Dynamics of primary cilia in endothelial and mesenchymal cells throughout mouse lung development

Abstract

Cilia are specialized structures found on a variety of mammalian cells, with variable roles in the transduction of mechanical and biological signals (by primary cilia, PC), as well as in the generation of fluid flow (by motile cilia). Their critical role in the establishment of a left–right axis in early development is well described, as well as in the defense immune function of multiciliated upper airway epithelium. By contrast, detailed analysis of the ciliary status of specific cell types during organogenesis and postnatal development has received less attention. In this study, we investigate the progression of ciliary status within the endothelium and mesenchyme of the lung. Remarkably, we find that pulmonary endothelial cells (ECs) lack PC at all stages of development, except in low numbers in the proximal portions of older pulmonary arteries. Mesenchymal cells, by contrast, widely exhibit PC in early development, and a large subset of PDGFRα+ fibroblasts maintain PC into adulthood. The dynamic and differential ciliation of multiple cellular populations in the developing lung both challenges prior assertions that PC are found on all cells and highlights a need to understand their spatiotemporal functions.

Key Findings

• Primary cilia are found broadly throughout early embryonic tissues.

• Primary cilia are observed in both epithelial and mesenchymal cells in the early lung.

• Pulmonary endothelial cells lack primary cilia during embryonic development.

• Maintenance of cilia in adult pulmonary PDGFRα+ fibroblasts.

1. INTRODUCTION

Cilia are highly specialized, hair‐like organelles found on a variety of mammalian cells. Cilia can be broadly defined as either non‐motile primary cilia (PC), which facilitate signal transduction by sensing chemical, biological, or mechanical stimuli, or as motile cilia, which function to generate directional fluid flow. ¹ Whether or not a cell constructs a cilium—and which kind it builds—likely reflects a need for a specific ciliary function.

The essential function of cilia is evident from the very beginning of embryonic development. The foundational example is found at the embryonic node in mice, which dictates the left–right asymmetry of the body plan. At the node, mechanisms of planar cell polarity in centrally located pit cells ensure that motile cilia are aligned to generate unidirectional fluid flow. That flow is then sensed by the immotile cilia on peripherally located crown cells, leading to asymmetrically induced calcium ion transients, followed by gene regulatory events such as Dand5 mRNA degradation. ² , ³ Beyond the establishment of overall internal organ sidedness, however, primary cilia have also been shown to be critical to the later development of several tissues, including heart, ⁴ , ⁵ bone, ⁶ , ⁷ , ⁸ kidney, ⁹ and brain. ¹⁰ , ¹¹

These later, organ‐specific developmental roles are evidenced by the wide‐ranging clinical presentations of the ciliopathies (diseases caused by mutations in known ciliary proteins), which cannot be fully explained by defective function in the primary cilia in the ventral node. From the study of polycystic kidney disease (PKD), Bardet‐Biedl syndrome, Meckel–Gruber syndrome, or Joubert's syndrome, defects in ciliary proteins have been linked to a wide range of multisystem birth defects. These include neural tube defects, congenital heart defects, respiratory problems, polydactyly, and more. ¹² Building from clinical observation, increasingly strong pre‐clinical evidence suggests that congenital heart disease is, at its core, a multigenic ciliopathy. ⁶ , ⁷ , ⁸ In the developing neural tube, PC functions as a sensory organelle and a signaling hub, relaying signals from a number of developmental pathways including the Sonic Hedgehog (Shh) pathway and regulating proper morphogenesis and patterning. ¹³ Deletion of Kif3a—a gene essential for cilia formation—in the developing pancreas results in severe pancreatic defects, including acinar‐to‐ductal metaplasia, lipomatosis, and fibrosis. ¹⁴

Another interesting aspect of ciliary biology is that they are remarkably transient, yet critical structures. Cilia are constructed from the docking of a primary centriole at the plasma membrane, which is then used as the foundation for microtubule extension. Hence, an actively dividing cell cannot maintain a mature cilium, as the same primary centriole that is used at the basal body of the cilium is needed to coordinate chromosomal separation and cytokinesis. The critical role that motile and primary cilia play in development—the most active period of cellular proliferation—is therefore somewhat paradoxical. When do they appear, and for what purpose?

As the different cellular lineages organize with respect to one another in an organ, the cellular antenna that is a primary cilium is well situated to detect and coordinate short‐range paracrine signals. Understanding which cells require primary or motile cilia can provide critical insight into these cross‐lineage organizational relationships. However, detailed temporospatial analysis of ciliary dynamics has thus far been largely limited to the epithelium—likely motivated by the clinical relevance of motile ciliary defects to the innate immune functions of lung bronchial epithelium. ¹⁰ , ¹¹ , ¹⁵

In this study, we define the temporal presence or absence of cilia on different cell types in the developing mouse lung. Using immunofluorescent antibody staining (IF), we find that pulmonary tissues display a high density of PC during early organogenesis. In addition, we identify the cells—epithelial, mesenchymal, or endothelial—that display cilia throughout development and into postnatal life. We find that lung endothelial cells (ECs) are almost completely devoid of PC at all stages of development examined, while mesenchymal cells initially express—but then steadily lose—their PC. A critical exception is alveolar PDGFRα+ fibroblasts, which retain cilia into adulthood. By understanding what pulmonary cells display cilia (and when), we hope to gain insight into their functional roles during organogenesis and the importance of cilia that are maintained into adulthood.

2. RESULTS

2.1. Primary cilia are present throughout multiple tissues in the early mouse embryo

To assess where primary cilia (PC) in the embryo might be positioned to play a role in tissue formation—beyond the well‐known establishment of laterality by the cilia at the ventral node (E7.5)—we looked for the presence of PC in a range of embryonic organs. We first stained sectioned mouse embryonic lungs using antibodies to ARL13B, which is a highly specific ciliary marker (Figure S1A). ²³ To assess which specific cell types might harbor PC, we also stained for endothelium using antibodies to the membrane‐bound glycoprotein endomucin (EMCN) and for epithelium using antibodies to the cell adhesion protein E‐cadherin (ECAD/CDH1). Our results showed the widespread presence of PC throughout embryonic organs at embryonic day 10.5 (E10.5, Figure S1A–F). We noted a high frequency of PC not only in the epithelium of tissues like lung (Figure S1A) and gut (Figure S1B) but also throughout the surrounding mesenchyme (asterisks). ARL13B+ structures could be observed scattered throughout all organs and tissues examined.

