Glucocorticoids Retain Bipotent Fibroblast Progenitors during Alveolar Septation in Mice

Abstract

Glucocorticoids have been widely used and exert pleiotropic effects on alveolar structure and function, but do not improve the long-term clinical outcomes for patients with bronchopulmonary dysplasia, emphysema, or interstitial lung diseases. Treatments that foster alveolar regeneration could substantially improve the long-term outcomes for such patients. One approach to alveolar regeneration is to stimulate and guide intrinsic alveolar progenitors along developmental pathways used during secondary septation. Other investigators and we have identified platelet-derived growth factor receptor-α–expressing fibroblast subpopulations that are alternatively skewed toward myofibroblast or lipofibroblast phenotypes. In this study, we administered either the glucocorticoid receptor agonist dexamethasone (Dex) or the antagonist mifepristone to mice during the first postnatal week and evaluated their effects on cellular proliferation and adoption of α-smooth muscle actin and lipid droplets (markers of the myofibroblast and lipofibroblast phenotypes, respectively). We observed that Dex increased the relative abundance of fibroblasts with progenitor characteristics, i.e., containing both α-smooth muscle actin and lipid droplets, uncoupling protein-1 (a marker of brown and beige adipocytes), delta-like ligand-1, and stem cell antigen-1. Dex enhanced signaling through the Smad1/5 pathway, which increased uncoupling protein-1 in a lung fibroblast progenitor cell line. We conclude that glucocorticoid receptor manipulation can sustain fibroblast plasticity, and posit that targeting downstream glucocorticoid responsive pathways could steer fibroblast progenitors along more desirable regenerative pathways.

Clinical Relevance

Current treatments for bronchopulmonary dysplasia, emphysema, and interstitial fibrosis do not restore damaged alveoli. This study shows how glucocorticoids regulate fibroblast progenitors, which contribute to alveolar septal formation during postnatal development. The findings identify glucocorticoid-responsive pathways that could be used to stimulate alveolar regeneration.

The unique contributions of individual pulmonary alveolar fibroblast subpopulations to air-space development and regeneration remain incompletely understood. Studies of hyperoxia-induced bronchopulmonary dysplasia in neonatal rats (1) or fibrotic injury in adult mice (2) have shown that structural abnormalities are ameliorated by steering fibroblasts toward a lipid-storage phenotype. Conversely, adoption of myofibroblast (MF) characteristics reduces the lipid-storage subpopulation (referred to as lipofibroblasts [LIFs]) (3). Recent studies using lineage tracing indicate that the fate of LIF progenitors is defined later than that of MFs, which arise from a prenatally defined population of sonic-hedgehog–responsive, Gli1-expressing progenitors (3). MF differentiation is dependent on transforming growth factor-β (TGF-β) signaling through TGF-β receptor 1 (Alk5), and disruption of Alk5 results in expansion of the LIF subpopulation (4). Manipulating the differentiation of one subpopulation influences that of the other, but it remains unclear whether this only occurs before birth or LIFs and MFs maintain postnatal plasticity. This is an important consideration when developing regenerative strategies for the treatment of destructive alveolar diseases such as bronchopulmonary dysplasia, emphysema, and interstitial fibrosis (5).

Mesenchymal progenitor cells with the potential to differentiate into adipocytes or another specialized cell (myocyte, oligodendrocyte, or adipocyte) express platelet-derived growth factor receptor-α (PDGFRα), which influences progenitor proliferation and differentiation (6–8). Furthermore, PDGFRα-expressing adipocyte precursors can assume characteristics of brown adipocytes and participate in the development of brown fat within deposits of white fat (beige fat) induced by cold exposure or β3-adrenergic stimulation (9). Using mice of different perinatal ages, investigators observed that PDGFRα is expressed in both LIFs (which stain for adipocyte-related differentiation protein [ADRP, also known as perilipin 2] and contain lipid droplets) and MFs (a larger proportion contain α-smooth muscle actin [α-SMA], and few contain ADRP) (10–12). Studies using flow cytometry to classify fibroblasts isolated from mice bearing green fluorescent protein (GFP) driven by the endogenous PDGFRα promoter (PDGFRα-GFP) identified fibroblasts with two levels of GFP intensity, which during alveolar development correlated with the abundance of PDGFRα molecules on the cell surface (13). During Postnatal Day 4 (P4) through P14, PDGFRα-GFP^low cells predominantly reside at the base of elongating secondary septa and store neutral lipids (14). PDGFR-GFP^high fibroblasts congregate at the alveolar entry ring (secondary septal tip) and exhibit more α-SMA. Flow cytometry has also been used to sort lung fibroblasts from adult mice based on their intensity of GFP fluorescence. In adult mice, PDGFRα-GFP^high cells demonstrated a gene expression profile for abundant production of extracellular structural proteins, whereas GFP^low lung fibroblasts showed higher expression of cell surface proteins involved in migration and cytoskeletal rearrangement (15). It remains unclear whether these differences in ontogeny and gene expression profiles result from divergent terminal differentiation along predefined pathways, adaptation to their current environment, or both.

