Altered Epithelial-Mesenchymal Progenitor States Lead to Matrix Deposition, Tissue Inflammation, and Transitional Epithelial State in Congenital Diaphragmatic Hernia

Abstract

Introduction:

Congenital diaphragmatic hernia (CDH) lungs are characterized by pulmonary hypertension and lung hypoplasia. We have used single cell RNA sequencing (scRNA-seq) to show that mesenchyme is perturbed in CDH, leading to disrupted epithelial-mesenchymal transition (EMT) dynamics and inflammatory signaling.

Methods:

Normal and CDH fetal rat lungs were harvested at E17, E19 and E21 – which correlate to pseudoglandular, canalicular, and saccular stages, respectively - and dissociated into single cell suspension. Seurat was used for single cell analysis. Cell types were identified by canonical genes and differential expression of genes were then analyzed. Findings were confirmed by staining. Score for mesenchymal versus epithelial-like characteristics in EMT was calculated.

Results:

During normal development, mesenchymal progenitors surround the developing airway undergoing EMT. At E17 in CDH, these cells downregulate Sox9, a plasticity marker, and upregulate extracellular matrix (ECM) proteins and TGFβ signaling molecules. CDH mesenchymal progenitors have an increased EMT score (p < 0.001), meaning more mesenchymal characteristics compared to normal lung. At E21, CDH mesenchyme upregulates TGFβ-2, TGFβR-2, and Smad2/3. CDH alveolar type 1 (AT1) and AT2 cells upregulate Krt8 and Krt18.

Discussion:

CDH lung mesenchymal progenitors attain mesenchymal-like characteristics prematurely and there is upregulation of ECM proteins when compared to normal lung. Moreover, CDH distal epithelial cells (Krt8/18+) enter a transitional state that is seen in fibrotic lung diseases. These findings represent imbalance of EMT, and thus dysregulation of key molecular pathways, which leads to poorly developed mesenchymal and epithelial structures that we speculate causes the lung hypoplasia found in CDH.

Introduction

Normal lung development is a complex process that occurs in utero and postnatally. There are three key stages that mark fetal lung development: pseudoglandular, canalicular, and saccular. During lung development, epithelial-mesenchymal transition (EMT) is a key biological process involved in maturation. EMT exists in both normal fetal lung development and lung pathology. It is the process by which epithelial cells lose epithelial-specific characteristics, like cell-cell adhesion, and gain mesenchymal characteristics, like migration and production of ECM components. This process is facilitated by coordinated crosstalk between multiple cell types in the lung [1].

Congenital diaphragmatic hernia (CDH) is a birth defect of unknown etiology characterized by incomplete diaphragm development and abnormal lung development including pulmonary hypoplasia and pulmonary hypertension. In CDH, incomplete closure of the diaphragm leads to herniation of abdominal contents into the thorax. Because lung hypoplasia and arteriole hypermuscularization occur bilaterally, current evidence supports a dual-hit hypothesis in which a molecular insult occurs early in development and mechanical compression acts as a second hit. The inciting molecular etiology of CDH is still unknown.

Although an exact molecular mechanism has not been elucidated, disruption in key regulators of branching morphogenesis is likely related to hypoplasia seen in CDH [2]. CDH has not been found to be cause by one gene alone and genetic mutations are heterogenous when it comes to diagnosis [3-5]. Genomic and molecular studies have demonstrated various pathways may be implicated in the development of CDH; these include molecular families critical for normal lung development such as WNT, BMP, FGF, TGFβ, and ROBO/SLIT [6]. Previous studies have found downregulation of Fgf18, Wnt7b, Wnt2, BMPR2, and BMP4 during the pseudoglandular and canalicular stages of hypoplastic lungs in nitrofen animal models [6-9]. A systematic review of the literature found that in addition to FGF and BMP pathways, other downregulated signaling pathways included Sonic Hedgehog (SHH) and retinoid acid (RA) signaling pathways, resulting in a delay in early epithelial differentiation, immature distal epithelium and dysfunctional mesenchyme [6]. ROBO/SLIT animal knockouts have diaphragmatic defects and thus studies focusing on this pathway suggest ROBO/SLIT may function as regulators of Sox9/Sox2 balance during development [2]. Specifically, one study found that inhibition of ROBO2 induced fetal lung growth in ex vivo lung explant cultures [2]. Studies focusing on other branching organs, like the mammary gland, have found ROBO/SLIT pathway balance may be involved in determining stem cell plasticity through interaction of the Wnt pathway [10,11]. Specifically, Slit2 has been found to oppose Wnt signaling and lead to cellular senescence and non-renewal [10]. These studies have inspired focus on the developing epithelium and mesenchyme, and some studies suggest that the abnormal mesenchymal development is responsible for pulmonary interstitial thickening and hypoplasia observed [12]. In addition, pulmonary hypertension seen in CDH lung is thought to be caused by aberrant smooth muscle cell development and hypermuscularization of arterioles, which develop from mesenchymal structures [13]. The aforementioned signaling pathways are of interest as they are involved in orchestrating complex crosstalk between the epithelium and mesenchyme [14].