To further evaluate distribution, we made a global quantification of total cilia and the total number of cells in multiple organs at E10.5. We counted the number of cells observed (using DAPI to identify nuclei of single cells) that displayed an associated ARL13B+ structure, as well as the total number of cells that did not. Overall, we found a high PC‐to‐cell ratio in multiple organs, particularly the lung (75%), gut (76%), and neural tube (71%), while tissues such as the liver and heart exhibited fewer cilia‐bearing cells (Figure S1G). Overall, these observations suggest a role for cilia during mammalian organogenesis, as they are widely distributed in different organs early during development, including the lung.

2.2. Primary cilia are present in multiple cell types in the embryonic lung

We next assess cilia in the budding lung more closely. Examination of E11 lungs revealed that cilia positive for ARL13B staining were widely distributed throughout the lung and present in both proximal and distal regions (Figure 1A). We next sought to determine the frequency of cilia on the main cell types—epithelial, mesenchymal, endothelial (Figure 1B–D).

FIGURE 1.

Presence of primary cilia of distinct lengths in lung epithelial and mesenchymal cells. At E11, immunofluorescent staining of developing lung tissue for PC with ARL13B (A). Higher magnification views of co‐staining for PCNT and ARL13B (green arrows) in epithelial (ECAD, B), mesenchymal (PDGFRα+, C), and endothelial (SOX17+, D) cells. Note presence of PCNT, but absence of ARL13B in the EC (red arrow) in (D). Adjacent non‐endothelial cells display structures that are positive for both PCNT and ARL13B (green arrow). The cilia expressed by epithelial cells (E) are significantly shorter than those expressed by mesenchymal cells (F). Mean epithelial ciliary length was 1.19 μm, while mean mesenchymal ciliary length was 1.92 μm (G). Cilia in E, F were assessed in distal lung sections. Scale bars are 100 μm (A) and 1 μm (B–F).

Focusing first on the pseudostratified epithelium of the distal lung bud, we found that each epithelial cell was associated with a single PC (Figure 1B). The presence of a single cilium in each cell was confirmed by co‐staining for ARL13B and pericentrin (PCNT), which localizes to the base of each PC. The columnar organization of the epithelium made it straightforward to identify the cell of origin for each cilium. However, the more complicated organization of both mesenchyme and endothelial cells (ECs) in distal portions of the lung was more challenging. Accuracy was further limited by a high density of embryonic tissue cells with a high nuclear‐to‐cytoplasmic ratio. To precisely assess the cell types that harbored cilia in these tissues, we sought to establish the polarity of each cilium by staining for cilia‐associated structures found at the proximal end of the structure. We did this by defining the 1:1 cilium: cell ratio using examination of cells in three dimensions, using Maximum Intensity Projection (MIP) reconstructions of Z‐resolution (0.5 µm slice depth). This facilitated the identification of cilia in mesenchymal cells (Figure 1C). Indeed, staining for the mesenchymal marker PDGFRα allowed us to ascertain that most mesenchymal cells within the lung bud displayed a single PC.

We next asked whether ECs also harbored cilia (Figure 1D). While PCs in epithelium and mesenchymal cells were pervasive and easily identifiable, it was more difficult to find PCs on ECs. Sox17+ ECs mostly exhibited PCNT staining with no associated ARL13B signal (red arrow), while cells in adjacent mesenchyme displayed clear colocalization of PCNT and ARL13B positivity (green arrow) (Figure 1D). Together, these data show that the PCs are differentially present among multiple cell types during early lung organogenesis.

Of note, the combined ARL13B and PCNT stain also allowed for evaluation of individual cilia length (Figure 1E,F). We found that PC in the epithelium (Figure 1E) displayed shorter cilia than those in the mesenchyme (Figure 1F,G). Given that ciliopathies present with shorter cilia, this differential length suggests that the epithelial primary cilia may have different signaling functions than those in the mesenchyme at this stage. ²⁴

To confirm that ARL13B marked bona fide cilia, we carried out costains for other known markers of PC, including IFT88, γ‐tubulin, and acetylated α‐tubulin (Figure 2A–F). ²⁵ , ²⁶ ARL13B does indeed label cilia that were also positive for these additional markers, at different stages and throughout the lung (Figure 2G–I). To further confirm cilia in the mesenchyme at this stage, we performed Transmission Electron Microscopy (TEM) on E12.5 lungs (Figure S2), which demonstrated PC located in a deep pocket on mesenchymal cells (Figure S2B), with the basal body and axoneme clearly visible (Figure S1B, inset).

FIGURE 2.

Confirmation of primary cilia identification with multiple immunofluorescent markers. Immunostaining of lung tissue at E12.5 (A–C') and P30 (D–F'). In all panels, ARL13B is shown in green, while other ciliary markers including acetyl‐α tubulin (A, A', D, D'), IFT88 (B, B', E, E'), and γ‐tubulin (C, C', F, F') are in red. (G) E12.5 images were selected from the distal lung bud epithelium and surrounding mesenchyme. P30 images were selected from the distal bronchioles (to show motile cilia on multiciliated epithelial cells) and the surrounding distal alveolar space. (H) Structure of a primary cilium and location of different substructures. (I) Localization of immunofluorescent markers used, showing proximodistal specific locations. Scale bar is 5 μm.

2.3. Cell type specific evaluation of cilia requires 3D analysis

The difficulty in clearly identifying endothelial cilia (Figure 1D) prompted us to assess additional ciliary markers to more clearly delineate the cell of origin for each PC. We therefore used GM130 to visualize the Golgi. Similar to PCNT, the Golgi is always associated with the base of the cilia (Figure 3A,B), but it additionally wraps around the nucleus of the cell of PC origin (yellow bracket). This shared contour allows for a more confident assignment of an individual cilium to its cell of origin.

