RNA sequencing indicates age-dependent shifts in the cardiac fibroblast transcriptome between fetal, neonatal, and adult developmental ages

Abstract

Cardiac fibroblasts are responsible for extracellular matrix turnover and repair in the cardiac environment and serve to help facilitate immune responses. However, it is well established that they have a significant phenotypic heterogeneity with respect to location, physiological conditions, and developmental age. The goal of this study was to provide an in-depth transcriptomic profile of cardiac fibroblasts derived from rat hearts at fetal, neonatal, and adult developmental ages to ascertain variations in gene expression that may drive functional differences in these cells at these specific stages of development. We performed RNA sequencing (RNA-seq) of cardiac fibroblasts isolated from fetal, neonatal, and adult rats and compared with the rat genome. Principal component analysis of RNA-seq data suggested that data variance was predominantly due to developmental age. Differential expression and gene set enrichment analysis against Gene Ontology and Kyoto Encyclopedia of Genes and Genomes datasets indicated an array of differences across developmental ages, including significant decreases in cardiac development and cardiac function-associated genes with age and a significant increase in immune- and inflammatory-associated functions, particularly immune cell signaling and cytokine and chemokine production, with respect to increasing developmental age. These results reinforce established evidence of diverse phenotypic heterogeneity of fibroblasts with respect to developmental age. Furthermore, based on our analysis of gene expression, age-specific alterations in cardiac fibroblasts may play a crucial role in observed differences in cardiac inflammation and immune response observed across developmental ages.

INTRODUCTION

Although cardiomyocytes (CMs) have historically been the therapeutic target for bioengineering approaches to heart repair (1, 2), noncardiomyocyte cell populations in the heart often play a critical role in the heart’s response to injury and, as such, have been subject to increased appreciation and study over the past few decades. Nonmyocyte cell populations, which include vascular cells, endothelial cells, immune cells, and cardiac fibroblasts (CFs), comprise ∼70% of the total cell population in the adult mammalian heart (3–5), and CFs are the major fraction of these nonmyocyte cells. CFs, in particular, play a major role in mediating cardiac homeostasis, facilitating the development and maturation of cardiac extracellular matrix (ECM), and promoting cardiomyocyte proliferation in utero (6), and yet they are also integral in the initiation and progression of various cardiac pathologies (7–10).

CFs are classically defined as mesenchymal-like cells responsible for extracellular matrix (ECM) production and remodeling (10). These cells are largely quiescent in the adult mammalian heart but exhibit extensive phenotypic plasticity, able to react to stimuli following injury and transition to a myofibroblast phenotype. In contrast to CFs in a resting state, myofibroblasts are highly active, migrating into and proliferating rapidly in the wound environment, where they extensively upregulate collagen I production to produce mature fibrotic scar tissue in adult mammals (9, 11). However, the role of CFs in responding to injury extends beyond ECM regulation and fibrosis. CFs are an important immune mediator and contributor to the inflammatory process (12). One study by Kawaguchi et al. (13) indicated that mouse CFs exposed to lipopolysaccharide (LPS), an inflammatory stimulus, promoted interleukin-1β (IL-1β, a major inflammatory mediator) expression in those cells, but not in CMs. The CFs were found to act as “sentinel” cells in the heart, deploying a broad range of chemokines including monocyte chemoattractant protein (MCP)-1, macrophage inflammatory protein-1, and RANTES (Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted) to aid immune cell influx into a wound environment (13, 14). The immune function of fibroblasts is broadly, and extensively, established in a recent paper by Krausgruber et al. (15) that presented a transcriptomic profile of mouse structural cells across organs, finding that fibroblasts, epithelial, and endothelial cells all present complex immune gene activity and regulation and these responses could be triggered in response to viral infection.

Despite their critical roles in the heart, CFs are loosely defined with respect to their phenotype, which varies considerably across developmental ages. The cell type traditionally considered a fibroblast is now understood to be a heterogeneous population of cells arising from different progenitors (16, 17), with a lack of specific markers to establish their identity in vitro, although CFs from both fetal and adult sources express periostin, vimentin, and discoidin domain receptor 2 (DDR2) (18, 19). Furthermore, through cross talk with cardiomyocytes, CFs influence cardiac response to injury differently with respect to developmental age. Fetal CFs are implicated in driving cardiomyocyte growth and proliferation via secretion of paracrine factors, whereas adult CF-secreted factors are implicated in inducing cardiac hypertrophy, findings that reflect the regenerative healing observed in fetal and neonatal hearts, and fibrosis and hypertrophy observed in injured adult cardiac tissue (17, 20). Furthermore, a single-cell RNA-seq study has established that transcriptomic changes in mouse CFs from a neonatal to adult state promote concomitant CM maturation, connecting age-specific CF phenotypic changes with cardiac maturation (21). However, despite the work noted above highlighting conclusive differences in fetal versus adult fibroblasts, there is scarce literature currently available with transcriptomic profiling of neonatal CFs, representing a gap in our understanding of this cell type through development. As there is a significant, rapid shift in ECM composition (22, 23) regenerative capacity (24) and mechanical function of the heart pre- to postbirth, there is considerable value in highlighting the neonatal phenotype.

Studies of CF behavior across various developmental ages are invaluable to further elucidating the identities and functions of this heterogeneous cell population and how it changes with age, particularly with respect to the healing response of cardiac tissue. Many analyses have investigated how CF developmental age impacts cardiac structure and function or induces functional change in CMs (2, 6). By contrast, an understanding of how CF gene expression related to immune infiltration and the inflammatory response change with age is not well-established, despite the critical roles of these cells in cardiac inflammation and immunity and how critical these processes are to the healing response (2, 14, 25, 26). In this study, we employ RNA-seq transcriptomic analysis to evaluate rat cardiac fibroblast gene expression at fetal, neonatal, and adult developmental ages. This transcriptomic profile presents a broad view of CFs across development and aging to facilitate age-specific analyses of these cells. However, we specifically investigate age-dependent changes in immune and inflammatory gene expression, speculating that CF involvement in these processes increases with development. We found that compared with postnatal day 1 neonatal CFs, fetal CFs display significantly downregulated immune- and inflammatory-implicated genes and pathways, whereas adult CFs markedly increased expression of immune and inflammatory proteins. Taken together, this suggests that CFs adopt an inflammatory/immune-support role rapidly after birth, which is further enhanced with continued development.

METHODS

Heart Harvest from Fetal, Neonatal, and Adult Rats

All animal procedures were performed in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee at Tufts University. Rat hearts harvested for cell isolation were from fetal, neonatal, and adult Sprague–Dawley rats. Pregnant dams (∼3 mo old) and adult rats (∼7 wk old) were deeply anesthetized with 3%–5% isoflurane and euthanized via heart removal. Fetal pups (embryonic day 18, E18) were isolated from the uterus, euthanized by conscious decapitation, and hearts were isolated. Neonatal pups (postnatal day 1, P1) were euthanized by conscious decapitation before heart harvest. Hearts were stored in an ice-cold solution of sterile phosphate-buffered saline (PBS) with 20 mM glucose and immediately used for cell isolation, as described below.

