confounding factors.

It is possible that varying etiology of the coarctation in part can explain that the transcriptomic profiles differ between patients and raises the question whether coarctations with similar etiology have similar gene expression patterns.

4.2. Inflammatory activity in the localized coarctation compared to proximal/distal part of aorta

In this study, we compared the gene expression in biopsies from the localized coarctation to the gene expression of distal or proximal biopsies. Studies published by other groups have not sequenced the localized coarctation, but the aortic isthmus or ascending aorta. This could be due to the difficulty in extracting RNA from the localized coarctation, due to low cell- and high fibrous content.

While it would have been valuable to be able to compare biopsies from patients with CoA to biopsies of healthy individuals, this was not an option due to the nature of the material analyzed. Instead, we chose to use the surrounding tissue as a reference to identify pathways characteristic of the localized coarctation.

In our study, we found an increase in complement activation and acute phase response via enhanced NFkB, EGR2/3 and Jun/Fos signaling in the localized coarctation, whereas STAT3 signaling was decreased. A central role for Jun signaling has also been suggested by LaDisa et al. The group performed gene expression analysis using microarray on the ascending aorta of male rabbits where a suture was used to mimic coarctation, comparing the expression profiles of biopsies from rabbits with and without this suture [27]. A later study, also by LaDisa et al., compared transcriptome profiles of human biopsies proximal and distal to the coarctation and reported a downregulation of NPR3 in proximal compared to distal biopsies [28]. We could not detect differential expression of NPR3 between our proximal and distal biopsies (Supplemental data). Of note, the patients in the study by LaDisa et al. were considerably older than the patients in our study (average age of 6.5 years, compared to 0.5 years). Seeing that age has a considerable impact on gene expression in our study, it cannot be excluded that age can explain the difference between the observed results.

A study by Liu et al. compared gene expression between biopsies from aortic isthmus of 12 children with and 10 children without CoA [19]. While our study did not identify large clusters of differentially expressed genes associated with extracellular matrix and smooth muscle cell differentiation as described by Liu et al., coarctation appeared to be associated with an upregulation of acute phase processes and complement activation in both data sets [19]. This indicates that while there are higher levels of complement activation and acute phase signaling in the coarctation, the surrounding tissue is also affected, although to a lesser extent. This effect on the surrounding tissue also entails that some characteristics of CoA are not detectable in this study, as they are present in both the localized coarctation and in the proximal/distal tissue used as a reference. An important such aspect appears to be a dysregulation of extracellular matrix, as this will affect both the structure of the vessel as well as the bioavailability of growth factors and cytokines – as these molecules become tethered to components of the extracellular matrix. The regulation of growth factors and cytokines will in turn affect cell proliferation and migration. Extracellular matrix accumulation is an integral part of the natural aortic maturation, and dysregulation of this process likely has an extensive effect on the vessel [29,30].Of note, the patients studied by Liu et al. were older than our cohort, with no patients below 3 months of age, which is important to keep in mind considering that patient age correlates with extracellular matrix accumulation.

It has previously been suggested that increased levels of proinflammatory cytokines are associated with vascular remodeling in CoA patients. Young adults who had undergone CoA repair during adolescence have been found to have increased serum levels of sICAM-1, sSVCAM-1, E-selectin and IL-1b as well as endothelial dysfunction and increased carotid intima-media thickness and stiffness [31]. Arterial abnormalities with proximal aortic stiffening and increased wall thickness also occur in patients who underwent repair at a younger age, though there is discrepant data concerning the presence of endothelial dysfunction [32].

A prior proteomics study by Skeffington et al. evaluated CoA material of 10 children operated beyond the neonatal period (mean 1.8 years) and compared tissue of patients with an associated bicuspid aortic valve (BAV) to those with a tricuspid aortic valve (TAV) [33]. The authors found that several canonical pathways involved in inflammation including acute phase response signaling, EIF2 signaling, IL12 and reactive oxygen species (ROS) were upregulated on a protein level in the BAV patients. They suggested that increased inflammation could be responsible for the higher rate of aortic valve related procedures in CoA patients with a BAV later in life. These results, in combination with the findings in our study, suggest that while coarctation itself gives rise to inflammation, the presence of BAV exacerbates it. A study by the same group performed a similar study on neonatal CoA patients (<3 weeks old) found that the presence of BAV was associated with altered expression of proteins involved in elastin fiber formation and oxidative stress, ie different pathways than in the older patient group [33], once again highlighting the impact of patient age on gene and protein expression.