FIGURE 3.

Identifying the ciliary cell of origin at the cellular level in embryonic tissues requires 3D analysis. Pericentrin (PCNT) marks the basal body of cilia (A). Golgi (GM130) also marks the base of a PC but provides contour to match the nucleus of the cell that expresses the cilium (B). The contour of GM130 allows for clear differentiation of the cell of origin, while high resolution (0.2–0.5 μm slices) confocal microscopy (C) and 3D modeling with Imaris (D) can overcome both high cell density and overlapping cells of different types (E, F). Scale bar is 5 μm.

To assess possible endothelial cilia, we additionally used the endothelial nuclear marker ERG, which is expressed in most ECs in the mid‐gestation stage lung tissue, ²⁷ in combination with ARL13B and GM130 (Figure 3C). However, we realized that for ECs, using maximum intensity projection (MIPs) still yielded artifactual data, as identifying the cell of origin for a single cilium depended on the precise z plane examined. As this method (MIP) compresses multiple visual slices, it often inaccurately showed ECs as having cilia. We therefore turned to analyzing ECs using 3‐dimensional reconstructions (Figure 3A'–D'). Single slices helped provide additional granularity (Figure 3E,F) and allowed us to analyze the cell of origin for each PC within the interstitium of the lung parenchyma. The methodology aided in the rigorous identification of embryonic lung EC and the possible presence of cilia. Using this approach, as well as careful analysis of 3D confocal Z‐stacks, we found that mouse embryonic lung ECs do not exhibit ciliary structures.

2.4. Lack of primary cilia in the pulmonary endothelium during development

Using the methods developed above, we assessed ciliary status in ECs throughout lung formation. We carried out 3D immunofluorescent staining of ARL13B in combination with SOX17, an EC nuclear marker, and GM130 for identification of Golgi. We focused on the distal lung during both pre‐ and post‐natal development.

Similar to our observations at E11 (Figure 1D), we found that ECs examined in 3D lacked PC at all developmental timepoints (Figure 4). Sectioned embryonic lung from E12.5 (Figure 4A), E16.5 (Figure 4B) and P0 (Figure 4C), as well as adult tissue P30 (Figure 4D) was stained for ARL13B, GM130, and SOX17. We assessed ECs located within the stroma of distal alveoli as we did in Figure 2 (see schematic in 2G). 3D renderings showed that SOX17+ nuclei were devoid of ciliary structures. Using ARL13B combined with GM130 allowed the determination of the cell of ciliary origin, and we showed that SOX17+ nuclei were not associated with PC (Figure 4A'–D'). We counted 1333 nuclei in 12 fields of view (FOV) in the lung periphery (n = 3), and we observed no association of ARL13B/GM130 with SOX17+ nuclei (Figure 4E,F). The complete absence of PC was noted in all embryonic ECs as well as mature ECs after the completion of alveologenesis (after P30). We did not attempt to differentiate between alveolar (aCap) or general (gCap) capillary cells in this study, as both cell types express ERG ²⁸ , ²⁹ and no ciliated ERG+ (or SOX17+) cells could be found at later time points (Figure 4C,D).

FIGURE 4.

Pulmonary endothelial cells lack primary cilia throughout development. Immunofluorescent staining of 10 μm slices of PFA‐fixed, paraffin‐embedded mouse lung tissue at the indicated developmental time points (A–D), followed by 3D rendering of the nuclei (A'–D') to determine each cilia's cell of origin. 3D models of nuclei, golgi, and cilia color to match the IF stains above. X and Y correspond to the X and Y plane of the above image, with the model rotated to highlight each cilia's correct relationship to the nucleus of its cell of origin. * and +identify the same cilia in each matching image set (A, A' for example). Quantification of total cells analyzed (E) reveal no ciliated endothelial cells among those analyzed at E12, E16.5, P0, or P30, while the mean EC and ciliated EC counts per FOV analyzed are shown in (F). Sections taken in regions similar to those in Figure 2G. Scale bar is 5 μm.

Because it has long been known that ciliary structure is dependent on the state of the cell cycle, ³⁰ we next examined whether perhaps the apparent lack of EC cilia was due to active cycling of ECs. To address this point, we co‐stained ECs with the endothelial marker ERG and the marker of cell proliferation Ki67 (which shows no staining in G0, punctate nuclear staining in G1 and G2, and strong nuclear staining in M phase). Actively dividing cells were easily identified (Figure 5A–D', white arrows). We found that while a small percentage of ECs at each stage examined were actively proliferating (Figure 5E), Erg+ cells that were not cycling also did not exhibit ARL13B+ structures (Figure 5A–D', green arrows). These findings indicate that the lack of endothelial PC is not likely representative of regression of cilia due to cell division.

FIGURE 5.

Absence of cilia in endothelial cells is not due to cilia regression during mitosis. Images showing lung endothelial cells (Erg, cyan) co‐stained for the proliferation marker Ki67 (yellow) (A–D). Panels showing Erg only staining (A'–D'). Note absence of ARL13B staining in dividing cells that are positive for Ki67 (white arrows). Conversely, note that numerous Erg+, but Ki67 negative cells (green arrows) do not display ARL13B+ cilia. (E) Quantification of ECs demonstrating Ki67 positivity. 1333 cells were counted. (F) Location of section for sections in (A–D').

While we found the distal lung endothelium was devoid of PC, ECs in larger vessels did indeed display PC, as previously reported. Whole mount staining of en face dissected branch pulmonary arteries revealed that the more proximal, large branch pulmonary arteries did contain ciliated ECs in the adult mouse (Figure 6A,C). Similarly, we were able to identify PC in ECs of the aortic arch (lesser curvature) (Figure 6B,D). These findings are in keeping with previous study reporting the presence of PC in the aorta—specifically, along the lesser curvature of the arch, where turbulent flow is present. ³¹ We confirmed this observation using TEM (Figure S3), which revealed the absence of PC in regions of laminar flow (descending aorta) (Figure S3A). Regions of disturbed flow, like the lesser curvature of the aortic arch, showed some occasional PC (Figure S3B,C). Hence, we were able to find some examples of endothelial PC, albeit in adult EC.