To address this issue, we manipulated glucocorticoid signaling in mice and examined the effects on LIF and MF phenotypes. Glucocorticoids are used clinically but have shown limited efficacy in treating bronchopulmonary dysplasia and pulmonary fibrosis. A better understanding of the effects of glucocorticoids on fibroblast subpopulations may enable us to identify signaling pathways that could be specifically targeted to circumvent some of the undesirable effects of long-term corticosteroid administration.

Both augmentation and suppression of signaling have been used to define the effects of glucocorticoids on alveolar septation. However, the findings are enigmatic because they vary with the timing of the stimuli, the cell populations that were targeted, and which cellular properties were studied. The glucocorticoid agonist dexamethasone (Dex) delays capillary maturation and produces transient alveolar wall thinning, but enhances the expression of elastin (16–18). Targeted deletion of the glucocorticoid receptor (GR) in mesenchymal cells was shown to increase the volume density (V_D) and proliferation of the airspace mesenchyme, diminish elastic fiber formation and epithelial maturation, and halt progression beyond the saccular stage, such that few mice survived after P1 (19). We hypothesized that the differential effects of glucocorticoids on the LIF and MF fibroblast subpopulations could explain why GR deletion increases alveolar mesenchymal cells but diminishes elastin, a major product of these cells.

Materials and Methods

A complete list of the materials used in this work and a more detailed explanation of the methods employed are provided in the online supplement.

Mice

Mice bearing the PDGFRα-GFP construct have been described previously (10). Production and nuclear localization of GFP are under the control of the endogenous pdgfrα promoter, and spatially and temporally recapitulate endogenous pdgfr-α expression (10). The mice were heterozygous for the pdgfrα-GFP allele and were phenotypically identical to wild-type (GFP⁻) mice. Members within 4 separate litters of mice were allocated to receive Dex or phosphate-buffered saline (in one litter, fibroblasts were isolated separately from two Dex-treated mice, yielding a total of five). In the Dex-treated mice, Dex was delivered subcutaneously at 100 ng/g body weight on P1 and at 50 ng/g on subsequent days. Mifepristone (Mfp) was dissolved in propylene glycol and 50 μg/g body weight was delivered subcutaneously to male mice from five separate litters, using other littermates as controls. To analyze proliferation, the pups were treated with 100 ng/g body weight of 5-ethynyl-2′deoxyuridine (EdU) 3 h before they were killed and lung fibroblasts were isolated (13). The protocols for animal use were approved by the Iowa City Veterans Affairs Medical Center Animal Use Committee (10).

Lung Inflation and Fixation

Lungs from mice bearing the PDGFRα-GFP insert at P8 and P12 were uniformly inflated (50 μl of fixative per gram body weight), fixed, and sectioned, and the Ki67antigen was analyzed by laser scanning confocal microscopy and enumerated using the optical fractionator stereological probe (14). Tissues used to analyze the gas-exchange surface area and alveolar wall thickness were embedded in London Resin (LR) White. They were analyzed using the Cycloids for Sv stereological probe to determine the gas-exchange surface area, and the Mertz probe to determine the arithmetic mean barrier thickness.

Isolation of Primary Mouse Fibroblasts and Flow Cytometry

Lung fibroblasts were isolated on P8 using a previously reported method involving digestion with collagenase and adherence to tissue-culture plastic for 1 h (10). Epithelial and endothelial cells comprised ∼ 2.5% and 1.6%, respectively, whereas macrophages were only detected in the PDGFRα-GFP⁻ population (10). Fibroblasts stained for ADRP, α-SMA, phosphorylated Smad1/5 (pSmad1/5), or uncoupling protein-1 (UCP1) were permeabilized before immunostaining (10). When staining for LipidTOX red (LTR) and stem cell antigen-1 (Sca1), the cells were not permeabilized. CD45⁺ cells, which had adhered to the culture plastic after 1 h, were excluded from analysis. Virtually all of the PDGFRα-expressing fibroblasts were in the CD45⁻ fraction. Forward scatter and side scatter were used to exclude small (presumably apoptotic) cells and aggregates. The background fluorescence from the corresponding immunoglobulin G isotype controls was subtracted to quantify the different fibroblast populations.

Isolation of Mitochondria for UCP1 Immunoblotting

Neonatal mouse lung fibroblasts (MLg) 2908 cells were subjected to adipogenic induction, and after a 16 h period without Dex, they were treated with 5 ng/ml TGF-β or 100 nM Dex, or remained untreated, and the mitochondria were isolated (20). After incubation with 0.5% n-octyl-B-D-glucopyranoside to solubilize UCP1, the samples were subjected to SDS-PAGE and immunoblotting. MLg 2908 cells were induced and similarly treated with TGF-β1 or Dex before immunoblotting for pSmad2, Smad2 pSmad1/5, Smad1/5, and β-tubulin.

