Pericytes are perivascular PDGF receptor-β+ (PDGFRβ+) stromal cells required for vasculogenesis and maintenance of microvascular homeostasis in many organs. Because of their unique juxtaposition to microvascular endothelium, lung PDGFRβ+ cells are well situated to detect proinflammatory molecules released following epithelial injury and promote acute inflammatory responses. Thus we hypothesized that these cells represent an unrecognized immune surveillance or injury-sentinel interstitial cell. To evaluate this hypothesis, we isolated PDGFRβ+ cells from murine lung and demonstrated that they have characteristics consistent with a pericyte population (referred to as pericyte-like cells for simplicity hereafter). We showed that pericyte-like cells expressed functional Toll-like receptors and upregulated chemokine expression following exposure to bronchoalveolar lavage fluid (BALF) collected from mice with sterile lung injury. Interestingly, BALF from mice without lung injury also induced chemokine expression in pericyte-like cells, suggesting that pericyte-like cells are primed to sense epithelial injury (permeability changes). Following LPS-induced lung inflammation, increased numbers of pericyte-like cells expressed IL-6, chemokine (C-X-C motif) ligand-1, chemokine (C-C motif) ligand 2/ monocyte chemotactic protein-1, and ICAM-1 in vivo. Sterile lung injury in pericyte-ablated mice was associated with decreased inflammation compared with normal mice. In summary, we found that pericyte-like cells are immune responsive and express diverse chemokines in response to lung injury in vitro and in vivo. Furthermore, pericyte-like cell ablation attenuated inflammation in sterile lung injury, suggesting that these cells play an important functional role in mediating lung inflammatory responses. We propose a model in which pericyte-like cells function as interstitial immune sentinels, detecting proinflammatory molecules released following epithelial barrier damage and participating in recruitment of circulating leukocytes.
- acute lung injury
- damage-associated molecular patterns
- innate immunity
via inhalation, the lungs are continuously exposed to a variety of potentially damaging microbes and pollutants, which can result in injury to airway and alveolar epithelia. Like other tissues, the lung includes a variety of innate immune responsive cells that express germline encoded pattern recognition receptors, which detect specific molecular patterns associated with pathogenic microbes [pathogen-associated molecular patterns (PAMPs)] or released by damaged cells [damage-associated molecular patterns (DAMPs); alarmins]. These pattern recognition receptors include Toll-like receptors (TLRs), the Nod-like receptors, the RIG family of receptors, and others (20).
However, to avoid chronic inflammation and progressive injury, it is critical that the lung only initiate a robust inflammatory response when a serious infection or tissue damage develops. One solution to moderate injury responses is for “sentinel” cells to exist in the lung interstitium, isolated from the alveolar compartment. Such a cell would be primed and ready to sense disruption of the mucosal barrier by detecting the passage of PAMPs and/or DAMPs into the interstitial space from the disrupted alveolar lumen. In turn, such interstitial sentinel cells would be able to initiate a systemic response to control infection and initiate repair of damaged tissue.
Recently, an in-depth microscopic study revealed substantial heterogeneity within the stromal cells of the mouse lung, including pericyte-like cells, as defined by their perivascular location, and expression of the pericyte-associated markers, platelet-derived growth factor receptor-β (PDGFRβ), nerve/glial antigen 2 (NG2, a.k.a., CSPG4), and desmin (15). Pericytes are mesenchyme-derived stromal cells that associate intimately with endothelial cells of capillaries and postcapillary venules (2). Traditional identification of pericytes required electron microscopy to confirm that the cells are embedded within the capillary basement membrane (16); however, many investigators now identify pericyte-like cells based on localization to the microvasculature and expression of pericyte-associated markers. The most common marker used to identify pericytes is PDGFRβ (2).
Using lineage tracing, our group recently reported that lung interstitial cells derived from prenatal mesenchyme FoxD1-expressing cells are consistent with pericytes (9). Furthermore, transcriptional profiling of these cells reveals distinct differences between collagen expressing, PDGFRα+ lung interstitial cells and FoxD1-derived putative pericytes, which have enriched expression of genes associated with innate immunity. In contrast, collagen-expressing PDGFRα+ cells were enriched for genes associated with formation of extracellular matrix.
We speculated that pericyte-like cells have a previously unrecognized function in the lung as interstitial innate immune sentinel cells that detect alveolar barrier disruption caused by infection or noninfectious injury. To further investigate the immune biology of pericyte-like cells, we examined their functional response to inflammatory stimuli in vitro and in vivo.
