Stromal heterogeneity in the adult lung delineated by single-cell genomics

Abstract

Stromal cells in the lung provide structural support to other cells and play critical roles in inflammation, repair, and fibrosis after injury. Recent technological advancements in single-cell genomics have tremendously improved our knowledge of stromal heterogeneity in the lung. Stromal heterogeneity in single-cell RNA sequencing data is often conserved across different studies despite the different annotation strategies. Spatial analyses suggest that each stromal subset is characterized by unique anatomic locations in the adult lung. This review overviews the stromal heterogeneity delineated by single-cell RNA sequencing studies and highlights the functional characteristics and locations of each population.

Graphical abstract

INTRODUCTION

Stromal cells are defined as structural cells in the connective tissue providing scaffolds to other types of cells by producing extracellular matrix (ECM). Typical stromal cells include fibroblasts, pericytes, and smooth muscle cells. In addition to their functions in normal tissue, stromal cells play essential roles in healing wounded areas or causing fibrosis by pathological deposition of ECM in chronic diseases. Growing evidence suggests that stromal cells interact with epithelial cells, immune cells, and other cell types to maintain normal tissue homeostasis and drive diseases (1–5). In this review, we focus on stromal cells in the adult lung.

Understanding cellular heterogeneity is a key to identifying specific biological processes and therapeutic targets as illustrated by the history of immunology classifying immune cell subsets in numerous contexts. However, the heterogeneity of stromal populations has always been an understudied area compared to that of immune or epithelial cells. The previous lack of knowledge in stromal heterogeneity might be partly due to the fact that researchers historically isolated stromal cells from tissues as adhesive cells in culture, by which stromal cells lose their in vivo characteristics (6, 7). Unlike immune cells that can be isolated without harsh dissociation, stromal cell studies are hampered by a lack of standardized tissue dissociation methods, which are limited by a trade-off between the complete digestion of ECM and the degradation of surface antigens.

Recent studies utilizing single-cell RNA sequencing (scRNA-seq) have greatly extended our understanding of cellular heterogeneity. However, challenges remained in revealing stromal heterogeneity when stromal cells are underrepresented in single-cell suspension because of insufficient dissociation, since the computational identification of subclusters relies on the number of events for the populations (8). To obtain more events for stromal cells, some studies adopted strategies to enrich stromal populations by flow cytometry or magnetic selection before scRNA-seq library generation, demonstrating previously unrecognized stromal heterogeneity in the lung (9, 10).

As the use of scRNA-seq became more common, however, another challenge arose in annotating stromal subclusters without a consensual nomenclature, making it difficult to compare multiple datasets. Various nomenclature strategies have been used, including accepting historical categorizations such as myofibroblast, naming clusters based on genes expressed, or annotating by cellular location. Despite the different strategies for annotation, cluster structures of scRNA-seq data have been largely conserved across different studies. Studies utilizing immunohistochemistry, in situ hybridization, genetic reporters, and spatial transcriptomics have shown that most stromal subclusters can be characterized by distinct anatomic locations (Fig. 1).

Figure 1.

Anatomic locations of stromal subsets in the normal lung. Adventitial fibroblasts are localized at adventitial cuff spaces around airways and arteries. Peribronchial fibroblasts are just beneath airway epithelial cells and intercalated with airway smooth muscle cells. Ductal myofibroblasts are localized at alveolar ducts, which are the intermediate spaces between airways and alveoli. Alveolar fibroblasts are localized inside alveolar walls. Pericytes are localized around capillary endothelial cells in alveolar walls.

This review summarizes the current understanding of the stromal heterogeneity delineated by single-cell genomics studies. We take advantage of conserved cluster structures in multiple scRNA-seq datasets and describe markers for mouse (Table 1, Fig. 2) and human (Table 2, Fig. 3) stromal populations in the lung. We further outline evidence supporting the putative anatomic location of each population and review evidence of the functional roles specific subsets play in lung homeostasis and disease.