Results

Glucocorticoid Signaling Regulates Alveolar Wall Thickness and Surface Area

Dex, a potent GR agonist, or Mfp, a GR as well as a progesterone receptor antagonist, was administered during P1–P7 or P1–P11 and the mice were killed on P8 or P12 (Figure 1A). Images of lungs that had been uniformly inflated with agarose during fixation are shown in the figure. Residual agarose accounts for the gray background in the air spaces. Casual inspection showed that Dex increased the V_D of the airspace at P8 and P12, and that the GR antagonist Mfp had the same effect at P8 (Figure 1B). Stereological analysis confirmed that the alveolar surface was smaller in both Dex- and Mfp-treated mice at P8 (Figure 1C). Mfp increased the mean barrier (alveolar wall) thickness at both ages, whereas wall thinning was only observed at P8 in Dex-treated mice (Figure 1D). Others have observed that pharmacological manipulation of the GR transiently alters the surface area, indicating that critical effects are manifest by P8, after which the GFP^low LIFs diminish (10, 16). The observation that both Dex and Mfp reduced the alveolar surface area indicates that septal outgrowth requires properly balanced glucocorticoid signaling, and that both excessive Dex and insufficient Mfp are detrimental.

Figure 1.

Perturbing glucocorticoid signaling disrupts alveolar structure. Dexamethasone (Dex) or mifepristone (Mfp) was administered, whereas controls were untreated, during the postnatal days shown in (A) (horizontal lines) and mice were killed (*) at Postnatal Day 8 (P8) or P12. (B) Lungs were uniformly inflated with low-melting-point agarose, fixed, embedded in London Resin White, sectioned, stained, and imaged (scale bars are 50 µm). Insets are lower magnification to show larger fields of view. (C) Combined alveolar surface (including both alveoli and alveolar ducts) for both lungs. Mean ± SEM, n = 4 mice per treatment. (D) Alveolar wall (arithmetic mean barrier) thickness was altered by both Mfp and Dex at P8, but only by Mfp at P12. Mean ± SEM, n = 4 mice for each treatment group, *P < 0.05 for Dex- or Mfp-treated mice compared with controls at the respective ages. EdU, ethinyl deoxyuridine; FACS, fluorescence-activated cell sorter.

Increased Proliferation Contributes to Increased Alveolar Wall Thickness in Mfp-Treated Mice

The increase in alveolar wall thickness after Mfp treatment could represent enhanced mesenchymal cell proliferation, as others have observed in GR-deleted mice (19). We examined proliferation in situ using the marker Ki67 or by pulse labeling with EdU before isolating fibroblasts at P8. Although the volume-corrected number of nonproliferating cells (GFP⁻ cells included resident cells in the alveolar wall that were not fibroblasts) was not altered, we observed an increase in Ki67⁺ GFP^low, but not GFP^high, fibroblasts after exposure to Mfp (Figures 2A and 2B). When we analyzed isolated fibroblasts using fluorescence-activated cell sorting, treatment with Mfp increased the proportion of EdU⁺ GFP^low fibroblasts at P8 (Figure 2C). Our prior studies showed that proliferation is sustained in PDGFRα-GFP fibroblasts, which bear the progenitor marker Sca1, and Dex increases the proportion of Sca1⁺ LFs within the GFP^high population (Figure 2D) (10). However, Dex reduced the size of the entire GFP^high population when expressed as a percentage of all CD45⁻ fibroblasts (Figure 3A). These observations suggest that Dex retains GFP^high fibroblasts as progenitors.

Figure 2.

Glucocorticoids differentially alter proliferation in PDGFRα-GFP–expressing subpopulations. Lung fibroblast proliferation was assessed by the abundance of Ki67 antigen or by flow cytometry at P8, using mice that were treated as shown in Fig. 1A. (A) Representative confocal images of lungs stained for Ki67 (blue) or the nuclear counterstain PoPo3 (red). PDGFRα-expressing cells exhibited GFP fluorescence (green, scale bar: 20 µm). (B) Ki67⁺ nuclei were enumerated within the GFP^high and GFP^low populations in control or Mfp-treated mice on P8. GFP⁻ includes alveolar cells other than fibroblasts, whereas only fibroblasts contained GFP. Mean ± SEM, n = 4 mice for each treatment group. *P < 0.05 for Mfp- or Dex-treated mice compared with controls. (C) EdU was administered to untreated (control) or Dex- or Mfp-treated mice on P8. Fibroblasts were isolated and subjected to flow cytometry, gating out CD45⁺ cells (macrophages or fibrocytes), so that the only GFP⁻ cells analyzed were fibroblasts. Mean ± SEM of the EdU⁺ fibroblasts as percent of CD45⁻ cells, n = 5 sets of 1 or 2 mice from 4 separate litters for each glucocorticoid treatment group. *P < 0.05 for Mfp-treated mice compared with controls. (D) Percent Sca1⁺, CD45⁻ within each population stratified by GFP intensity. Mean ± SEM, n = 4 control and Dex-treated littermates derived from 4 separate litters. *P < 0.05 for Dex-treated mice compared with controls. GFP, green fluorescent protein; PDGFRα, PDGFR, platelet-derived growth factor receptor-α.