MATERIALS AND METHODS
Animals and Reagents
The University of Washington Institutional Animal Care and Use Committee approved all experiments involving the use of mice described here. C57BL/6J mice, BAC transgenic mice expressing DsRed fluorescent protein under the regulation of the NG2/Cspg4 promoter (NG2DsRedBAC), mice expressing a GFP-Cre fusion protein under the regulation of the FoxD1 promoter (FoxD1GC), mice expressing the Cre-dependent TdTomato reporter at the Rosa26 locus (R26TdT-R), and mice with Cre-dependent expression of simian diphtheria toxin receptor (iDTR) at the Rosa26 locus (R26-iDTR) were purchased from Jackson Laboratories (Bar Harbor, ME). Myd88−/− mice on a C57BL/6 background were originally developed by Shizuo Akira (Osaka University). Mice with TdTomato expression restricted to FoxD1 lineage stromal cells were generated by crossing FoxD1GC mice with R26TDT-R mice. Mice with diphtheria toxin (DT) receptor expression restricted to FoxD1 lineage stromal cells were generated by crossing FoxD1GC mice with R26-iDTR mice.
Antibodies for immunofluorescence included anti-PDGFRβ (clone Y92), anti-CD146 (clone EPR3208), and anti-NG2 (EMD Millipore, Billerica, MA), and rat anti-CD31 (clone 390) (eBioscience, San Diego, CA). Alexa Fluor 488-conjugated goat anti-rabbit IgG, Alexa Fluor 568-conjugated goat anti-rat IgG, and Zenon Alexa Fluor 647 Rabbit IgG labeling kit were from Life Technologies (Carlsbad, CA). Antibodies for magnetic-activated cell sorting included phycoerythrin (PE)-conjugated anti-PDGFRβ (clone APB5), and biotin-conjugated anti-CD31 (clone 390), anti-CD45 (clone 30-F11), and anti-CD326 (clone G8.8) (eBioscience). Magnetic columns were from Miltenyi Biotec (San Diego, CA). Liberase TL Research Grade enzyme mix and DNase I were purchased from Roche Applied Science (Pleasanton, CA). All TLR ligands were purchased from InvivoGen (San Diego, CA). Recombinant IL-1β was purchased from PeproTech (Rocky Hill, NJ). Recombinant human Fas ligand (rhFasL) was from Enzo Life Sciences (Farmingdale, NY). Diptheria toxin was purchased from Sigma Aldrich (St. Louis, MO). Primer sequences were designed using PrimerQuest Design Tool software from Integrated DNA Technologies (Coralville, IA). ELISAs for detecting mouse IL-6, chemokine (C-X-C motif) ligand (CXCL) 1/keratinocyte-derived chemokine (KC), and CXCL2/macrophage inflammatory protein 2 (MIP-2) were purchased from R&D Systems (Minneapolis, MN). Trypsin immobilized on magnetic beads was purchased from Clontech (Mountain View, CA).
Tissue Preparation and Immunofluorescence
Pulmonary vasculature was flushed via the right ventricle with cold PBS, and lungs were inflated with and embedded in optical cutting temperature medium. Ten-micrometer sections were made on a Leica CM1950 cryostat (Buffalo Grove, IL) and mounted on Fisher Plus microscope slides. Sections were air-dried and stored at −80°C.
Tissue sections were stained with anti-PDGFRβ (1:100) and anti-CD31 (1:100), followed by anti-rabbit IgG (1:500) and anti-rat IgG (1:200) or with anti-CD146 (1:100) directly conjugated to Alexa Fluor 647, counterstained with 4,6-diamidino-2-phenylindole (DAPI) and imaged using a Nikon A1R Confocal Microscope (Melville, NY).
Primary Cell Isolation and Culture
Single-cell preparations from whole lung digests were obtained as previously described (9). Single-cell preparations were resuspended in culture media and cultured in a T75 flask coated with 0.2% gelatin for 3–4 days. Following expansion in culture, cells were negatively selected for CD45, CD31, and CD326 expression and positively selected for PDGFRβ. PDGFRβ+ cells were then expanded in culture and used by passage 3. Cultured cells were recovered with Accutase (Innovative Cell Technologies), washed once with magnetic-activated cell sorting buffer (PBS, 0.5% BSA, 2 mM EDTA, pH 7.2), and labeled with biotin-conjugated anti-CD45, anti-CD31, and anti-CD326 for 15 min. Labeled cells were depleted by incubation with anti-biotin microbeads, followed by passage over a magnetized LD column. Unlabeled cells were collected and incubated with PE-conjugated anti-PDGFRβ. PDGFRβ+ cells were collected by incubation with anti-PE microbeads and passage over a magnetized MS column. Retained, PDGFRβ+ cells were collected and cultured. For select experiments, nonretained PDGFRβ- stromal cells were collected and cultured. All stromal cells were cultured on 0.2% gelatin-coated plates with DMEM/F-12 supplemented with 10% FBS, 1% insulin-transferrin-selenium, and 1% penicillin/streptomycin (culture media).