Table 1.

Markers of murine stromal subsets

	Inmt	Itga8	Col13a1	Scube2	Saa3	Cthrc1	Pi16	Hhip	Aspn	Fgf18	Lgr6	Myh11	Notch3	Higd1b
Alveolar fibroblast	High	High	High	High	–	–	–	Low/–	Low/–	–	–	–	Low/–	–
Inflammatory fibroblast (bleo)	Int	Int	Int	Int	High	–	–	–	Low/–	–	–	–	Low/–	–
Fibrotic fibroblast (bleo)	Low	Int	–	–	–	High	–	–	Int	Low	–	Low	Low	–
Adventitial fibroblast	Int	Int/Low	Low/–	–	–	–	High	Low/–	Int	–	–	–	Low/–	–
Ductal myofibroblast	Low	Low	Low/–	–	–	–	–	High	High	High	High	Low	Low/–	–
Airway smooth muscle cell	Low	Int	Low/–	–	–	–	–	High	High	Int	High	High	Low/–	–
Vascular smooth muscle cell	Low	High	–	–	–	–	–	–	Int	–	Low/–	High	High	Low/–
Pericyte	Low	Low	Low/–	–	–	–	–	–	Low	–	–	Int	High	High

bleo, Bleomycin.

Figure 2.

Markers of murine stromal cell subsets. A: mouse stromal cells from lungs on days 0, 7, 14, and 21 after bleomycin treatment (from Ref. 4, GSE210341) were reannotated. ASM, airway smooth muscle; VSM, vascular smooth muscle. B: gene expressions of markers for stromal subsets (see also Table 1) visualized in uniform manifold approximation and projection (UMAP) space.

Table 2.

Markers of human stromal subsets

	INMT	ITGA8	COL13A1	SFRP2	CCL2	CTHRC1	PI16	MFAP5	LGR5	ASPN	WIF1	FGF18	LGR6	MYH11	NOTCH3	HIGD1B
Alveolar fibroblast	High	High	High	–	Low/–	–	–	–	–	–	–	–	–	–	–	–
Inflammatory fibroblast (IPF)	Int	Int	Low	High	High	Low/–	–	–	–	Int	–	Low/–	–	–	–	–
Fibrotic fibroblast (IPF)	Low	Int	Low	Int/Low	Int/Low	High	–	–	–	High	–	Low/–	–	Low	Low	–
Adventitial fibroblast	Low	Low/–	–	High/Int	High/Int	–	High	High	–	Int	–	–	–	–	–	–
Peribronchial fibroblast	Int	Int	Int	–	High	–	–	–	High	Int	–	–	–	–
Ductal myofibroblast	Int	Low	Low	–	Int/–	–	–	–	–	High	High	High	–	Low	–	–
Airway smooth muscle cell	Int	Int	–	–	Low/–	–	–	–	–	Int	–	High	High	High	–	–
Vascular smooth muscle cell	Int	Int	–	–	Low/–	–	–	–	–	Low	–	–	Int	High	High	Low/–
Pericyte	Int	Low	–	–	–	Low/–	–	–	–	Int	–	–	–	Int	High	High

IPF, idiopathic pulmonary fibrosis.

Figure 3.

Markers of human stromal cell subsets. A: human stromal cells from Ref. 11 (GSE132771) were subsetted to include only normal and idiopathic pulmonary fibrosis (IPF) lungs, and the clusters were reannotated. B: gene expressions of markers for stromal subsets (see also Table 2) visualized in uniform manifold approximation and projection (UMAP) space.

