Evidence for reprogramming of monocytes into reparative alveolar macrophages in vivo by targeting PDE4b

Abstract

Increased lung vascular permeability and neutrophilic inflammation are hallmarks of acute lung injury. Alveolar macrophages (AMϕ), the predominant sentinel cell type in the airspace, die in massive numbers while fending off pathogens. Recent studies indicate that the AMϕ pool is replenished by airspace-recruited monocytes, but the mechanisms instructing the conversion of recruited monocytes into reparative AMϕ remain elusive. Cyclic AMP (cAMP) is a vascular barrier protective and immunosuppressive second messenger in the lung. Here, we subjected mice expressing GFP under the control of the Lysozyme-M promoter (LysM-GFP mice) to the LPS model of rapidly resolving lung injury to address the impact of mechanisms determining cAMP levels in AMϕ and regulation of mobilization of the reparative AMϕ-pool. RNA-seq analysis of flow-sorted Mϕ identified phosphodiesterase 4b (PDE4b) as the top LPS-responsive cAMP-regulating gene. We observed that PDE4b expression markedly increased at the time of peak injury (4 h) and then decreased to below the basal level during the resolution phase (24 h). Activation of transcription factor NFATc2 was required for the transcription of PDE4b in Mϕ. Inhibition of PDE4 activity at the time of peak injury, using intratracheal rolipram, increased cAMP levels, augmented the reparative AMϕ pool, and resolved lung injury. This response was not seen following conditional depletion of monocytes, thus establishing airspace-recruited PDE4b-sensitive monocytes as the source of reparative AMϕ. Interestingly, adoptive transfer of rolipram-educated AMϕ into injured mice resolved lung edema. We propose suppression of PDE4b as an effective approach to promote reparative AMϕ generation from monocytes for lung repair.

INTRODUCTION

Macrophages (Mϕ) are known to be the most plastic immune cell type, laying on a continuum of phenotypic states, from being entirely antiinflammatory (M2) to entirely proinflammatory (M1), in a balance crucial for coordinating the initiation, progression, and ultimate resolution of inflammatory injury (1) . The alveolar macrophages (AMϕ) (CD11c⁺/SiglecF⁺/CD11b⁻) are the most prominent resident sentinel Mϕ in the lung. Under basal conditions, AMϕ maintain the antiinflammatory niche in the lung by responding to environmental pollutants and pathogens encountered over the course of breathing (2). However, during pathological infections, AMϕ rapidly trigger the generation of proinflammatory cytokines and migration of neutrophils into the airspace using a repertoire of pathogen-associated molecular pattern receptors, namely, the lipopolysaccharide (LPS) receptor, Toll-like receptor 4 (TLR4) (3). AMϕ undergo apoptosis while fending off pathogens and can also acquire an inflammatory lineage, leading to sustained inflammatory signaling, loss of lung barrier function, and neutrophilic inflammation, hallmarks of acute lung injury (ALI) (4, 5). Thus, a reparative AMϕ pool must be continuously replenished to restore lung homeostasis while inducing host-defense.

During development, AMϕ are formed from fetal liver monocytes, which enter the alveoli after birth (6). AMϕ are then thought to be slowly replaced through local proliferation under basal conditions with minimal contribution from monocytes (7). However, following injury, airspace-recruited monocytes replenish the AMϕ pool by forming monocyte-derived Mϕ (Mo-Mϕ) (8). In this context, Mo-Mϕ can acquire either a proresolving or inflammatory phenotype depending on the lung niche (1, 9). For example, during pulmonary fibrosis, AMϕ were shown to be replaced by inflammatory Mo-Mϕ, leading to persistent lung damage (10). However, little is known about the molecular determinants of Mo-Mϕ differentiation into proresolving AMϕ, leading to tissue repair.

Cyclic AMP (cAMP) is a well-known second messenger, which plays a critical role in regulating the reparative function of several cell types present in the lung, including Mϕ (11, 12). Increased cAMP shifts Mϕ towards an antiinflammatory, M2 phenotype, in vitro (13, 14). We have previously shown that cAMP functions by suppressing activation of the calcium-dependent transcription factor, NFAT, in AMϕ, leading to the dampening of inflammatory cytokine production and resolution of lung injury (15). In addition, cAMP was shown to bind to the NLR family pyrin domain containing 3 (NRLP3), directly reducing inflammasome generation and inflammation (16). cAMP levels are regulated by a family of phosphodiesterases (PDE) that cleave cAMP into AMP (17). There are several unanswered questions concerning cAMP metabolism in the setting of lung injury, including how the cAMP level is regulated in AMϕ and whether deliberately augmenting the cAMP level in AMϕ can reprogram airspace-recruited Mo-Mϕ into the proresolution AMϕ pool in vivo.

To resolve these questions, we took advantage of a mouse that expresses GFP under the control of the Lysozyme-M promoter, giving all myeloid cells acquire a green color (18), and a model of rapidly resolving lung injury (post LPS challenge). Using panRNA sequencing, we identified PDE4b as the top LPS-responsive cAMP regulating gene. We show that LPS induces PDE4b expression at the time of peak injury via activation of the transcription factor NFATc2. PDE4b belongs to the PDE4 family, which specifically degrades cAMP (17). Intriguingly, we show that inhibition of PDE4b at the time of peak injury, using the well-established small molecule inhibitor, rolipram (17, 19), markedly increased the pool of reparative AMϕ derived from Mo-Mϕ, leading to rapid resolution of lung injury. Rolipram-educated AMϕ retained their antiinflammatory function because the adoptive transfer of these Mϕ into injured mice repaired the lung damage. Thus, this study establishes the principle that the programming of Mo-Mϕ into reparative AMϕ through suppression of PDE4b stimulates rapid lung repair.

METHODS AND PROCEDURES

Experimental Animals

All animal work was approved by the Institutional Animal Care and Use Committee of University of Illinois. CD11b-DTR (Stock No. 006000) and wildtype (Stock No. 000664) mouse strains were obtained from Jackson Laboratory (Bar Harbor, ME). LysM-GFP mice were provided by Dr. Klaus Ley (La Jolla Institute for Immunology, UCSD). Breeding colonies of all strains were maintained in a pathogen-free housing facility at the University of Illinois at Chicago. All mouse strains were of a C57Blk/6J background. Both male and female (6–8-wk-old) mouse were used in a pairwise manner throughout the study.

Drug Administration

Rolipram (Cat. No. R6520, Millipore Sigma, Burlington, MA) or PBS vehicle was administered via noninvasive i.t. (intratracheal) instillation. Rolipram was prepared as a stock solution in DMSO and further diluted to working concentration with sterile PBS. Diphtheria toxin isolated from Corynebacterium diphtheriae (Cat. No. D0564, Millipore Sigma) dissolved in sterile PBS was delivered via intraperitoneal injection. BrdU, 5-bromo-2′-deoxyuridine, (Cat. No. B5002, Sigma-Aldrich, St. Louis, MO) was dissolved in PBS (15 mg/mL) and 100 µL was injected intraperitoneally.