PDGFRβ+ cells cultured on collagen I-coated glass coverslips were fixed in 4% paraformaldehyde for 15 min, washed, blocked in 10% goat serum, and incubated with primary antibodies (PDGFRβ, 1:100; CD146, 1:100, or NG2, 1:100) for 1 h, followed by incubation with a secondary anti-rabbit IgG Alexa Fluor 488 antibody (1:500) for 30 min. Coverslips were mounted with Vectashield containing DAPI (Vector Laboratories, Burlingame, CA). Images were acquired with a Nikon TiE Inverted Widefield Fluorescent Microscope.
Primary lung endothelial cells (17) and bone marrow-derived macrophages (1) were isolated and cultured as previously described. MLE-12 epithelial cells were purchased from ATCC and cultured according to vendor instructions.
Cultured lung PDGFRβ+ stromal cells were collected using Accutase and incubated with Fc Block (BD Biosciences), followed by labeling with anti-CD146-AF647 and anti-PDGFRβ-PE. Fixable Viability Dye eFluor 450 was used to identify live cells. Negative gates were set to exclude 95% of isotype antibody exposed and unstained cells, using fluorophore-conjugated isotype antibodies and unstained cells, respectively. Flow cytometry was performed on an LSR II Flow Cytometer (BD Biosciences) and analyzed using FlowJo Software (version 9.4, Ashland, OR).
Assessment of Functional TLR Expression and Response to Bronchoalveolar Lavage Fluid
Lung PDGFRβ+ cells were cultured on collagen I-coated 12-well plates at a density of 25,000 cells/well. At 90% confluency, cells were washed and incubated overnight in serum-free DMEM/F-12 medium and evaluated by the following.
Serum-deprived cells were exposed to murine recombinant IL-1β or various TLR ligands (Table 1) diluted in DMEM/F-12 medium for 18 h, and RNA were collected.
Bronchoalveolar lavage fluid exposure.
Serum-deprived cells were exposed to bronchoalveolar lavage fluid (BALF) collected from injured or uninjured mice diluted 1:1 in serum-free DMEM/F-12 for 18 h. Cell culture supernatant and RNA were then collected. For some experiments, BALF was incubated overnight with or without immobilized trypsin. The following morning, trypsin was removed using a magnet before incubation with the cells.
To collect BALF from mice with sterile lung injury, we used a model of combined rhFasL administration and mechanical ventilation (MV). Briefly, 8-wk-old mice were anesthetized with isoflurane and administered 50-µl rhFasL via oropharyngeal aspiration (12.5 ng/g body weight at a concentration of 6.25 ng/µl). After 18 h, mice were anesthetized with isoflurane and intubated via tracheotomy. Mice were paralyzed with pancuronium and ventilated on a FlexiVent system (SciReq, Montreal, QC) for 5 h at a tidal volume of 20 ml/kg, respiratory rate of 120 breaths/min, inspired O2 fraction of 0.5, and an end-expiratory pressure of 0. At the conclusion of the experiment, mice were euthanized, and bronchoalveolar lavage was performed with 1 ml of serum-free DMEM/F-12. As controls, BALF was obtained from uninjured mice. BALF was spun at 500 g for 10 min to remove any cells and debris.