MYOFIBROBLASTS

Myofibroblasts were originally described as activated fibroblasts expressing contractile genes in granulation tissues after skin wounds (12). Since their discovery, myofibroblasts have been widely accepted as the major population of fibroblasts responsible for both ECM production and tissue contraction in normal wound healing and pathological tissue fibrosis. Decades of studies focused on myofibroblasts have made substantial contributions to the current understanding of wound healing and tissue fibrosis. Although many scRNA-seq studies confirm that contractile genes are upregulated in profibrotic fibroblast populations compared to their precursors, increasing evidence suggests that contractile features are not always associated with profibrotic features. Extensive scRNA-seq data show that in the lung the fibroblast subset expressing the highest levels of ECM mRNAs is not the only population with increased expression of contractile proteins (13–15). Acta2, which is often used as a marker of myofibroblasts, is transcriptionally expressed in almost all stromal populations at the steady state, with the highest expression in smooth muscle cells, although the formation of stress fibers with ACTA2 protein might not be as ubiquitous as transcripts (16, 17). In addition, postnatally developing lungs see the emergence of myofibroblast populations, which express high levels of contractile genes and are shown to contribute to secondary septum formation but are not characterized by profibrotic functions (18). We therefore prefer not to continue to use the general term “myofibroblast” and restrict this term to a subpopulation of fibroblasts that express high levels of ACTA2 and are present around alveolar ducts in both healthy and injured lungs.

ALVEOLAR FIBROBLASTS

Alveoli are the parenchyma of the lung where the gas exchange happens. The alveolar surface is mostly covered with thin and flat type 1 epithelial cells (AT1), whereas cuboidal alveolar type 2 epithelial cells (AT2) produce surfactant and are stem cells in alveoli (1, 19). Alveolar fibroblasts reside inside alveolar walls. Early studies using electron microscopy showed that alveolar fibroblasts make direct contact with AT2 cells (20, 21). Wnt secretion from alveolar fibroblasts maintains the AT2 stem cell function (22). Whereas alveolar fibroblast markers were also reported earlier in bulk RNA-seq of a population purified based on the brightness of Pdgfra-GFP (23), a cluster identified by scRNA-seq was attributed to fibroblasts inside alveolar walls by using in situ hybridization of Npnt and Ces1d in murine lungs (11). Our group generated a CreER mouse strain for Scube2, a specific marker of murine alveolar fibroblasts, to further show that this population provides niches for AT2 cells in the normal lung and differentiates into injury-induced fibroblasts after fibrotic alveolar injury (4, 24). Some studies suggest that there is further heterogeneity among alveolar fibroblasts, mostly characterized by quantitative differences in some genes such as Bmper and Brinp1 (25, 26), although locational and functional distinctions of these alveolar fibroblast subsets are still unclear. Whereas SCUBE2 is not highly expressed in alveolar fibroblasts in the adult human lung, INMT, COL13A1, ITGA8, CES1 (Ces1d in mice), WNT2, and FGFR4 can be used as common markers of alveolar fibroblasts in humans and mice (9, 11, 27–29). Fibroblasts in alveolar walls have sometimes been described as lipofibroblasts since they store lipid droplets (1, 30, 31). The literature suggests the mobilization of triglyceride from lipofibroblasts to AT2 cells (32). Plin2, also known as adipose differentiation-related protein (ADRP), is a marker of lipid droplet formation and is highly expressed in murine alveolar fibroblasts (11). Although human alveolar fibroblasts do express PLIN2, adventitial fibroblasts and subpleural fibroblasts have higher PLIN2 expression in humans (11, 33). Of note, Plin2 is broadly expressed in many cell types including macrophages and AT2 cells in the lung, making it difficult to use Plin2 as a specific feature of any fibroblasts (11, 34).