Complement activation is part of the inflammatory cascade and has previously been described in myocardial ischemia [34,35], where the complement activation and the generation of ROS help to initiate the inflammatory response. Our findings of upregulation of complement factor D (CFD), C1q and complement C3 indicate activation of the classical and alternative complement pathways in the coarctation.

Most neonates in our study were exposed to preoperative Prostaglandin E1 (n = 14/17), which can trigger an inflammatory response. However, the changes we found were localized to the specific tissue of the coarctation, compared to the other parts of the resected surrounding aortic tissue. Thus, the prostaglandin infusion was not the cause of this focal inflammatory response.

4.3. Inflammation and sex

Although inflammation was elevated in the localized coarctation for all patients, females tended to have higher inflammation compared to males. Males, on the other hand, appeared to have a response more skewed towards interferon signaling. This difference was mediated by diverging expression of members of the S100 family and associated with factors involved in estrogen and nitric oxide (NO) signaling.

In a recent study analyzing blood biomarkers in CoA patients by Hlebowicz et al., adult males but not females with a history of repaired CoA were observed to have significantly lower levels of TNF receptor 1 (TNF-R1) and TNF superfamily member 10B (TRAIL-R2) than healthy controls [36]. As TNF-R1 signaling has been suggested to inhibit arteriogenesis and angiogenesis [7], it has been suggested that possible changes in apoptosis mediated by TNF-R1 may play a role in the development of CoA. Of note, Luo et al. could not see this effect in female mice, and TNF-R1 has been shown to differ between sexes [37].

We identified factors associated with estrogen signaling to be differentially expressed in the aortic and coarctation tissue of males and females. In humans, estrogen has been reported to play a protective role in hypertension [38]. According to our pathway analysis, estrogen signaling was associated with the regulation of inflammation and antiviral responses via the S100 family.

S100 proteins are a large subfamily of low molecular weight calcium binding proteins, consisting of several isoforms with structural differences. The S100 proteins are required for inflammation and cellular homeostasis and are an important part of the innate immune response [39]. There is growing experimental and clinical evidence that the S100 protein family is involved in the occurrence and development of various cardiovascular diseases [40].

S100 signaling was differentially regulated in males compared to females, with males having increased expression of S100A8/A9 and S100A12 leading to IRF activation and interferon production while females had higher expression of S100A1 and S100A7, leading to higher inflammation.

It has previously been reported that males have a higher protein expression of S100A8/A9 in atherosclerosis plaques compared to females [41], which is in accordance with our observations for the localized coarctation. Absence of S100A1 in the vasculature has been reported to be associated with a hypertensive phenotype in mouse models [42]. In addition, in these models, the contribution of S100A1 in the regulation of blood pressure has been proposed to be sex specific, with males being more affected.

The group of S100 isoforms are clustered on chromosome locus 1q21, and we identified this locus to have significant sex-specific differential expression.

23.4 % of patients with a reported microdeletion of 1q21 have been shown to have cardiac abnormalities [43]. Interestingly, an earlier genomewide association (GWAS) study by Saw et al. (2020) showed an association of locus 1q21.21q21.3 (rs12740679) among patients with aortic diseases and the authors suggested a pathogenetic role of ADAMTSL4 [44], which is expressed in the medial layer of the arterial wall [45]. SNPs associated with spontaneous coronary aortic dissection (SCAD) have also been reported in 1q21.2, further underlining the importance of this locus in regulation and pathogenesis of the aorta [46].