Assessment of In Vivo Lung PDGFRβ+ Stromal Cell Activation
To examine the in vivo functional response of PDGFRβ+ stromal cells to specific TLR agonist, C57BL/6J mice 8–12 wk of age were given 2.5 mg/kg of lipopolysaccharide (LPS, E. coli 055:B5, Sigma-Aldrich) or an equal volume of sterile PBS via oropharyngeal aspiration, as previously described (11). Six hours following LPS instillation, mice were euthanized, and a single-cell dispersion of lung cells was prepared by enzymatic digestion as above. Red blood cells were lysed by resuspending cells in 9 ml of dH2O for 30 s, followed by addition of 1 ml of 10× PBS. Cells were then collected at 400 g, 4°C × 5 min, and resuspended at 107 cells/ml FACS buffer (Dubecco's PBS, 2 mM EDTA, 0.1% BSA) for immunolabeling. Following incubation with Mouse BD Fc Block (1 μg per 1 × 106 cells) for 10 min at 4°C, cells were labeled with PE-conjugated antibodies against mouse CD31, CD45, and CD326 for 30 min at 4°C. After wash, cells were fixed and permeabilized using ABD Serotec reagents A/B per manufacturer’s instructions, and cells were stained for PDGFRβ, IL-6, CXCL1, chemokine (C-C motif) ligand 2 (CCL2)/monocyte chemotactic protein (MCP)-1, and CD54/ICAM-1 for 30 min at room temperature, followed by staining with appropriate fluorophore-conjugated secondary antibody for 30 min at room temperature. Samples were resuspended in 1% paraformaldehyde and kept in the dark at 4°C until analysis within 24 h. Cells were analyzed using FACSCanto RUO (BD Boioscience). PDGFRβ+ stromal cells were defined by CD31−, CD45−, CD326−, and PDGFRβ+ gating. We then assessed the level of CD54/ICAM-1, CCL2/MCP-1, IL-6, and CXCL1 expression in the PDGFRβ+ stromal population.
Assessment of Acute Lung Injury Following Pericyte Ablation
We examined acute lung injury in FoxD1GC;Rs26-iDTR bitransgenic mice using a bleomycin-induced lung injury model. Our laboratory previously described FoxD1-derived cells as stromal cells that are enriched for PDGFRβ expression in the lungs (9). Since FoxD1 is only expressed during embryogenesis and is silenced in adults and during lung injury, we used the FoxD1GC transgene to drive the expression of simian diphtheria toxin receptor (iDTR) in the FoxD1-lineage stromal cells in bitransgenic FoxD1GC;Rs26-iDTR mice. FoxD1-lineage cells, enriched in PDGFRβ expression, become susceptible to ablation when exposed to DT in this model. We administered DT to the lungs directly by oropharyngeal aspiration (1 ng/g mouse on day 1 and day 2) to ablate PDGFRβ-enriched cells in FoxD1GC;Rs26-iDTR bitransgenic mice, as previously described (10). Littermate Rs26-iDTR transgenic mice without the FoxD1GC transgene were also administered DT at the same dose and schedule (control). On day 3, both groups were administered bleomycin (1.3 U/kg mouse) by intratracheal instillation, and mice were harvested on day 5 (48 h following bleomycin) when we expected to see evidence of acute lung injury in the bleomycin model.
All data are presented as means ± SE. Data from TLR agonist experiments were compared by one-way ANOVA. For the bronchoalveolar lavage stimulation experiments, data were analyzed by two-way ANOVA with the main effects being genotype or condition. Bonferroni’s multiple-comparison test was used for post hoc analysis. A P value ≤ 0.5 was used to determine statistical significance. Analysis was done using GraphPad Prism 5 (La Jolla, CA).
Identification of Putative Pericytes In Vivo
We performed immunofluorescence studies in uninjured C57BL/6 mice. PDGFRβ+ cells present in the alveolar interstitium were closely associated with, but did not colocalize to, CD31+ cells (Fig. 1, A and B). PDGFRβ did colocalize with the pericyte-associated marker CD146 (3, 6) (Fig. 1, C–E). We also evaluated for the presence of NG2, which is expressed on pericytes associated with arterioles and capillaries, but not on pericytes located in postcapillary venules (13). Using BAC transgenic mice that express dsRed fluorescent protein regulated by the NG2 promoter, we found that NG2 was expressed in a subset of PDGFRβ+ cells in the alveolar interstitium (Fig. 1F). These results are consistent with prior reports of pericyte-like cells in the lungs of mice (9, 15).