Epithelial-mesenchymal transition (EMT) is a biological process that exists in both normal fetal lung development and lung pathology, such as COPD, cancer, and fibrosis. EMT is not one-directional. This process is highly plastic and recent evidence argues cells can transition between three states – epithelial (E), hybrid (E/M), or mesenchymal (M)—depending on surrounding cues [1,15]. Normal mesenchymal progenitor cells in the lung exist in a hybrid state and are seen at the lung bud tip in branching morphogenesis. Evidence exists that TGFβ, WNT, and NOTCH signaling networks regulate EMT/MET [1,15,16]. EMT can also be driven by mechanical factors, such as ECM density and mechanical stress; specifically high stress areas are preferential to mesenchymal states [17,18]. Despite the complexity of EMT, there appears to be a regulatory network, specifically the TGFβ/ZEB/miRNA 200 family, that functions as one of the master regulators of EMT. TGFβ is essential for both the establishment and maintenance of EMT through upregulation of Zeb1 and Zeb2 [19,20]. Zeb1/2 suppresses the epithelial phenotype by suppression of epithelial phenotype genes, like cadherin. In contrast, high levels of miRNA 200 support the epithelial phenotype by suppression of Zeb1/2 expression. Studies have found autocrine TGFβ is required for Zeb1/2 expression and maintenance of mesenchymal state [19,20]. Thus, the ratio of Zeb and miR-200 influences the switch between epithelial, hybrid, and mesenchymal states. Collectively, these findings demonstrate that epithelial cell plasticity is controlled by an autocrine TGFβ/ZEB/miR-200 signaling network.

We have used single cell RNA sequencing (scRNA-seq) to acquire evidence that the mesenchyme in developing lung tissue is preferentially perturbed in CDH, and that this injury occurs prior to epithelial deficit. Given our understanding of the implicated pathways in CDH, the primary aim of this study was to understand differences in cellular composition during lung development in normal and CDH lung. The secondary aim was to identify cell populations that had differential gene expression in CDH lung compared to normal lung. ScRNA-seq has expanded our ability to understand multicellular systems but has not been previously performed on an animal model with congenital diaphragmatic hernia at multiple timepoints during lung development. Specifically, our data show that CDH is associated with altered epithelial-mesenchymal transition (EMT) dynamics during early lung development, namely the pseudoglandular stage. We suspect this disruption leads to interrupted mesenchymal and epithelial development leading to lung hypoplasia.

Previous work by our group has demonstrated reversal of lung hypoplasia in a CDH rat model using particle-based epigenetic therapy [21]. The findings in our current study suggest that mesenchymal progenitors may be an ideal target for clinical intervention and warrant further investigation.

Methods

Animal care and rat CDH model

Wild type time dated Sprague-Dawley rats were obtained from Charles River Laboratories (Wilmington, MA) at gestational day 8 (E8). Female rats were fed 100 mg of nitrofen (Cat# 33374, Sigma Aldrich, St. Louis, MO) dissolved in 1 mL of olive oil on embryonic day 9 (E9) to induce congenital diaphragmatic hernia, lung hypoplasia, and pulmonary hypertension. All animal use was in accordance with the guidelines of the Animal Care and Use Committee of Yale University [6].

Lung tissue extraction

Lung samples were analyzed from E17, E19, and E21 as these correspond to distinct lung developmental stages, respectively, pseudoglandular, canalicular, and saccular stages (Fig. 1a). For each gestational age, two dams (Wild-type time-dated Sprague-Dawley rats) were used, one as a control and one which had nitrofen administered and had pups with CDH. Dams were anesthetized with inhaled isoflurane and given intraperitoneal injections of sodium heparin (0.15 ml at 1000 U/ml). Fetuses were delivered by cesarian section and a microdissection scope was used to extract fetal lungs. For healthy controls, a clamshell incision was made and the heart was perfused with heparin solution (100 U/ml). For CDH fetuses, a transverse abdominal incision was made to inspect for diaphragmatic defect. Once confirmed, lung extraction and cardiac perfusion was done. Lungs were carefully dissected from surrounding structures. At E17 and E19, two lung samples for each condition were harvested. At E21, four normal lung samples and two CDH lung samples were obtained. The number of lung samples for CDH was limited by the presence of a diaphragmatic defect. All CDH cases were diaphragmatic agenesis, which is the most severe phenotype. This was identified through direct visual inspection of the diaphragmatic defect prior to lung sample collection. Lung samples were taken after hernia confirmed and fetuses were sacrificed. All CDH cases were left-sided. Remaining fetal lungs were reserved for histology.

Figure 1.

Characterization of cell class and type across E17, E19, and E21. (A) Schematic of nitrofen administration, time points sampled, and corresponding gestational age in rat and human. (B) Workflow for collecting fetal lung tissue for single-cell dissociation. (C) Proportion by cell class between normal and CDH lung at each timepoint. (D) Embedding of cell class with specific markers, timepoints, and condition showing shifts on cell phenotype between normal and CDH lung. Proportions of cell class across all timepoints in both normal and CDH lung.