While CoA is more common in males than in females, at a 2.5:1 ratio [47], there are reports that females with isolated CoA may have an increased mortality, even after repair. Sex has been reported to affect both incidence and pathogenesis of several aortic diseases [48] and the observed differences are often attributed to the X chromosome. The role of the X chromosome in the development of CoA is illustrated by the fact that CoA affects 12 % of women with Turner syndrome [49]. In the present study, the X chromosome was significantly over-represented among genes significantly upregulated in females compared to males, which is not surprising. The cluster of genes on 1q21 found to be overrepresented among genes differentially expressed between males and females contained both up- and downregulated genes. Although there are previous reports of some of the genes in the region to be associated with sex specific differences in signaling, the exact relationship between this cytogenetic region and sex and how this relates to aortic diseases, including CoA, remains to be elucidated.

4.4. Fibrosis and age

Factors associated with aorta maturation and extracellular matrix accumulation were found upregulated in patients <3 months old compared to patients >3 months old. During the neonatal period, the aortic wall stiffens through the accumulation of extracellular matrix. This is mediated by the production of collagen and other matrix components mainly by resident fibroblasts, although other cell types such as Sca-1(+) progenitor cells and bone marrow-derived infiltrating fibrocytes may also contribute [50].

Our data indicated that the increased expression of matrix proteins was mediated by several different pathways, including canonical TGF-β signaling, Wnt signaling and the Hippo and Hedgehog pathways. In addition, PXDN, a peroxidase produced by myofibroblasts that has been shown to contribute to extracellular matrix accumulation [51] was significantly upregulated.

The Hedgehog pathway has long been known to be vital for the formation and development of the aorta [52]. Our data indicated increased Hedgehog signaling in patients <3 months old, which is compatible with the ongoing aortic maturation in these individuals.

Expression of the adhesive glycoprotein THBS1, as well as its receptor components was significantly increased in patients <3 months. It has previously been shown that THBS1 expression is activated by mechanical cues derived from pulsatile blood flow and pressure [53]. THBS1 mediates cell-to-cell and cell-to-matrix interactions by binding to fibrinogen, fibronectin, laminin, collagens and integrin αVβ1 and is associated with both Hippo and TGF-β signaling. Our data indicate that both Hippo signaling and the canonical TGF-β pathway, via Smad3, are more active in patients <3 months old. TGF-β canonical signaling via Smad3 has a central role in fibrosis [54]. In some contexts, TGF-β has been reported to stimulate IL-11 production, leading to fibrosis [55]. IL-11 receptor IL11RA was significantly upregulated in biopsies from patients <3 months old. IL-11 itself, however, had near undetectable levels of expression in the biopsies. This is in accordance with reports that cells in the vascular wall of healthy humans have low IL-11 expression [55].

Interestingly, TGF-β expression as well as aortic stiffness on a physiological level have been shown to be increased in patients with repaired CoA [56]. In addition, TGF-β signaling plays an important role in the pathogenesis of aortopathies such as Loeys-Dietz, and Marfan syndrome [57].

We also observed evidence of activation of the canonical Wnt/β-catenin pathway in biopsies from <3 months old patients. Wnt signaling is critical for developmental processes, including cell proliferation, differentiation, and tissue patterning. Typically, Wnt signaling activity is only present during early cardiovascular development and not present in the cardiovascular system of healthy adults, with re-activation observed only during different pathologies [58]. Usually, TGF-β signaling is thought to inhibit the Wnt pathway. However, during response to vascular injury canonical TGF-β signaling and elevated Smad3 have been shown to enhance the canonical Wnt/beta-catenin pathway instead [59]. It is likely that a similar mechanism, leading to concurrent activation of both pathways, is active during the remodeling process around the aortic isthmus.

We could also observe increased activity of Notch3 signaling, leading to the Ephrin signaling pathway. While Notch signaling has been shown to promote arterial differentiation of endothelial cells, Ephrin signaling controls the repulsive sorting of arterial- and venous-fated endothelial cells into their respective arteries and veins [60]. Together, these pathways appear to be involved in the ongoing aortal development in young patients.

In contrast to the pathways mentioned above, the JAK/STAT pathway appeared to be less active in biopsies from <3 months old patients, and instead upregulated in the older patients. The JAK/STAT pathway has been reported to be activated by chronic inflammation, profibrotic cytokines and growth factors in the context of hypertension, where it leads to aortic remodeling [61].

In the clinical context, these results can indicate a predominant activity of fibrotic remodeling with fibroblast proliferation and markers of arterial endothelial cell differentiation in our younger patients operated for CoA.

