Development of an Optimized Culture System for Generating Mouse Alveolar Macrophage–like Cells

This article is featured in Top Reads, p.1495

Tissue resident macrophages are the innate immune cells that play vital roles in pathogens defense and maintenance of tissue homeostasis. They are located in various anatomical tissues and have distinct biological features and functions (1). Alveolar macrophages (AMs) are the largest population of innate immune cells in the alveolar airspace (2). Due to their location, AMs directly contact inhaled airborne particles and microbes and become the first actors of the pulmonary immune response (3). AMs maintain normal surfactant homeostasis that is essential for gas exchange in the lung. Pulmonary alveolar proteinosis is a rare fatal syndrome characterized by the abnormal accumulation of alveolar surfactant (4). When the development or function of AMs is impaired, pulmonary alveolar proteinosis occurs, which may result in respiratory failure and increased risks for pulmonary infection and fibrosis (4).

Cytokines, such as GM-CSF and TGF-β, as well as transcription regulators, including PU.1 and peroxisome proliferator–activated receptor γ (PPAR-γ), are required for the development of AMs, whereas M-CSF is largely dispensable (5–9). AMs were originally proposed to have a fetal origin and are largely maintained by local self-renewal throughout life with minimal contribution from postnatal monocytes. However, emerging evidence demonstrates that the contribution of postnatal monocytes to AM compartment slowly increases with age (∼70% at 36 wk) (10). Indeed, the AM pool can be restored by the postnatal monocytes under the condition of lung infection, injury, or in irradiation chimeras (11–14).

A growing body of literature documented that embryonic and postnatal monocyte-derived AMs exhibited a very similar gene expression profile (with only ∼0.1% differential genes) (15, 16). However, they play distinct roles in lung diseases. For example, a recent study shows that influenza virus–induced postnatal monocyte-derived AMs, but not resident AMs, confer prolonged antiviral protection through IL-6 production (13). Moreover, some studies reported a profibrotic role of postnatal monocyte-derived AMs (14, 17). However, another study reported that postnatal monocyte-derived AMs were crucial for the resolution of pulmonary fibrosis through ApoE production (18). Thus, the roles of postnatal monocyte-derived AMs in tissue homeostasis and diseases remain unclear, and more research is needed to clarify this issue.

Although an immortalized AM cell line MH-S has been used in many studies of AM development and functions (19), these cells lack defining characteristics of primary AM surface markers, such as sialic acid–binding Ig-like lectin-F (Siglec-F) (20). Many studies also used primary AMs isolated from bronchoalveolar lavage fluid (BALF); however, a large number of mouse donors were required to obtain sufficient numbers of cells in those experiments. In addition, other studies also used primary AMs isolated from lung tissues; however, the isolation procedures for primary AMs involve enzyme digestion and flow cytometric sorting, which may change the features of those AMs. In this study, we have established, to our knowledge, a novel culture system for generating macrophages with the features of primary AMs, named AM-like cells, from cells in adult BM or fetal liver using a combination of GM-CSF, TGF-β, and PPAR-γ agonist rosiglitazone (referred to hereafter as GTR). Furthermore, we found that this culture system was also suitable for the expansion of BALF-derived AMs (BALF-AMs) in vitro. The AM-like cells derived from adult BM mostly resembled the cultured BALF-AMs and could be easily obtained in sufficient numbers that will allow genetic manipulations and functional studies of AMs at the molecular level. Thus, we provide a powerful tool for studying the development and functions of AMs.

Six- to eight-week-old C57BL/6 mice were housed under specific pathogen-free conditions at the Laboratory Animal Research Center of Tsinghua University. Cas9 knockin mice were a kind gift from Dr. Dong Chen (Institute for Immunology Tsinghua University). All animal experimental protocols were approved by the Institutional Animal Care and Use Committee of Tsinghua University.