Induction and Assessment of Lung Vascular Permeability

To induce acute lung injury, mice were exposed to a nebulized 1 mg/mL solution of lyophilized Escherichia coli LPS (Cat. No. 2880, Sigma-Aldrich) dissolved in sterile saline for 50 min in an enclosed space. Lung edema and vascular permeability were quantified by measuring Evans Blue-tagged albumin extravasation (EBAE) and lung wet-to-dry weight ratio. For EBAE, 100 µL Evan’s blue albumin was injected retro-orbitally 45 min before the mouse was euthanized. Right lung lobes were extracted from control and LPS exposed mice and used for Evans blue albumin extravasation whereas the left lung lobe was excised and completely dried in an oven at 55°C overnight for calculation of lung wet-dry-weight ratio as previously described (9, 15).

Florescence-Activated Cell Sorting

Lungs were surgically harvested following perfusion of the heart with cold PBS. Lung tissue was minced and enzymatically digested with 1 mg/mL collagenase A (Cat. No. 10103586001, Roche, Indianapolis) for 50 min at 37°C. Digested tissue was washed twice following centrifugation at 1,250 rpm for 4 min and cells were resuspended in FACS buffer (1% BSA in PBS). A single cell suspension was prepared by forcing minced tissue through a narrow cannula as previously described (9). The washed cell suspension was passed through a 75-μm nylon filter. Red blood cells were lysed using RBC lysis buffer which was diluted 10× (Cat. No. 11814389001, Invitrogen). The cell suspension was then incubated with Fc blocking CD13/CD32 antibody (Cat. No. 14–0161-82, Invitrogen) for 30 min on ice followed by the addition of fluorochrome-labeled antibody cocktail for 30 min on ice. Dead cells were identified using 7AAD (Cat. No. 420404, Biolegend, San Diego, CA) staining. Samples were washed 2× with 1 mL FACS sample buffer and fixed using IC Fixation buffer (Cat. No. FB001, Invitrogen). Samples were analyzed using a CytoFlex flow cytometer (Beckman Coulter, Brae, CA). All data were processed using Flow Jo software (TreeStar, Inc., Ashland, OR). All antibodies used for flow cytometry were specific for mouse antigens and are listed in Table 1.

Table 1.

Fluorescently labeled antibodies used in flow cytometry experiments

Target	Clone No.	Fluorochrome	Company	Cat. No.	Host Species	Dilution	RRID
CD64	X54-5/7.1	PEcy7	Biolegend	139314	Mouse	1/200	AB_2563904
CD64	X54-5/7.1	APC	Biolegend	139306	Mouse	1/200	AB_11219391
CD11b	M1/70	APC	eBioscience	17-0112-82	Rat	1/200	AB_469343
CD11b	M1/70	ef450	eBioscience	48-0112-82	Rat	1/200	AB_1582236
CX3CR1	SAA011F11	PEcy7	Biolegend	149016	Mouse	1/70	AB_2565700
Ly6c	HK1.4	PERCPcy5.5	eBioscience	45-5932-82	Rat	1/400	AB_2723343
Ly6g	1A8	BV785	Biolegend	127645	Rat	1/200	AB_2566317
SiglecF	S17007L	PE	Biolegend	155506	Rat	1/400	AB_2750235
SiglecF	S17007L	APC	Biolegend	155508	Rat	1/400	AB_2750237
CD45	30F11	PEcy7	Biolegend	103114	Rat	1/200	AB_312979
BrdU	Bu20a	APC	Biolegend	339808	Mouse	1/20	AB_959349
CD62L	MEL-14	APC	Biolegend	104411	Rat	1/200	AB_313098

Bronchoalveolar Lavage

Bronchioalveolar lavage (BAL) was performed on euthanized mice using an 18-gauge needle, as previously described (9, 15). Briefly, a tracheotomy was performed, and 1 mL of cold PBS was slowly injected into the lungs, aspirated back into the syringe, and then collected. This procedure was repeated for a total of 4 mL of fluid. BAL was then centrifuged at 1,250 rpm for 10 min and the supernatant was discarded. Cell pellets were used in further experiments.

Noninvasive Intratracheal Instillation of Rolipram or Cells

Monocytes or macrophages suspended in 50 µL volume were endotracheally instilled in the airway. Briefly, mice were anesthetized by injecting ketamine and xylazine (100 mg/kg and 12 mg/kg) and hung in the supine position with the help of thread on the upper jaw teeth as previously described (9). The tongue was grasped with the help of forceps and pulled outside towards the forward direction after which 50 µL cell suspension or rolipram was injected into the oropharynx. Nostrils were closed for 5–10 s and observed for the inhalation of cells/drug suspension by the mice. The nostrils were then opened, and the animal returned to its cage.

Isolation and Culture of Bone Marrow-Derived Macrophages

Mouse bone marrow-derived macrophages (BMDM) were isolated from wildtype mice as described (9). Mouse femur and tibia were flushed with RPMI media containing 1% antibiotic/antimycotic, 10% fetal bovine serum (FBS), and 25 ng/mL M-CSF (Cat. No. 416-ML, R&D Systems, Minneapolis, IN) and cultured at 37°C in a 5% CO₂ incubator. On the third day, the media was replaced with fresh M-CSF free RPMI containing 10% FBS and cells were incubated for additional 2 days. For each experiment, BMDMs were serum-starved in RPMI media containing 1% FBS for 30 min before stimulation with LPS (1 µg/mL). For inhibitor experiments, serum-starved cells were preconditioned with 25 µM INCA-6 (Cat. No. 48040, Millipore Sigma) for 15 min before LPS stimulation.

Monocyte Isolation from Bone Marrow

Bone marrow was harvested as described above and incubated with magnetic microbeads and isolated using a magnetic separator, as detailed in Joshi et al. (9). Purified CD11b⁺ monocytes were isolated using CD11b magnetic microbeads (Cat. No. 130-049-601, Miltenyi Biotec, Cologne, Germany).

ELISA Assays

BAL fluid from each mouse was collected as described previously (15). The BAL fluid was directly poured onto plastic dishes. After 1 h, cells were vigorously rinsed with sterile PBS to remove nonadherent cells. cAMP levels were determined using ELISA (Cat. No. CA201-1KT, Sigma-Aldrich) following the manufacturer’s recommendation.

Immunoblotting

Cells were lysed in 2× Laemmli buffer. For each sample, 20 µL of lysate were run on SDS-PAGE gels and transferred to a nitrocellulose membrane. Membranes were Western blotted using primary antibodies as listed in Table 2, diluted in TBST with 3% BSA. Anti-rabbit (Cat. No. sc-2357, Santa Cruz Biotechnology) or anti-mouse (Cat. No. sc-516102, Santa Cruz Biotechnology) IgG-HRP (1:10,000 dilution) were used as secondary antibodies. β-actin was used as a loading control. Chemiluminescent signal was recorded on the ChemiDoc XRS Biorad Imager (Biorad Laboratories, Hercules, CA) and data were analyzed using ImageJ software (NIH, Bethesda, MD). All antibodies used for immunoblotting are listed in Table 2.

Table 2.

Unconjugated antibodies used in histology experiments

Target	Company	Cat. No.	Host Species	Dilution	RRID
PDE4b	Invitrogen	MA5-25677	Mouse	1/1,000	AB_2723449
PDE4d	Proteintech	12918-1-AP	Rabbit	1/1,000	AB_2161097
GFP	GenTex	GTX113617	Rabbit	1/50	AB_1950371
p-NFATc2 ser54	Invitrogen	44-944G	Rabbit	1/1,000	AB_2533803
NFATc2	Cell Signaling	4389S	Rabbit	1/1,000	AB_1950418
Actin	Thermo Fisher	MA5-15739-HRP	Rabbit	1/1,000	AB_2537667

PDE4b, phosphodiesterase 4b.