FIGURE 6.

Presence of cilia in the large pulmonary arteries and aorta. Aorta and pulmonary artery of mice (n = 3, P60) were perfused and fixed with PFA prior to en face staining for EC borders (VECAD, cyan) and primary cilia (ARL13B, pink), with Golgi (GM130, yellow) to clarify cell of origin. Multiple cilia were found in both pulmonary artery (A) and the lesser curvature of the aorta (B). Schematics showing location of stained ECs (boxed areas) in main pulmonary artery (C, blue box) and in the aortic arch (D, red box). Scale bar is 10 μm.

This study identifies a proximal‐distal gradient of endothelial PC in large vessels of the adult mouse, with select subsets of ECs in the large arteries displaying PC, whereas ECs of the distal capillaries do not.

2.5. Maintenance of primary cilia in PDGFRα+ alveolar fibroblasts

Finally, because PCs are observed in early lung mesenchyme, we assessed the presence of PCs in derivatives of the mesenchymal cell population at P30, with a focus on the distal alveolae. Using DNAH11 as a reliable marker of motile cilia, ²⁵ we confirmed the presence of DNAH11 in motile epithelial cilia (Figure S4A) in order to show its absence in the primary cilia identified in the alveolar region (Figure S4B). In this region, PDGFRα + cells comprise the majority of the lung parenchyma and are critical units of lung function. ³² Given the complex three‐dimensional architecture of the mature alveolus, we examined markers specific to all major cell types to assess which cells exhibit a primary cilium at the alveolar level. We found that while PDGFRα expressing cells exhibited single PCs (Figure 7H), neither pericytes (PDGFRβ+) (Figure 7B), aCAP cells (CA4+) (Figure 7C), ECs (SOX17+) (Figure 7D), AT1 cells (HOPX+) (Figure 7E), AT2 cells (LAMP3+) (Figure 7E), nor lymphatic ECs (PROX1+) (Figure 7G) displayed PCs. Interestingly, the mesothelium also displayed primary cilia (Figure S5). In sum, the mesenchymal cell population diversifies greatly as development progresses, but the PC presence appears to be primarily maintained in adulthood by PDGFRα + alveolar fibroblasts.

FIGURE 7.

Maintenance of primary cilia by PDGFRα + myofibroblasts in the adult at P30. Immunofluorescent analysis of adult mouse distal lung to identify cell types that harbor cilia. Figure organization is shown in (A), with alveolar spaces outlined in orange and indicated by asterisks. In each image, yellow staining marks the cell type of interest. The yellow inset shows cells positive for cell type specific marker as indicated, while the red inset shows a ciliated cell. Immunofluorescent analysis revealed that pericytes (B), alveolar capillary (aCap) cells (C), ECs (D), AT1 cells (E), AT2 cells (F), and lymphatic ECs (G) all lack PC. Co‐localization of PC with is observed only with PDGFRα+ fibroblasts (H). Scale bar is 10 μm.

2.6. Temporal regulation of mesenchymal ciliogenesis as analyzed by scRNAseq

To further gain insight into the specific mesenchymal cell types that might harbor PC in the mouse lung, we carried out an analysis of previously published scRNAseq datasets ¹⁸ , ¹⁹ , ²⁰ (Figure 8A). Focusing on the mesenchymal subset, UMAP analysis was able to delineate all the expected cell types, including airway smooth muscle cells, fibroblasts, and secondary crest myofibroblasts (SCMFs) (Figure 8B). We also resolved cell types by developmental stage (E12, E17, P7, P15, P42) (Figure 8C).

FIGURE 8.

Mesenchymal Arl13b expression can be found in Pdgfrα‐expressing fibroblasts throughout lung development. (A) Previously published lung development scRNAseq datasets were integrated and the mesenchyme was included as a subset. (B) UMAP of mesenchymal cell types (C) UMAP of developmental timepoints. (D) Feature plots show expression of Pdgfrα and Arl13b in the lung mesenchyme across developmental timepoints. Whole‐mount images (40 μm z‐stacks) from P7 (E) and adult (F) PdgfrαH2B:GFP reporter mice shows nuclear ARL13B expression in Pdgfrα‐expressing fibroblasts (magenta arrows). Scale bar is 50 μm.

Using the single transcript Arl13b to assess the correlation between RNAseq and our immunofluorescent analysis of mesenchymal cells, we examined clusters that represented various subsets of stromal cells, including early mesenchyme, fibroblasts, and various types of smooth muscle and myofibroblasts. Examining embryonic stages (E12 and E17), we found that Pdgfrα was highly transcribed in adventitial and alveolar fibroblasts, early proliferating mesenchyme, and ductal fibroblasts (Figure 8D). Postnatally, Pdgfrα expression peaked in proliferating and SCMF mesenchymal populations, as well as in adventitial fibroblasts (Figure 8F). After P15, elevated expression was observed mainly in alveolar fibroblasts. Strikingly, these were all the same populations that displayed high levels of Arl13b transcripts, reflecting our immunofluorescent findings.

To assess whether scRNAseq data would confirm the lack of PC in the endothelium, we created an endothelial‐specific subset from the same data (Figure S6). This analysis revealed sporadic cells that expressed Arl13b, with a higher number of Arl13b + cells in the gCap cells of the older (P42) animals (Figure S6D). We then asked if these Arl13b + gCap cells were also positive for Ift88 and Tubg1, as we had seen clearly by immunofluorescence in the ciliated epithelium and mesenchyme. By creating a sequential subset of Arl13b, Ift88, and Tugb1, we were able to investigate the ability of a limited gene set to identify ciliated cells (Figure S7A). We noted that these three transcripts were not present in similar abundance within the same cells. The “ciliated” subgroup of epithelial cells was the only group for whom even co‐expression of two different transcripts (Arl13b + and Ift88) did not reduce the proportion of positive cells to a negligible value (Figure S7B). In fact, even the epithelial cell population as a whole was reduced from 10,901 cells to only 22 cells that co‐expressed Arl13b, Ift88, and Tubg1.