Characterization of Cultured Lung PDGFRβ+ Stromal Cells
PDGFRβ+ lung cells isolated by magnetic sorting were expanded in cell culture for two passages before analysis of expression of pericyte-associated markers. By immunofluorescence, the majority of cultured cells expressed PDGFRβ (Fig. 2A) and CD146 (Fig. 2B). In contrast, no CD45+ or CD31+ cells were detected (data not shown). Cultured PDGFRβ+ cells isolated by magnetic sorting from three different lung digests were also analyzed by flow cytometry, and we found that 86.4 ± 1.5% of cultured cells were both PDGFRβ+ and CD146+, 12.4 ± 1.3% were CD146+ only, 0.5 ± 0.1% were PDGFRβ+ only, and 0.7 ± 0.2% were negative for both CD146 and PDGFRβ (Fig. 2C). All cells were CD45−, CD31−, and Epcam/CD326−. Additionally, a subset of cells was NG2+, which suggests that the cultured cells were isolated from both precapillary/capillary and from postcapillary locations (Fig. 2D). Lung PDGFRβ+ cells were also isolated by magnetic sorting and cultured from mice that express TdTomato red fluorescent protein restricted to FoxD1-derived cells (Foxd1-GC;R26RTdT) (8, 9). These cells demonstrated uniform expression of red fluorescent protein (Fig. 2E). Finally, we evaluated functional expression of PDGFRβ in isolated cells by quantifying angiopoietin-1 (Angpt1) mRNA transcript level following addition of various amounts of PDGF-BB in culture medium for 1 wk. Cultured pericytes are reported to express angiopoietin-1 (19). PDGF-BB resulted in a dose-dependent increase in Angpt1 (Fig. 2F). Together, these data strongly suggested that the isolated lung PDGFRβ+ cell population was enriched for putative pericytes. Furthermore, isolated cells cultured over several passages retained phenotypic characteristics of pericytes. However, given the inability to confirm that the isolated cell population arises from microvascular-associated stromal cells, we refer to these cells as pericyte-like cells.
Proinflammatory Activation of Pericyte-like Cells In Vitro
To evaluate for functional expression of TLRs, lung pericyte-like cells were incubated with various concentrations of ligands for different TLRs or with IL-1β. RNA was collected and assessed for expression of the neutrophil chemokines Cxcl1 (KC) and Cxcl2 (MIP-2) (Fig. 3, A and B), the lymphocyte chemokine, Cxcl10 (interferon-γ-induced protein 10) (Fig. 3C), the monocyte chemokine, Ccl2 (MCP-1) (Fig. 3D), Tnf (Fig. 3E), and Icam1 (Fig. 3F). We found that lung PDGFRβ+ cells had increased chemokine expression when exposed to ligands for IL-1r, TLR1, TLR2, TLR4, TLR6, and TLR7, suggesting that these cells expressed a panel of functional TLRs. In contrast, there were minimal transcriptional responses to flagellin, a ligand for TLR5, or to unmethylated CpG, a ligand for TLR9.
Based on the findings that pericyte-like cells exhibit functional TLR responses, we hypothesized that these cells function as interstitial sentinel cells that respond to effector mediators such as DAMPs released from the alveolar compartment into the interstitial space following alveolar epithelial injury and barrier disruption. To test this hypothesis, we collected BALF from uninjured mice and from mice with sterile lung injury induced by oropharyngeal aspiration of rhFasL, followed 18 h later by 6 h of MV. This model, which does not rely on PAMP-mediated TLR activation, resulted in significant sterile lung injury, as demonstrated by neutrophilic inflammation and increased alveolar-capillary permeability (Fig. 4A).
We incubated cultured pericyte-like cells with BALF from uninjured mice and from mice with FasL/MV-induced sterile lung injury. We found that BALF from uninjured mice induced transcription of the chemokines Cxcl1, Ccl2, and Il6 (Fig. 4B). This proinflammatory transcriptional response was further increased if lung PDGFRβ+ cells were incubated with BALF from mice with sterile FasL/MV lung injury. Lung PDGFRβ+ cells isolated from mice lacking the MyD88 adapter protein, which is required for intracellular signaling by a subset of the TLR/IL-1R family of receptors, had minimal transcriptional response to BALF from uninjured mice and an attenuated response to BALF from mice with sterile lung injury. These findings indicate that lung PDGFRβ+ stromal cells respond to factors constitutively present in the alveolar luminal space via a MyD88-dependent mechanism.
Because BALF from mice without sterile lung injury activated lung PDGFRβ+ cells, we assessed if other cell types, namely mouse alveolar epithelial cells (MLE12 cells), primary mouse lung endothelial cells, primary mouse PDGFRβ negative stromal cells, and mouse bone marrow-derived macrophages, were also stimulated by BALF at the 24-h time point. Although all of these cells upregulated chemokine expression in response to LPS, their transcriptional response to BALF collected from normal, uninjured mice was minimal (Fig. 4C). Additional measurements of IL-6, CXCL1/KC, and CCL2/MCP-1 in cell culture supernatant from PDGFRβ+ stromal cells confirmed that exposure to BALF from uninjured mice increased secretion of these cytokines (Fig. 5). Neither IL-6 nor CXCL1/KC was detectable in BALF from uninjured mice in the absence of incubation with pericyte-like cells (data not shown). These data suggest that factors present in the normal, uninjured alveolar compartment act as proinflammatory stimulants in pericyte-like cells specifically but not in other cell types, further supporting the hypothesis that pericyte-like cells may play a unique role in sensing epithelial barrier disruption and the inflammatory response that ensues. To further investigate whether proinflammatory factor(s) present in normal, uninjured BALF are protein(s), we exposed pericyte-like cells to trypsin-digested BALF and media-only control. Pretreatment of BALF with trypsin digestion completely abrogated the proinflammatory stimulus on pericyte-like cells, suggesting that the proinflammatory mediator present in normal BALF is of protein origin (Fig. 5).