Quantitative Real-Time PCR

Total RNA was collected from the BMDM or flow-sorted macrophages using TRIzol reagent (Cat. No. 15596026, Invitrogen) as previously described (9, 15). RNA was quantified using Biodrop and reverse transcription reaction was carried to generate cDNA using the High-Capacity cDNA Reverse Transcription Kit (Cat. No. 4368814, Applied Biosystems, Foster City, CA,) according to the manufacturer’s instructions. Primer sequences are listed in Table 3.

Table 3.

Primer sequences used in RT-QPCR experiments

Gene	Forward Primer Sequence (5′–3′)	Reverse Primer Sequence (3′–5′)
Actin	CGTTGACATCCGTAAAGACCT C	AGCCACCGATCCACACAGA
PDE4a	CAAGCGCCAGAAGCAGAG	CATAGTCTTCAGGTCAGCCAGA
PDE4b	AATGTGGCTGGGTACTCACA	AAGGTGTCAGATGAGATTTTAAACG
PDE4c	ATGGGGACTTGATGTGTTCA	TCTTGAGGAGGTCTCGTTCC
PDE4d	CGTTTTCCGAATAGCAGAGC	TTTTAAACGTTTTTAACAAATCTCG
IL-6	AGTCCGGAGAGGAGACTTCA	TTGCCATTGCACAACTCTTT
iNOS2	TGCATGGACCAGTATAAGGCAAGC	GCTTCTGGTCGATGTCATGAGCAA
IFN-y	TGAACGCTACACACTGCATCTTGG	CGACTCCTTTTCCGCTTCCTGAG

All primer pairs are specific for mouse. PDE4a, phosphodiesterase 4a.

Immunohistochemistry and Microscopy

Lungs were harvested following perfusion with cold PBS and embedded in optimal cutting temperature compound (OCT) (Cat. No. 23-730-571, Fisher Scientific) and frozen at −80°C before cryosectioning by the UIC Histology Research Core (9). Each 12-micron section was fixed in 2% paraformaldehyde and stained with appropriate primary and secondary antibodies or fluorochrome tagged primary antibody. Lung sections and cytospin slides were analyzed using an LSM880 confocal microscope (Carl Zeiss) equipped with 63× Plan Apochromat oil objective (NA 1.4.) Hematoxylin/eosin staining was performed using Hema 3 Stat pack (Fisherbrand, Cat. No. 123-869) according to manufacturer’s instructions, slides were analyzed on an Echo Rebel hybrid visible light microscope (Echo, San Diego, CA).

Phagocytosis Assay

To assess phagocytic capacity, BAL from indicated mice was plated on 35 mm glass-bottom dishes containing RPMI media and 10% FBS. After 1 h, all nonadherent cells were removed and adherent cells were rinsed vigorously with PBS. Cells were incubated with fluorescently labeled latex beads (20) (Cat. No. L2778, Millipore Sigma) for 2 h after which cells were washed with PBS to remove residual beads and fixed with 2% paraformaldehyde solution. Phagocytosis was analyzed using an LSM880 confocal microscope (Carl Zeiss, Jena, Germany).

Luciferase Assay

Gaussia luciferase construct containing the PDE4b promotor was obtained from Genecopoeia (Rockville, MD) (Product ID: MPRM43245) (15). BMDM were transfected with 1.5 µg plasmid using Amaxa Nucleofector electroporation system (Cat. No. AAB-1001, Lonza, Basel, Switzerland) and stimulated with LPS and INCA-6. Cell lysates were analyzed using Secrete-pair dual luminescence assay kit (Genecopoeia, Cat. No. LF031) using the manufacturer’s protocol.

Oligo-Immunoprecipitation

Wildtype BMDM pretreated with INCA-6 and LPS as indicated were washed with PBS and lysed using RIPA lysis buffer. An equal amount of lysate protein (100–150 µg) was incubated with Poly IC compound (Sigma-Aldrich, Cat. No. I4023) (40 ng/µL) in 500 µL binding buffer (HEPES 12 mM, Tris 4 mM, KCl 60 mM, glycerol 5% of final volume, EDTA 500uM, DTT 1 mM, and protease inhibitor cocktail). Annealed biotinylated oligo (2 µm) was then added and the mixture was rotated at 4°C overnight. The next day, protein-oligo complexes were incubated with prewashed Streptavidin beads for an additional 4–6 h. Complexes were centrifuged, washed four times, and resuspended in 25–30 µL of 4× Laemmli buffer. These complexes were heated at 90–95°C for 10 min and Western blotted as described above. Oligo sequences are listed in Table 4.

Table 4.

Oligonucleotide sequences for oligo-immunoprecipitation experiment

	Sequence
Oligos (containing NFATc2 binding sites)
Oligo forward	GACTTGTCCTGAGAAGTGAAAAATAGCTTTCCAATCACAGCTAGATATATGATCTTGAAT
Oligo reverse	ATTCAAGATCATATATCTAGCTGTGATTGGAAAGCTATTTTTCACTTCTCAGGACAAGTC
Control oligos (no NFATc2 binding sites)
Oligo forward	GACTTGTCCTGAGAAGTGATAAAAAGCTTTCCAAGAACAGCTAGATATATGATCTTGAAT
Oligo reverse	ATTCAAGATCATATATCTAGCTGTGACCAAAAAGCAAAAAGTCACTTCTCAGGACAAGTC

RNA Sequencing and Bioinformatics

RNA from flow-sorted cells was collected and submitted to Northwestern University NUseq research core for Illumina NextSeq 50075SE High Output RNA sequencing and bioinformatics analysis. Analysis of the PDE4b promoter was done using Gene Runner (https://generunner.net) and the eukaryotic promoter database (https://epd.epfl.ch//index.php). These data have been deposited at the Gene Expression Omnibus under accession number GSE174532.

Statistical Analysis

Results are expressed as means ± SD from three to five independent experiments. One-way ANOVA followed by post hoc Tukey’s multiple comparisons test and unpaired parametric t test was used to compare groups using Graph Pad Prism v. 7.0 (GraphPad Prism Software, Inc, San Diego, CA). Data are represented as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 to denote significant difference between treatment and vehicle control groups at the same timepoint, whereas #P < 0.05, ##P < 0.01, ###P < 0.001, and ####P < 0.0001 to denote significant difference between baseline and after LPS exposure.