PE-conjugated anti-mouse Siglec-F Ab was purchased from BD Biosciences (552126; San Jose, CA). PE-Cy7–conjugated anti-mouse CD11c (25-0114-82), allophycocyanin-conjugated anti-mouse F4/80 (17-4801-82), allophycocyanin-conjugated anti-mouse CD80 (17-0801-82) Abs, and 7-aminoactinomycin D (7-AAD) viability staining solution (00-6993-50) were purchased from eBioscience (San Diego, CA). BV605-conjugated anti-mouse CD11b (101257), allophycocyanin-Cy7–conjugated anti-mouse CD45 (103116), PE-conjugated anti-mouse CD86 (105007), FITC-conjugated anti-mouse MHC II (107606), and PE-conjugated anti-mouse MerTk (151506) Abs were obtained from BioLegend (San Diego, CA). Rosiglitazone (R2408) and LPS (L2880) were obtained from Sigma-Aldrich (St. Louis, MO). 15d-PGJ2 (18570) and 10-Nitrolinoleate (10037) were obtained from Cayman (Ann Arbor, MI). Murine GM-CSF (96-315-03-100), murine M-CSF (315-02), murine IL-3 (213-13), murine IL-6 (216-16), murine stem cell factor (250-03), and human TGF-β cytokines (100-21c) were obtained from PeproTech (Rocky Hill, NJ, USA). CpG oligodeoxynucleotide 1668 (tlrl-1668-1) and Pam3CSK4 (tlrl-pms) were obtained from InvivoGen (San Diego, CA). R848 was obtained from Enzo Life Sciences (ALX-420-038-M005; Farmingdale, NY).

The mouse BM cells were isolated, and RBCs were removed by RBC lysis buffer (NH₄Cl). These cells were strained through a 70-μm filter. BM cells were then seeded at a density of 5 × 10⁵ cells/well in 12-well plates and maintained in DMEM (5.5 mM glucose, 11885084; Life Technologies) containing 10% FBS, penicillin (100 U/ml), streptomycin (100 U/ml), 20 ng/ml GM-CSF, and 2 ng/ml TGF-β. At day 7, the culture medium was refreshed with 20 ng/ml GM-CSF, 2 ng/ml TGF-β, and 0.1 μM PPAR-γ agonist rosiglitazone. At day 9, nonadherent cells were discarded, and adherent macrophages were detached by Accutase (A6964; Sigma-Aldrich) or cell scraper and collected as AM-like cells. Similarly, for fetal liver cell-derived AM-like cells, mouse fetal livers were isolated from embryonic days 15–16 embryos, then grinded and strained through a 40-μm filter. RBCs were removed by RBC lysis buffer (NH₄Cl). After centrifugation, fetal liver cells were collected. Then the cells were seeded at a density of 2 × 10⁶ cells/well in 12-well plates and maintained in 1640 complete medium with GM-CSF and TGF-β for 7 d and refreshed by the medium with GM-CSF, TGF-β, and 0.1 μM PPAR-γ agonist rosiglitazone for 2 more days.

The mouse was euthanized; then the skin and muscles were removed to expose the trachea. A small incision was made below the larynx, and this incision was inserted by a blunted 20-G cannula (201483; Braun, Mersongen, Germany) toward the lungs. Then 0.8 ml PBS with 0.4 mM EDTA and 1.5% FBS were admitted into the lungs by syringe. The lungs were washed 10 times, and BALF was collected. RBCs were removed by RBC lysis buffer (NH₄Cl), and the remaining cells were filtered through a 70-μm cell strainer. After centrifugation, BALF-AMs were collected. For in vitro cultivation of GTR-BALF-AMs, BALF-AM cells were seeded at a density of 0.5 × 10⁵ cells/well in 12-well plates and maintained in GlutaMAX supplement DMEM (5.5 mM glucose, 10567014; Life Technologies) containing 10% FBS, penicillin (100 U/ml), and streptomycin (100 U/ml), in the presence of 20 ng/ml GM-CSF, 2 ng/ml TGF-β, and 0.1 μM rosiglitazone for 3 or 9 d (the medium was changed every 2 days).

Total RBC-removed bone marrow (BM) cells were cultivated at a density of 2 × 10⁶ cells/ml and maintained in DMEM supplemented with 10% FBS in the presence of 20 ng/ml M-CSF. After 5–7 d, floating cells were discarded, and the attached cells were detached by Accutase and collected as BM-derived macrophages (BMDMs).