Cell isolation

Enzyme solution was made with DMEM containing 1 mg/mL Collagenase/Dispase (Roche), 3 U/mL Elastase (Worthington), and 20 U/mL DNAse (Worthington) and pre-heated to 37 C. Lungs were saturated with enzyme and placed in enzyme solution in shaking water bath at 37 C and 60 rpm for 20 min. Lung tissue was mechanically dissociated by pushing though a 100um strainer until mostly connective tissue remained. Strainer was then rinsed with ice-cold DMEM containing 10% FBS. The resulting tissue solution below the strainer was spun down, the supernatant discarded, and the pellet resuspended in red cell lysis buffer at a 2:1 ratio with pellet volume and rocked manually for 2 min. This was then diluted with 0.04% BSA in PBS (0.4 mg/mL) and spun down a second time. The pellet was resuspended in 0.04% BSA in PBS and filtered through a 70um filter and spun a third time. A final resuspension in 0.04% BSA in PBS was then performed and the solution passed through a 40um filter. The resulting cell suspension was counted, assessed for viability, and serially diluted to 2.4x10⁶ cells/ml for single cell library preparation.

10× Genomics library construction

Single-cell suspensions were processed using the 10x Genomics Single Cell 3′ RNA-seq kit. Gene expression libraries were prepared according to the manufacturer’s protocol. Final libraries were sequenced to reach an approximate depth of 50,000 reads/cell. FastQ files were processed through CellRanger for alignment, UMI counting, and generation of feature-barcode matrices.

Computational analysis

Analysis was performed using R studio and Seurat (version 4.2). Each timepoint containing both conditions was analyzed separately initially and then merged for subclass and timepoint analysis. Samples were filtered to exclude cells with >15% mitochondrial expression. After normalization, scaling, and identification of variable features, principal component analysis was performed. Communities of cells were calculated using the FindNeighbors() function and clusters were created using the Louvain algorithm (FindClusters() function). The RunUMAP function was used to embed the data [22]. Marker genes were identified and ranked using a Wilcoxon rank-sum test within the FindAllMarkers() function (Supplemental Tables 1-3). Ratio of non-zero gene expression between populations was used to rank markers by specificity.

Statistical analysis of differential gene expression

Raw single-cell RNA sequencing data from normal and CDH fetal lung samples were processed using the Seurat R package. Quality control was performed to exclude low-quality cells based on total RNA counts, detected features, and mitochondrial transcript content (<15%). Following filtration, samples were merged into a single Seurat object, with metadata annotated by experimental condition (CDH vs. Normal). Gene expression values were normalized using the LogNormalize method, in which each gene’s expression is scaled by the total expression per cell, multiplied by a scale factor of 10,000, and log-transformed. The data were subsequently scaled to mitigate the influence of highly expressed genes on downstream analyses.

Highly variable genes were identified using the variance-stabilizing transformation (VST) method, selecting the top 2,000 variable features for dimensionality reduction. Principal component analysis (PCA) was applied to capture the major sources of transcriptional variation, and significant principal components were selected based on elbow plots and heatmap inspection. Cell clustering was performed using Seurat’s graph-based approach, wherein FindNeighbors() constructs a K-nearest neighbor graph and FindClusters() applies the Louvain algorithm for community detection. The resulting clusters were visualized using Uniform Manifold Approximation and Projection (UMAP) to project the data into two dimensions.

Differential gene expression analysis between clusters was conducted using Seurat’s FindAllMarkers() function, which employs the Wilcoxon rank-sum test to compare expression distributions between cells in each cluster and all other clusters. Only genes expressed in at least 10% of cells in either population (min.pct = 0.1) and exhibiting a log fold-change greater than 0.1 were included. P-values were adjusted for multiple testing using Bonferroni correction to control the false discovery rate. Cluster-specific marker genes were ranked by log fold-change and expression prevalence. Clusters characterized by high ribosomal or mitochondrial gene expression, indicative of stressed or dying cells, were excluded from subsequent analyses.

We created marker lists for each cell type comparing normal lung to CDH. To examine condition-specific differences, we compared gene expression directly between conditions. Only genes upregulated in one condition over the other were retained (only.pos = TRUE). For each gene, a power score was calculated as the product of its expression ratio and average log2 fold change (ratio * avg_log2FC), providing a combined metric of expression strength and specificity.

To refine the list of condition-specific markers, several categories of likely irrelevant or uninformative genes were removed. This included hemoglobin genes (e.g., Hba, Hb), ribosomal proteins (Rps, Rpl), and uncharacterized or low-confidence transcripts (e.g., those starting with AABR, LOC, RGD, AC1). Finally, the top genes with the highest expression ratio between conditions were selected as the most enriched and potentially biologically relevant condition-specific markers.

These genes were compiled into a list for downstream visualization and interpretation of the molecular differences between normal and CDH-affected cells. These thresholds were used for all cell types and all timepoints to create differential gene expression lists.

Patterns of interest were visualized via violin plots made using ggplot2 and heatmaps made using ComplexHeatmap [23,24].

Calculation of epithelial-mesenchymal transition directionality score

To assess epithelial-to-mesenchymal transition (EMT) at the single-cell level, we computed an EMT directionality score based on the relative expression of curated epithelial and mesenchymal gene sets. The epithelial gene set included Cdh1, Epcam, Cldn2, Ocln, and Krt18, while the mesenchymal gene set comprised Vim, Zeb1, Snai1, Twist1, Tjp1, Cdh2, Fn1, Snai2, Zeb2, Pxn, and Col1a1 [25-27].