Figure 3.

Perturbing glucocorticoid signaling alters lipid storage in the GFP^high population. (A) Flow cytometry showed that Dex, but not Mfp, altered the relative abundance of the three lung fibroblast (LF) populations defined by PDGFRα-GFP intensity: GFP⁻, GFP^low, and GFP^high. Mean ± SEM, n = 5 control and treated littermates from 4 litters. *P < 0.05, size of GFP^low or GFP^high component compared with control. *P < 0.05 for GFP^high, comparing Dex-treated mice with controls. (B) Effects of Mfp and Dex on the proportions of neutral lipid–containing fibroblasts (stained with LipidTox red [LTR]) within each subpopulation stratified by intensity of GFP-fluorescence. Mean ± SEM, n = 5 mice, for each treatment, from 4 separate litters. *P < 0.05 comparing Mfp-treated mice with controls. (C) Using the cells shown in B, the LTR⁺ CD45⁻ fibroblasts were gated into two populations based on the intensity of LTR fluorescence. The distribution of GFP^low and GFP^high cells in LTR^low and LTR^high populations is shown. Mean ± SEM, n = 5, same mice as in B. (D) Analysis similar to that shown in B, except that lipid droplets were identified by perilipin-2 (ADRP). Mean ± SEM, n = 5 mice, for each treatment, from 4 separate litters. *P < 0.05 comparing Mfp- or Dex-treated mice with littermate controls.

PDGFRα GFP^high Fibroblasts Exhibit LIF Characteristics after Treatment with Mfp or Dex

PDGFRα⁺ mesenchymal progenitors retain a less differentiated, progenitor state in other organs; for example, both preadipocytes and myoblasts can assume either a lipid-storage or muscle-like phenotype (21). Although Mfp is a GR antagonist, its functional effects are not always converse to those of Dex (22). Our readouts were sensitive to the size (only LTR) and the abundance (both LTR and ADRP) of lipid droplets. These readouts reflect not only the immediate effects on GR but also differentiation (lipid droplets were observed both in GFP^high progenitors and in mature GFP^low fibroblasts), the balance between fibroblast lipogenesis and lipolysis, and the response to hormones such as insulin (22). These factors are addressed in more detail in the Discussion.

Mfp increased lipid storage in the GFP^high population (Figures 3B and 3D). Mfp modestly increased the proportion of LTR⁺ GFP^high fibroblasts (P = 0.045, a small but statistically significant difference). Gating on LTR fluorescence intensity, Dex increased the per-cell abundance of lipids in both the GFP^low and GFP^high populations (Figure 3C). Mfp increased the proportion of GFP^low fibroblasts, whereas Dex decreased the proportion of GFP^high fibroblasts that contained α-SMA (Figure 4A). This indicates that glucocorticoid signaling influences the relative abundance of lipid droplets and α-SMA in PDGFRα-expressing fibroblasts. Whereas Mfp did not alter the proportions of ADRP and α-SMA double-positive cells compared with controls (data not shown), Dex increased the proportion of GFP^high fibroblasts that contained both ADRP and α-SMA (Figure 4B). This is consistent with GFP^high fibroblasts retaining their bipotent progenitor status and delaying commitment to differentiated MFs.

Figure 4.

Perturbing glucocorticoid signaling alters the proportions of α-smooth muscle actin (α-SMA)-containing fibroblasts. Fibroblasts were isolated from mice treated with Mfp or Dex during P1–P7, or untreated littermate controls (see Fig. 1), and subjected to flow cytometry. (A) The proportions of CD45⁻ GFP^low and GFP^high fibroblasts that contained α-SMA are shown. (B) A minority of fibroblasts contained both α-SMA and ADRP, and Dex increased the proportion of double-positive cells in the GFP^high population. Mean ± SEM, n = 5 treated mice from 4 litters each for Mfp and Dex with 5 littermate controls, which differed for Mfp and Dex exposures. *P < 0.05 comparing Dex- or Mfp-treated mice with the respective control populations stratified by GFP fluorescence intensity.