4.5. Analysis of ductus arteriosus tissue

The DA is a vessel that bridges the systemic circulation with the pulmonary circulation during fetal life. Closure of the DA occurs spontaneously after birth in most infants. Functional closure of the DA usually occurs within the first few hours after birth, whereas the anatomical closure process takes several weeks to complete [62]. In neonates with severe CoA, the DA is needed to provide blood to the systemic circulation, and when the ductus closes, the patient risks circulatory shock. Therefore, Prostaglandin E1 is administered to relax smooth muscle cells in the DA and the isthmus area to restore the blood flow until the CoA can be operated and the aorta repaired.

In our few biopsies from ductus tissue, we found an upregulation in genes involved in mitosis. This is likely a result of smooth muscle cell proliferation, as proliferation and migration of medial smooth muscle cells into the subendothelial region has been described to initiate thickening of the intimal cushions and thereby anatomical closure of the DA [63]. In a model using swine, proliferation was found to increase in ductus tissue 48 h after birth, becoming most prominent at 3 days, and decreasing at 5 days to disappear 2 weeks after birth [63]. In our study, the 4 ductus tissues analyzed were derived from patients 5–12 days old, meaning that increased proliferation is in accordance with the observations in the swine model.

Of note, gene expression in the DA has been shown to vary between closing and PDA [64] and may also be affected by Prostaglandin E1 administration. All of the patients in this study where ductus tissue was available had ongoing Prostaglandin E1 treatment, and none were reported to have PDA.

5. Study limitations

This study is explorative in nature and has several limitations. One major limitation is the absence of a cohort of healthy controls. This, however, is not possible as healthy infants do not undergo aortic surgery and postmortem samples would not serve as adequate controls. The patient group is heterogenous, both concerning age and clinical presentation with some patients having extracardiac manifestations. Our ability to perform external validation of our findings was limited by the scarcity of previously published gene expression data on relevant tissue, and a lack of this type of data for neonates. Our results using RNAseq have not been verified with alternative methods. More research is needed to validate our findings.

6. Conclusions

There is an inflammatory process in the localized coarctation, characterized by an acute response and complement activation. The surrounding tissue is also affected, but to a lesser extent. The transcriptome of the coarctation area is largely dictated by patient characteristics such as sex and age. Males tended to have a response slightly more skewed towards interferon, while females had a response more skewed towards inflammation. This difference appeared to be partly mediated by contrasting expression of genes located on 1q21 between the sexes and initiated by differences in estrogen and NO signaling. Younger patients (<3 months old) had a much more active fibrinogenesis than older patients. Finally, there was an upregulation of cell proliferation in the ductus of young patients (5–12 days old), signifying an ongoing anatomical closure of the ductus.

Taken together, our results offer insight into the complexity of the gene expression of the coarctation area and underline the importance of controlling for patient sex and age in future CoA studies.

Funding

This research was funded by ALF Grants, Region Östergötland, grant number RÖ-974174, RÖ-966451, RÖ-417841, RÖ-392391. Medical research council of Southeast Sweden, grant numbers FORSS-937721, FORSS-982242, FORSS-941313, FORSS-909301. The Swedish Heart Lung Foundation, grant number 20220418.

Declaration of generative AI and AI-assisted technologies in the writing process

AI-assisted technology was not used in the preparation of this work.

CRediT authorship contribution statement

Rada Ellegård: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Torsten Malm: Writing – review & editing, Writing – original draft, Supervision, Software, Resources, Methodology, Investigation. Constance G. Weismann: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization. Eva Fernlund: Writing – review & editing, Writing – original draft, Software, Methodology, Investigation, Funding acquisition, Conceptualization. Anneli Nordén Björnlert: Writing – review & editing, Validation. Hanna Klang Årstrand: Writing – review & editing. Katarina Ellnebo-Svedlund: Validation. Cecilia Gunnarsson: Writing – review & editing, Writing – original draft, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary figures

Data availability

Transcriptome data (raw counts as well as DESeq2 statistical analysis) has been made available by submission to the gene expression omnibus (GEO), GSE record GSE263570.