For each cell, we calculated module scores using the AddModuleScore() function in Seurat (v4.0), which estimates the average expression of a given gene set subtracted by the average expression of randomly selected control gene sets matched for expression levels [28]. The epithelial and mesenchymal scores were computed separately.

The EMT directionality score was then defined as the difference between the mesenchymal and epithelial module scores:

$$EMTScor{e}_{directional}=MesenchymalScore-EpithelialScore$$

A higher EMT directionality score reflects a shift toward a mesenchymal transcriptional program.

Histological staining and immunohistochemistry

Fetal lung samples were fixed in formalin (10% NBF) for 24 h and then submitted to Yale Pathology Tissue Services for paraffin embedding, sectioning, and hematoxylin and eosin staining (H&E), Van Gieson (EVG), and Masson’s Trichrome (Masson Tri). Unstained sections were heated and de-waxed in xylene and ethanol. They were soaked in citrate buffer (0.01 M citric acid, 0.05% Tween 20, distilled water, pH 6.0) for 20 min at 75 C. After cooling, sections were permeabilized using permealization buffer (0.2% Triton X-100 in PBS) and rinsed in PBS. They were blocked with blocking buffer (5% BSA and 0.75% glycine in PBS) and incubated overnight in 4 C with primary antibodies using appropriate dilutions. Slides were rinsed with PBS and incubated with secondary antibodies at 1:500 dilution for 1 h at room temperature. Sections were then stained with DAPI and mounted with polyvinyl alcohol mounting medium.

Results

Quality control, description of classes, cell types, and differential patterns during lung development

To assess the cellular landscape and heterogeneity of cell types throughout lung development in normal and CDH lung at three distinct stages – pseudoglandular, canalicular, and saccular – single cell suspensions were created from whole lung tissue (Fig. 1a). Both left and right lungs were used from normal and CDH lungs (Fig. 1b). After quality control filtration, normalization, and scaling (see methods), 81,851 samples (cells) across 22,698 genes (RNA) were recovered. Mean mitochondrial fraction across dataset was 3.9%, indicating excellent cell quality. We categorized cell populations into 5 main cell classes: mesenchyme, epithelium, endothelium, immune cells, and erythroblasts/erythrocytes. Erythroblasts/erythrocytes were removed from analysis. Cell classes and types were identified using known canonical markers (Table 1). To first identify cell classes (epithelial, endothelial, mesenchymal, and immune), we used expression of Epcam, Cdh5, Col1a1, and Ptprc, respectively (Fig. 1c, d).

Table 1.

Genes used for cell cluster allocation and cell population identification.

	Cell cluster	Gene	Reference
Mesenchymal		Col1a1	[29]
	Fibroblasts	Fgf10, Itga8, Col1a1, Pdgf	[29]
	Proliferating	Kif	[30]
	Early Mesenchyme	Wnt2	[31]
	Pericytes	Gucy1a1, Gucy2a1	[32,33]
	Mesenchymal Progenitors	Twist1, Prrx1, Col1a1, Col3a1, Col11a1, Wnt2, Sox9	[27,31,34]
	Mesothelial	Wt1	[35,36]
	Neural Crest Cells (NCCs)	Sox10	[37]
	Neuroendocrine Cells (NECs)	Tlx2, Chga, Ascl	[38]
	Adventitial Fibroblasts	Vcam1, Ogn, Itga8, Fbln5, Thbd, Pdgfra, Ptn	[29,39]
	Smooth Muscle Cells	Acta2, Myh11	[29,40]
Epithelial		Epcam
	Alveolar Type 2 (AT2)	Sftpc	[41,42]
	Proximal Epithelial	Sox2	[31,43]
	Distal Epithelial	Sox9	[44]
	Alveolar Type 1 (AT1)	Aqp4, Pdpn, Ager	[45]
	Ciliated Epithelium	Tuba, Dyn	[46]
	Non-Ciliated Secretory	Scgb3a2	[42]
Immune		Ptprc	[47]
	Macrophages	Cd163	[48]
	Antigen Presenting Cells (APCs)	Rt (MHC II complex)	[49]
	Dendritic Cells	Siglech
	B Cells	Cd24	[50]
	T Cells	Cd3	[51]
	Neutrophils	S100a	[52]
Endothelial		Cdh5	[53]
	Endothelial	Cdh5	[53]

Mesenchymal cells were the predominant cell type across all developmental timepoints and in both conditions (Fig. 1c). Endothelial, epithelial, and immune cell compartment fraction increased during development (Fig. 1c). Clustering and differential expression analysis revealed multiple cell types with distinct phenotypic character across all timepoints and between normal and CDH lung (Fig. 1d).