PDGFRα-Expressing Fibroblasts Exhibit Characteristics of Brown Adipocytes

Others have shown that C3H10T1/2 mesenchymal progenitor cells alternatively differentiate to smooth-muscle– and brown-adipocyte–like phenotypes (23). The observation of a similar relationship in PDGFRα⁺ alveolar fibroblasts (10) prompted us to examine their similarity to brown adipocytes using the marker UCP1. In controls without manipulation of the GR, UCP1 gene expression was higher in the GFP^low population, which also exhibited higher expression of PRDM16 (Figure 5A), an important regulator of preadipocyte differentiation toward brown adipocytes. Flow-cytometric analysis showed that Mfp did not alter the proportions of UCP1⁺ cells within the three populations (data not shown), whereas Dex decreased the proportion of UCP1⁺ fibroblasts in the GFP^low GFP^high population (Figure 5B). Both UCP1 mRNA (Figure 5A) and protein (Figure 5B) were highest in the GFP^low population of control fibroblasts. Both Mfp and Dex increased the proportion of ADRP, UCP1 double-positive cells in the GFP^high population (Figure 5C), which resulted in part from the higher prevalence of ADRP⁺ cells in the GFP^high population (Figure 3D). Dex diminished the proportion of ADRP, UCP1 double-positive cells in the GFP^low population (Figure 5C). We also observed that Mfp increased the proportion of GFP^low fibroblasts that contained both α-SMA⁺ and UCP1⁺ (Figure 5D) and increased the proportion of GFP^high fibroblasts that contained α-SMA, ADRP, and UCP1 (Figure 5E). Therefore, as with beige adipocytes (24, 25), manipulation of glucocorticoids lessened the distinction between ADRP⁺ GFP^low and α-SMA⁺ GFP^high fibroblasts, and conferred a more adipocytic phenotype to PDGFRα-expressing fibroblasts, which also contain α-SMA.

Figure 5.

GFP^high fibroblasts that acquire more lipid droplets exhibit characteristics of brown adipocytes. (A) Fibroblasts were isolated from PDGFRα-GFP mice (control, not treated) and subjected to flow-cytometric sorting, gating on CD45⁻ cells, and their GFP fluorescence intensity. Using quantitative RT-PCR, uncoupling protein-1 (UCP1, a marker of brown adipocytes) and PR domain containing 16 (Prdm16), mRNA from the GFP^low and GFP^high populations was normalized to mRNA from the CD45⁻, GFP⁻ population from the same fibroblast isolation. Mean ± SEM, n = 5 mice, all from separate litters. *P < 0.05 comparing GFP^low and GFP^high. (B) Fibroblasts were isolated from mice treated with Dex during P1–P7 and subjected to flow cytometry after staining for UCP1, ADRP, and α-SMA. Mean ± SEM, n = 6 mice for each treatment group from 4 separate litters. *P < 0.05 for GFP^high, comparing Dex-treated mice with untreated controls. (C) Dex and Mfp altered the proportions of fibroblasts that stained positively for ADRP as well as UCP1. Mean ± SEM, n = 5 mice for each treatment group from 4 separate litters. *P < 0.05 for GFP^low or GFP^high, comparing Dex- or Mfp-treated mice with controls. (D) Dex and Mfp altered the proportions of fibroblasts that stained positively for both α-SMA and UCP1. Mean ± SEM, n = 5. Control and Mfp- or Dex-treated mice are the same as those shown in C. (E) Proportions of cells that stained positively for α-SMA, ADRP, and UCP1 within the GFP⁻, GFP^low, and GFP^high populations in fibroblasts isolated from control and Dex-treated mice. Mean ± SEM, n = 5 using the same mice shown in D.

Dex Stimulates Phosphorylation of Smad1/5 in the Context of Increased UCP1

Dex augments Smad1/5 phosphorylation in NIH3T3 cells and adult human primary lung fibroblasts (26). Others have shown that that the TGF-β homolog BMP7 enhances phosphorylation of pSmad1/5, inducing C3H10T1/2 cells to acquire characteristics of brown adipocytes, including increased UCP1 (23). Therefore, we examined whether Dex augments Smad1/5 phosphorylation and induces UCP1 in the progenitor-like (bipotent differentiation to MF or adipocyte characteristics) MLg 2908 lung fibroblast cell line. Dex increased Smad1/5 phosphorylation (Figures 6A and 6B) and UCP1 protein in isolated mitochondria (Figure 6C). To determine whether Dex alters Smad1/5 signaling in vivo, we administered Dex to mice during P1–P8, and using immunostaining and flow cytometry, we observed that the proportion of pSmad1/5⁺ GFP^high fibroblasts (73.6 ± 3.9, mean ± SEM, n = 5) was significantly higher than in controls (63.8 ± 3.0, P < 0.05). Gating on the CD45⁻ fibroblasts, which were also ADRP⁺, demonstrated a more substantial effect of Dex on the proportion of GFP^high fibroblasts containing pSmad1/5 (Figure 6D). Therefore, as in preadipocytes, Smad1/5 signaling was accompanied by a shift of differentiation toward the LIF phenotype.

Figure 6.