Nonadherent cells were discarded, and adherent macrophages were collected by cell scraper. Then the cells were blocked by anti-CD16/32 Ab for 15 min at 4°C. Furthermore, the cells were stained by anti-CD11c, anti-Siglec-F, anti-F4/80, anti-CD11b, and anti-CD45 for 30 min at 4°C. After staining, cells were washed and resuspended by buffered saline solution containing 5 mM EDTA with 7-AAD. The AM-like cells (CD45⁺ 7-AAD⁻ CD11c⁺ Siglec-F⁺) were analyzed by flow cytometer.

About 2 × 10⁴ cells were collected and cytospun onto the glass slides. The slides were then stained by Wright-Giemsa staining (32884; Sigma-Aldrich) following the manufacturer’s instructions. The morphology of stained cells was observed under a microscope (OLYMPUS IX71).

Before stimulation, AM-like cells, GTR-BALF-AMs, and BALF-AMs were allowed to rest for 4 h in DMEM complete medium without GTR. Then the cells were seeded at a density of 1 × 10⁵ cells/well in 48-well plates and then stimulated with 100 ng/ml LPS, 0.5 μM CpG oligodeoxynucleotide 1668, 300 ng/ml Pam3CSK4, or 10 μg/ml R848, respectively, for 6 h. Total RNA from the stimulated cells was extracted using TRIzol Reagent (15596018; Invitrogen, Carlsbad, CA), and cDNA was synthesized using PrimeScript RT Master Mix (RR036A; TaKaRa, Otsu, Shiga, Japan). Quantitative RT-PCRs (qRT-PCRs) using SYBR Premix Ex Taq (A25742; Applied Biosystems, San Francisco, CA) were run in an ABI 7900HT machine. The qRT-PCR primer sequences were obtained from primer bank (https://pga.mgh.harvard.edu/primerbank/) and are listed in Supplemental Table I. The PCR efficiency was verified by performing standard curve analysis, which indicated high PCR efficiency (87–105%) could be achieved.

BALF-AM, GTR-BALF-AM, or AM-like cells were seeded at a density of 1 × 10⁵ cells/well in 48-well plates and incubated with 2 μm fluorescent red latex beads (L3030; Sigma-Aldrich) at 37°C for 20 or 45 min. Then, cells were washed with PBS and digested with Accutase at 37°C for 20 min; the cells, which had phagocytosed beads, were then detached and analyzed by flow cytometer.

BALF-AMs, GTR-BALF-AMs, or AM-like cells were seeded at a density of 0.5 × 10⁵ cells/well in 12-well plates in the presence of 20 ng/ml GM-CSF, 2 ng/ml TGF-β, and 0.1 μM rosiglitazone for 9 d (the medium was changed every 2 days). At day 9, these cells were analyzed for EdU incorporation, following the manufacturer’s instructions (C10338-2; RiboBio, Guangzhou, China).

The BM cells from Cas9 knockin mice were extracted, and RBCs were removed by RBC lysis buffer (NH₄Cl). Lin⁻ cells were isolated using negative magnetic cell sorting with Ab cocktails (anti-TER119, anti-CD2, anti-CD3, anti-CD45R, anti-CD8, and anti-Ly6G). The retroviruses were generated in Plate-E cells by transfecting pRVKM-U6-CMV-mAmetrine plasmids (originated from pRVKM-U6-CMV-hCD2 plasmids, a kind gift from Dr. Dong Chen, Tsinghua University). Before retroviral transduction, Lin⁻ cells were supplemented with 100 ng/ml stem cell factor, 10 ng/ml IL-6, and 6 ng/ml IL-3 overnight; then, these cells were seeded at a density of 5 × 10⁵ cells/well in 24-well nontreated plates and infected with the retroviruses harboring guide RNA (gRNA) target Siglec-F (single guide RNA sequences: 5′-CCGCGGAACTGCAACTGCCA-3′) using Retronectin (T100B; Takara, Otsu, Shiga, Japan). Twenty-four to forty-eight hours postinfection, these cells were seeded at a density of 1 × 10⁵ cells/well in 12-well plates and stimulated with 20 ng/ml GM-CSF and 2 ng/ml TGF-β for 7 d and then refreshed by the medium with GM-CSF, TGF-β, and 0.1 μM PPAR-γ agonist rosiglitazone for 2 more days. Then, the expression of Siglec-F was determined by flow cytometric analysis.