Identification & characterization of perturbed mesenchymal progenitor population in CDH at pseudoglandular stage

In CDH, mesenchymal progenitors at E17 undergoing EMT have downregulation of Sox9. We analyzed all cell types at E17 and based on the expression of Wnt2 and its associated ligand Sfrp, identified a group of cells we have labeled as mesenchymal progenitors (Fig. 2a). We analyzed the top markers in this cell population in both conditions and found they were EMT-related genes, such as Twist1, Prrx1, and Sox9. The combination of expression of mesenchymal specific genes (Wnt2) and EMT-related genes, led to the naming of this group as mesenchymal progenitors. This group of cells existed as two distinct populations in the embedding space, where one cluster was particularly enriched in CDH lung compared to normal lung (Fig. 2b). We also looked at differential gene expression between these cells and all other cell types in both tissues and found that the cluster of cells predominantly found in CDH had downregulation of Sox9 (Fig. 2b).

Figure 2.

Mesenchymal progenitors surrounding airways of developing lung at E17 have loss of Sox9 and increased mesenchymal characteristics in CDH compared to normal lung. (A) UMAP embedding of all cell types at E17. (B) UMAP embedding of all cell types at E17 highlighting mesenchymal progenitor population primarily present in CDH lung tissue (top, CDH = peach, Normal = navy) and FeaturePlot for Sox9 showing downregulation in corresponding group of mesenchymal progenitors in CDH tissue (circled, bottom). (C) Immunohistochemistry staining for Sox9 in normal lung (left) and CDH lung (right) showing absence of Sox9 fluorescence surrounding developing airways in CDH (dotted line). (D) UMAP embedding of mesenchymal progenitors emphasizing two distinct populations. (E) Violin plot with EMT directionality score showing increased mesenchymal-like characteristics in CDH mesenchymal progenitors (Genes used to calculate mesenchymal score: Vim, Zeb1, Snai1, Twist1, Tjp1,Cdh2, Fn1, Snai2, Zeb2, Pxn, Col1a1; genes used to calculate epithelial score: Cdh1, Epcam, Cldn2, Ocln, Krt18).

Using immunohistochemistry, we localized aforementioned mesenchymal progenitors surrounding the developing airway. We performed immunohistochemistry staining for Sox9 in normal lung and CDH lung to identify the location of these mesenchymal progenitors and confirm our single cell RNA-seq findings. In both normal and CDH lung, there is Sox9 positivity as expected in the distal epithelium. In normal lung, there is a group of Sox9+ cells surrounding the developing airway (Fig. 2c, dashed line). These cells are present from the proximal airway to the distal airway up until terminal branching, where Sox9 positivity is seen in the epithelium (Fig. 2c, dashed line). In CDH lung, Sox9 positivity was reduced in cells surrounding the developing airways (Fig. 2c, dashed line).

At E17, mesenchymal progenitors undergoing EMT have increased mesenchymal-like characteristics in CDH. Given the observed specificity of EMT markers to mesenchymal progenitors, we calculated an EMT directionality score that calculates the difference in average expression of mesenchymal and epithelial genes. We have used this score to evaluate the degree of mesenchymal-like shift present in this cell population. A higher EMT directionality score reflects a shift toward a mesenchymal transcriptional character. In CDH lung, mesenchymal progenitors with Sox9 downregulation have an increased EMT score, representing a shift toward mesenchymal-like character (p < 0.001, Fig. 2e).

During pseudoglandular stage mesenchymal progenitors surrounding airways in CDH upregulate ECM protein synthesis, wnt signaling molecules, and TGFβ signaling molecules

We identified genes differentially expressed between CDH and normal lung in mesenchymal progenitors at E17 (Fig. 3a). We observed upregulation in CDH of various laminins, integrin, lumican, decorin, elastin, fibulin, and fibrillin (Fig. 3b) [54,55]. There was also upregulation of various collagens in CDH (Fig. 3c).

Figure 3.

Mesenchymal progenitors in CDH upregulate genes for ECM composition and molecules involved in pathways important for branching morphogenesis in CDH lung. (A) Heatmap of mesenchymal progenitors at E17 highlighting differential gene expression between CDH and normal lung. (B-C) In CDH, mesenchymal progenitors have increased production of ECM composition molecules such as laminins, decorin, elastin, fibrin, fibrillin, integrins, lumican, and multiple collagens. (D) Proteins important in TGFβ regulation and signaling (Smad1 and Ltbp3/4), (E) Robo pathway (Robo2 and Slit 2/3), and (F) Wnt and NOTCH signaling (Sfrp1/2, Fzd1, Dlk). (G) Representative images of H&E, Verhoeff’s Van Gieson (EVG), and Masson’s Trichrome (Masson Tri) for normal and CDH tissue at E17. E17 CDH tissue displays a more pronounced elastin composition as demonstrated by the darker appearance of the stain, compared to normal. The same is demonstrated for trichrome staining, with E17 CDH tissue showing an upregulation in collagen fibers (blue). Scale bars for individual images and respective insets are reported on each panel.

Mesenchymal progenitors in CDH exhibit dysregulation of signaling pathways important for branching morphogenesis. In CDH, TGFβ ligand Smad1 and its modulator Ltbp3 and 4 are upregulated (Fig. 3d). Robo2 and its ligands Slit 2 and 3 were also upregulated by CDH progenitors (Fig. 3e). Wnt ligands Sfrp 1 and 2, and Fzd1 were upregulated in CDH (Fig. 3f). Finally, Notch ligand Dlk1 was upregulated by progenitors in CDH (Fig. 3f).