RESULTS

LPS Induces a Dynamic Shift in Lung AMϕ during Injury and Repair

We have shown that nebulized LPS induces inflammatory lung vascular injury within 4 h and that the injury resolves within the next 24 to 48 h (21). Thus, we implemented this model of ALI in LysM-GFP mice (18), in which all myeloid cells express GFP (green), and gating strategy described in Fig. 1 and Fig. 2A to investigate dynamic alterations in lung-resident Mϕ phenotype over the course of injury and repair. As expected, LPS induced lung edema 4 h postinjury, which was sustained for up to 16 h and then resolved by 24 h (Fig. 2B). We found that AMϕ, operationally defined as GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺, accounted for ∼60% of total lung GFP⁺/Ly6g⁻/CD64⁺ cells under basal conditions (Fig. 2, C and D). We grouped the other pool of resident Mϕ, namely the interstitial Mϕ (IMϕ) and monocyte-derived Mϕ (Mo-Mϕ), as (CD11b⁺/SiglecF⁻) together and found them to constitute ∼40% of the total number of macrophages in naïve lungs (Fig. 2, C and E). The AMϕ pool sharply declined to ∼20% at 16 and 24 h (Fig. 2, C and D). We also noted that at 16 h (overlapping with the injury phase), the SiglecF⁺ population bifurcated into two subpopulations, namely, CD11b^med and CD11b^high (16 h) (Fig. 2, C and F). In the resolution phase, i.e., 24 h post LPS challenge, the SiglecF⁺ pool shrank to ∼20% of the total Mϕ pool and almost entirely switched to CD11b⁺ expression (Fig. 2, C, D, and F). IMϕ and Mo-Mϕ remained SiglecF⁻/CD11b⁺ in all phases of injury and repair and proportionally increased to ∼80% of the total number of Mϕ at 16 h and 24 h (Fig. 2, C and E). To examine the distinction between these two populations, we next flow-sorted AMϕ from mice 16 h after LPS challenge and found that CD11b^lo and CD11b^hi AMϕ were structurally distinct, with most of the CD11b^hi AMϕ proving to be smaller and rounder in shape and higher in cytoplasmic density than CD11b^lo AMϕ, which were bigger than CD11b^hi AMϕ (arrow) and had a well-defined cytoplasm (Fig. 2G).

Figure 1.

Gating strategy for assessing pulmonary macrophage populations in a LysM-GFP mouse. Single cell lung suspension is first gated to remove debris and doublets. Cells that are positive for 7AAD (dead cells) are excluded, whereas 7AAD negative cells are gated onto GFP for selecting myeloid cells. Neutrophils (Ly6g⁺) cells are next excluded from myeloid cells whereas CD64⁺ Mϕ and monocytes are further gated onto CD11b and SiglecF to isolate AMϕ (SiglecF⁺) and IM/Mo-Mϕ (SiglecF⁻/CD11b⁺). AMϕ, alveolar macrophages; LysM-GFP, lysozyme-M promoter green fluorescent protein.

Figure 2.

Tracing the phenotypic shift of pulmonary macrophages following LPS-induced acute lung injury. A: schematics of assessment of lung injury and Mϕ phenotype following LPS-induced injury. B: LysM-GFP mice were exposed to nebulized LPS (1 mg/mL) for 45 min. Lungs were harvested at indicated times and lung injury was determined by measuring lung wet-dry ratio. n = 5 mice/group. FACS experiments were performed 5 times independently. C–F: lung cells were stained with SiglecF, CD11b, CD64, and Ly6g antibodies and gated as described in Fig. 1. Flow cytometry analysis was performed to determine AMϕ or IMϕ/Mo-Mϕ population. n = 5 mice/group. A representative dot plot of pulmonary Mϕ subpopulations at indicated time is shown in C. D: plot shows AMϕ number (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) as % of total Mϕ pool (GFP⁺/Ly6g⁻/CD64⁺). E: plot shows IMϕ/Mo-Mϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁻/CD11b⁺) as a percentage of total Mϕ pool. F: histogram shows CD11b surface expression on AMϕ at indicated time points following LPS challenge, histogram representative of five independent experiments. G: representative image of flow sorted AMϕ were cytospun onto glass bottom coverslips. Cells were stained with hematoxylin and eosin and analyzed using a light microscope. Arrows indicate structurally distinct alveolar macrophages. Scale bar, 70 µm. Data in B, D, and E are represented as individual scatter along with means ± SD, whereas ###P < 0.001 and ####P < 0.0001 denote significant difference between LPS exposed and 0 h unexposed mice. AMϕ, alveolar macrophages; IMϕ, interstitial Mϕ; LPS, lipopolysaccharide; LysM-GFP, lysozyme-M promoter green fluorescent protein.

LPS Induces PDE4b Expression in AMϕ at the Time of Peak Injury

To assess how the phenotypic shift in Mϕ is related to alteration in their transcriptomics, we flow-sorted GFP⁺/Ly6g⁻/CD64⁺ lung cells at baseline, time of peak injury, and resolution phase and performed panRNA sequencing. We focused on genes associated with cAMP signaling due to its role in regulating antiinflammatory and proresolving macrophage signaling (14, 22). We found PDE4b, which specifically degrades cAMP (17), to be one of the most highly expressed cAMP-related genes in Mϕ (Fig. 3A). Interestingly, PDE4b expression sharply increased at the time of peak injury and then dropped below baseline during the resolution phase (Fig. 3A). We validated these findings using flow-sorted AMϕ and IM/Mo-Mϕ. As shown in Fig. 3, B and C, at the time of peak injury, PDE4b mRNA increased ∼3-fold in AMϕ but only ∼1.8 fold in IM/Mo-Mϕ. In both populations, PDE4b expression decreased to below basal levels within 16–24 h. There are four members of the PDE4 gene family (13): PDE4a, PDE4b, PDE4c, and PDE4d. However, PDE4b has been indicated to play an essential role in inflammatory lung disease (23). Indeed, a comparison of the RT-QPCR critical threshold (CT) values, where lower CT values indicate greater mRNA expression, for each PDE4 gene in AMϕ by RT-QPCR confirmed PDE4b to be a highly expressed PDE4 family member (Fig. 3D). Whereas PDE4d was modestly expressed in AMϕ, we failed to detect PDE4a and PDE4c mRNA (Fig. 3D). In IMϕ and Mo-Mϕ, we only detected PDE4b mRNA but not PDE4a, PDE4c, or PDE4d (Fig. 3E). We also performed Western blotting of AMϕ using PDE4b and PDE4d antibodies. As expected, we failed to detect PDE4d expression in AMϕ (Fig. 3F) corroborating mRNA findings shown in Fig. 3D.

Figure 3.

LPS upregulates phosphodiesterase 4b in alveolar macrophage during injury. A: heatmap of cAMP-regulating genes in GFP^+/Ly6g⁻/CD64⁺ Mϕ using bulk RNA sequencing at indicated time points after LPS inhalation. PDE4b (arrow) is the highly expressed and LPS responsive gene during lung injury. B and C: validation of PDE4b mRNA expression in flow sorted AMϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) (B) and IM/Mo-Mϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁻/CD11b⁺) (C) using RT-QPCR (n = 3 sort/group), using β-actin as a housekeeping gene. D and E: plots show CT values of PDE4 genes in sorted AMϕ (D) and IM/Mo-Mϕ (E) using RT-QPCR (n = 3 sort/group). Each sorting experiment was performed independently. F: Western blot showing PDE4b and PDE4d protein expression in AMϕ pooled from 5 mice (n = 3). Data in B–E are shown as means ± SD and represented as #P < 0.05 and ##P < 0.01 to denote significant difference between LPS exposed and unexposed cells. AMϕ, alveolar macrophages; cAMP, cyclic AMP; IMϕ, interstitial Mϕ; LPS, lipopolysaccharide; PDE4b, phosphodiesterase 4b.