Statistical analysis was performed with GraphPad Prism 8, and statistical significance between two groups was determined by unpaired two-tailed Student t tests or one-way ANOVA followed by Sidak test for multiple comparisons with 95% confidence intervals.

The characteristic cell surface phenotype of AMs is CD11c^highSiglecF^highCD11b^low. It is known that cytokines GM-CSF and TGF-β are essential for the development of AMs, and PPAR-γ signaling is particularly important for AM differentiation. In an attempt to develop a system for generating AM-like cells in vitro, mouse BM cells were cultured with a combination of factors known to be required for AM development, 20 ng/ml GM-CSF and 2 ng/ml TGF-β for 9 d, with 0.1 μM PPAR-γ agonist rosiglitazone (GTR) added at day 7 (Fig. 1A). The adherent cells generated in culture were then collected and the phenotype analyzed by flow cytometer. BALF-AMs were collected and used as primary AM control. As shown in (Fig. 1B, various combinations of cytokines were tested for induction of AM differentiation. The combination of GTR was found to efficiently induce the differentiation of cells resembling AMs (named AM-like cells hereafter), with the surface expression of high levels of AM markers, such as Siglec-F, CD11c, and F4/80, and a yield of 87% of cultured cells being AM-like cells. Interestingly, both GM-CSF plus TGF-β–treated-cells and GTR-treated cells coexpressed higher levels of Siglec-F and CD11c and lower levels of CD11b, whereas GM-CSF alone and GM-CSF plus rosiglitazone (referred to hereafter as GR)-treated cells coexpressed lower levels of Siglec-F and higher levels of CD11b (Fig. 1B, 1C). In addition, all these cells expressed mature macrophages marker F4/80 (Fig. 1B).

This GTR-supplemented in vitro culture system was highly efficient in producing large numbers of AM-like cells from BM cells (Fig. 1D). Furthermore, light microscopic analysis revealed that only when TGF-β was added did the cultured cells display morphological changes from an irregular shape, which is the typical appearance of GM-CSF–cultured BMDMs, to a rounded appearance, which is similar to that of BALF-AMs (Fig. 1E). These results demonstrated that a combination of GM-CSF, TGF-β, and rosiglitazone could efficiently induce the generation of AM-like cells in vitro.

To further determine whether the AM-like cells generated in this culture system have similar gene expression profiles to that of primary AMs, we examined the expression of AM-specific genes (21) in AM-like cells, including carbonic anhydrase 4 (Car4), placenta-expressed transcript 1 (Plet1), eosinophil-associated RNase A family member 1 (Ear1), cell death–inducing DNA fragmentation factor A–like effector c (Cidec), and cytokeratin 19 (Krt19) by qRT-PCR. The expression levels of Car4 and Plet1 mRNA were comparable between BALF-AM and AM-like cells (Fig. 2A). Although the expression levels of Cidec, Ear1, and Krt19 mRNA in AM-like cells were moderate compared with that of BALF-AMs, these gene transcripts could only be induced under the culture conditions with GR or GTR (Fig. 2B). In contrast, GM-CSF alone did not induce the expression of these genes (Fig. 2B). In addition, AM-like cells expressed many AM-specific genes (21), which were not expressed by M-CSF–cultured BMDMs (Fig. 2C). We also confirmed these results using Gapdh as another reference gene (Supplemental Fig. 1). Moreover, unlike BMDMs, which expressed high levels of genes, including Cd14, Csf1r, Irf7, and Mpo, AM-like cells expressed very low levels of those genes, indicating that AM-like cells more closely resembled the primary AMs rather than BMDMs (Fig. 2D). Furthermore, BALF-AMs, AM-like cells, and BMDMs all expressed comparable levels of the myeloid master regulator PU.1, encoded by Spi1 (Fig. 2E).