H&E, Verhoeff’s Van Gieson (EVG), and Masson’s Trichrome (Masson Tri) stains for normal and CDH tissue at E17 displayed a more pronounced elastin composition as demonstrated by the darker appearance of the stain, compared to normal (Fig. 3g). The same is demonstrated for trichrome staining, with E17 CDH tissue showing an upregulation in collagen fibers (blue).

During saccular stage mesenchymal cells in CDH differentially express TGFβ-2 and snai proteins

In CDH, there were phenotypic differences at E21 in fibroblasts, SMCs/myofibroblasts, and pericytes (Fig. 4a). To characterize these differences, we compared gene expression in the mesenchyme between CDH and normal lung. We found that in CDH, TGFβ-2, TGFβR-2, Smad2, Smad3, Snai1, and Snai2 were upregulated compared to normal lung (Fig. 4c-e).

Figure 4.

Mesenchymal cells at E21 in CDH express pro-fibrotic genes. (A) UMAP embedding of all mesenchymal cells at E21 with corresponding plot showing cell representation by condition. (B) Barplots with proportion of each cell type within the mesenchyme in normal and CDH lung. In CDH, mesenchymal cells upregulate (C) TGFβ signaling, (D) FGF signaling, (E) EMT proteins, and (F-J) Dysregulated signaling of TGFβ, FGF, and EMT is primarily seen in fibroblasts, SMCs/myofibroblasts, and adventitial fibroblasts. (K) Representative images of H&E, EVG, and Masson’s Trichrome (Masson Tri) for normal and CDH tissue at E21. Compared to normal, E21 CDH lung shows an upregulation of collagen (Tri) and elastin (EVG) production, but less than E17 CDH tissue (Fig. 3B).

TGFβ signaling is dysregulated in CDH. In CDH, TGFβ-2 and TGFRβ-2 were upregulated by fibroblasts, adventitial fibroblasts, and proliferating fibroblasts (Fig. 4f). A greater proportion of SMCs/myofibroblasts expressed TGFβ-2 in CDH compared to normal, although the overall expression was greater in normal SMCs. With regards to its receptor, TGFRβ-2, a greater proportion of adventitial fibroblasts expressed it in CDH when compared to normal (Fig. 4f). In CDH, Acta2+ pericytes upregulated TGFRβ-2 when compared to normal (Fig. 4f). In CDH, Smad2 and Smad3 were increased in virtually all mesenchymal cells, namely fibroblasts, adventitial fibroblasts, SMCs/myofibroblasts, pericytes, and the mesothelium (Fig. 4g).

Fgf signaling was dysregulated in CDH compared to normal. Fgf18 was downregulated in the mesenchyme of CDH lung but increased in SMCs/myofibroblasts (Fig. 4i). Fgfr2, however, was noted to be increased overall, specifically in SMCs/myofibroblasts, fibroblasts, and the proliferating fibroblast population [56].

In CDH, fibroblasts, pericytes, and smooth muscle cells continue to undergo EMT. Snai1 was increased in fibroblasts, SMCs/myofibroblasts, adventitial fibroblasts, and Pdgfr + pericytes (Fig. 4j). Snai2 was increased in fibroblasts and proliferating mesenchyme (Fig. 4j).

H&E, EVG, and Masson’s Trichrome staining was done for normal and CDH tissue at E21. Compared to normal, CDH lung shows an upregulation of collagen and elastin production (Fig. 4k).

During saccular stage distal epithelial cells in CDH upregulate genes related to cell transitional states and regeneration

We analyzed the epithelium at E21 and noted phenotypic differences in the distal epithelium (Fig. 5a). The proportion of AT2 cells in CDH (15%) compared to normal (28%) was lower by almost 50% (Fig. 5b). There was also a lower proportion of ciliated cells in CDH lung (3%) compared to normal (5%) (Fig. 5b). To better characterize these differences, we compared gene expression in the epithelium between CDH and normal lung.

Figure 5.

Distal epithelial cells at E21 in CDH upregulate genes related to fibrosis, and epithelial regeneration. (A) UMAP embedding of all epithelial cells at E21 with corresponding plot showing cell representation by condition. (B) Barplots with proportion of each cell type within the epithelium in normal and CDH lung showing decrease in AT2 cell proportion in CDH (15%) versus normal (28%). In CDH, the epithelium is in a (C) transitional epithelial stem cell state, and shows (D) upregulation of TGFβ pathway, and (E) profibrotic molecules. (F-H) In CDH, distal epithelial cells (AT1 and AT1 cells) are responsible for dysregulated gene expression. (I) Immunohistochemical staining of paraffin-embedded E21 lung tissue (CDH and Normal) demonstrating differential Krt8 expression. In CDH lungs, Krt8 is upregulated in the alveolar regions, marking ATI and ATII cells, whereas in normal lungs, Krt8 expression is largely restricted to the large airways. Vimentin (Vim) staining is included for comparison. Arrows indicate Krt8+ alveolar cells in CDH lungs and Krt8+ airway cells in normal lungs. Scale bar: 250 μm.

In CDH, there was upregulation of Krt8 and Krt18 (Fig. 5c) [57,58]. Specifically, Krt8 and Krt18 were upregulated in AT1 and AT2 cells (Fig. 5f).