Primary Cardiac Fibroblast Isolation and Culture

Cardiac fibroblasts were isolated from fetal, neonatal, and adult rat hearts following previously described methods (27). After euthanasia and heart harvest, adult heart left ventricles were separated from the heart and minced to pieces less than 1 mm³ in size. Fetal and neonatal hearts were minced whole. Minced tissue underwent three to five serial digestions in collagenase type 2 (Worthington Biochemical Corp, Lakewood, NJ) in sterile PBS with 20 mM glucose and passed through a 40-µm cell strainer and centrifuged. CFs were separated from other cardiac cells via collection as fast-adhering cells in a 30-min preplating in Dulbecco’s Modified Eagle’s Medium (DMEM) with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (Invitrogen), and 0.1 mM ascorbic acid, followed by removal of nonadherent cells and reapplication of media (2). Fibroblasts for RNA-seq were expanded to confluency in culture and used immediately for RNA isolation. CFs for all other experiments were used at passage 2–4.

Immunofluorescence

Immunofluorescent imaging was used to validate that the cells used in culture were predominantly cardiac fibroblasts and not from another nonmyocyte cell lineage. To confirm this, cells were stained for vimentin (a broad marker for fibroblasts but also expressed in endothelial cells; Refs. 28 and 29), and either CD31 or CD45 to identify endothelial or hematopoietic cells, respectively (5, 30). CFs at passage 2–4 were seeded at a density of 50,000 cells/cm² for 24 h and then fixed for 10 min with ice-cold 4% paraformaldehyde. Cells were permeabilized with 0.05% Triton X-100 in PBS for 10 min at room temperature, rinsed in PBS, and blocked with 5% normal donkey serum and 1% bovine serum albumin (BSA) in PBS overnight at 4°C. After blocking, cells were incubated with primary antibodies, against vimentin (Cell Signaling Technology, #5741, dilution 1:200), CD31 (Abcam, ab119339, dilution 1:500), and CD45 (Abcam, ab10558, dilution 1:250) diluted in 1% BSA in PBS overnight, again at 4°C. Cells were rinsed in PBS with 0.1% Tween-20 (PBST) before incubation with fluorescently labeled secondary antibodies (Thermo Fisher Scientific; Donkey Anti-Rabbit 488-A21206, Donkey Anti-Mouse 488-A10037) for 2 h, followed by Hoescht (10 µg/mL) and phalloidin (Invitrogen 588-A12380, diluted according to the manufacturer’s instructions, ∼125 µg/mL) for 30 min at room temperature. Cells were rinsed in PBST again and immediately imaged on a Keyence BZ-X710-fluorescent microscope.

RNA Preparation for Sequencing

P0 neonatal, fetal, and adult Sprague–Dawley rat cardiac fibroblasts were cultured to confluency in T75 flasks, in serum-containing media with 0.1 mM ascorbic acid. Upon reaching confluent density, the cells were lysed in 1 mL of ice-cold TRIzol reagent (Invitrogen) and RNA was isolated via manufacturer’s instructions: briefly, 200 µL of chloroform was added to each milliliter of TRIzol, mixed vigorously, incubated on ice for 2–5 min, and centrifuged at 4°C at 12,000 g for 15 min. The aqueous upper layer containing RNA was isolated and purified of chemical contaminants and DNA via processing in a SurePrep RNA Cleanup and Concentration Kit, and treatment with Thermo Fisher Scientific RapidOut DNA Removal Kit, both following manufacturer’s protocols. RNA was then quantified on a Nanodrop 2000 UV-vis spectrophotometer (Thermo Fisher Scientific) and stored at −80°C until further use.

RNA Sequencing

Purified RNA samples from fetal, neonatal, and adult rat hearts (n = 3 for each age group) were delivered to the Tufts University Genomics Core Facility for quality certification, cDNA library construction (Illumina TruSeq Stranded mRNA), and sequencing. Sequencing was performed on the Illumina HiSeq 2500 platform (50-bp single-end lane). Sequencing data in fastq format can be accessed in the Gene Expression Omnibus database (GEO) with the accession number GSE162277. The alignment of sequenced reads to the rat reference genome (Rn6 rat genome from the UCSC Genome Browser) was performed in several steps. First, sequenced data were adapter trimmed using Trimmomatic 0.39 (31). Pre- and posttrimming, raw sequence data were quality validated using FASTQC (32). Trimmed and quality-assessed data were loaded into STAR (v. 2.5.2b) (33) for read alignment to the Rn6 rat genome. STAR-aligned reads stored as BAM files loaded into featureCounts (34) for read counting. Raw counts were imported into R for differential gene expression analysis.

Differential Gene Expression Analysis

Raw counts were imported into R for gene expression analysis, using the DESeq2 package, using the workflow described in the DESeq2 manual (35). Briefly, DESeq2 normalizes counts via a median of ratios method, in which counts are divided by sample-specific size factors determined by the median ratio of gene counts, relative to the geometric mean per gene (36). Principal component analysis (PCA) was performed with the plotPCA function in R, using natural log-transformed expression data from DESeq2. Normalized gene expression data from the three groups was output into a hierarchical clustering heatmap with the pheatmap package in R. These analyses were done to check for batch effects and to broadly evaluate the gene expression data across the three age groups. Calculation of number of reads per sample (in millions of reads), correlation between samples, analysis of loadings of individual genes in the principal components, and clustering of individual genes with respect to principal components was performed in pcaExplorer (37), which utilizes the DESeq2-transformed data in its gene set analyses.

The neonatal cardiac fibroblast group (NCF) was assigned as the control condition for analysis, and differential expression comparisons to adult cardiac fibroblasts (ACF) and fetal cardiac fibroblasts (FCF) were calculated, although the comparison case ACF versus FCF was also analyzed and included in results. The full datasets preranked by log₂FC were imported into the clusterProfiler package in R (38), which includes methods incorporating Gene Set Enrichment Analysis (GSEA) (39) and analyzed against Gene Ontology (GO) (40) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (41) databases to identify enriched gene sets and biological pathways. GSEA was run with gene set size thresholds of a minimum of 3 and maximum of 500, P value cutoff of 0.05, and P value adjustment using the Benjamini–Hochberg method (42). Separately, expression results were filtered against a cutoff log₂FC and adjusted P value [abs(logFC) ≥ 1.5, P_adj < 0.05] to create datasets of significantly differentially expressed genes for downstream evaluation.