We also found that AM-like cells expressed higher levels of Pparg and the alternative macrophages activation signature genes, such as Ym1, Mrc1, and Marco (Fig. 3A). Because tissue resident macrophages could self-renew in vivo (12, 22), we next examined the expression of self-renewal–associated genes in AM-like cells. Expectedly, similar to BALF-AM, AM-like cells expressed higher levels of mRNA of pro-self-renewal genes Klf4 and Myc, but lower levels of self-renewal inhibition genes Maf and Mafb, than that of BMDMs (Fig. 3B). We also confirmed that these AM-like cells did not express markers specific for other tissue macrophages, such as GATA binding protein 6 for peritoneal macrophages, spalt-like transcription factor 1 (Sall1) for microglia, and Langerin for Langerhans cells (data not shown). Taken together, these results demonstrated that the combination of GM-CSF, TGF-β, and PPAR-γ agonist rosiglitazone promoted AM-like cells differentiation.

Macrophages play pivotal roles in modulating inflammatory response in various organs, including lung (23). To confirm whether these AM-like cells generated in the new culture system also have similar functional capacities to that of BALF-AMs, we stimulated AM-like cells with various TLR ligands, such as Pam3CSK4 for activating TLR2, LPS for TLR4, R848 for TLR7/8, and CpG for TLR9. Similar to BALF-AMs, on stimulations, AM-like cells upregulated the expression of Il1b, Il6, Isg15, and Tnfa mRNA. Interestingly, the levels of mRNA of Il1b, Il6, and Isg15 expressed by AM-like cells were even higher than that of BALF-AMs, while the Tnfa mRNA level was lower than that of BALF-AMs (Fig. 4A). These results demonstrated that AM-like cells and BALF-AMs are similar in responding to stimulations by these TLR ligands and in production of inflammatory cytokines/factors, although some minor differences were also observed.

Phagocytosis is a functional hallmark of macrophages (24). We then measured the phagocytic capacity of AM-like cells by incubating with fluorescence-labeled latex beads followed by flow cytometric and light microscopic analysis. As shown in (Fig. 4B, 80.1% AM-like cells phagocytosed beads at 20 min and 92.9% at 45 min. The phagocytic efficiency of AM-like cells was even higher than that of BALF-AMs, which were 36.9% and 70.7% at 20 and 45 min, respectively. The functional differences between BALF-AM and AM-like cells could be because of the in vitro cultivation environment of AM-like cells; we therefore tested whether similar functional capacities could be observed after BALF-AMs were subjected into the same culture condition. We isolated BALF-AMs and cultured them for 9 d in the presence of GTR (referred to hereafter as GTR-BALF-AM). The functions of GTR-BALF-AM were then analyzed and compared with that of AM-like cells. First, we compared the mRNA levels of cytokines expressed by AM-like cells and GTR-BALF-AMs on activation by TLR ligands. As expected, AM-like cells expressed similar levels of Il1b mRNA after activation by LPS and R848 (both at 6 h) and Tnfa mRNA by LPS and R848 (at 1 and 6 h, respectively) to that of GTR-BALF-AMs (Fig. 5A). Furthermore, we compared the phagocytic capacity of AM-like cells, GTR-BALF-AMs, and BALF-AMs. Interestingly, compared with that of BALF-AMs, the GTR-BALF-AMs showed enhanced phagocytic capacity and were more similar to that of AM-like cells (Fig. 5B). Moreover, the morphology of AM-like cells was more similar to that of GTR-BALF-AMs by Giemsa staining (Fig. 5C). These results indicated that AM-like cells were more similar to GTR-BALF-AM in terms of their TLR ligands sensing and phagocytic capacity.

To further compare the phenotype among these cells, we analyzed the expression of additional surface markers and the proliferative capacity of these cells. Similar to BALF-AMs, both AM-like cells and GTR-BALF-AMs expressed pan-macrophage markers CD64 and MerTK (Supplemental Fig. 2A). Interestingly, both AM-like cells and GTR-BALF-AMs showed lower expression of activation markers CD80 and CD86 than BALF-AMs (Supplemental Fig. 2B), whereas AM-like cells, but not GTR-BALF-AMs, showed a higher proliferative potential than BALF-AMs in in vitro EdU assay (Supplemental Fig. 2C).