ADVENTITIAL FIBROBLASTS

Pulmonary arteries run alongside the airways in the structures called bronchovascular bundles. Adventitial cuff spaces are interstitial spaces around vessels and airways inside bronchovascular bundles in the lung (35). It has been known that fibroblasts and immune cells reside in adventitial cuff spaces (36). A cluster in scRNA-seq of lung fibroblasts expresses Il33 and Pi16 as their specific markers, and Dahlgren et al. (37) showed that Il33+ Pdgfra+ fibroblasts predominantly reside in the adventitial cuff spaces, providing niches for immune cells. In situ hybridization or protein staining of Pi16 further confirmed Pi16+ fibroblasts as adventitial fibroblasts (4, 11, 27, 38). In the mouse, Col14a1 is most highly expressed in adventitial fibroblasts (9). PI16+ fibroblasts can be universally found in many organs (39). MFAP5 is a specific marker for human adventitial fibroblasts, and its presence in human adventitial cuff spaces has been shown by high-throughput in situ hybridization and spatial transcriptomics (15, 40). In addition to providing niches to immune cells, how adventitial fibroblasts interact with surrounding cells in homeostasis and diseases is an active area of research. It is also of interest to know whether there is cross-organ heterogeneity in adventitial fibroblasts or whether there is heterogeneity inside lungs (e.g., adventitial fibroblasts around arteries vs. veins).

PERIBRONCHIAL FIBROBLASTS

Fibroblasts just beneath airway epithelial cells and intercalated with airway smooth muscle cells have been reported by the histological assessment of Lgr5-CreER, Axin2-CreER, and Col1a1-GFP mice (10, 11, 41), although Lgr5 is also expressed in ductal myofibroblasts in mice. In the murine scRNA-seq data, there are two Pi16+ fibroblast populations: one expresses Il33, and the other expresses Adh7 (11). Since Adh7+ fibroblasts also express Lgr5 and low levels of Hhip, markers of fibroblasts intercalated with airway smooth muscle cells (11, 41), whether Adh7+ fibroblasts in the mouse are peribronchial fibroblasts needs further investigation. With the lack of a specific mouse tool to target this population, the functions of peribronchial fibroblasts remain to be elucidated. In humans, LGR5+ PDGFRA+ fibroblasts are observed just beneath airway epithelial cells by in situ hybridization (29), although LGR5+ fibroblasts in humans do not express PI16. Madissoon et al. (40) confirmed peribronchial fibroblasts as a distinct population expressing LGR5, F13A1, and TMEM132C in humans by Visium spatial transcriptomics.

DUCTAL MYOFIBROBLASTS

Myofibroblasts emerge in the developing lung to form secondary septa. Fgf18-CreER marks these myofibroblasts (42). Although Pdgfra^high myofibroblasts (or secondary crest myofibroblasts) in alveoli undergo apoptosis once secondary septation is complete, some Fgf18-CreER-labeled myofibroblasts persist in alveolar ducts, which are the intermediate spaces between bronchioles and alveoli (42). Narvaez Del Pilar et al. (38) showed that a cluster (which is also marked by Fgf18) of myofibroblasts is principally located at alveolar ducts by using Cdh4-CreER, Crh-Cre, Lgr6-CreER, and Hhip protein staining, illuminating that ductal myofibroblasts wrap alveolar ducts in a similar fashion to airway smooth muscle cells wrapping larger conducting airways. Although it is established that the temporal emergence of alveolar myofibroblasts in the developing lung is necessary for secondary septum formation, the homeostatic roles of ductal myofibroblasts persisting in the adult lung are unknown. Ductal myofibroblasts share some markers with airway smooth muscle cells, such as Lgr6, but express lower levels of contractile genes (38, 43, 44). Our group initially annotated this cluster in our scRNA-seq data as “peribronchial fibroblasts” based on the enrichment of Hhip transcripts in the peribronchial areas (11, 45). However, the peribronchial Hhip transcripts could also be marking airway smooth muscle cells. It is reported that some peribronchial fibroblasts intercalated with airway smooth muscle cells are Pdgfra+ (10, 29, 37), whereas the Fgf18+ Aspn+ ductal myofibroblasts are characterized by low or negative Pdgfra expression. In humans, the presence of WIF1+ ASPN+ FGF18+ WNT5A+ myofibroblasts in alveolar ducts was confirmed with in situ hybridization (15, 27). Of note, in humans HHIP is expressed in AT2 cells but not in ductal myofibroblasts (27).