We then broadened our examination further to include all subsets—endothelial, mesenchymal, epithelial, and immune cells (Figure S8). Beyond the findings in the mesenchymal subsets, we noted similar patterns of Arl13b+, Ift88, and Tugb1 expression in epithelial cells, secondary crest myofibroblasts, and alveolar fibroblasts (Figure S8E). These scRNAseq data reveal the overall patterns but do not facilitate the identification of ciliated cells for downstream pathway analysis as a clue to specific ciliary function.

Given the clear overlap of Arl13b expression with that of Pdgfrα, and the immunofluorescent data from Figure 7, we asked whether the Pdgfrα population of cells bears PC in the postnatal lung mesenchyme. Using a PdgfrαH2B:GFP mouse line, we confirmed that the Pdgfrα‐positive cells indeed give rise to cells displaying clear ARL13B+ structures at both P7 (Figure 8E) and adulthood (Figure 8F). We found that in P7 and P42 mice (Figure 8E,F), approximately 57% and 65% (respectively) of all Pdgfrα H2B‐GFP+ cells displayed ARL13B+ ciliary structures (counting 114 and 188 cells over 5 fields of view for each time point).

Together, these data reveal a dynamic spatiotemporal arrangement of primary and motile cilia throughout the development of the murine lung (Figure 9). The unique absence of primary cilia from ECs throughout development highlights the cell‐ and context‐specific signaling in which primary cilia participate. Importantly, we also identify a population of PDGFRα+ alveolar fibroblasts that maintain primary cilia into adulthood.

FIGURE 9.

Summary of dynamics of ciliary presence in the developing and mature lung. (A) During embryogenesis, PC are observed along the apical surface of the lung epithelial cells (distal epithelium in dark green, and proximal bronchial epithelium in light green). In addition, cilia are observed on most mesenchymal (yellow), but not endothelial (blue) cells. (B) In the adult, PC are exclusively observed on PDGFRα+ cells within the stroma of the alveoli, but not PDGFRβ+ (orange), AT1 (light green) or AT2 (dark green). PC are not observed in the ECs (blue) of the aCaps. Some PC are found in the aorta and large pulmonary arteries, and multiciliated cells are found in epithelium (white) of the bronchus.

3. DISCUSSION

Here, we report the dynamic and differential appearance of cilia in epithelial, mesenchymal, and endothelial cell (EC) types within the developing mouse lung. We use immunofluorescent staining of markers for the ciliary protein ARL13B, as well as co‐staining with pericentrin (PCNT) and GM130, to identify the cell of origin of primary cilia (PC). We confirm that these ARL13B+ structures are cilia using other known ciliary markers. We examine forming lung tissues from E10.5, throughout development, and into adulthood. We find that, contrary to our initial expectations, ECs within peripheral lung tissues never display PC. Endothelial PC are exclusively observed in large vessels, such as the pulmonary arteries and the midline thoracic dorsal aortae. Conversely, cilia are found dynamically in epithelial and mesenchymal cell types. Intriguingly, we observe PC within the pulmonary stroma, primarily on PDGFRα + cells. We do not observe PC on numerous other cell types, including alveolar capillary (aCAP) cells, lymphatic ECs, AT1 and AT2 cells, and pericytes. To our knowledge, this is the first high‐resolution characterization of cilia within the developing lung. These findings underscore the dynamic and cell‐specific presence of PC in the lung and suggest important signaling pathways that are active within a subpopulation of the stroma.

3.1. What role do cilia play in pulmonary organogenesis?

Our study aims to bridge two key findings well established in the literature—paracrine signaling and cilia signal transduction—and map these events at a cellular level in the developing lung. First, multiple studies have demonstrated the critical role of hedgehog, Wnt, and PDGF signaling in the normal development of the lung, as they coordinate cellular outgrowth and differentiation to form a branched and vascularized epithelial structure. ³³ , ³⁴ Second, numerous studies have identified the ciliary localization of hedgehog, Wnt, and PDGF signaling receptors and downstream effectors, and the role of cilia in effective signal amplification. ³⁵ The role of this remarkable cellular structure in coordinating these pathways is yet unknown in lung development. It is likely that PC acts as “antennae,” sensing and transducing multiple paracrine signals that guide cell differentiation and embryonic morphogenesis. In this study, we show they are abundant throughout the lung.

We note that the markers used in this study do not definitively distinguish these embryonic structures as either primary or motile cilia. Only careful transmission electron microscopy (TEM) can do this by assessing the structure of the (9 + 2) outer microtubule doublets with two central microtubules, something not possible with techniques used here. Generally, motile cilia are associated with the expression of FoxJ1 in the cell body and DNAH11 in the cilium, among other markers (Figure S4). ²⁵ Further detailed studies would be needed to confirm the precise types of cilia in embryonic tissues; however, given those markers used here, we are likely observing the widespread presence of PC in embryonic lung tissues.

3.2. Widespread absence of pulmonary endothelial cilia

ECs are highly adapted to sense and withstand the significant mechanical forces induced by blood flow (shear, stretch, pressure). PCs are known to participate in force sensing and transduction, ³⁶ and so we expected to find cilia in pulmonary endothelial cells. Indeed, the presence of ciliated endothelial cells has been thoroughly demonstrated in the zebrafish. ³⁷ , ³⁸ , ³⁹ Because EC PCs are observed in vitro under static or low shear conditions, but are either sheared off ⁴⁰ or disassembled ⁴¹ by the onset of high shear stress flow, we predicted that cilia would be present in the capillaries and absent in the larger arteries. Surprisingly, the exact reverse was true.