Dex increases UCP1 in cultured MLg 2908 cells and pSMAD1/5 both in culture and in vivo. (A) MLg 2908 neonatal mouse lung fibroblasts were induced to assume a lipid-storage phenotype and then exposed to medium alone (control) or supplemented with either transforming growth factor β 1 (TGF-β1) or Dex and then harvested for Western immunoblotting. The blots were probed for either pSmad2 and Smad2 or pSmad1/5 and Smad1/5, followed by β-tubulin. C, uninduced control; IC, induced control. (B) The density of pSmads in immunoblots is expressed relative to the abundance of the corresponding Smad for each sample and normalized to the ratio for the control. Mean ± SEM, n = 4 separate experiments, *P < 0.05 comparing Smad1/5 for Dex-treated MLg with controls. (C) MLg cells were induced and stimulated as in A before mitochondrial isolation and Western immunoblotting. The density of UCP1 relative to the respective induced control (set to a density of one) is shown for three separate experiments. Mean ± SEM, *P < 0.05 comparing Dex-treated mice with controls. (D) On P8, fibroblasts were isolated from a separate cohort of controls or mice that had been treated with Dex as shown in Fig. 1A. (A) After staining for pSmad1/5 and ADRP, flow cytometry was performed, gating on CD45⁻ cells. Mean ± SEM, n = 5 mice for each treatment group from four different litters. *P < 0.05 comparing the GFP^high population from Dex-treated mice with controls. HSP60, heat shock protein 60; pSMAD, phosphorylated form of a homolog of the Drosophila family of proteins including SMA (small body size) and mothers against decapentaplegia (MAD).

Dex Retains Lung Fibroblast Progenitors

Dex increased the proportion of GFP^high LF that contained both ADRP and α-SMA (Figure 4B), and a larger proportion were Sca1⁺ (Figure 2D), suggesting that GFP^high cells retained progenitor characteristics with markers of both MF and LIF. We posited that Dex may restrain the divergence into LIF and MF phenotypes by holding more fibroblasts in a bipotent state. Therefore, we examined the expression of sterol receptor binding protein-1 (SREBP1), which indicates a transition to an adipocyte-like phenotype, and delta-like ligand-1, splice product-C (Dlk1C), a progenitor marker that is regulated by PDGFRα during mouse alveolar development (10, 27). Treatment with Dex increased Dlk1C and decreased SRBP1 mRNA relative to untreated controls, consistent with retention of progenitor characteristics (Figure 7).

Figure 7.

Dex increases the retention of fibroblasts as progenitors. Fibroblasts were isolated from control and Dex-treated PDGFRα-GFP mice and subjected to flow-cytometric sorting, gating on CD45⁻ cells and their GFP fluorescence intensity. Using quantitative RT-PCR, (A) sterol receptor binding protein-1 (SREBP1) and (B) delta-like ligand-1, splice product-C (Dlk1C) mRNA was analyzed and the cycle threshold (CT) values were normalized to the CT value for a single normalizing RNA sample, which was used in all analyses. n = 5 separate cell isolations from control and Dex-treated mice, which were obtained from different litters. Mean ± SEM, *P < 0.05 comparing Dex-treated GFP^high mice with controls.

Our findings illustrate how glucocorticoids influence the balance between LIFs and MFs, and sustain progenitors, which share characteristics of both phenotypes. Lung fibroblasts contain the beige adipocyte progenitor marker UCP1, which is augmented by Dex in the PDGFRα-GFP^high population. They also illustrate how the effects of glucocorticoids depend on the differentiation status, and that timing and context are essential for targeting their effects on alveolar development or regeneration.

Discussion

Glucocorticoids Complexly Regulate Lipid-Droplet Metabolism

We used two different markers to identify lipid-storage cells by flow cytometry. LTR accumulates inside lipid droplets, where it binds to triglycerides and cholesterol esters. ADRP (perilipin 2) is located on the outer phospholipid coating rather than in the interior of lipid droplets. Therefore, the LTR intensity depends on the droplet volume, whereas the anti-ADRP fluorescence intensity depends on the droplet surface area. Like other cells, LIFs contain multiple droplets of various sizes (28). The size of the lipid droplets is controlled by the balance between lipid esterification and lipolysis, and therefore is sensitive to agents such as Mfp and Dex, which modulate lipolysis (28). We observed that both Dex and Mfp increased the proportions of ADRP⁺ (Figure 3D) and UCP1⁺, ADRP⁺ double-positive fibroblasts (Figure 5C) in the GFP^high population. Dex increased the intensity of LTR fluorescence per cell (Figure 3C), which is consistent with an increased aggregate volume of intracellular lipid droplets. Dex also increased the proportion of LTR⁺ cells in the GFP^high population (Figure 3B). Mfp increased the proportion of ADRP⁺ fibroblasts (dependent on increased surface area and not droplet volume) in the GFP^high population (Figure 3B). Mfp decreased the lipid droplet size (with a corresponding increase in the surface/volume ratio) in adipocytes of mice receiving a high-fat diet (29, 30). If Mfp similarly affects fibroblast lipid droplets, one would expect an increased proportion of ADRP⁺, but not LTR+, fibroblasts, which is what we observed. Further complexity arises from differential effects of Dex on preadipocytes versus adipocytes. Dex increased proliferation, accelerated differentiation, and increased the UCP1 gene expression of brown preadipocytes (31). However, Dex antagonized the cyclic adenosine monophosphate (cAMP)-mediated stimulation of oxygen consumption and UCP1 expression in differentiated brown adipocytes (32), and increased lipolysis in mature white adipocytes (33). Although Dex and Mfp exert opposing effects on the GR, this does not consistently translate into obverse findings at the cellular and organ levels.