NFATc2 Induces PDE4b Expression during Lung Injury

Although PDE4b is well studied in the setting of various lung diseases (24–26), the mechanisms regulating PDE4b expression during lung injury are unknown. We, therefore, analyzed the PDE4b promotor to identify key transcription factors that may induce PDE4b during injury. We found that the PDE4b promoter contains multiple binding sites for the Ca²⁺-dependent transcription factor NFAT near the transcription start site (Fig. 4A). We recently showed that LPS increases intracellular Ca²⁺, which then activates NFATc2 (15). Activated NFATc2 then cooperates with NF-κB to augment inflammatory signaling in AMϕ, thereby impairing the resolution of lung injury. Hence, we tested the possibility that NFATc2 synthesizes PDE4b to counter the antiinflammatory activity of cAMP generated during injury. For this purpose, we used bone marrow-derived macrophages (BMDM) and the small molecule inhibitor INCA-6, which inhibits NFAT transcriptional activity (27). LPS increased NFATc2 ser54 phosphorylation, a measure of NFATc2 activity in the nucleus (27), within half an hour, whereas preconditioning cells with INCA-6 for 15 min before LPS exposure prevented NFAT phosphorylation at phospho-serine54 (Fig. 4, B and C). LPS also increased PDE4b protein expression in association with increase in NFATc2 activity in an INCA-6 sensitive manner (Fig. 4, B and D). Moreover, we also found that LPS also increased PDE4b mRNA and protein expression when BMDM were stimulated over a longer time course (Fig. 4, E–G). However, LPS failed to induce PDE4b mRNA and protein expression in BMDM-treated INCA-6 (Fig. 4, E–G).

Figure 4.

NFAT upregulates PDE4b expression in macrophages. A: representation of mouse PDE4b promoter region with three NFAT binding sites. Arrow indicates transcription start sites. B–D: BMDM were preincubated with 25 µM INCA-6 for 15 min followed by addition of 1 µg/mL LPS for another 15 min. NAFTc2 activity was determined using phosphoNFATc2 Ser54 antibody. In addition, PDE4b protein expression was determined using PDE4b antibody. B shows a representative immune blot whereas plots C and D show fold increase in NFATc2 phosphorylation (p-NFATc2) and PDE4b protein expression taking no LPS (-) as the control, experiments performed three times independently, n = 3. E–G: BMDMs were exposed to LPS for indicated timepoints and PDE4b expression at the level of protein (E and F) and mRNA (G) was determined using Western blotting (E and F) or RT-QPCR (G) (n = 3), β-actin was used as a loading control in E. In G, PDE4b mRNA was normalized taking β-actin mRNA was taken as a housekeeping gene. E shows a representative blot whereas F shows fold increase in PDE4b expression normalized against no LPS (-). Each experiment was repeated three times independently. H: lysates from unstimulated or LPS stimulated BMDM treated without or with INCA-6 were incubated with biotinylated PDE4b promoter oligos containing NFATc2 binding sites after which complexes were pulled down using streptavidin beads. In parallel, lysates were incubated with oligos containing mutated NFATc2 binding sites and immunocomplexes were pulled down. Controls included lysates incubated with IgG and PDE4b oligos. A representative Western blot is shown from experiment that was repeated three times independently. I: densitometry of NFAT binding to PDE4b promoter at indicated times following LPS stimulation of BMDM (n = 3). J: BMDM were transfected with empty vector or PDE4b luciferase promoter construct. After 48 h, cells were exposed to LPS after with or without preexposure with INCA-6. The experiments were repeated three time independently. Data are shown as means ± SD and represented as *P < 0.05 and ****P < 0.0001 to denote significant difference between treatment conditions at the same timepoint whereas ####P < 0.0001 denotes significant difference between a treatment condition and unstimulated controls. BMDM, bone marrow-derived macrophages; LPS, lipopolysaccharide; PDE4b, phosphodiesterase 4b.

We next used oligo immunoprecipitation and Western blotting to investigate whether NFATc2 directly binds to the PDE4b promoter following LPS challenge. To this end, BMDM were left unstimulated or stimulated with LPS for the indicated times, after which cell lysates were incubated with biotinylated oligonucleotide sequences matching the PDE4b promoter or oligonucleotide sequences lacking NFATc2 bindings sites. We found that LPS increased the binding of NFATc2 to the PDE4b promoter (Fig. 4, H and I). However, this binding activity was not observed in cells pretreated with INCA-6 or PDE4b promoter constructs lacking the NFATc2 binding sites (Fig. 4, H and I).

To further corroborate these findings, we transfected BMDM with a PDE4b luciferase promoter construct and determined the increase in promoter activity induced by LPS. Again, we found that LPS markedly increased NFATc2 promoter activity, but this response was abolished in BMDM pretreated with INCA-6 (Fig. 4J).

Inhibition of PDE4b Augments AMϕ Pool Truncating Inflammatory Lung Injury

CD11b expressing AMϕ have been linked to persistent inflammatory states in both mice and humans (28–30). However, we showed that at 24 h, 80% of the AMϕ population remained CD11b^high (Fig. 2, C and D) despite the decline in PDE4b expression and resolution of lung injury. We, therefore, surmised that the SiglecF⁺ CD11b⁺ pool also contained a cAMP-sensitive pool of reparative AMϕ. To test this possibility, we inhibited PDE4 activity using rolipram, a well-established inhibitor of PDE4 (31), in a therapeutic model of lung injury.

We determined whether preventing cAMP degradation would skew the SiglecF⁺/CD11b⁺ pool to acquire a homeostatic AMϕ signature (SiglecF⁺/CD11b⁻) and whether this phenotypic shift would also promote resolution of lung injury. Thus, we administered rolipram intratracheally (2.5 mg/kg) 2 h after administration of LPS and assessed AMϕ phenotype and lung edema (Fig. 5A). We found that the AMϕ population (SiglecF⁺) constituted a larger percentage of the total resident Mϕ pool in rolipram-treated lungs, increasing by 50% at both 16 h and 24 h but not at 4 h (Fig. 5, B and C). Rolipram given alone had no effect on AMϕ phenotypic shift (Fig. 5, D and E). Moreover, rolipram treatment after LPS exposure shifted the AMϕ pool from SiglecF⁺/Cd11b⁺ to SiglecF⁺/Cd11b⁻ (Fig. 5, F and G). To determine if this increase in AMϕ population was due to proportional changes in AMϕ versus IMϕ/Mo-Mϕ, we quantified the absolute number of both populations in the lung at each timepoint. Although rolipram did not affect the number of IMϕ/Mo-Mϕ, it did increase the absolute AMϕ number in the lung at 16 and 24 h post LPS challenge (Fig. 5, H and I).

Figure 5.

Inhibition of PDE4 restores CD11b⁻ alveolar macrophages in the lungs at peak of injury. A: schematics of assessment of AMϕ phenotype and lung injury following LPS inhalation and noninvasive intratracheal administration of 2.5 mg/kg rolipram or vehicle. B: a representative dot plot following gating of GFP⁺/Ly6g⁻/CD64⁺ cells. C: AMϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) as a proportion of the total lung Mϕ pool (GFP⁺/Ly6g⁻/CD64⁺) with and without rolipram treatment. Experiments were repeated four times independently, n = 5 mice/group. D and E: AMϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) as a proportion of the total lung Mϕ pool (GFP⁺/Ly6g⁻/CD64⁺) from mice receiving rolipram alone. Representative dot plot (D) with quantification (E). Experiments repeated twice independently, n = 4 mice/group. F: percent change in CD11b⁻ AMϕ following rolipram treatment. G: representative histogram comparing shift in CD11b cell-surface expression on AMϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) following rolipram treatment at each timepoint. Experiments were performed 5 times independently, n = 5 mice/group. H and I: quantitation of total number of AMϕ (H) or IM/Mo-Mϕ (I). Experiments were performed four times independently, n = 4 mice/group. Data are shown as means ± SD and represented as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 to denote significant difference between treatment and vehicle control groups at the same timepoint. AMϕ, alveolar macrophages; LPS, lipopolysaccharide; PDE4, phosphodiesterase 4.