RT-qPCR Gene Expression Validation

CFs at passage 2–4 were lysed and RNA isolated and purified using a SurePrep DNA/RNA/Protein Concentration Kit, following manufacturer’s protocols. RNA was quantified on a Nanodrop and stored at −80°C until use. Following RNA purification, RNA was converted to cDNA using a Thermo Fisher Scientific High Capacity cDNA Reverse Transcriptase Kit and cDNA was stored at −80°C until further analysis. Reverse transcriptase quantitative PCR (RT-qPCR) was performed using an Applied Biosystems TaqMan Gene Expression Master Mix, carried out on a BioRad CFX96 Real-Time System using TaqMan PCR primers (Thermo Fisher Scientific), according to manufacturer instructions using 50 µg of cDNA per reaction. The genes assayed were COL1A1 (Collagen 1 α1 chain, Rn01463848_m1), MMP2 (matrix metalloproteinase 2, Rn01538170_m1), MMP9 (Rn00579162_m1), IL-6 (interleukin-6, Rn01410330_m1), myosin heavy chain β (MYC7, Rn01488777_g1), and GAPDH (Rn01775763_g1). Genes of interest were normalized to GAPDH expression, with comparative C_T used for analysis using neonatal cardiac fibroblasts as a control group.

Statistical Analysis

Statistical analysis of differential expression data was performed entirely in the R programming environment. As stated above, when identifying significantly expressed genes, a cutoff adjusted P value of 0.05 was applied. For qPCR evaluation, statistical significance was determined via a one-tailed t test, with significance set at P < 0.05. Statistical analysis and data visualization of qPCR data was done in the GraphPad Prism environment.

RESULTS

Fibroblast Sequencing, Culture, and Characterization

Cardiac fibroblasts (CFs) were isolated from fetal (E18), neonatal (P1), and young adult (∼7-wk-old, at which point the rats are sexually mature; Ref. 43) rat hearts. After preplating the isolated mixed cardiac cells (44, 45), isolates were cultured to confluency and used for RNA sequencing or expanded out to passages 2–4 for additional experiments and culture validation (Fig. 1). RNA-seq mapping to the Rn6 rat genome yielded a total of 17,337 discrete genes mapped. Differential expression analysis yielded a total of 12,786 genes of interest between comparison conditions (adult vs. fetal, adult vs. neonatal, and fetal vs. neonatal) (Fig. 1).

Figure 1.

Pipeline for RNA isolation and differential expression analysis. RNA is mapped to Rn6 rat genome via STAR and reads counted, yielding 17,337 distinct observations. These are normalized and used in differential expression analysis in DESeq2 and then used directly in Gene Set Enrichment Analysis (GSEA) for enrichment analysis or filtered via log fold change (LogFC) and adjusted P value to evaluate significantly expressed genes. Inset: schematic depicting isolation of three distinct ages of rat cardiac fibroblasts, followed by culture and RNA isolation. [Created with BioRender.com and published with permission.]

CFs grown in culture were imaged following immunofluorescence labeling, and fibroblast lineage was confirmed by demonstration that the cells were negative for markers of endothelial (CD31) (Fig. 2, D–F) or hematopoietic (CD45) lineages (Fig. 2, G–I), respectively (30). Cells were positive for vimentin (Fig. 2, A–C), a broad CF marker (28, 46). Cultured fibroblasts all displayed similar elongated morphology and positive vimentin expression throughout (Fig. 2).

Figure 2.

Immunofluorescence microscopy characterization of primary P2–P4 cardiac fibroblast cultures. Vimentin staining of fetal (A), neonatal (B), and adult (C) cells to identify cardiac fibroblasts. CD-31 staining of fetal (D), neonatal (E), and adult (F) cells to identify endothelial-lineage cells in culture. CD-45 staining of fetal (G), neonatal (H), and adult (I) cells to identify hematopoietic-lineage cells in culture. Scale bars = 100 µm.

Clustering Analysis Reveals Broad Variances in Gene Expression Based on CF Age

Sequencing data of fetal, neonatal, and adult CFs (groups referred henceforth as FCF, NCF, and ACF, respectively) were analyzed and normalized before differential expression analysis was executed, all in the DESeq2 package in R. Sample-to-sample Manhattan distance was computed and a heatmap showing sample distance and correlation was produced (Fig. 3A).

Figure 3.

Normalized RNA-seq sample data from adult cardiac fibroblasts (ACFs), neonatal cardiac fibroblasts (NCFs), and fetal cardiac fibroblasts (FCFs). A: sample-to-sample distance heatmap utilizing Manhattan-distance calculation, showing close clustering of FCF and NCF samples. Heatmap produced in pcaExplorer in R. B: heatmap of normalized gene counts from all identified genes, showing clustering of genes in the three age groups. Genes are row-normalized, and heatmap color is based on z-score. Heatmap produced with R software.

The heatmap indicated a stronger correlation between the FCF and NCF samples. FCF samples had the lowest level of similarity to ACF samples, followed by NCFs, suggesting that the sample variability was strongly age dependent. This was supported by unsupervised hierarchical clustering of the DESeq2-normalized gene set across all samples, visually presented in the heatmap in Fig. 3B FCFs and NCFs clustered together, whereas ACFs diverged greatly in their gene expression compared with the two younger CF sources.

Principal component analysis (PCA) was evaluated with respect to samples (Fig. 4A) and genes (Fig. 4, B and C). The sample PCA supported the hypothesis that variation in samples was primarily due to age, as it indicated sample distribution along the first principal component (constituting ∼92% of total variance in the dataset) coincided with the associated age of the fibroblast sample source (Fig. 4A).

Figure 4.

Principal component analysis with associated loadings with respect to gene expression in fetal, neonatal, and adult rats. A: principal component analysis (PCA) of samples with respect to first two principal components, showing broad distribution with respect to age on the first PC axis. All graphs were generated in pcaExplorer package in R. B: loadings plot for top/bottom 10 highest contributing genes of first and second principal components. C: PCA of top 50 genes contributing most variance to the first two principal components, showing relationship and location with respect to fetal (FCF), neonatal (NCF), and adult (ACF) samples. ACF, adult cardiac fibroblast; FDF, fetal cardiac fibroblast; NCF, neonatal cardiac fibroblast; PC, principal component.

Analysis of the 50 genes that most significantly contributed to the PCA variance produced two distinct clusters, with respect to the first principal component, which appeared to be associated with CF age. This suggested greater variance between ACFs and the two juvenile CF groups (FCF and NCF), with the juvenile groups associated with genes that clustered closer together. Genes associated with the ACF cluster were broadly immune associated, including chemokines Ccl22, Ccl17, and Ccl3 and cell surface receptor Cd74. The NCF/FCF cluster was observed to be predominantly associated with genes related to cardiac and embryonic development, including Myh6, Myh7, Nppa, Myl2, Myl3, and Actc1 (Fig. 4, B and C).