AIRWAY SMOOTH MUSCLE CELLS

Airway smooth muscle cells are observed as cells highly positive for ACTA2 staining around airways and known to modulate structural and functional responses of airways by their contractility and through interactions with surrounding cells (46). In scRNA-seq, smooth muscle populations, including vascular smooth muscle cells, can be easily distinguished by high expressions of contractile genes such as ACTA2, MYH11, ACTG2, and ACTC1. Key observations that allowed us to distinguish airway smooth muscle cells from vascular smooth muscle cells in scRNA-seq came from studies showing that Lgr6-CreER and Crh-Cre mark airway smooth muscle but not vascular smooth muscle cells (38, 41). scRNA-seq data indicate that some vascular smooth muscle cells also express lower levels of Lgr6 (44), but the expression might not be high enough to drive Cre recombination in Lgr6-CreER mice. In uniform manifold approximation and projection (UMAP) plots of murine scRNA-seq data, airway smooth muscle cells usually show up as a continuous population of ductal myofibroblasts (43, 44). Studies using Lgr6-CreER showed that Lgr6+ airway smooth muscle cells promote the regeneration of airway epithelial cells after injury by secreting FGF10 (41, 47).

VASCULAR SMOOTH MUSCLE CELLS

Vascular smooth muscle cells express high levels of contractile genes such as ACTA2 and MYH11 like airway smooth muscle cells. They support the vessel wall structure and modulate vascular tone to maintain intravascular pressure (48). Some UMAP plots of scRNA-seq data show vascular smooth muscle cells as a continuous population with pericytes, which is corroborated by some shared markers with pericytes including NOTCH3 and MCAM (43, 44). NOTCH3 is expressed in vascular smooth muscle cells and maintains their phenotypic stability (49). Notch3-CreER marks at least a subset of vascular smooth muscle cells in the lung (50). Another marker shown to be expressed in vascular but not airway smooth muscle is Ebf1 (51). Muhl et al. (52) showed that Cspg4-DsRed marks arterial smooth muscle cells but not venous smooth muscle cells. The transcriptional differences between arterial and venous smooth muscle cells seem to be mostly quantitative, but arterial smooth muscle cells have higher Actc1 expression, and venous smooth muscle cells have higher Olfml2b expression in murine scRNA-seq data (52).

PERICYTES

Capillaries traverse alveolar walls and are critical for gas exchange. Pericytes, which are characterized as Pdgfrb^high cells, have been suggested to be closely localized with capillary endothelial cells. Anatomically, pericytes in alveolar walls directly contact endothelial cells and are localized inside endothelial basement membranes, whereas fibroblasts are on the opposite side of endothelial basement membranes (20). It is suggested that pericytes regulate vascular homeostasis and inflammation based on studies with Cspg4 (NG2)-CreER, FoxD1-Cre, or Pdgfrb-CreER, although vascular smooth muscle cells also express these markers (53–56). Klouda et al. (57) recently identified Higd1b as a specific marker of pericytes on human and mouse scRNA-seq data and subsequently generated Higd1b-CreER to specifically label pericytes around capillaries. In fibrotic lung injury, Higd1b-CreER-labeled pericytes upregulate Acta2 but rarely become Cthrc1+ fibrotic fibroblasts (24). Other pericyte markers include COX4I2 and NDUFA4L2 (4).

INJURY-INDUCED INFLAMMATORY FIBROBLAST POPULATIONS

The lung is an epithelial organ and is continuously exposed to external stimuli that cause inflammation and tissue damage. Recent studies showed that fibroblasts in the lung respond to damage signals and become injury-induced fibroblasts to modulate inflammatory, fibrotic, and regenerative responses. In fibrotic injury, alveolar fibroblasts sequentially differentiate into inflammatory and fibrotic fibroblasts (Fig. 4) (4). However, whether the inflammatory state is a necessary step to becoming fibrotic fibroblasts is an open question. In the mouse, inflammatory fibroblasts are marked by Saa3 and Lcn2 expression, which are induced by IL-1β in vitro (4).