Lung Pericyte-like Cells Are Activated and Regulate Inflammation and Vascular Barrier Integrity During Lung Injury
To determine whether the proinflammatory phenotype in pericyte-like cells can be replicated in vivo, mice were challenged with intratracheal LPS. After 6 h, mice were euthanized, and a monocellular suspension was analyzed by flow cytometry. About 20% of PDGFRβ+/CD45−/CD31−/CD326− cells collected from control (PBS-exposed) lungs expressed the leukocyte adhesion molecule, ICAM1/CD54, 3% expressed IL-6, 17% expressed CXCL1, and <3% expressed CCL2/MCP-1 protein. In contrast, LPS-exposed mice had significantly higher percentages of PDGFRβ+/CD45−/CD31−/CD326− cells that expressed ICAM-1/CD54 (46%), IL-6 (17%), CXCL1 (40%), and CCL2/MCP-1 (12%) compared with PBS-exposed mice (Fig. 6).
While these results demonstrated pericyte-like cells are capable of mounting an inflammatory response, whether these responses have functional relevance in acute lung injury remains unknown. We previously demonstrated the ability to ablate ~40–50% of pericyte-like cells (both by PDGFRβ+ staining and FoxD1-lineage tracing) by oropharyngeal aspiration of DT in FoxD1GC;Rs26-iDTR bitransgenic mice. In these mice, FoxD1-lineage cells express the iDTR and are sensitive to ablation by DT, and the ablation was seen up to 7 days following administration of DT (10). To assess the functional relevance of pericyte-like cells in acute lung inflammation, we examined whether the ablation of pericyte-like cells attenuated acute lung inflammation induced by bleomycin, a third model of sterile lung injury (Fig. 7A). DT (1 ng/g) was administered via oropharyngeal aspiration to FoxD1GC;Rs26-iDTR bitransgenic mice (Cre+) and to Cre− littermate controls (controls) on 2 consecutive days. One day following DT administration, bleomycin (1.3 U/kg) was delivered by intratracheal instillation to induce sterile lung injury, and injured mice were harvested 48 h following bleomycin instillation. DT administration at this dosage without bleomycin resulted in modest elevations of white blood cells (WBC) in BALF to ~20 × 104 cells/ml WBC in both Cre+ and control mice (Fig. 7B). A modest elevation of total protein in the range of 0.3–0.4 mg/ml in the BALF was also observed in both Cre+ and controls (Fig. 7B). These results are consistent with our prior observations in the DT-ablation model, where DT by itself induced a modest inflammatory response in the lungs. When these DT-exposed mice were challenged with bleomycin, controls showed an even higher BALF WBC count at ~37 × 104 cells/ml, whereas Cre+ mice showed an attenuated BALF WBC count at 25 × 104 cells/ml. Similarly, controls showed further elevations in total protein in the BALF (0.9 mg/ml), whereas Cre+ mice had attenuated BALF total protein (0.5 mg/ml) (Fig. 7B). When we examined the levels of IL-6, CXCL-1, and CCL2 in the BALF from DT-exposed and bleomycin-injured mice, Cre+ animals showed lower levels of IL-6 and CCL2 compared with controls, whereas the CXCL-1 level was elevated compared with controls (Fig. 7C).
In this study, we examined whether pericyte-like cells represent a previously unrecognized subset of immune-responsive cells in the lung. The primary findings of this study are as follows: 1) confirmation of pericyte-like cells in the lung (9, 15) based on expression of pericyte-associated markers and close association with CD31+ endothelial cells in the alveolar microvasculature; 2) enrichment and culture of pericyte-like cells that are phenotypically consistent with pericytes; 3) pericyte-like cells express multiple TLRs; 4) a variety of proinflammatory genes are induced in pericyte-like cells in response to BALF from both uninjured and injured mice; and 5) in vivo, pericyte-like cells are activated during lung injury and contribute to the development of leukocytic inflammation and vascular barrier dysfunction.