Differential expression analysis further supported the transcriptomic variance between adult CFs and neonatal and fetal CFs: post hoc filtering (log₂FC > 1.5, P_adj > 0.05) produced 2,794 significant differentially expressed genes in ACFs versus FCFs and 2003 significant differentially expressed genes in ACFs versus NCFs, versus 784 genes differentially expressed between FCFs and NCFs (Fig. 1). The top 25 most significantly upregulated and downregulated genes from each comparison are presented in Table 1.

Table 1.

List of 25 most significantly upregulated and 25 most significantly downregulated genes from Fetal vs. Neonatal, Adult vs. Neonatal, and Adult vs. Fetal comparisons

Fetal vs. Neo		Adult vs. Neo		Adult vs. Fetal
Genes	log2FoldChange	Genes	log2FoldChange	Genes	log2FoldChange
Upregulated
Upk3bl	5.919187	Ccl22	10.28234	Clec10a	11.58615
Tbx4	5.711532	Cd74	9.176081	Serpina3n	11.07736
Hoxb7	5.665582	Ccl17	8.783643	Asgr2	10.40074
Igfbp5	5.59857	Asgr2	8.239064	Ccl22	10.36736
Myog	5.505488	Serpina3n	8.080356	Ccl17	9.922081
Dhrs9	5.362969	Mmp7	7.744266	Pla2g2a	9.609374
Tenm4	5.193547	Pla2g2a	7.512271	Akr1c14	9.354025
Gldn	5.130201	RT1-Ba	7.50239	Ccl4	9.265844
Hoxb5	5.100569	Irg1	7.486625	Ccl3	9.027888
Mfap2	5.052972	Chi3l1	7.470716	LOC24906	9.026394
B3gnt5	4.977134	Rgs9	7.464147	Nrros	8.722336
Fendrr	4.923848	Clec10a	7.399204	Cd300lf	8.613316
Mybpc1	4.773853	Cst7	7.368542	Chi3l1	8.451384
Hoxb8	4.760618	Plet1	7.300047	Serping1	8.380459
Stmn2	4.745606	Ccl4	6.978046	Arhgap9	8.349699
Mstn	4.73928	Irf4	6.931538	Ch25h	8.338975
Wnt16	4.575537	RT1-Da	6.926247	Arhgap25	8.31574
Ppp2r2c	4.567238	Milr1	6.76698	Stap1	8.28389
Cdh8	4.46549	Ccl20	6.640746	Irg1	8.245699
Chrna1	4.405294	RT1-Bb	6.541552	Lbp	8.199643
Tnnc2	4.373668	Ccl9	6.532978	Mmp7	8.183537
Fam150a	4.267047	Klkb1	6.498334	Irf4	8.140108
Krt15	4.174717	Ccl3	6.464289	Atp6v0d2	8.12108
Glp1r	4.155335	C3	6.40599	Mcemp1	8.063229
Hmga2	4.136327	Atp6v0d2	6.403729	Ccl7	8.046836
Downregulated
Akr1c14	−5.566468285	Myh7	−11.30766504	Myh7	−12.48441723
LOC24906	−4.802399558	Actc1	−11.19711678	Actc1	−12.31689974
Figf	−4.586929021	Myh6	−10.97836111	Myh6	−11.8069547
Mrgprf	−4.403228606	Nppa	−10.7449663	Myl2	−11.49213976
Clec10a	−4.186947411	Myl2	−9.924278413	Nppa	−11.4027724
Cpxm2	−4.011002786	Myl3	−9.308555082	Tnni1	−10.60890482
Slc47a1	−3.90131184	Sh3bgr	−8.980427683	Sh3bgr	−10.41421189
Dpysl4	−3.846230821	Tnni1	−8.978986082	Myl3	−10.37610834
Mnda	−3.770006226	Csrp3	−8.766944839	Csrp3	−9.909283343
Scn7a	−3.679321021	Acan	−8.489888441	H19	−9.635111663
Acot1	−3.536174658	Actn2	−8.410946329	Actn2	−9.256275492
Mfrp	−3.489356287	Nrap	−8.299109006	Smpx	−9.170751224
Ceacam1	−3.467994934	Casq2	−8.111326289	Tnnc1	−9.157876842
Mageb4	−3.46534003	Smpx	−7.849079831	Unc45b	−9.056882513
Nrros	−3.444183605	Myl7	−7.810108718	Nrap	−8.993720954
Upk1a	−3.432813169	Pln	−7.761398734	Casq2	−8.713779547
Car1	−3.422033795	Unc45b	−7.643272419	Synpo2l	−8.684113331
Sepp1	−3.373467452	Tnnc1	−7.550449335	Acan	−8.642440402
Abca8a	−3.270056117	Nppb	−7.548924581	Ryr2	−8.530608706
Lbp	−3.26664172	Reg3b	−7.544072558	Nppb	−8.448727672
Krt24	−3.239943831	Xirp2	−7.541372318	Alpk3	−8.439659492
Pmfbp1	−3.198896228	Ryr2	−7.398237956	Dusp27	−8.40961941
Tlr7	−3.187350765	Ldb3	−7.351657713	Pln	−8.380649219
Clec7a	−3.179790031	Synpo2l	−7.268467268	Ldb3	−8.365997859
Akr1c1	−3.122931155	Tnni3	−7.157721088	Reg3b	−8.229578394

Neo, neonatal.

Additional descriptions of functions of the top 10 most significantly upregulated and downregulated genes from each comparison are presented in Supplemental Table S1; all supplemental material is available at https://doi.org/10.6084/m9.figshare.14823915.v2. A preliminary investigation of gene distribution within these comparisons was conducted using volcano plots generated in the EnhancedVolcano R package, using a log fold change (logFC) cutoff of 1.5 and P value cutoff of 0.001 as parameters. The plot of FCFs versus NCFs indicated that immune-associated genes such as Ifitm1 and Lbp and matrix regulation-associated genes such as Ctsc, Figf, and Itgb3 were downregulated in FCFs compared with NCFs, whereas developmental genes such as TBX4, Mdk, Hoxb7, and Hoxb8 were upregulated (Fig. 5A and Table 1). ACF comparisons to NCFs and FCFs yielded similar gene distributions in the volcano plots (Fig. 5, B and C). In both comparisons, cardiac development-associated genes such as Actc1, Actg2, Tnnc1, and Myh7 were downregulated in ACFs versus the younger group, whereas immune- and inflammatory-associated genes such as Tgfb1, Ccl3, and Ccl4 were observed in ACFs to be comparatively upregulated (Fig. 5, B and C, and Table 1).

Figure 5.