We next determined if the shift in AMϕ towards the CD11b⁻ AMϕ lineage reverses lung injury. We first determined cAMP levels in sorted AMϕ following LPS challenge. We found that rolipram transiently increased cAMP levels in AMϕ by 3-fold within 4 h, which declined to control levels at 16 h (Fig. 6A). At 4 h, rolipram-treated mice developed lung injury like control mice post LPS challenge (Fig. 6, B and C). Intriguingly, at 16 h, mice receiving rolipram showed an 80% decrease in lung edema formation and albumin accumulation after LPS challenge relative to control mice (Fig. 6, B and C). Rolipram treatment alone had no effect on edema formation (Fig. 6D). Inhibition of PDE4b also massively decreased neutrophil extravasation in the alveolar space at 16 h (Fig. 6, E and F) without affecting the percentage of peripheral lung neutrophils (Fig. 6G). Also, rolipram-treated AMϕ acquired the antiinflammatory lineage as revealed by markedly suppressed expression of iNOS, IL-6, and IFNγ (Fig. 6H). Because AMϕ promote tissue repair by phagocytizing and clearing debris (32), we also assessed whether rolipram treatment enhances AMϕ phagocytic function. To do this, we exposed AMϕ isolated from vehicle or rolipram-treated lungs to fluorescently labeled latex beads and after 45 min determined the phagocytic index. We found that rolipram increased AMϕ phagocytic function by ∼1.5 fold (Fig. 6, I and J). These results indicate that boosting cAMP levels at the time of peak injury shifts AMϕ into the antiinflammatory and CD11b^- lineage, thus accelerating the resolution of lung injury.

Figure 6.

Inhibition of PDE4 rapidly resolves LPS-induced lung vascular inflammatory injury. A: plot shows changes in cAMP levels in AMϕ post LPS and rolipram treatment determined as described in methods. (n = 3). B and C: mice exposed to nebulized LPS (1 mg/mL) for 50 min received rolipram as shown in Fig. 4A. One hour before the mice were euthanized at indicated times, Evans blue-labeled albumin was injected retro-orbitally into each mouse. Lung vascular inflammatory injury was determined by measuring lung wet-dry ratio (B) and albumin influx (C). The plot shows individual values along with means ± SD. Experiments were performed five times independently, n = 5 mice/group. D: lung wet to dry weight ratio from mice receiving intratracheal rolipram alone. Means ± SD. Experiments were performed twice independently, n = 4 mice/group. E and F: neutrophil count was performed (per field) on hematoxylin and eosin-stained BAL fluid from the indicated conditions. The plot shows individual values with means ± SD. Scale bar, 120 µm. Experiments were performed three times independently. G: dot plot shows neutrophils (GFP⁺/Ly6g⁺) in LPS exposed lungs with or without rolipram treatment as a percentage of total GFP⁺ cells (n = 5 mice/group) in whole lung. H: mRNA expression of proinflammatory genes in isolated AMϕ using RT-QPCR (n = 3). Experiments were performed twice independently. β-actin was used as an internal control. mRNA expression is shown as fold change following rolipram treatment. I and J: AMϕ were isolated from bronchoalveolar lavage and allowed to adhere on glass bottom dishes for 1 h. Cells were washed vigorously to remove unadhered cells. Fluorescently labeled latex beads were added at a ratio of 10:1 beads/AMϕ. Cells were fixed after 2 h and images were acquired using a confocal microscope. Phagocytosis index was measured as the average number of beads/cell in each visual field (5 fields/condition, ∼10 cells/field, taken from two independent experiments). Scale bar, 25 µm. Data are represented as individual scatter along with means ± SD with *P < 0.05 denoting significant difference between treatment and vehicle controls. Data are shown as means ± SD and represented as *P < 0.05 and ****P < 0.0001 to denote significant difference between treatment and vehicle control groups at the same timepoint and ###P < 0.001 and ####P < 0.0001 to denote significant difference between experimental conditions and basal cells. AMϕ, alveolar macrophages; BAL, bronchioalveolar lavage; cAMP, cyclic AMP; LPS, lipopolysaccharide; PDE4, phosphodiesterase 4.

Adoptive Transfer of cAMP-Programmed Alveolar Macrophages Resolves Lung Injury

To examine whether the cAMP-programmed AMϕ population could be used as a cell-therapy in a mouse model of ALI, we adoptively transferred rolipram-educated AMϕ 2 h after exposing WT mice to LPS (Fig. 7A) (9, 15). To this end, we flow-sorted AMϕ from LysM-GFP mice having received vehicle control (CC) or rolipram (RC) 16 h post LPS injury. Sorted AMϕ (2 × 10⁵ cells) were adoptively transferred into a wildtype mouse by the intratracheal route 2h after LPS inhalation (Fig. 7A). Mice receiving no cells (NC) with or without LPS were used as controls. Lung injury was assessed 16 h after adoptive transfer of AMϕ. We found that adoptive transfer of cAMP-programmed AMϕ promoted the resolution of lung injury, whereas this response was not observed in mice receiving no cells or vehicle alone (Fig. 7B). We confirmed the presence of adoptively transferred LysM-GFP AMϕ by immunostaining lung sections with SiglecF antibody (Fig. 7C). These findings indicate that rolipram-educated AMϕ could repair lung damage upon transplantation in a diseased mouse.

Figure 7.

Adoptive transfer of rolipram educated alveolar macrophages resolves lung injury in injured mice. A: schematic of AMϕ adoptive transfer. LysM-GFP mice exposed to LPS were treated without or with rolipram. At 16 h post LPS challenge, AMϕ were isolated via flow-sorting from vehicle (control cells, CC) or rolipram-treated (RC) lungs. These AMϕ were adoptively transferred into nonGFP mice that had been exposed to LPS 2 h before receiving cells, control mice received no cells (NC). B: lung edema was measured at 14 h post adoptive transfer of indicated cells, or 16 h post initial LPS exposure. The plot shows individual values along with means ± SD. Experiments were performed three times independently, n = 5. C: lung from uncolored mice receiving LysM-GFP flow sorted AMϕ were sectioned. Tissue was stained with DAPI and labeled antibodies for GFP and SiglecF to establish that LysM-GFP AMϕ sorted cells were present in lungs 14 h after adoptive transfer. Scale bar, 20 µm. Image representative of two independent experiments. Data are represented as *P < 0.05 and ***P < 0.001 to denote significant difference between treatment and vehicle control groups at the same timepoint. AMϕ, alveolar macrophages; LPS, lipopolysaccharide; LysM-GFP, lysozyme-M promoter green fluorescent protein.

Recruited Monocytes Are the Source of Proresolving AMϕ

AMϕ originate from embryonic precursors that self-maintain throughout their entire lifespan by proliferation (6). Evidence also indicates that during acute lung injury the AMϕ pool can be replaced by monocytes recruited into the airspace (8). We, therefore, assessed if rolipram induces proliferation of AMϕ or stimulates differentiation of recruited monocytes into new AMϕ. We injected BrdU into mice receiving vehicle or rolipram 14 h before euthanizing them and assessed AMϕ proliferation using a labeled BrdU antibody. As shown in Fig. 8, A and B, rolipram did not affect AMϕ proliferation, thus ruling out AMϕ expansion as the cause of the increase in AMϕ number.