Pericytes are a stromal cell subpopulation closely associated with the abluminal surface of capillaries and postcapillary venules in all organs. The primary functions attributed to pericytes are the maintenance of microvascular development and stability. There has been relatively little data examining pericytes in the lung. Pericytes have been previously identified in the lungs of humans, dogs, guinea pigs, and rats by traditional electron microscopy studies, which identified cells associated with the abluminal surface of the pulmonary microvasculature and partially embedded within the microvascular basement membrane (22). More recently, Rock et al. (15) identified putative pericytes in mouse lungs using confocal microscopy to identify PDGFRβ+, NG2+, and desmin+ cells closely associated with CD31+ endothelial cells in the alveolar interstitium. Our group recently described FoxD1-derived cells in the alveolar interstitium of mice that are consistent with a lung pericyte-like population (9). The histological data presented in this report provide further confirmation of the presence of pericyte-like cells in the lung parenchyma of mice, allowing the opportunity to further study the potential functions of this cell population in lung biology.
Recent studies in cremaster muscle and dermis suggest a previously unrecognized role of pericyte-like cells in regulation of tissue leukocyte trafficking during inflammation and injury (14, 18). Our laboratory recently demonstrated that FoxD1-derived pericyte-like cells have baseline increased expression of immune-associated transcripts compared with collagen-expressing lung fibroblasts (9), suggesting that pericyte-like cells may have an unanticipated role in lung immune function similar to that seen in skin and cremaster muscle. Using isolated and expanded cells in vitro, we demonstrate for the first time that cultured pericyte-like cells, which are phenotypically consistent with pericytes, express a range of functional TLRs, further supporting a possible role for this cell population in innate immune responses in the lung. Furthermore, following TLR activation, these cells upregulate the leukocyte adhesion molecule, Icam1/CD54, and chemokines for multiple classes of leukocytes, including polymorphonuclear cells (Cxcl2/Mip2), monocytes (Ccl2/Mcp1), and lymphocytes (Cxcl10/Ip10).
Localization of pericyte-like cells to the lung interstitium suggests that these cells may be important in detecting either microbial products (PAMPs) or DAMPs (or danger signals, alarmins) released from the alveolar compartment in the setting of epithelial disruption. To explore this hypothesis, we utilized three different models of sterile lung injury. The first model combined two clinically relevant factors, soluble FasL and MV. The FasL/Fas system has been implicated in multiple studies as a potential contributor to epithelial barrier disruption during acute respiratory distress syndrome (4, 7, 12). This sterile model of lung injury was chosen because it was not dependent on specific TLR activation by any individual PAMP, but instead would result in significant sterile cell injury and release of cell DAMPs, similar to what might occur in clinical acute respiratory distress syndrome. Pericyte-like cells exposed to BALF collected from mice with FasL/MV-induced sterile lung injury increased chemokine expression, consistent with a role of PDGFRβ+ cells as lung interstitial immune sentinel cells. Furthermore, the finding that BALF from uninjured mice also selectively activated pericyte-like cells suggests component(s) found in the normal alveolar lumen may act as a pericyte-selective danger signal with translocation into the interstitial compartment. While we have yet to identify the key factors in the alveolar space that stimulate an inflammatory phenotype in pericyte-like cells, we speculate these factors are proteins that are susceptible to trypsin digest based on our observations. Chemokine upregulation was significantly attenuated in pericyte-like cells lacking the TLR adapter protein, MyD88, implicating potential DAMP-TLR interaction as a mechanism for cellular activation. To confirm activation of pericyte-like cells, in vivo, we used a common model of sterile lung inflammation induced by acute lung exposure to LPS and found increased expression of ICAM-1, IL-6, CXCL1, and CCL2 in CD45−, CD31−, CD326−, and PDGFRβ+ cells, supporting the validity of the in vitro findings that pericyte-like cells can upregulate a variety of proinflammatory molecules in response to lung inflammation.