Volcano plots of significant differentially expressed genes. A: fetal vs. neonatal, B: adult vs. neonatal, C: adult vs. fetal, with boundaries set at log fold change (logFC) of 1.5, and P value threshold of 10⁻¹⁰. Genes color-coded yellow indicate significant LogFC but not statistically significant. Purple genes indicate statistical significance but genes under absolute value of LogFC threshold. Brown color-coded genes meet LogFC threshold and statistical significance. Volcano plots generated in EnhancedVolcano package in R. NS, no significance.

Differential Expression Indicates Age-Specific Shifts in Immune Signaling and Cardiac Development

Differentially expressed gene lists from the three separate comparisons (FCF vs. NCF, ACF vs. NCF, ACF vs. FCF) were ranked based on fold change and subsequently analyzed using Gene Set Enrichment Analysis (GSEA) methodology (39), within the clusterProfiler package in R (38). The full lists of differentially expressed genes identified in these analyses are available in Supplemental Table S3 (Fetal vs. Neonatal), Supplemental Table S5 (Adult vs. Neonatal), and Supplemental Table S7 (Adult vs. Fetal). These datasets were subjected to analysis utilizing Gene Ontology (GO, analyzing for Biological Process, Cellular Component, and Molecular Function terms) (40) and Kyoto Encyclopedia of Genes and Genomes (KEGG) curated datasets (41). Analysis of fetal CFs versus neonatal CFs against GO datasets yielded an array of significantly downregulated ontologies associated with immune response and immune cell signaling (Fig. 6A and Table 2), including “leukocyte-mediated immunity,” “positive regulation of immune response,” and “activation of immune response.”

Figure 6.

Analysis of functionally enriched Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) gene sets comparing three ages of cardiac fibroblasts, via Gene Set Enrichment Analysis (GSEA). Dot color is indicative of −log₁₀(q value), and size is indicative to number of identified core-enriched genes. Gene ratio is the ratio of identified core-enriched genes vs. the total number of genes associated with that curated gene set. A–C: dot plots indicating top 5 upregulated and downregulated general GO gene sets for fetal vs. neonatal, adult vs. neonatal, and adult vs. fetal, respectively. GSEA analysis cutoff P value was 0.05. D–F: dot plots indicating top 5 upregulated and downregulated KEGG gene sets for fetal vs. neonatal, adult vs. neonatal, and adult vs. fetal, respectively. GSEA analysis parameters were set to filter out gene sets with less than 3 genes or greater than 500, or functionally enriched gene sets with P value > 0.05. Graphs produced in the R programming environment using the clusterProfiler software package.

Table 2.

Ten most significant (based on q value) GO and KEGG results from GSEA analysis of Fetal vs. Neonatal differential expression

GO	ID	Description	setSize	NES	q Values	Count	GeneRatio
BP	GO:0048706	Embryonic skeletal system development	98	2.211	1.22E-07	41	0.418
BP	GO:0048704	Embryonic skeletal system morphogenesis	77	2.164	2.57E-06	27	0.351
BP	GO:0030239	Myofibril assembly	63	2.133	1.94E-05	31	0.492
BP	GO:0045214	Sarcomere organization	47	2.116	0.000121	25	0.532
BP	GO:0010226	Response to lithium ion	38	2.110	0.000178	11	0.289
BP	GO:0002443	Leukocyte-media on of immune response	457	−2.256	5.37E-08	144	0.315
BP	GO:0050778	Positive regulation of immune response	334	−2.285	5.37E-08	107	0.320
BP	GO:0045087	Innate immune response	372	−2.287	5.37E-08	144	0.387
BP	GO:0002253	Activation of immune response	174	−2.305	5.37E-08	51	0.293

KEGG	ID	Description	setSize	NES	q Values	Count	GeneRatio
	rno04260	Cardiac muscle contraction	70	1.775	0.006495	33	0.471
	rno05217	Basal cell carcinoma	44	1.742	0.0082	14	0.318
	rno04390	Hippo signaling pathway	128	1.687	0.003496	28	0.219
	rno04080	Neuroactive ligand-receptor interaction	192	1.672	0.000687	78	0.406
	rno03040	Spliceosome	113	1.659	0.006495	79	0.699
	rno05152	Tuberculosis	137	−2.018	4.61E-06	45	0.328
	rno04145	Phagosome	131	−2.038	2.95E-06	48	0.366
	rno04621	NOD-like receptor signaling pathway	130	−2.110	1.80E-06	55	0.423
	rno05323	Rheumatoid arthritis	65	−2.192	4.61E-06	28	0.431
	rno04620	Toll-like receptor signaling pathway	77	−2.216	2.95E-06	24	0.312

GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; NES, Normalized Enrichment Score; NOD, nucleotide binding oligomerization domain.

Core-enriched genes in these datasets included proinflammatory chemokines such as Ccl3, Ccl4, Ccl11 and immune-associated receptors such as Ddx58 and Lbp. This downregulation in immune-associated ontologies was supported by KEGG results, which indicated downregulation of Nod-like receptor signaling pathways and Toll-like receptor signaling pathways (47) (Fig. 6D, Table 2, and Supplemental Table S2).

Upregulated GO terms in FCFs versus NCFs included developmental terms such as “embryonic skeletal system morphogenesis” and cardiac tissue-associated ontologies including “myofibril assembly” and “sarcomere organization” (Fig. 6A), supported by KEGG pathways including “cardiac muscle contraction” and “Hippo signaling pathway.” Identified core-enriched genes included homeobox proteins (Hoxb5/7/8, Nkx2.5), T-box transcription factors (Tbx1/15), and several cardiac tissue proteins including Actc1, Actn2, and Myh6 (Fig. 6D, Table 2, and Supplemental Table S2).

Similar patterns emerged in analysis of adult CFs when compared with both neonatal CFs and fetal CFs. Differential expression in adult versus neonatal and adult versus fetal indicated upregulation of ontology terms associated with adaptive and innate immune response, leukocyte function, and regulation of immune response (Fig. 6, B and C) and in both cases, KEGG pathways including “cytokine-cytokine receptor interaction” and “viral protein interaction with cytokine and cytokine receptor” were upregulated (Fig. 6, E and F). This was supported by examination of core enriched genes: an array of chemokines including Ccl3, Ccl4, Ccl7, Ccl11, and Ccl22 were identified as enriched in both adult versus neonatal (Table 3 and Supplemental Table S4) and adult versus fetal (Table 4 and Supplemental Table S6) analyses, suggesting significant variances in chemokine expression with respect to age.

Table 3.