Figure 8.

Rolipram programs monocytes differentiation into reparative alveolar macrophages. A and B: LysM-GFP mice were injected with BrdU 2 h after LPS inhalation. Lung cells were then stained with labeled anti-BrdU antibody and BrdU incorporation was determined in AMϕ (GFP⁺/Ly6g⁻/CD64⁺/SiglecF⁺) at 16 h was using FACS analysis. Histogram (A) shows BrdU⁺ AMϕ in control or rolipram administered mice to show the proportion of proliferating cells. Experiments were performed twice independently, n = 3. C: schematic of rolipram treatment in CD11b-DTR mice. CD11b-DTR mice were exposed to LPS as above. After 2 h, mice received DT (25 ng/g of mouse intraperitoneally). Simultaneously mice were injected with CD11b⁺ monocytes harvested from LysM-GFP mouse bone marrow intravenously along with rolipram as above. Lung edema was assessed at 16 h. D: FACS dot plot of CD45⁺/Ly6g⁻/CD62L⁻/CD11b⁺/CX3CR1⁺/Ly6c⁺ cells in CD11b-DTR mouse lungs at 16 h with and without DT injection showing depletion of monocytes by DT. Figure representative of 3 independent experiments. E: lung edema determined by measuring lung wet-to-dry lung weight ratio. n = 5 mice/group. F: FACS dot plot of CD45⁺/CD64⁺ BAL cells isolated after 16 h from CD11b-DTR mice receiving LPS and DT, with and without intravenous. LysM-GFP⁺ monocytes. FACS dot plots are representative of 3 independent experiments. G: BAL cells obtained from monocyte depleted mice following PBS or adoptive transfer of LysM-GFP monocytes were stained with SiglecF and CD11b antibodies to confirm AMϕ phenotype (merge yellow). A representative image is shown from experiments that were performed twice. Scale bar, 50 µm. Data are shown as means ± SD and represented as ***P < 0.001, and ****P < 0.0001 to denote significant difference between treatment and vehicle control groups at the same timepoint. AMϕ, alveolar macrophages; BAL, bronchioalveolar lavage; LPS, lipopolysaccharide; LysM-GFP, lysozyme-M promoter green fluorescent protein.

To determine whether monocytes contribute to the proresolution AMϕ pool after rolipram treatment, we used a mouse model that expresses the human diphtheria-toxin receptor (DTR) under the control of the macrophage-monocyte-specific CD11b promotor (33). In these mice, diphtheria toxin (DT) administration leads to a rapid depletion of monocytes (referred to as Mϕ^dep mice) without altering neutrophils, which also express CD11b (9). Thus, we administered DT to CD11b-DTR mice 2 h after initiation of LPS inhalation followed by intratracheal delivery of rolipram (Fig. 8C). As expected, DT induced depletion of monocytes (Fig. 8D) but had no effect on neutrophils (data not shown). DT also did not alter the gain of CD11b expression in AMϕ (data not shown). Interestingly, rolipram failed to resolve lung injury in Mϕ^dep mice (Fig. 8E). However, adoptive transfer of DT-resistant CD11b⁺ monocytes, harvested from LysM-GFP bone marrow, into Mϕ^dep mice at the same timepoint as DT and rolipram administration resolved injury (Fig. 8E). We also collected BAL from mice receiving DT and intravenous CD11b⁺ LysM-GFP⁺ monocytes at 16 h and quantified adoptively transferred cells using FACS. We show that BAL contains 6% recruited macrophages (GFP⁺/CD64⁺ cells). Interestingly, of these GFP⁺ macrophages in the airspace, ∼14% also acquire SiglecF, whereas the rest remained CD11b⁺. Resident AMϕ (nonGFP⁺) cells under the same conditions amounted to ∼90% (Fig. 8F). In addition, confocal analysis of BAL cells confirmed the presence of GFP⁺/SiglecF⁺/CD11b⁺ cells in the airspace (Fig. 8G). These findings demonstrate that enhancing cAMP levels mobilizes and programs monocytes toward a reparative AMϕ phenotype.

DISCUSSION

The current study addressed four fundamental aspects of AMϕ biology in a mouse model of rapidly resolving lung injury. First, PDE4b was identified as the predominant cAMP regulating enzyme expressed in lung resident macrophages and was shown to be upregulated in AMϕ at the peak of lung injury following LPS challenge via the Ca²⁺-responsive transcription factor NFATc2. Second, we showed that this increase in PDE4b expression was associated with a shift of the AMϕ population toward more CD11b⁺ AMϕ. Third, cAMP rose in AMϕ when PDE4b was inhibited, thus confirming the physiological relevance of this PDE4 family member. Fourth, selective inhibition of PDE4b in AMϕ stimulated reparative AMϕ generation in a manner dependent on recruitment and reprogramming of monocytes and resolved lung injury in vivo. Thus, the data demonstrate a key role of PDE4b activity in regulating the in vivo programming of Mo-Mϕ into reparative AMϕ generation and point to the targeting of PDE4b activity to rapidly augment the generation of immunosuppressive and reparative AMϕ capable of repairing lung damage endogenously or after adoptive transfer.

In the naïve lung, AMϕ are antiinflammatory and exhibit various phenotypic and genetic markers that are distinct from those of IMϕ or Mo-Mϕ. However, evidence indicates that during lung injury such a distinction between AMϕ and other lung macrophages is seldom achievable because AMϕ and other Mϕ lie along a phenotypic continuum with complementary functions (28, 34). In this context, we showed that AMϕ acquire the CD11b⁺ lineage and became inflammatory, leading to increased lung vascular injury, as described in a number of other studies (28, 35, 36). However, we also showed that a significant pool of AMϕ remain CD11b⁺ despite the return of lung fluid-homeostasis to the basal state. Given that CD11b⁺ AMϕ were not different from normal subjects in patients suffering from COPD, one possibility may be that CD11b⁺ AMϕ surface expression need not be interpreted as definitively “inflammatory” (37). Another possibility may be that the subpopulation of CD11b⁻ AMϕ present during injury (20% of the total) was sufficient to restore the antiinflammatory lung niche.

To resolve this conundrum, we focused on mechanisms regulating the AMϕ proresolving lineage. We recently showed that cAMP is an important second messenger that switches proinflammatory macrophages into antiinflammatory macrophages (15). In addition, elevation of cAMP has been shown to suppress several lung diseases ranging from asthma to lung injury (26, 38, 39). Our RNAseq data show that PDE4b is the most abundantly expressed cAMP degrading phosphodiesterase in lung macrophages even under basal conditions, indicating that cAMP levels are finely tuned in lung macrophages for host-defense functions that arise during normal breathing (40). We also showed that PDE4b expression increased during injury, but then declined to below basal level during the resolution phase. In line with this concept, we showed that, although LPS increased cAMP levels in sorted AMϕ by a factor of 2, inhibition of PDE4b increased the cAMP level by a factor of ∼5. However, we showed that the 5-fold increase in cAMP level did not alter the development of lung injury at 4 h but rather induced rapid resolution of lung injury at 16 h. Resolution of lung injury was accompanied by marked attenuation of neutrophil influx into the airspace and cytokine generation by AMϕ. Importantly, we showed that inhibition of PDE4b subverted the AMϕ lineage from CD11b⁺ to CD11b⁻. Much like the findings reported here, several studies have shown that PDE4b is expressed in immune cells such as eosinophils, monocytes, and neutrophils, and inhibition of PDE4 suppresses lung injury, COPD, and airway responsiveness (38, 39). However, whether PDE4b controls the AMϕ lineage remains unclear. To our knowledge, we are the first to define a novel role of PDE4b in directly controlling AMϕ lineage. We demonstrate that PDE4b expression and thus activity controls the retention of the CD11b⁻ AMϕ lineage, leading to resolution of lung injury and inflammation. This is significant because current doctrine tends to regard macrophage polarization, studied mostly ex vivo, as the basis of AMϕ inflammatory-antiinflammatory lineage.