Shear stress in the pulmonary arteries increases dramatically after birth. It then decreases somewhat with the closure of the ductus arteriosus and is further modulated by the continued vascular development and alveolarization that takes place in the first 6 months of human life. The effects of these hemodynamic changes on endothelial ciliation are not known but will be investigated in our future study. Recent computational modeling of shear stress in the pulmonary vasculature provides evidence that, in fact, shear in the distal arteries is much higher than in the proximal arteries, which may explain our preliminary findings on this issue. ⁴² This model (higher shear stress in smaller vessels leads to loss of PC) would support a hypothesis that shear forces determine EC ciliary presence in the mouse. However, blood flow is quite minimal in the early stages of development examined in our study—at which point ECs are still devoid of PC. More study is needed to understand how a primary cilium can be maintained under variable shear stresses in vivo—and what purpose (mechanosensitive or otherwise) the PC serves in the larger arteries.

The single cell RNA sequencing data (Figure S6) reveal scattered ARL13B expression in ECs across time points (albeit with very little expression at E12). This expression could indicate either (i) cilia formation that we failed to see by IF, (ii) Arl13b transcription and localization not associated with cilia, ⁴³ (iii) contamination by ambient RNA, or (iv) imperfect separation of closely associated endothelial and mesenchymal cells during processing for scRNAseq, as previously reported in brain parenchyma. ⁴⁴ Indeed, prior study has claimed the identification of primary cilia in a single human pulmonary arterial EC (Figure 1H from Lee 2018). ⁴⁵ While it is difficult to tell, this EC appears to be in a larger artery. While we are confident in our analysis of the IF stains, further study is needed to clarify the expression and role of ciliary genes and proteins in ECs in the larger arteries of the adult human lung.

3.3. Mesenchymal cilia: What is their function?

The ubiquity of PC in the mesenchyme of the developing lung that we found is remarkable and highlights yet again the necessity of in vivo studies. In vitro, PC are typically only found on cultured cells that have been serum‐starved in order to enter a quiescent state. Pulmonary mesenchymal cells, by contrast, are actively growing, migrating, and dividing during active organogenesis, and are thus anything but static or quiescent. Future studies will be needed to investigate the presence of PC throughout the cell cycle and the role cilia play in cellular migration or differentiation during organogenesis.

3.4. Single cell analysis reveals lung mesenchyme heterogeneity and identity of ciliated cells

Understanding exactly what cell type builds and maintains PC in the lung is likely to be critical for a mechanistic understanding of ciliopathy phenotypes of the lung. As the current study demonstrates, it is no small feat identifying with clarity the spatiotemporal celloforigin of the cilia. At this time, there is no clear gene signature for cells with PC that could be used to track their status using single cell sequencing methods. We have chosen to examine ARL13Bas a proxy for ciliated cells. However, transcript counts of Arl13b are not necessarily true hallmarks of a ciliated cell—as this measure likely reflects cells in the process of active cilium generation. However, our immunofluorescent localization pointed to the lung stroma as a likely population of cells harboring PC. This, in parallel with our single cell analysis of the lung mesenchyme, allowed us to begin painting a picture of which cell populations construct and maintain PC over developmental time. Lack of maintenance of ciliary and peri‐ciliary factors arising from transcriptional regulation has also been recently reported in differentiated granule cells of the cerebellum that lack cilia. ⁴⁶

It is important to note that postnatal development continues in many organs—and it is critical for the lung. After birth, alveologenesis continues up to P36 in mice to create the mature, functional lung. Secondary crest myofibroblasts (SCMFs) play a critical role in alveologenesis by providing the contractile force to generate the “secondary crest” that grows into a new, mature alveolar wall, as well as providing crucial paracrine signals during septae formation. ⁴⁷ , ⁴⁸ Building on the previously reported importance of PDGFRα signaling to SCMFs, we now find that SCMFs widely and robustly exhibit PC. We are currently investigating the role of various ciliary signaling pathways in this population. Additionally, the fact that a remarkably active, contractile cell population like the SCMFs all displays PC suggests a role in directed cellular migration and cell–cell signaling. We might soon come to recognize the PC not just as an antenna, but also a rudder, steering the cellular ship and keeping it on course.

4. EXPERIMENTAL PROCEDURES

4.1. Mice and embryo handling

Experiments were performed in accordance with protocols approved by the University of Texas Southwestern Medical Center (UTSW) Institutional Animal Care and Use Committee. Both male and female E10.5–E18.5 CD1 embryos and postnatal tissues were collected and dissected in phosphate‐buffered saline (PBS). Animals were euthanized with CO₂ followed by cervical dislocation, after which the removal of lung tissue was performed directly. The Pdgfrα‐H2B‐eGFP mice were obtained from The Jackson Laboratory (Stock #007669).

4.2. Dissection, sectioning, and immunofluorescent analysis of lung sections

E12.5–E18.5 embryos were dissected and fixed in 4% paraformaldehyde (PFA) in PBS overnight at 4°C. Lungs of postnatal stages P0 and P7 were removed and placed directly into 4% PFA and fixed overnight. Lungs of postnatal stages P14 to P60 were inflated with intratracheal instillation of 4% PFA to 20 cm H₂0 pressure. The trachea was tied off and the whole lung was placed in 4% PFA for overnight fixation, after which individual lung lobes were dissected for further processing. Following dissection, lungs were processed for either cryosectioning or paraffin sectioning. In all immunofluorescent assays, coronal (frontal) sections of lungs were analyzed. The location of regions assessed is shown in all figures as boxed areas, showing the region along the proximodistal axis of the lung, and whether the region is more central or peripheral. Analysis of epithelium, mesenchyme, or endothelial cells is noted.