Figure 4.

Lineage development of alveolar fibroblasts after fibrotic injury. After fibrotic alveolar injury, inflammatory cytokines such as IL-1β are produced by macrophages, which induce inflammatory fibroblasts from alveolar fibroblasts. Inflammatory fibroblasts produce inflammatory chemokines to recruit immune cells such as monocytes. In the fibrotic phase, fibrotic fibroblasts induced by transforming growth factor (TGF)-β proliferate, migrate into the alveolar lumen, and produce extracellular matrix (ECM) proteins to cause intra-alveolar fibrosis and alveolar collapse.

Inflammatory fibroblasts produce inflammatory chemokines such as Ccl2 or Cxcl12 (4). In human fibrosis, inflammatory fibroblasts express high levels of SFRP2 and PTGDS (4, 15). Human inflammatory fibroblast markers in interstitial lung diseases have significant overlap with those of adventitial fibroblasts, although ITGA8 expression can distinguish inflammatory fibroblasts (4). We also reported another type of inflammatory fibroblasts characterized by interferon-induced genes in humans (4). Sfrp1+ fibroblasts reported as transitional fibroblasts after bleomycin injury in mice partially overlap with Saa3+ inflammatory fibroblasts, but Sfrp1 is expressed in a broader population of fibroblasts (4, 58). Inflammatory fibroblasts also emerge in other lung injury models such as the surfactant protein-C mutation model, Hermansky–Pudlak syndrome, butylated hydroxytoluene (BHT)-induced injury, genetic AT1 cell ablation, and AT2-specific Mdm2 knockout (59–61). The influenza infection model induces damage-responsive or interferon-responsive fibroblasts, although developmental and functional differences of these populations in flu should be further explored (3). Another recent study showed that blocking Notch signaling in Pdgfra+ fibroblasts suppresses dysplastic Krt5+ epithelium formation after flu infection, demonstrating that injury-induced fibroblasts control regenerative responses after epithelial damage (62).

FIBROTIC FIBROBLASTS

Accumulation of myofibroblasts expressing high levels of collagens has long been described in pulmonary fibrosis, although as discussed above ACTA2 alone is insufficient to specifically identify profibrotic fibroblasts (11). We revealed that this profibrotic fibroblast population can be identified by the high expression of CTHRC1 in scRNA-seq data, and CTHRC1+ fibroblasts are localized in fibroblastic foci in idiopathic pulmonary fibrosis (IPF) (11). As CTHRC1 is one of the genes strongly induced by transforming growth factor (TGF)-β, CTHRC1+ fibrotic fibroblasts are also found in cancer, scleroderma-associated fibrosis, COVID-associated pneumonia, and influenza infection (11, 63–66). Blocking TGF-β signaling in alveolar fibroblasts after bleomycin injury suppressed the induction of Cthrc1+ fibroblasts, resulting in the almost complete blockade of fibrosis (4). Those Cthrc1+ fibroblasts are observed as cells that migrate across the basal lamina into the alveolar lumen to cause intra-alveolar fibrosis(Fig. 4) (4). Blocking the induction of fibrotic fibroblasts also causes an increase in the expression of markers of inflammatory fibroblasts and exacerbates inflammation, suggesting that TGF-β signaling in fibroblasts is a critical switch that shuts down lung inflammation and promotes fibrosis (4). Runx1 and Runx2 are upregulated in fibrotic fibroblasts and are also suggested as key regulators of fibrotic fibroblasts (24, 61), although other populations, such as leukocytes and epithelial cells, also express Runx1 and Runx2. Other markers highly expressed in profibrotic fibroblasts include COL1A1 and POSTN. Although Spp1 is also highly upregulated in bleomycin-induced fibrosis in mice, Spp1^high fibroblasts do not overlap with Cthrc1+ fibroblasts in scRNA-seq data, and their functional role is yet to be determined (4, 58).