How then does LPS increase PDE4b expression to drive AMϕ into the inflammatory lineage, thus leading to lung injury? It is known that LPS via TLR4 induces proinflammatory cytokine generation through NF-κB (15, 41). However, our studies showed that inhibition of NFAT using a small molecule inhibitor suppressed PDE4b expression. Furthermore, using an oligonucleotide ChIP assay as well as PDE4b luciferase promoter activity, we showed that the binding of NFAT, specifically NFATc2, to the PDE4b promoter was required for LPS induction of PDE4b transcription. NFAT is a Ca²⁺-dependent transcription factor well known to regulate T cell function (42). However, we have shown that LPS activates thrombin generation in Mϕ which then induces Ca²⁺ entry through the TRPV4 channel to promote NFAT activity. Activated NFAT promotes NF-κB binding to cytokine promoters leading to increased inflammatory cytokine generation and thereby impairs resolution of inflammation (15). In the current study, we showed that PDE4b inhibition at the time of peak lung injury augmented cAMP levels, which then suppressed AMϕ inflammatory function. LPS induction of NFATc2 activity and thereby PDE4b transcription in AMϕ could explain the observed alteration in cAMP levels in AMϕ and the increase in inflammatory signaling. Thus, activation of the NFATc2-PDE4b circuit ‘tunes’ the AMϕ inflammatory lineage in response to LPS.

PDE4b inhibition affects multiple cell types in the lungs (43). One possibility is that the observed antiinflammatory AMϕ lineage observed after PDE4b inhibition may be the result of suppression of PDE4b activity in other cell types. However, we showed that adoptive transfer of rolipram-treated AMϕ but not vehicle-treated AMϕ repaired lung damage in LPS-exposed mice. We also showed that rolipram instructed AMϕ were more phagocytic. We inferred from these findings that PDE4b inhibition not only mobilizes reparative AMϕ during injury but also instructs these AMϕ to retain their reparative capacity in a transplantable model.

It is known that during acute lung injury, the resident AMϕ die off through induced programmed cell death (4, 44). One notion is that AMϕ can proliferate to restore AMϕ number during specific inflammatory conditions but the evidence in support of this idea is limited (45). Alternatively, new AMϕ can be generated by the differentiation of newly arrived monocytes in the airspace (9). We showed that rolipram did not increase AMϕ proliferation compared with control conditions, thus ruling out the proliferation of AMϕ as the basis for the restoration of AMϕ number after injury. CD11b-DTR mice have been previously utilized to conditionally ablate monocytes (9, 33), allowing us to demonstrate that recruited monocytes were responsible for generating reparative AMϕ. We showed that PDE4b inhibition by rolipram did not resolve lung injury in monocyte depleted mice (CD11b-DTR mice post DT challenge). However, adoptive transfer of bone marrow monocytes back into monocyte depleted mice rescued lung injury resolving effects of rolipram. Unlike previous studies which showed that recruited monocytes became inflammatory AMϕ in the lung (8), we demonstrated that airspace recruited monocytes formed reparative macrophages. BAL from mice receiving DT and intravenous CD11b⁺ LysM-GFP⁺ monocytes showed evidence of recruitment and programming of monocytes into macrophages. Interestingly, a minute pool of these GFP⁺ Mo-Mϕ (GFP⁺/CD64⁺ cells) also acquired SiglecF while remaining CD11b⁺. As rolipram failed to resolve injury in monocyte depleted mice, our study demonstrated that recruited monocytes form resolving AMϕ and may serve to create a niche for educating the existing resident AMϕ population to acquire an antiinflammatory phenotype. In the absence of the educational niche provided by monocytes, rolipram alone is insufficient to drive resident AMϕ to a proresolving phenotype.

A caveat in using CD11b-DTR mice is that DT can also deplete other CD11b⁺ myeloid cells such as neutrophils, IMϕ, and dendritic cells (DCs). We have previously used CD11b-DTR mice to deplete both IMϕ and monocytes without affecting neutrophils (9). Previous studies similarly demonstrated that DT had no effect on neutrophils (33). However, the effect of DT on CD11b⁺ dendritic cells remain controversial. Although some authors show no alteration in DC after DT injection (33), others have showed partial depletion of DC after DT (46). Thus, further studies will be required to fully delineate the time course and mechanism of conversion of airspace recruited monocytes/macrophages into AMϕ and the involvement of other lung myeloid populations such as DC and IMϕ in this process.

The basis of cAMP activation of reparative AMϕ generation during lung injury is not clear. cAMP is known to activate protein kinase A (PKA) (47). PKA phosphorylates cAMP response element-binding protein (CREB), leading to its nuclear transport and transcriptional activity. Thus, a likely scenario is that cAMP activates CREB, which then programs the subsequent differentiation of monocytes into reparative AMϕ. This notion is supported by studies in which CREB is shown to induce macrophage polarization into the antiinflammatory lineage in vitro (48–50).

In summary, the present study has identified NFATc2-PDE4b signaling in AMϕ as a key mechanism orchestrating extremely tight control of cAMP levels in AMϕ and their antiinflammatory lineage. We show that LPS activates PDE4b transcription via NFATc2, leading to the generation of inflammatory AMϕ, neutrophil influx, and inflammatory lung injury. Inhibition of PDE4b, therefore, increases cAMP, which mobilizes and instructs the differentiation of monocytes into reparative AMϕ, thus resolving lung vascular inflammatory injury. Owing to the pivotal role of PDE4b in monocyte differentiation, we suggest that inhibition of PDE4b in AMϕ is a potentially useful approach for the rapid generation of antiinflammatory AMϕ that subsequently prevent or resolve ALI.

GRANTS

This work was supported by National Institutes of Health National Heart, Lung, and Blood Institute Grants HL060678, HL84153, and HL137179. I. Rochford was partially supported by the T32 Lung Biology and Pathology Training Grant HL007829.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

I.R. and D.M. conceived and designed research; I.R., J.C.J., S.R., M.A., M.Z.A., L.Y., and S.B. performed experiments; I.R., J.C.J., S.R., M.A., M.Z.A., S.B., and D.M. analyzed data; I.R. and D.M. interpreted results of experiments; I.R., S.R., and D.M. prepared figures; I.R. and D.M. drafted manuscript; I.R., J.C.J., M.A., and D.M. edited and revised manuscript; I.R., J.C.J., S.R., M.A., Y.L., M.Z.A., S.B., and D.M. approved final version of manuscript.