Following three washes in PBS, immunofluorescent stains were then performed as previously described. ⁴⁹ In brief, tissues were dehydrated in an 25%–50%–75%–100% ethanol series, cleared in xylene, and embedded in Paraplast (Leica) as previously described. ¹⁶ Embedded tissues were sectioned at 10 μm thickness on a HistoCore MULTICUT semi‐automated rotary microtome (Leica) onto SuperfrostPlus glass slides (Fisher). Sections were permeabilized for 10 min in 0.2% Triton X‐100 in PBS, then subjected to antigen retrieval with R‐buffer B in a 2100 Retriever (Electron Microscopy Sciences). Slides were blocked with CAS‐block (Invitrogen), and primary antibody incubations were carried out at 4°C overnight (for antibody information and dilutions, see Table S1).

Alternatively, for imaging of thick lung sections, the fixed embryos or lung tissue were fixed as described above, washed three times in PBS, and then embedded in 4% agarose. These blocks were sectioned at 150 or 200 μm on a Leica VT1200 S vibratome. Vibratome sections were permeabilized with 0.3% Triton X‐100 for 15 min, washed with PBS, blocked with CAS‐block, and stained with primary antibodies overnight. The tissue was mounted on SuperfrostPlus glass slides with 0.25 mm deep iSpacer® double‐sided sticky wells.

For all immunofluorescent assays, coronal (frontal) sections of lungs were analyzed. The location of regions assessed is shown in all figures as boxed areas, showing the region along the proximodistal axis of the lung and whether the region is more central or peripheral. Unless otherwise specified, more distal regions of the lung parenchyma were chosen (see Figure 2G schematics) at different stages of development (Figure 4). Analysis of epithelium, mesenchyme, or endothelial cells is noted. Endothelial cells of the larger arteries, arterioles, and veins were included when present, but they represent a small fraction of the total number of endothelial cells analyzed.

Images for all types of sections were obtained with either an A1R Nikon confocal microscope or a Nikon CSU‐W1 dual‐camera inverted spinning‐disc confocal microscope.

Lung sections from Pdgfrα‐H2B‐eGFP mice were prepared for whole‐mount staining as previously described. ¹⁷ Briefly, the thick lung sections were permeabilized with PBS + 1% Triton X‐100 and blocked with PBS + 0.3% Triton X‐100 + 1% bovine serum albumin. Tissue sections were stained with primary antibodies for 48 h, followed by overnight secondary antibody incubation. The samples were further cleared with Scale A2 for 1 week prior to imaging on a Leica Sp8 confocal microscope. Images were processed using ImageJ (FIJI) and the presented images are z‐stack projections of 40 μm containing optical slices of 1 μm.

4.3. Electron microscopy

Whole E12.5 embryos were fixed overnight in 4% PFA and sectioned at 100 μm. Sections were submitted to the UTSW Electron Microscopy core for further processing with standard protocols. Slices were imaged on a JEOL 1400+ transmission electron microscope.

4.4. Cilia presence and length quantifications

Cilia were assessed using FIJI. To quantify the percentage of endothelial cells (ECs) harboring cilia, 100 ECs were assessed in each of three different fields of view (FOV). Visual sections of 5–10 mm thickness had to be deconvoluted to distinguish individual cells. Ciliary length was measured utilizing both manual measurements and custom macros that trace the length of the ARL13b channel signal in three dimensions, beginning at the edge of the pericentrin (PCNT) signal and extending to the opposite end of the ARL13b signal. The measurements provided at this stage were taken from mesenchymal and epithelial cells at intervals along the length of the lung buds, both at the sides and the leading edge. Co‐localization of ARL13b with IFT88 and acetylated alpha‐tubulin was used to demonstrate that ARL13b staining reached the distal end of the cilium and supported the accuracy of ciliary length measurements. For cell type‐specific analysis, 3D analysis of cells that were positive for GM130 and ARL13B was necessary to establish the cell from which ciliary structures originated. Careful analysis of 10 μm sections using z‐stacks of 0.2 or 0.5 μm slice depths was performed on all ECs, and all cases wherein an EC appeared to be ciliated were referred for 3D rendering and analysis (see Figure 2G schematics) at different stages of development (Figure 4). 3D reconstructions were generated using Imaris 10.0 (Oxford Instruments). Surfaces were drawn using intensity thresholding, followed by manual tracing and editing to clearly demarcate nuclear borders. All ECs (macro and microvascular) were included in this analysis. Models were generated for all ECs where the cilia's originating cell could not be determined by visual sectioning of image Z‐stacks obtained by confocal microscopy.

4.5. Data analysis and visualization

Data were both analyzed and plotted in Graphpad PRISM 10.2.3. Statistical tests (indicated in the relevant methods sections and figure legends) utilized were the parametric unpaired Student's t‐test and the 2‐Way ANOVA with Tukey's multiple comparisons test. Statistical significance is noted as follows: ns = p > 0.05; * = 0.01 < p < 0.05; ** = 0.001 < p < 0.01; *** = 0.0001 < p < 0.001; **** = p < 0.0001. Images were analyzed in FIJI. Figures and models were made using Microsoft Powerpoint, data were partially analyzed in Microsoft Excel, and text was written in Microsoft Word.

4.6. scRNAseq data analysis

Raw sequencing reads from previously published single‐cell RNA‐sequencing (scRNA‐seq) data from previous studies ¹⁸ , ¹⁹ , ²⁰ were downloaded from the NIH GEO accessions GSE180822, GSE149563, and GSE165063, respectively (Table S2). Reads were mapped to the GRCm10 mouse reference genome using STAR‐solo. ²¹ Gene‐by‐cell counts matrices were subsequently ambient‐RNA corrected, integrated, filtered, and projected using tools in the scverse toolchain ²² as described in Reference 17. Cell type annotations were made manually by examining the expression of known marker genes for each respective cell type within groups of cells within single/highly similar clusters.