Ten most significant (based on q value) GO and KEGG results from GSEA analysis of Adult vs. Neonatal differential expression

GO	ID	Description	setSize	NES	q Values	Count	GeneRatio
BP	GO:0045087	Innate immune response	372	2.720	5.36E-09	165	0.444
BP	GO:0030595	Leukocyte chemotaxis	164	2.654	5.36E-09	92	0.561
BP	GO:0097529	Myeloid leukocyte migration	157	2.646	5.36E-09	86	0.548
BP	GO:0050776	Regulation of immune response	457	2.640	5.36E-09	179	0.392
BP	GO:0002250	Adaptive immune response	256	2.634	5.36E-09	119	0.465
BP	GO:0006941	Striated muscle contraction	150	−2.749	5.36E-09	61	0.407
CC	GO:0030018	Z disc	108	−2.783	5.36E-09	57	0.528
CC	GO:0031674	I band	118	−2.857	5.36E-09	50	0.424
CC	GO:0043292	Contractile fiber	200	−2.957	5.36E-09	81	0.405
CC	GO:0030017	Sarcomere	168	−2.976	5.36E-09	72	0.429

KEGG	ID	Description	setSize	NES	q Values	Count	GeneRatio
	rno04061	Viral protein interaction with cytokine and cytokine receptor	54	2.580	1.29E-09	42	0.778
	rno04610	Complement and coagulation cascades	54	2.465	1.29E-09	32	0.593
	rno05152	Tuberculosis	137	2.444	1.29E-09	57	0.416
	rno04612	Antigen processing and presentation	60	2.415	1.29E-09	40	0.667
	rno04060	Cytokine-cytokine receptor interaction	168	2.412	1.29E-09	79	0.470
	rno04022	cGMP-PKG signaling pathway	137	−1.917	5.01E-06	42	0.307
	rno05412	Arrhythmogenic right ventricular cardiomyopathy	54	−2.082	4.89E-05	25	0.463
	rno04261	Adrenergic signaling in cardiomyocytes	130	−2.201	8.18E-09	44	0.338
	rno05410	Hypertrophic cardiomyopathy	71	−2.340	2.37E-08	31	0.437
	rno04260	Cardiac muscle contraction	70	−2.342	1.28E-08	26	0.371

cGMP-PKG, cyclic guanosine monophosphate-dependent protein kinase G; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; NES, normalized enrichment score.

Table 4.

Ten most significant (based on q-value) GO and KEGG results from GSEA analysis of Adult vs. Fetal differential expression

GO	ID	Description	setSize	NES	q Values	Count	GeneRatio
BP	GO:0045087	Innate immune response	372	2.856	5.38E-09	171	0.460
BP	GO:0002250	Adaptive immune response	256	2.701	5.38E-09	127	0.496
BP	GO:0050776	Regulation of immune response	457	2.701	5.38E-09	180	0.394
BP	GO:0097529	Myeloid leukocyte migration	157	2.669	5.38E-09	83	0.529
BP	GO:0002443	Leukocyte mediated immunity	269	2.667	5.38E-09	123	0.457
BP	GO:0055001	Muscle cell development	181	−2.730	5.38E-09	69	0.381
BP	GO:0055002	Striated muscle cell development	169	−2.756	5.38E-09	66	0.391
CC	GO:0031674	I band	118	−2.766	5.38E-09	59	0.500
CC	GO:0043292	Contractile fiber	200	−2.815	5.38E-09	78	0.390
CC	GO:0030017	Sarcomere	168	−2.871	5.38E-09	70	0.417

KEGG	ID	Description	setSize	NES	q Values	Count	GeneRatio
	rno04061	Viral protein interaction with cytokine and cytokine receptor	54	2.571	1.18E-09	37	0.685
	rno05133	Pertussis	63	2.521	1.18E-09	33	0.524
	rno05152	Tuberculosis	137	2.507	1.18E-09	54	0.394
	rno04060	Cytokine-cytokine receptor interaction	168	2.477	1.18E-09	90	0.536
	rno04145	Phagosome	131	2.468	1.18E-09	53	0.405
	rno04360	Axon guidance	150	−1.758	6.84E-05	61	0.407
	rno05412	Arrhythmogenic right ventricular cardiomyopathy	54	−2.165	2.62E-06	22	0.407
	rno04261	Adrenergic signaling in cardiomyocytes	130	−2.184	4.19E-09	46	0.354
	rno05410	Hypertrophic cardiomyopathy	71	−2.312	7.39E-09	30	0.423
	rno04260	Cardiac muscle contraction	70	−2.407	1.18E-09	31	0.443

GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; NES, normalized enrichment score.

Downregulation of cardiac function and cardiac development terms were also evident in both comparisons. GO terms indicated as functionally downregulated by the analyses included “contractile fiber,” “striated muscle contraction,” and “sarcomere” in the adult versus neonatal comparison, and similarly “muscle cell development” and “striated muscle development” were downregulated in the adult versus fetal analysis (Fig. 6, B and C). KEGG terms-associated cardiac function including “cardiac muscle contraction” and “adrenergic signaling in cardiomyocytes” were similarly downregulated in both adult versus neonatal and adult versus fetal comparisons. This was supported by examination of core enriched genes for these ontologies, which recapitulated this downregulation of cardiac functional and developmental genes: in both cases genes such as Myh6/7, Actn1, Actn2, and Actc1, among several others, were identified (Tables 3 and 4 and Supplemental Tables S4 and S6). Taken together, these data underscore a transcriptomic shift from more development-supporting genes to predominantly immune- and inflammatory-specific genes. This shift is evident in comparison of young animal-sourced cardiac fibroblasts (NCF and FCF) to ACFs, but transcriptomic differences also exist between FCFs and NCFs as well.

RT-qPCR Validates Differential Expression Differences in CFs

Reverse transcriptase quantitative PCR (RT-qPCR) was used to analyze FCFs, NCFs, and ACFs to validate the RNA-seq results for selected genes. Results show relative expression normalized to GAPDH for ACFs and FCFs, normalized against results from NCFs. The genes evaluated were Il6, Tbx5, Myh7, Mmp9, Mmp2, and Col1a1. IL-6 was selected as Interleukin-6 was shown to be comparatively upregulated in ACF versus NCF (Supplemental Table S5) and is an important inflammatory cytokine. Tbx5 (T-box transcription factor 5) is a critical cardiac development protein (48) and was observed to be upregulated in FCFs versus NCFs (Supplemental Table S3). Myh7 (encoding a myosin heavy chain β isoform) was observed to be markedly downregulated in ACFs versus NCFs (Supplemental Table S5). Mmp2, Mmp9, and Col1a1, encoding matrix metalloproteinase-2 and -9, and collagen type I, alpha-1, are all critical in cardiac matrix regulation and crucial proteins expressed by CFs. Of these, Mmp9 was shown to be upregulated in ACFs versus NCFs, whereas Mmp2 and Col1a1 were observed to be marginally downregulated in both ACFs and FCFs compared with NCFs (Fig. 7A).

Figure 7.