Because M-CSF–cultured BMDMs are often used as the in vitro cell model to study the functions of macrophages, we therefore compared the functions of AM-like cells, BALF-AMs, and BMDMs. As shown in the Supplemental Fig. 3A, AM-like cells were larger and rounder than BMDMs in morphology. We also found that both BALF-AMs and AM-like cells were slightly less efficient in phagocytic activity than BMDMs (Supplemental Fig. 3B). In addition, both BALF-AMs and AM-like cells expressed higher levels of Il6 and Tnfα, but lower levels of Nos2, a classical macrophage activation marker, than BMDMs on LPS stimulation (Supplemental Fig. 3C). These results further suggested that compared with BMDMs, AM-like cells serve as a better in vitro cellular model for studying the function of AMs and AM-like cells generated in our new culture system and exhibited phenotypic, genetic, and functional features of primary AM cells.

It has been shown that fetal liver monocytes are the major precursors of AMs (16); we therefore tested the AM-like cell culture system for AM differentiation from fetal liver precursors. Mouse fetal liver cells were isolated and subjected to the cultures with the combination of GTR. Flow cytometric analysis of the culture products showed that a large proportion (77.4%) of GTR-treated cells coexpressed AM markers Siglec-F and CD11c, while only 28.2% of cells from GM-CSF and 49.7% from GM-CSF plus TGF-β–supplemented cultures did so (Fig. 6A). In addition, the level of CD11b was lowest in GTR-treated cells (Fig. 6A, bottom panels). Moreover, the GTR-supplemented culture system was the most efficient in producing AM-like cells compared with other culture conditions (Fig. 6A, right panel).

It is well-known that AMs have the capacity to self-renew and expand. We then examined whether this GTR culture system could also support the expansion of BALF-AMs in vitro. As shown in (Fig. 6B, GTR could promote the proliferation of BALF-AMs, regardless of initial seeding density. More importantly, BALF-AMs could be expanded about up to 10-fold after 7 d and more than 20-fold after 9 d (Fig. 6B). Although GM-CSF is often used to expand BALF-AMs in vitro, it is not clear whether those expanded BALF-AMs maintained expression of the AM-specific genes. To test this, we harvested BALF-AMs and cultured them for 3 and 9 d under the GTR condition, then assessed the expression of AM-specific genes (Car4, Cidec, Ear1, and Krt19) by qRT-PCR. As shown in (Fig. 6C, consistent with the results of BM cell cultures for AM-like cells, only the combination of GTR stimulation, rather than GM-CSF alone, could maintain expression of most AM-specific genes, whereas culture with GM-CSF alone resulted in the loss of expression of these genes except Car4 (Fig. 6C). Taken together, these results confirm that the AM-like cells culture system is also sufficient for generating AM-like cells from fetal liver precursors and for expansion of BALF-AMs in vitro. This system therefore provided potential for generating AM-like cells from both fetal and adult precursors and for investigating the functional differences of AMs of different origins.

To further evaluate the applicability of the AM-like cell culture system, we used CRISPR-Cas9 knockin mice to generate AM-like cells with Siglecf gene deletion. The gRNAs for Siglecf were designed; in brief, BM lineage-negative (lin⁻) cells from CRISPR-Cas9 knockin mice were infected with retrovirus harboring gRNA target Siglecf, then cultured with the combination of GM-CSF plus TGF-β and rosiglitazone for generating AM-like cells (Fig. 7A). The cultured cells were then harvested and analyzed by flow cytometry. As shown in (Fig. 7B, when gRNA for Siglec-F was introduced into BM lin⁻ cells, the surface expression of Siglec-F on the cells generated in culture could no longer be detected, indicating a successful deletion of Siglecf gene in these cells. Overall, our data demonstrated that this AM-like cell culture system could serve as a powerful tool for genetic manipulation of AMs and for further studies of the molecular regulations of development and functions of AMs.