AUTHOR CONTRIBUTIONS

Conceptualization: Stephen Spurgin, Ondine Cleaver; Methodology: Stephen Spurgin, Ondine Cleaver; Formal analysis: Stephen Spurgin, Ange Michelle Nguimtsop; Investigation: Stephen Spurgin, Ange Michelle Nguimtsop, Jocelynda Salvador, Fatima N. Chaudhry, Sylvia N. Michki; Resources: Ondine Cleaver; Data curation: Stephen Spurgin, Ange Michelle Nguimtsop, Fatima N. Chaudhry, Sylvia N. Michki; Writing—original draft: Stephen Spurgin, Ondine Cleaver; Writing—review & editing: Stephen Spurgin, Ange Michelle Nguimtsop, Fatima N. Chaudhry, Sylvia N. Michki, M. Luisa Iruela‐Arispe, Jarod A. Zepp, Saikat Mukhopadhyay, Ondine Cleaver; Visualization: Stephen Spurgin, Ange Michelle Nguimtsop; Supervision: Saikat Mukhopadhyay, Ondine Cleaver; Project administration: Stephen Spurgin; Funding acquisition: Ondine Cleaver.

FUNDING INFORMATION

This research was supported in large part by a grant from the Leducq Research Foundation grant (21CVD03 to M. Luisa Iruela‐Arispe and Ondine Cleaver). In addition, the following grants were instrumental to the study: National Heart, Lung and Blood institute (R35HL140014 to M. Luisa Iruela‐Arispe, R00HL141684 to Jarod A. Zepp and HL113498 to Ondine Cleaver): ATS/Alveolar Capillary Dysplasia (ACDMPV) Research Grant (23‐24PACDA12 to Stephen Spurgin) and Pediatric Scientist Development Program (3K12HD000850‐38S1 to Stephen Spurgin): and the National Institute of General Medical Sciences (R35GM119461 to Jarod A. Zepp and 1R35GM144136 to Saikat Mukhopadhyay). Deposited in PMC for immediate release.

Supporting information

FIGURE S1. Widespread, multi‐organ presence of primary cilia during organogenesis. Immunofluorescent staining of 10 μm slices of paraffin‐fixed, whole E10.5 embryos (n = 3) were used for whole‐organ assessments of overall ciliary density in the all cells of developing lung (A), gut (B), liver (C), cardiac ventricle (D), atrium (E), and neural tube (F). Total number of cilia (ARL13B, white) relative to the total number of nuclei (DAPI) was quantified in ×3 digitally magnified images of ×20 scans (G). Scale bar is 50 μm.

FIGURE S2. Confirmation of mesenchymal primary cilia in the lung by electron microscopy. Whole E12.5 embryos were fixed in 4% PFA for TEM. (A) Interface of pseudostratified pulmonary epithelium and mesenchyme shown. (B) High magnification view of a primary cilium in the lung mesenchyme. Inset shows detail of a primary cilium (white arrow = basal body, black arrow = ciliary axoneme, sectioned). The cilium sits within a deep ciliary pocket. Scale bars sizes as indicated.

FIGURE S3. Confirmation of endothelial primary cilia in the aorta by electron microscopy. (A) Adult aortic ECs in the region of the descending aorta (P60) were not found to express PC. The lesser curvature of the aortic arch contains some ECs with primary cilia, shown at (B) low magnification and (C) high magnification. Scale bar is 1 μm.

FIGURE S4. Expression of DNAH11 specifically within motile cilia. Immunofluorescent staining of P30 mouse tissue for DNAH11, a key component of motile cilia that is not present in primary cilia, shows clearly that the mesenchymal cilia identified in this study do not have DNAH11 and are likely non‐motile. (A) Motile cluster of cilia showing co‐expression of DNAH11 and ARL13B in upper airway epithelial cells. (B) Primary cilia in alveolar epithelium (more distal) showing lack of DNAH11 expression. Note nearby multiciliated cell at top of panel in B. Insets (dotted white line) show both red and green channels in A', A", B', B", showing details of multiciliated and primary ciliary structures, as well as presence or absence of staining. Scale bar is 5 μm.

FIGURE S5. Mesothelial cells lining the periphery of the lung harbor PC. Immunofluorescent analysis of lung mesothelium at various stages of development revealed ciliated mesothelial cells. Not all mesothelial cells appear ciliated in these 10 μm slices. FOXF1 is expressed in ciliated mesothelial cells at E12.5 (A), E16.5 (B), and P60 (C). Red arrows highlight primary cilia (ARL13B, magenta).

FIGURE S6. Capillary Arl13b expression found in the developing endothelium. (A) Previously published lung development scRNAseq datasets were integrated and the endothelium was subset. (B) UMAP of endothelial cell types (C) UMAP of developmental timepoints. (D) Feature plots show expression of Erg and Arl13b in the lung endothelium across developmental timepoints.

FIGURE S7. Limited ability of gene sets to identify ciliated cells in single cell RNA sequencing. (A) By sequentially identifying cells with any expression of Arl13b, Ift88, and Tugb1, we assessed whether single cell data could be used for subsequent pathway analysis of ciliated cells. (B) Cells identified as ciliated (*) by overall pathway analysis (Negretti 2021) only showed active transcription of Arl13b in 50% of cells, while only 41% of cells were double positive for Arl13b and Ift88, and 2% were positive for all three (triple positive cells, circled and indicated with asterisk). (C) Given clear co‐localization in cilia in our immunofluorescent analysis, scRNAseq data was not used for subsequent analysis of known gene ontology and pathway analysis.

FIGURE S8. Single cell mRNA sequencing is useful to highlight active ciliogenesis but not static ciliary presence. (A) Previously published lung development scRNA seq datasets were integrated. (B) UMAP of developmental time points (C) UMAP of major cell groups. (D) UMAP of detailed cellular subgroups. (E) Feature plots show expression of Tubg1, Ift88, and Arl13b in the lung mesenchyme across developmental timepoints. Arrows highlight Arl13b expression in ciliated epithelial cells (1), secondary crest myofibroblasts (2), and alveolar fibroblasts (3).

TABLE S1. Antibody information.

TABLE S2. RNAseq data availability.

Spurgin S, Nguimtsop AM, Chaudhry FN, et al. Dynamics of primary cilia in endothelial and mesenchymal cells throughout mouse lung development. Developmental Dynamics. 2025;254(11):1187‐1202. doi: 10.1002/dvdy.70008