PLASTICITY

One open question in the field is how plastic these stromal subsets are in the normal lung. What if one population of fibroblasts is lost? Do they self-recover within the population? Or can some other population transdifferentiate to replenish the lost population? Although there are some indications that universal fibroblasts can become tissue-specialized fibroblasts (39), current evidence for transdifferentiation is mostly from pseudotime analyses of scRNA-seq data, and this should be tested by specific lineage-tracing tools. On the other hand, the literature indicates that multiple stromal subsets can adopt injury-induced states. We showed that fibroblast subsets, including alveolar, adventitial, and ductal fibroblasts, can upregulate profibrotic markers when they are intratracheally transferred into injured lungs (11). scRNA-seq analyses also showed that profibrotic genes can be upregulated not only in alveolar but also adventitial or ductal fibroblasts after bleomycin injury (67), suggesting that each fibroblast subset is responsive to injury-induced signals such as TGF-β. Therefore, fibroblast subsets responsible for inflammation and fibrosis are likely determined by locations and types of injury. Another open area of research is the fate of injury-induced fibroblasts. Profibrotic fibroblasts are classically believed to undergo apoptosis in the resolution phase of wound healing (68). Studies using Acta2-CreER indicate that Acta2-CreER-labeled fibrotic fibroblasts can revert to the normal fibroblast state (31, 69). Since Acta2 is not specific to fibrotic fibroblasts, the fate of injury-induced fibroblast subsets should be further tested with more specific tools.

CHALLENGES IN HUMAN STROMAL CELL IDENTIFICATION

Human scRNA-seq experiments are influenced by many more variables than murine experiments. In addition to the diverse genetic backgrounds of humans, protocols for sample collection, sample storage, tissue dissociation, and enrichment strategies are different in each study. Since the size of human lungs is large, the difference in sampling positions, such as proximal or distal parts, may also bias cellular composition. Moreover, one peculiar donor sometimes can drive an entire cluster in scRNA-seq data (70). Because of these difficulties, further studies are needed to characterize some human stromal populations. For example, fibromyocytes described in Travaglini et al. (27) from in situ hybridization of ACTG2 and ASPN share the nuclear shape and some marker genes with airway smooth muscle cells, and their distinction would benefit from further exploration. PLIN2+ subpleural fibroblasts were identified in the lungs of idiopathic pulmonary fibrosis (IPF) patients (15, 33). It will also be of interest to characterize this population in the normal lung. Label transfer tools, such as Human Lung Cell Atlas (HLCA) and LungMAP single-cell reference (CellRef) (71, 72), currently have disagreements regarding stromal cell annotation, and such tools will benefit from a standardized nomenclature based on cellular locations of subsets as illustrated in this review. It should also be noted that currently available scRNA-seq data come from a relatively small group of murine models and a relatively small number of human fibrotic lung diseases. It seems highly likely that additional interesting and functionally important subsets of fibroblasts will be seen in other models and human fibrosing lung diseases.

CONCLUSIONS

Recent progress in single-cell genomics technology has provided a plethora of new information about the heterogeneity of stromal cells in the normal and diseased lung. Emerging spatial studies and functional studies defining the unique anatomic locations and functional roles of most stromal subsets provide a framework for the relatively simple (but obviously provisional) classification based on anatomic location and (when known) function outlined in this review. We expect that the emergence of new datasets in multiple human lung diseases and additional models over the next few years will provide further clarification of the roles and locations of the subsets described here and identify additional functionally important fibroblast states. We hope that the simple approach to naming stromal subsets outlined in this review can serve as a scaffold upon which emerging insights can be added.

DATA AVAILABILITY

scRNA-seq analyses were performed with publicly available data from accessions GSE132771 (human) and GSE210341(mouse). The reannotated Seurat objects and codes used are available from Zenodo (doi: 10.5281/zenodo.15330956).