RT-qPCR validation of differentially expressed genes in adult and fetal cardiac fibroblasts (CFs), with respect to neonatal CFs. A: Log₂FC differential gene expression from RNA-seq data for genes selected for validation via qPCR. B: IL-6 gene expression in adult CFs vs. fetal CFs, normalized to neonatal CF controls and relative to GAPDH expression. C–G: T-box transcription factor 5 (TBX5), myosin heavy chain 7 (MYH7), matrix metalloproteinase 9 (MMP9), MMP2, and collagen 1 α1 (COL1A1) expression levels, respectively, in adult CFs vs. fetal CFs. Statistical analysis: one-tail t test, with cutoff P value threshold of 0.05. *P < 0.05, ***P < 0.001, ns, no significance. Error bars indicate standard deviation.

RT-qPCR results recapitulated these observed expressed differences in part: IL-6 had a relative (to NCF, normalized to GAPDH) expression value of 2.72 ± 0.54 (standard deviation) in ACFs versus expression in FCFs of 1.64 ± 1.67 (P = 0.1320) (Fig. 7B). Tbx5 was significantly less expressed (5.33 ± 1.65) in ACFs versus FCFs (36.38 ± 28.04, P = 0.0345) (Fig. 7C) consistent with trends observed in the RNA-seq results, whereas Myh7 showed variable, low expression in both cases (ACF: 0.96 ± 0.52, FCF: 1.34 ± 0.58, P = 0.1821) (Fig. 7D). Mmp9 (ACF: 29.06 ± 17.67, FCF: 9.74 ± 3.33, P = 0.0376), Mmp2 (ACF: 4.33 ± 0.58, FCF: 1.76 ± 0.38, P = 0.0002), and Col1a1 (ACF: 1.73 ± 0.10, FCF: 1.39 ± 0.30, P = 0.0387) all had significantly higher expression in ACFs versus FCFs (Fig. 7, E–G). For further comparison with the differential expression between Adult versus Fetal CFs, relative expression of the genes measured in Fig. 7 in Adult CFs, normalized to Fetal CF values, is presented in Supplemental Fig. S1.

DISCUSSION

Analysis of the cardiac fibroblast transcriptome in fibroblasts isolated from fetal, neonatal, and adult developmental ages indicates broad differences in gene expression with respect to age. These data offer new insights into the potential functions of CFs at these developmental ages while reinforcing age-specific gene expression differences in CFs and cardiac tissue that have been previously uncovered in the literature. More specifically, these data support previous conclusions that indicate that fetal CFs present a gene expression profile reflecting functions in support of embryonic growth and cardiac development, whereas their adult CF counterparts upregulate genes more reflective of immune cell communication and trafficking and cardiac homeostatic maintenance (15, 20, 49). Our data both support and expand on these studies by incorporating neonatal CFs into the analysis to evaluate fetal-to-neonatal differences. This establishes a clearer timeline of the phenotypic shifts in CFs, from the primarily developmental role of fetal CFs (20, 49), transitioning to a phenotype associated with rapid expansion and maturation in the neonatal heart to an adult phenotype characterized by immune signaling and tissue maintenance (21, 50, 51).

RNA-Seq Confirms Cardiac Fibroblast Immune Function Is Age Dependent

Although historically the function of CFs was considered to exclusively ECM maintenance, it is now known that these cells possess varied functions fundamental to cardiac function (21, 46, 49), and in particular, are vital to cardiac immune and inflammatory processes. In this respect, CFs secrete an array of growth factors, chemokines, and cytokines that facilitate invasion of immune cells such as macrophages and granulocytes (15, 51–54), including MCP-1, MIP-1α, MIP-1β, IP-10, and RANTES (52, 55).

RNA-seq results in this study reinforces the hypothesis of an immune-cross talk role for cardiac fibroblasts, focusing on how this role varies across organismal development. The lack of a strong immune system in the fetal and neonatal mammalian heart is well-established and hypothesized to contribute to the regenerative wound healing observed in fetal and neonatal mammals (56, 57). It is similarly understood that maturation of the immune system and expression of inflammatory cytokines increases rapidly during postnatal development (58), but an in-depth examination of how the cardiac fibroblast transcriptome changes in an immune context, and how these changes may contribute to immune maturation in the heart has, to our knowledge, not been thoroughly explored. The transcriptomic data presented herein point toward more immune cross talk and inflammatory signaling in CFs, with respect to increasing developmental age, and is worth further investigation and validation. Understanding how CF impact inflammation and immunity at different developmental ages may support the development of regenerative therapies and engineered cardiac tissues by uncovering ways to minimize the tissue inflammatory response and aggressive immune response typical of fibrotic wound repair following injury in adults (10, 59, 60).

We observed an upregulation of genes responsible for macrophage invasion (Ccl3/Ccl4) (61) (Tables 3 and 4 and Supplemental Tables S5 and S7), with respect to developmental age, in addition to increases in the expression of inflammatory mediators (IL-1β, IL-6) in adult CFs versus neonatal and fetal CFs (Supplemental Tables S5 and S7). Furthermore, gene ontology analysis results support the hypothesis that immune and inflammatory gene expression differences exist between fetal and neonatal CFs, suggesting cardiac immune maturation progresses rapidly after birth. These included fetal downregulation (compared with neonatal and adult CFs) of immune response-associated gene ontology terms (Fig. 6A), downregulation of Nod-like and Toll-like receptor signaling pathways (Fig. 6D), which have both been established as expressed in cardiac fibroblasts and are implicated in triggering an immune response to pathogen-associated molecular patterns (14, 52). We also observed a comparative upregulation of chemokines in adult CFs versus fetal CFs, including Ccl3/4 immune-associated genes like Ddx58 (an innate immune response receptor essential in identifying viral-infected cells) (62) and Lbp, which binds to bacterial lipopolysaccharides and initiates an immune response (63) (Table 1).

Cardiogenic Developmental/Regenerative Potential of Cardiac Fibroblasts Is Rapidly Lost After Birth

Cardiac fibroblasts, like their counterparts in other bodily tissues, play a vital role in tissue morphology, extracellular matrix regulation and upkeep, mechanical support, and inflammatory response in vivo (10, 14, 19, 64–67). Beyond this, CFs are also implicated in the development and maturation of the heart in utero and assist in driving cardiomyocyte proliferation in that regard. This was well-established in work by Ieda et al. (6) in 2008, in studies that determined that fibroblast-derived proteins including fibronectin and periostin stimulated cardiomyocyte proliferation via β1-integrin signaling.