In CDH, there was also increased expression of TGFβ-2, TGFRβ-2,a, Serpin6a, and Serpin6b (Fig. 5d, e) [59]. TGFβ-2 was upregulated in AT1 cells and its receptor, TGFβR-2, was upregulated in both AT1 and AT2 cells, and to a lesser degree, ciliated cells (Fig. 5g). Pdgfa was upregulated in CDH AT1, non-ciliated secretory, and ciliated cells. Serpin6a and Serpin6b were upregulated in CDH by both AT1 and AT2 cells (Fig. 5h).

We performed immunohistochemical staining of paraffin-embedded E21 lung tissue (CDH and Normal) demonstrating differential Krt8 expression. In CDH lungs, Krt8 is upregulated in the alveolar regions, marking AT1 and AT2 cells (marked by arrows), whereas in normal lungs, Krt8 expression is largely restricted to the large airways. Vimentin (Vim) staining is included for comparison (Fig. 5i).

Discussion

In CDH, there is a mesenchymal progenitor population present in early lung development that surrounds proximal airways. We found that this cell population gains a more mesenchymal-like characteristic when compared to normal fetal lung. In both conditions, these cells are undergoing EMT as evidenced by Twist1 and Prrx1 expression. EMT is a highly plastic process [1,15]. Normal mesenchymal progenitor cells in the lung exist in a hybrid state and are seen at the lung bud tip in branching morphogenesis [1,15]. Our findings suggest the mesenchymal progenitors are cells undergoing EMT in the developing lung bud [1,15].

In CDH, these cells downregulate Sox9, an important marker of cell plasticity, and upregulate ECM matrix proteins. EMT scoring analysis revealed Sox9- mesenchymal progenitors in CDH have increased expression of mesenchymal-specific genes compared to normal lung, largely driven by an increase in ECM protein expression. This difference in ECM protein expression in CDH suggests progenitors are entering a mesenchymal-like state compared to normal lung, which may represent altered regulation of EMT. Dysregulation of EMT and ECM production has implications for both mesenchymal and epithelial lung development. In addition to upregulation of ECM proteins, this cell population also secretes pro-fibroblastic cues in our CDH model via TGFβ pathway. This phenomenon has been observed in other study models of CDH. Kunisaki et al, compared molecular and cellular differences between normal and CDH lung cells in a human organoid model, and reported findings that are consistent with our study, including impaired growth of mesenchymal progenitors. They also demonstrated significant downregulation of Sox9, exacerbated by mechanical stress in lung organoids. These organoid models also had upregulation of pro-fibrotic markers and ECM proteins [60]. We speculate that imbalance of EMT and thus dysregulation of key molecular pathways in CDH, in particular TGFβ signaling, leads to poorly development mesenchymal structures and a transitional distal epithelial state that causes lung hypoplasia.

In CDH, there is dysregulation of the TGFβ pathway at both the pseudoglandular and saccular stage. Early in development, Ltbp4 and SMAD1 proteins are increased by mesenchymal progenitors. Latent transforming growth factor β-binding protein (Ltbp4) is an essential regulator of TGFβ activation and plays a central role in regulating inflammation and fibrosis by binding TGFβ in the ECM [61]. Studies suggest Ltbp4 is crucial in fibroblast activation and can lead to fibrosis [62]. In our study, increased Ltbp4 production could be a contributing factor to a fibrotic niche. SMAD proteins are downstream signaling proteins of TGFβ activation and are increased by multiple mesenchymal cells at E17 and E21, suggesting increased activation of the TGFβ pathway [36]. These findings suggest a fibrotic-like environment initiated by mesenchymal progenitors that persists throughout lung development in CDH.

Epithelial cells in CDH exhibited a transitional regenerative state. AT1 and AT2 cells in CDH upregulate markers of cellular transitional state (Krt8/Krt18) and pro-inflammatory molecules, such as Pdgfa and Serpin proteins [59]. We speculate these inflammatory mediators were induced by changes in the microenvironment by mesenchymal progenitors. Of particular interest is the presence of AT1 and AT2 cells that express Krt8 and Krt18, as co-expression has been found to represent a transitional stem cell state that precedes the regeneration of AT1 cells in lung injury [57]. Previous studies have suggested Krt8/Krt18+ epithelial cells persist in fibrotic states due to TGFβ activation [57,58]. The persistence of transitional epithelial stem cell states disrupts terminal differentiation [57]. We speculate this leads to lung hypoplasia in CDH. Moreover, CDH epithelial cells expressed various inflammatory genes associated with senescence, including Serpin6a and 6b, Tgfβ, and Pdgfa. We hypothesize that these inflammatory mediators contribute to the initiation and progression of the disease and that activation of TGFβ further exacerbates a fibrotic niche. A similar phenomenon has been proposed in bleomycin-induced interstitial lung disease of murine models [59]. To further support the observed inflammatory phenomenon, a recent study by Dylong et al. demonstrated aberrant activation of the NF-κB signaling pathway, a key regulator of inflammation, in multiple CDH models, including nitrofen-exposed fetal rat lungs, lung explants, and human fetal CDH lungs [63]. They found that phosphorylated p65, a transactivator of NF-κB, was significantly enriched in the airway epithelium of hypoplastic lungs across developmental stages [63]. Importantly, they also showed that abnormal branching morphogenesis and NF-κB–mediated inflammation could be rescued via antenatal treatment with dexamethasone or targeted NF-κB inhibition. These findings reinforce the idea that inflammation in CDH is not a late or secondary feature, but rather an intrinsic and persistent aspect of disease pathophysiology that may directly impair lung growth. Several other studies have reinforced the role of NF-κB dysregulation in the pathogenesis of CDH [64-68], including Antounians et al. who showed CDH fetal rat hypoplastic lungs have an inflammatory signature with high density of macrophages and up-regulation of biological pathways including NF-kB and TNF-alpha that are involved in inflammatory and innate immune response [69].