Glucocorticoids Have Time-Sensitive and Context-Dependent Effects on Alveolar Septation

The effects of glucocorticoids on alveolar development are influenced by the age at which they are administered. Massaro and coworkers (34) showed that administering 0.1 μg of Dex (which reduced body weight by 10%) to rats between P4 and P13 diminished the alveolar surface area at P14, and it remained lower than in saline-treated controls at P60. The alveolar wall interstitial thickness, the V_D of lipid-laden and non–lipid-laden interstitial cells, and the V_D of fibroblast lipid droplets were all reduced at P14 (18). Tschanz and colleagues (35), and later Roth-Kleiner and colleagues (17), showed that administering Dex from P1 through P4 produced septal thinning and diminished alveolar capillaries, both of which normalized after P13. This abbreviated administration of Dex only transiently interrupted the proliferation and apoptosis of septal cells at P4 and diminished elastin and tenascin-C at P6 (36). These studies identified a critical time during which Dex reversibly interrupts mesenchymal cell functions, and coincides with the postnatal period in which LIFs maximally proliferate and accumulate neutral lipids (14, 37). The study presented here explains some of these time-sensitive effects of glucocorticoids during secondary septation. First, proliferation of the GFP^low population was lower in control than in Mfp-exposed fibroblasts at P8 (Figure 2B). Therefore, the GFP^low population is particularly susceptible to manipulation of GR signaling before P5. Second, although GFP^high fibroblasts diminished as a proportion of CD45⁻ cells (Figure 3A), the GFP^high, ADRP⁺, α-SMA⁺, and UCP1⁺ progenitor populations increased (Figure 5E). Retained progenitors may contribute to the “catch-up” septal formation that is observed when Dex is withdrawn before P5 (17, 35). Ntokou and coworkers (12) reported that only fibroblasts that had been lineage marked on or before P2 were observed in both lipid-laden and soma-expressing populations at P7. This suggests that by preserving PDGFRα-expressing progenitors, transient (P1–P4) glucocorticoid administration may enable alveolarization to recover.

Significance of UCP1 in Lung Fibroblast Progenitors

UCP1-expressing adipocytes arise from at least two progenitor populations. Classical brown adipocytes are defined embryonically and arise along myotomes from Pax7⁺, Myf5⁺ precursors, which may also differentiate into myocytes (38). Beige (sometimes termed brite) adipocytes arise within white adipose tissue from progenitors with characteristics of smooth muscle cells (i.e., they are PDGFRα⁺ and α-SMA⁺, and some are Myh11⁺). Whereas UCP1 is constitutively expressed by brown adipocytes, expression by beige adipocytes requires induction from cold exposure or β3-adrenergic stimulation (25). Adrenergic induction of UCP1 in beige adipocytes follows expansion of the adipose stromal vascular cell population and is dependent on vascular endothelial growth factor (VEGF)-A and vascular endothelial growth factor receptor-2 (VEGF receptor-2 or VEGFR2) (39). Relevant to our findings is the novel observation of Seki and associates (40) that the adipose stromal vascular endothelium produces PDGF-CC (a PDGFRα ligand), which stimulates UCP1 gene expression in PDGFRα⁺, CD34⁺, and Sca1⁺ perivascular mesenchymal cells. We previously found that some PDGFRα⁺ lung fibroblasts express CD34 and Sca1 at P8 (10), and have now shown that Dex expands the Sca1⁺, GFP^high subpopulation and increases the proportion of these cells that express UCP1. Likewise, we previously showed that targeted pdgfrα deletion increased Dlk1C mRNA (10), and have now observed that Dex increases Dlk1C, consistent with their progenitor state. Therefore, like beige adipocyte precursors, PDGFRα-expressing lung fibroblast progenitors retain characteristics of both smooth-muscle and lipid-storage cells (41). Precisely what drives PDGFRα-expressing fibroblasts to express UCP1 remains unclear, because they are in a warm environment and β3-adrenergic receptors have not been observed in lung fibroblasts (42). UCP1 more likely marks progenitor cells, which have not fully committed to either an MF or a lipid-storage phenotype.