Based on the in vitro and in vivo findings that pericytes were able to elaborate proinflammatory cytokines, we further hypothesized that PDGFRβ+ cells play a functional role in the acute inflammatory and injury responses in the lung. To test this hypothesis, we used the bitransgenic FoxD1GC;Rs26-iDTR mice to selectively ablate pericyte-like cells in the lung. We previously showed FoxD1-lineage cells in the lung are enriched in PDGFRβ expression and label a population of cells localized to the perivascular niche. We used the FoxD1GC to drive expression of DT receptor in this model and delivered DT to ablate sensitive FoxD1-derived cells via oropharyngeal aspiration. We observed attenuation of acute lung inflammation in mice where pericyte-like cells have been ablated, using a third model of bleomycin-induced sterile lung injury. This model was chosen to study the functional role of pericyte-like cells in response to epithelial injury, as opposed to specific PAMP/TLR-mediated inflammation. These results provided further support that pericyte-like cells play an important functional role in regulating the acute inflammatory responses in the lung. Curiously, CXCL1 levels were actually elevated in the BALF from bleomycin-injured, pericyte-ablated mice, even though our in vivo and in vitro data show pericyte-like cells upregulate CXCL1 in response to acute lung injury. This may indicate that CXCL1 expression and regulation are more complex, and other cell types such as alveolar epithelial cells may play more important functional roles in CXCL1 expression in vivo.
Together, these data lead us to hypothesize that lung pericytes function as interstitial sentinel cells that detect PAMPs and DAMPs released from the alveolar compartment, following disruption of the alveolar epithelial barrier (Fig. 8). Such a system would allow the lung to maintain a low inflammatory state during routine exposures to inhaled microbes or toxins but rapidly mount a systemic inflammatory response in the event of a breach in the alveolar epithelial barrier. By utilizing three different models of sterile lung injury, we demonstrate that the inflammatory response in pericyte-like cells are not unique to one specific sterile lung injury model, but is more likely a broad biological response to sterile injury and inflammation in the alveolar compartment.
There are several limitations to this work. The primary limitation is the inability to conclusively identify pericytes as the innate immune responsive cell in the pericyte-like cell population. This would require in vivo immunolocalization and quantitation of proinflammatory molecules in pericytes by electron microscopy. However, demonstration of multiple pericyte-associated markers and derivation from FoxD1 expressing mesenchymal cells strongly support that the pericyte-like population is enriched for pericytes. The use of DT to deplete pericyte-like cells resulted in reduced inflammation and permeability, following sterile lung injury. This suggested that pericyte-like cells play a functional role in regulating responses to lung injury. However, since DT induces apoptosis in sensitive cells, and efferocytosis is reported to promote an anti-inflammatory phenotype in macrophages (21, 23), we cannot exclude the possibility that the observed phenotype is partially the result of macrophage phenotype reprogramming following efferocytosis of apoptotic pericyte-like cells. Finally, these studies are limited to mouse lung pericyte-like cells. However, a recent report of functional TLR4 expression on human brain microvascular pericytes suggests that immune functions of pericyte populations are not limited to mice (5).
In summary, we found that lung pericyte-like cells expressed functional TLRs, were transcriptionally activated by BALF collected from mice with and without lung injury via both MyD88-dependent and -independent mechanisms, and were activated in a mouse model of acute lung injury. More importantly, selective ablation of pericytes from mouse lung using a diphtheria-toxin model resulted in attenuated inflammation in a sterile lung injury model, suggesting that pericytes play an important functional role in regulating acute lung inflammation. These data combined suggest that pericyte-like cells in the lung may be a previously unrecognized innate immune sentinel cell that can orchestrate inflammatory responses following epithelial injury.
This work was supported by a Research Grant from the American Heart Association (W. A. Altemeier); National Heart, Lung, and Blood Institute Grants HL-122895 (W. A. Altemeier) and HL-127075 (C. F. Hung); and a Provost Grant from the University of Washington (W. A. Altemeier).
J. S. Duffield owns stock in Biogen, Inc., which works on anti-inflammatory therapeutics.
C.F.H., K.L.M., R.B., and B.L.M. performed experiments; C.F.H., K.L.M., B.L.M., R.B. and P.C. analyzed data; C.F.H., K.L.M., T.S.H., P.C., L.M.S., W.C.L., J.S.D., and W.A.A. interpreted results of experiments; C.F.H. and K.L.M. prepared figures; C.F.H., K.L.M., W.C.L., T.S.H., and W.A.A. drafted manuscript; C.F.H., K.L.M., T.S.H., W.C.P., P.C., L.M.S., W.C.L., J.S.D., and W.A.A. edited and revised manuscript; C.F.H., K.L.M., R.B., B.L.M., T.S.H., W.C.P., P.C., L.M.S., W.C.L., J.S.D., and W.A.A. approved final version of manuscript.
Present addresses: W. C. Parks, P. Chen, and R. Brauer, Cedars-Sinai Medical Center, Steven Spielberg Bld, 8723 Alden Dr. Los Angeles, CA 90048; L. M. Schnapp, Medical University of South Carolina; 169 Ashley Ave., Charleston, SC 29425.
- Copyright © 2017 the American Physiological Society