Mesenchymal progenitors in CDH upregulate proteins important for ECM production and structure. Multiple collagens are upregulated by this cell population in CDH. Over-production of collagen is associated with multiple lung pathologies, such as fibrosis [40]. Collagen production in the adult lung is typically produced by fibroblasts or myofibroblasts in response to lung injury, however, we observe overexpression of collagen early in lung development in CDH. Laminins, which are essential components of the basement membrane, also upregulated by CDH mesenchymal progenitors. In normal lung development, different laminin isotypes play a role in lung tissue structure and branching morphogenesis and thus different developmental abnormalities of the lung may result from dysregulation of each [70-72]. In our study, we see upregulation of laminin alpha-2 which is important for smooth muscle cell differentiation [71]. Lumican, a proteoglycan important for lung homeostasis, is also upregulated by CDH mesenchymal progenitors. In human studies, the presence and quantity of lumican has been linked to severity of fibrotic and inflammatory processes [73]. It also plays a role in ECM matrix composition, which influences lung mechanics, much like collagen and elastin and may play a role in the observed pro-fibrotic niche [74]. Decorin (Dcn) is also upregulated in and highly specific to CDH lung mesenchymal progenitors. Dcn has been described in the cancer literature to be play an anti-metastatic role by antagonizing TGFβ signaling. Dcn expressing mesenchymal stem cells attenuate acute lung inflammation and inhibit fibrosis [75]. The production of decorin by mesenchymal progenitors in normal lung is low. This suggests that decorin in CDH is upregulated in response to altered TGFβ signaling seen. The upregulation of aforementioned ECM proteins in CDH suggests aberrant mesenchymal progenitors surrounding the developing airway are contributing to changes in ECM matrix structure and thus mechanics as early as the pseudoglandular stage.

There are notable limitations and challenges to this study. One limitation is the small sample size. Even though our cell count and quality was excellent, given it is fetal tissue, cell recovery is smaller than adult rat lung. Although single-cell RNA-sequencing provides a powerful view of cellular transcription, it typically captures only a subset of the transcriptome. This can affect detection of lowly expressed genes or subtle expression differences. The enzymatic and mechanical dissociation required to isolate single cells can introduce bias by preferentially capturing certain cell types while underrepresenting or damaging others, such as fragile or tightly adherent cell populations. More samples should be repeated to improve reproducibility. Another limitation is lack of validation of certain genes of interest at the protein level. This is an ongoing work in progress in our laboratory. The primary challenge has been finding validated rat antibodies for certain genes. Further studies are required to further elucidate ligand-receptor relationships not uncovered via phenotypic data analysis. Dissociation of tissue into single cells disrupts spatial information, making it difficult to infer cell-cell interactions, tissue architecture, or niche-specific functions without complementary spatial transcriptomic data. In our lab, we are currently replicating the discussed timepoints and conditions using spatial-omics. In addition, single cell RNA-sequencing of a fetal rat lung atlas has not previously been described. Because we have labeled cell types based on referenced datasets and previous studies conducted on mouse or human lungs with known markers, our dataset may not fully represent the diversity or novelty of unknown fetal lung cell populations. This can lead to misclassification or incomplete labeling of cell states.

Despite advances in care for babies with CDH, there is tremendous morbidity and mortality due to underdeveloped lungs, termed pulmonary hypoplasia, and abnormal blood flow to the lungs, termed pulmonary hypertension. These babies are critically ill from the moment they take their first breath and it would be ideal to be able to improve lung structure and function in utero before the lungs are required for gas exchange. It is our long-term hope that understanding the molecular signals driving both normal fetal lung development and pulmonary hypoplasia in CDH will allow for targeted molecular treatments.

To date, CDH is not treated as a fibrotic or inflammatory disease and we suspect that insight from this study might be valuable for future therapeutic development. Our group has completed proof of concept studies that demonstrate improved lung structure following particle-based epigenetic therapy targeting TGFβ signaling in this CDH rat model [21]. A more complete understanding of the molecular mechanisms driving lung development may allow us to target and influence the specific cell populations that drive abnormal fetal lung development in CDH. In addition, an understanding of the coordination of fetal lung development will be important in the tissue engineering space. We intend, in future work, to use the data present in this manuscript to design targeted therapeutics capable of guiding lung development as a holistic multicellular system. Success in this area would advance pulmonary regenerative medicine as a translational, data-driven discipline.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15513815.2025.2585371.