Contributions of Fibroblast Plasticity to Alveolar Repair and Regeneration

In mice and rats, neutral lipid droplets appear during the late canalicular stage and remain abundant through the first two postnatal weeks (14, 37, 43). Although their presence in the lungs of human newborns remains controversial, lipid-laden fibroblasts contribute to alveolar repair and regeneration in models of human disease (44). Other investigators have studied murine fibroblast subpopulations during compensatory right lung growth (CLG) after a left pneumonectomy. It was found that in adult mice, a larger proportion of PDGFRα-GFP^low fibroblasts contained α-SMA, and this population increased after a pneumonectomy (11). Administration of rosiglitazone, a peroxisome proliferator-activated recptor-γ (PPARγ) agonist that promotes lipid accumulation in LIFs, increased the abundance of GFP^high, α-SMA^low fibroblasts. In a follow-up study, the same group explored differences between the GFP^low and GFP^high populations in sham-operated and pneumonectomized mice (15). In the sham-operated mice, the lipid-laden cells also expressed CD34. Pneumonectomy reduced the proportion of lipid-laden cells and increased the proportion of α-SMA⁺ cells within the CD34⁺ population. Adrenalectomy influenced CLG much like targeted GR deletion did during the saccular stage of development. Adrenalectomized rats exhibited a thickened interstitium with more fibroblasts during CLG compared with pneumonectomized controls with intact adrenal glands (45). Therefore, pneumonectomy modifies the CD34⁺ fibroblast population by reducing the proportion that contains lipid droplets, increasing the proportion that contains α-SMA, and increasing the expression of the extracellular matrix proteins tenascin C and periostin, consistent with a more synthetic phenotype. Recently, El Agha and associates (46) lineage traced the lipid-storage and myofibroblastic phenotypes in adult mice. They demonstrated that after bleomycin administration, fibroblasts that were lineage labeled for α-SMA acquired lipid droplets, suggesting that adult lung fibroblasts retain plasticity and may have salutary effects after lung injury.

PDGFRα Marks Progenitors, but Does Not Exclusively Control their Differentiation

Our study also clarifies how PDGFRα signaling contributes to the balance between progenitors and differentiated alveolar septal MFs. Although alveolar fibroblast subpopulations can be distinguished by their apparent level of PDGFRα gene transcription (with GFP as a reporter of endogenous PDGFRα gene expression), it remains unclear how well GFP intensity correlates with PDGFRα signaling activity. During secondary septation, the GFP intensity correlates with the abundance of CD140a (PDGFRα) on the cell surface and the abundance of PDGFRα mRNA (10), and the PDGFR-kinase inhibitor imatinib primarily suppresses the proliferation of GFP^high fibroblasts (10). However, factors in addition to gene expression may regulate PDGFRα signaling and impact the fibroblast phenotype. In murine neonates and adults, GFP intensity correlates with quantitative rather than qualitative differences in lipid storage or α-SMA, as both the GFP^low and GFP^high populations contain cells exhibiting both lipid droplets and α-SMA (15). Gene expression profiling in adult mice characterized which genes are more highly expressed in GFP^low (vinculin and integrin α8, consistent with a contractile phenotype) and GFP^high (the extracellular matrix proteins periostin, collagen 3a1, fibrillin 1, consistent with a synthetic phenotype) populations (15). However, PDGFRα signaling is also regulated post-transcriptionally. Studies using oligodendrocyte precursor cells (which are similar to cancer stem cells) from the periphery of glioblastoma multiforme tumors showed that PDGFRα signal transduction preserves “stemness” manifested as self-renewal, retention of markers of multipotency, and invasiveness (47). Other examples of post-transcriptional regulation include (1) the abundance of PDGFRα on the cell surface, which is regulated by endosomal recycling and exosomal shedding (48, 49); (b) the location on the cell surface (i.e., whether PDGFRα localizes to membrane lipid rafts) (48); and (3) the proximity to phosphatase inhibitors, which are also regulated by recycling and membrane location (50). These factors influence how PDGFRα impacts oligodendrocyte precursor cell stemness and differentiation, and could also influence PDGFRα function in alveolar fibroblasts (47). Additional experimentation is required to understand the impact of these various pathways on alveolar fibroblasts.

Potential Clinical Application

Lineage-tracing studies have demonstrated the importance of Smad2 signaling for the differentiation of Shh-responsive, PDGFRα⁺, mesenchymal progenitors destined to be MFs (3). A second population of PDGFRα⁺ cells, from the same mesenchymal lineage, can differentiate postnatally into a lipid-storage phenotype in the absence of Smad2 signaling (4). Our studies suggest that a portion of this second population maintains bipotency at least through P8, combining characteristics of both MFs (α-SMA) and LIFs (ADRP, UCP1). These less committed cells more likely retain the progenitor markers Dlk1C and Sca1, and contain both α-SMA and lipid droplets, if Dex is administered during this early postnatal time window. Administering glucocorticoids to newborns with bronchopulmonary dysplasia or adults with pulmonary fibrosis has not improved their clinical outcome. However, our studies suggest that Dex sustains plasticity, and that targeting additional downstream pathways could steer fibroblasts away from fibrogenesis toward a more salutary phenotype. Further investigation is required to identify and modify these incompletely defined pathways.