Although the important role of AMs in defending against pathogens and maintaining lung tissue homeostasis has been recognized for quite a long time, as of now a good in vitro culture system for generating AMs is still lacking. Various culture systems for expanding AMs have been reported in previous studies; however, whether the cultured “AMs” still maintained the identity of primary AMs has not been thoroughly characterized (25, 26). Our study is the first, to our knowledge, to describe a novel culture system for generating AM-like cells in vitro from adult mouse BM and fetal liver precursors. We demonstrated that the synergistic effects of GM-CSF, TGF-β, and PPAR-γ signaling were essential for the development of AMs in vitro. These AM-like cells, resembling BALF-AMs, displayed a characteristic round shape morphology and expressed AM-specific genes, such as Sigelcf, Car4, Ear1, Plet1, Krt19, and Cidec, that were not expressed by other tissue macrophages. Functionally, the AM-like cells showed similar capacities of producing inflammatory cytokines and phagocytosis.

In addition, we showed that this culture system could also efficiently promote the production of AM-like cells from fetal liver precursors. Although the expression of many AM-specific genes by fetal liver-derived AMs closely resembled that of adult BM-derived AMs, their functional differences have been suggested (13, 14, 17). Our new AM culture system would provide a useful tool for further characterizing and distinguishing these functional differences between fetal and adult AMs.

AMs can be continuously regenerated in lung tissues because of their self-renewal ability. Previous studies often used GM-CSF, an essential factor for AM development, to expand BALF-AMs in vitro (22, 27–29). However, our study demonstrated that GM-CSF alone was not sufficient to maintain the expression of AM-specific genes, such as Cidec, Ear1, and Krt19, which were downregulated during in vitro cultures with or without GM-CSF. Only the combination of GM-CSF plus TGF-β and rosiglitazone could maintain the identity of BALF-AMs in vitro. This suggests that all these signaling pathways are required and always activated during AM differentiation in vivo.

We also found that both AM-like cells and GTR-BALF-AMs, although displaying blasted-like morphology, showed lower levels of activation markers CD80 and CD86 expression than BALF-AM. It suggests that although BLAF-AMs in the lung environment are exposed to particles or microorganisms in the alveolar space constantly, they therefore express higher levels of activation markers CD80 and CD86. Consistently, previous studies also reported that AMs did express CD80 and CD86 in vivo under steady state (30, 31). The lower expression of CD80 and CD86 by AM-like cells and GTR-BALF-AMs may be explained as they had been cultured for a period under sterile conditions, without any stimulation by microorganisms.

Another key point revealed by this study is the requirement for activation of the specific transcription factor PPAR-γ, for the differentiation of AM-like cells in vitro. Addition of PPAR-γ agonist rosiglitazone more efficiently promoted AM-like cell differentiation in culture. However, it is still not clear which is the endogenous ligand for PPAR-γ activation in lung, even though several endogenous ligands for PPAR-γ, such as 15d-PGJ2, nitrolinoleic acid, nitro-oleic acid, lysophosphatidic acid, and so on have been reported (32–35). Interestingly, we found that both 15d-PGJ2 and nitrolinoleic acid could also induce the expression of AM-specific genes such as Ear1 and Krt19 in vitro (Supplemental Fig. S4). The endogenous ligands for PPAR-γ in lung tissue are yet to be identified.

Further, to test whether this AM culture system could be applicable for genetic manipulations of AMs, we used the CRISPR-CAS9 system to delete one of the AM marker genes Siglecf in AM-like cells derived from BMs of Cas9 knockin mice. The result showed that this genetic manipulation had been successful, indicating the potential applications of the AM-like cell culture system for studying the molecular mechanisms of AM development and functions and AM deficiency–associated pulmonary diseases. Overall, our study provided, for the first time to our knowledge, a novel AM culture system that can serve as a very useful tool for future studies of AM differentiation and function.

We thank the Laboratory Animal Research Center (Tsinghua University, Beijing, China) for technical support.

The authors have no financial conflicts of interest.