It is well understood that cardiomyocyte proliferation and cardiac regenerative wound healing diminishes rapidly after birth (21, 22, 24, 67) and is reflected in a reduced regenerative potential in cardiac fibroblasts, which begins to express a more profibrotic phenotype (2, 68, 69). Our study reinforces the age-dependent transition to a fibrotic phenotype in CFs and suggests that it begins rapidly after birth, as we observed significant downregulation of several immune-associated GO and KEGG pathways in fetal CFs versus neonatal CFs (Fig. 6, A and D, and Supplemental Table S2), implying neonatal CFs begin to upregulate genes necessary to mount significant immune recruitment to initiate an inflammatory response immediately after birth. Our data also indicated functional upregulation of genes associated with muscle cell development in fetal and neonatal CFs versus adult CFs (Fig. 6, B and C, and Supplemental Tables S4 and S6), which may support the conclusion that early-development CFs play a direct role in the functional development of cardiac tissue. This was particularly evident in FCFs, within which development-associated transcription factors including Tbx4, Hoxb5, and Hoxb7 were upregulated, and ontological terms associated with morphogenesis and development were more functionally upregulated than their neonatal counterparts (Fig. 6, A and D, Table 1, and Supplemental Table S1). This builds upon previous work investigating transcription factors in cardiac fibroblasts, which concluded that cardiogenic genes including Tbx20 and Gata4 expressed by adult fibroblasts contribute to cardiac repair (70).

The presence of cardiomyocyte structural genes including Actc1, Actn2, Myh6, and Myh7 was also observed to be upregulated in FCFs and NCFs versus ACFs, which was surprising, as these proteins are conventionally considered cardiomyocyte specific (70, 71). Although we cannot rule out that this may be due to low-level presence of CMs and CM progenitor cells in our cultures, there is evidence that CFs, particularly fetal derived, express these genes at low levels: a previous microarray assay of fetal versus adult mouse cardiac fibroblast gene expression indicated similar differential expression of these genes with expression levels consistent with what was observed in this study, specifically upregulation of Actc1, Actn2, Myh6, and Myh7 in fetal CFs versus adult CFs (6).

Differences in CF expression of ECM and ECM-associated proteins is also strongly linked to the transition to profibrotic wound healing in the heart. In previous studies of fetal versus adult cardiac fibroblasts, gene expression analysis indicated enrichment of fibronectin (Fn), collagen (Col) genes, heparin-binding EGF-like growth factor (HBEGF), periostin (Postn), and tenascin C (Tnc) in fetal cardiac fibroblasts but not adult cardiac fibroblasts, an expression profile associated with increased cardiomyocyte proliferation and regenerative healing (20, 72). Within our differential expression analysis, we observed an upregulation of Col9a1, Col26a1, and Col4a4 and Tnc in fetal CFs versus neonatal CFs (Supplemental Table S3). Of the collagens, both Col9 (73) and Col4 (74) have potential links to cardiac development. Tnc, the gene for protein tenascin-C, is predominantly expressed in embryonic tissue and implicated in tissue morphogenesis and remodeling via modulation of cell adhesion to ECM substrates (75). Taken together, these data provide more evidence that the loss of regenerative phenotype in CFs occurs rapidly after birth.

Study Limitations

This study methodology contains several limitations that should be addressed by future investigations. First, we exclusively explored the transcriptomic profile of cardiac fibroblasts, which represents the gene expression profile of CFs at distinct developmental ages but does not suggest a mechanism for observed differences. Direct evaluations of genes identified must be performed to definitively identify specific factors responsible for the phenotypic differences in CFs.

Second, the protocols used to isolate cardiac fibroblasts from its resident tissue differed slightly across developmental ages; specifically, although neonatal and adult CFs were from ventricular tissue, fetal CFs were isolated from the whole fetal heart, which was necessary to produce a sufficient cell yield. However, cardiac fibroblast phenotype can vary with respect to location in the heart (76) and incorporation of location-dependent CF RNA-seq analyses would likely be of value.

Third, we isolated fibroblasts using a preplate protocol described herein, in methods. This method was chosen to ensure our results represented the full composition of cardiac fibroblasts, which can differ with respect to location, state, and developmental lineage (77), and present varying biomarkers that would prevent more specific isolation methods such as fluorescent-activated cell sorting (FACS) (28, 78–80). Preplating likely introduces some cellular contamination from CMs and other support cell types. Further evaluation of CFs using FACS sorting for rigorous identification of cell identity and single-cell RNA-seq-associated studies may be valuable to identify crucial cell-cell heterogeneity associated with CFs and how this likely differs across developmental age and may be partly responsible for observations presented in this study (21, 28, 81). We also note that the qPCR work was completed with fibroblasts cultured in vitro at low passage numbers versus the RNA-seq analysis that was done with confluent P0 CFs. This could explain why the gene expression in the qPCR study does not fully recapitulate the RNA-seq, as CFs are established to alter their phenotype across extended culture and passages (82).

Conclusions

This transcriptomic analysis of the rat cardiac fibroblast at three discrete developmental ages (fetal, neonatal, and adult) using RNA-seq provides significant insights into CF transcriptional heterogeneity. Immune function-related transcriptomic evaluations presented herein suggest that age-specific CF heterogeneity may be partly responsible for differences in cardiac inflammation and immune response observed across developmental ages in vivo. The results reinforce established knowledge of fibroblast phenotypic heterogeneity across developmental ages, including marked upregulation of cytokines and chemokines in adult CFs versus fetal and neonatal CFs, and an age-dependent transition to a more profibrotic, proinflammatory gene expression profile versus the proregenerative gene expression observed in fetal and early-neonatal CFs. The results further suggest that this transition away from a developmental, proregenerative phenotype becomes acute quickly after birth.

ETHICAL APPROVALS

All animal procedures were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals, carried out in accordance with ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines, and approved by the Institutional Animal Care and Use Committee at Tufts University.

DATA AVAILABILITY

The data that support this study are available at the Gene Expression Omnibus (GEO) repository, GSE162277, https://www-ncbi-nlm-nih-gov.ezproxy.u-pec.fr/geo/query/acc.cgi?acc=GSE162277.

SUPPLEMENTAL DATA

Supplemental Tables S1–S7 and Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.14823915.v2.

GRANTS

This work was supported by the United States Department of Defense (W81XWH-16-1-0304 to L. D. Black), the National Science Foundation (NSF#1603524 to L. D. Black), the American Heart Association (18PRE33960362 to L. R. Perreault), the NIH (R00-CA207866-05 to M. J. Oudin), and Tufts University (Start-up funds from the School of Engineering to M. J. Oudin and a Breast Cancer Alliance Young Investigator Grant to M. J. Oudin and T. T. Le).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.T.L., M.J.O., and L.D.B. conceived and designed research; L.R.P., T.T.L., and L.D.B. performed experiments; L.R.P. analyzed data; L.R.P., T.T.L., and L.D.B. interpreted results of experiments; L.R.P. prepared figures; L.R.P. drafted manuscript; L.R.P., T.T.L., M.J.O., and L.D.B. edited and revised manuscript; T.T.L. and L.D.B. approved final version of manuscript.