A new tractable method for generating human alveolar macrophage-like cells in vitro to study lung inflammatory processes and diseases

ABSTRACT Alveolar macrophages (AMs) are unique lung resident cells that contact airborne pathogens and environmental particulates. The contribution of human AMs (HAMs) to pulmonary diseases remains poorly understood due to the difficulty in accessing them from human donors and their rapid phenotypic change during in vitro culture. Thus, there remains an unmet need for cost-effective methods for generating and/or differentiating primary cells into a HAM phenotype, particularly important for translational and clinical studies. We developed cell culture conditions that mimic the lung alveolar environment in humans using lung lipids, that is, Infasurf (calfactant, natural bovine surfactant) and lung-associated cytokines (granulocyte macrophage colony-stimulating factor, transforming growth factor-β, and interleukin 10) that facilitate the conversion of blood-obtained monocytes to an AM-like (AML) phenotype and function in tissue culture. Similar to HAM, AML cells are particularly susceptible to both Mycobacterium tuberculosis and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections. This study reveals the importance of alveolar space components in the development and maintenance of HAM phenotype and function and provides a readily accessible model to study HAM in infectious and inflammatory disease processes, as well as therapies and vaccines. IMPORTANCE Millions die annually from respiratory disorders. Lower respiratory track gas-exchanging alveoli maintain a precarious balance between fighting invaders and minimizing tissue damage. Key players herein are resident AMs. However, there are no easily accessible in vitro models of HAMs, presenting a huge scientific challenge. Here, we present a novel model for generating AML cells based on differentiating blood monocytes in a defined lung component cocktail. This model is non-invasive, significantly less costly than performing a bronchoalveolar lavage, yields more AML cells than HAMs per donor, and retains their phenotype in culture. We have applied this model to early studies of M. tuberculosis and SARS-CoV-2. This model will significantly advance respiratory biology research.

constant for human cell AM development among donors than the other components of alveolar lining fluid.
Infasurf, GM-CSF, TGF-β, and IL-10 signaling resulted in upregulation of PPAR-γ, a signature transcription factor essential for AM development. Human AML cells exhibited light and electron microscopy morphology resembling HAM, including the appearance of lipid body inclusions, some appearing as lamellar bodies. Expression of a gene set unique to HAM as well as global transcriptomic analysis by RNA-seq revealed expression profiles of AML cells related to freshly obtained HAM, including increased expression of key AM transcription factors and PPAR-γ, TGF-β, and GM-CSF signaling pathways. In addition, AML cells showed increased OxPhos and mitochondrial respi ration and reduced glycolysis, similar to what is reported for AMs (12). AML cells had increased expression of CD206, macrophage receptor with collagenous structure (MARCO) and CD11c, and reduced CD36 expression. Culturing AML macrophages in the lung component cocktail after macrophage adherence maintained the AML phenotype over time in culture. Importantly, similar to HAM, AML cells were particularly susceptible to the airborne pathogens, M.tb and SARS-CoV-2. Thus, we present a novel model for generating AML cells, which is minimally invasive, significantly less costly, results in more AML cells relative to HAM recovered from one person, and can be maintained in culture. Individual components of the cocktail alone cannot generate AML cells. We present a promising model to study HAM in a variety of lung inflammation contexts.

Collection and isolation of HAM
Fresh HAMs were isolated and cultured from BAL of healthy donors as described previously (17) and used for the respective studies. See the detailed supplemental methods section.

In vitro development of AML cells and MDM from human PBMCs
PBMCs were cultured for 6 days to allow for the differentiation of monocytes into untreated MDM or treated AML cells. To generate AML cells, Infasurf (100 µg/mL), GM-CSF (10 ng/mL), TGF-β (5 ng/mL), and IL-10 (5 ng/mL) were added on days 0, 2, and 4 (three doses of "ALL cocktail," treatment with all four components). For some experiments, Infasurf, GM-CSF, TGF-β, and IL-10 were only added on day 0 (one dose of ALL cocktail). In other experiments, we analyzed the role of individual components of the ALL cocktail. On day 6, both control MDM and AML cells were harvested and adhered to tissue culture dishes for 2 hours in RPMI 1640 with 10% fresh autologous serum, lymphocytes were washed away, and then all experiments were performed. We also determined the requirement of the continuous addition of ALL cocktail after differentiation to retain the AML cell phenotype. The complete protocol and surfactant component information are in the detailed supplemental methods section.

Isolation of human monocytes by magnetic sorting and development of AML cells
as described above to allow for the differentiation of purified monocytes into AML cells or untreated MDM. AML and MDM cell lysates and supernatants were used for quantita tive RT-PCR (qRT-PCR) and Luminex assay, respectively, to compare differentiated AML cells from PBMCs or freshly isolated monocytes. See more information in the detailed supplemental methods section.

Cytospin analysis
Single-cell suspensions of freshly isolated HAM, cultured AML or MDM cells (5 × 10 4 ) were placed in a cytofunnel and centrifuged at 150× g for 5 minutes onto cytoslides that were dried and stained with HEMA 3 differential staining. The slides were examined with a Motic AE2000 inverted microscope. See details in the detailed supplemental methods section.

Transmission electron microscopy (TEM)
AML and MDM cells were fixed with 4% formaldehyde and 1% glutaraldehyde in phosphate buffer (Invitrogen) overnight at 4°C. The samples were processed and imaged using a JEOL 1400 TEM. Expanded protocol is given in the detailed supplemental methods section.

Multicolor confocal microscopy
AML and MDM cells were stained with fluorescence-conjugated antibodies or control antibodies, and stained slides were visualized with a Zeiss LSM 800 confocal microscope. See information in the detailed supplemental methods section.

Bead cell association study
AML and MDM cells were incubated with non-opsonized FluoSpheres Sulfate Micro spheres, 1.0 µm, yellow-green fluorescent F8852-beads (Invitrogen). Stained slides were visualized with a Zeiss LSM 800 confocal microscope. The cells were counted based on DAPI staining. The number of fluorescent beads was also counted and shown as beads/ macrophages. For further information, refer to the detailed supplemental methods section.

RNA-seq and analyses
Freshly isolated HAMs and 2-hour-adherent MDM and AML cells were lysed in TRIzol, and RNA was isolated using Direct-zol RNA Miniprep kit, R2052 (Zymo Research) as per the manufacturer's instructions. RNA sequencing was carried out using the HiSeq 3000 platform (Illumina). The detailed protocol and data analysis are elaborated in the detailed supplemental methods section.

Western blot analysis
AML and MDM cells were lysed with NE-PER Nuclear and Cytoplasmic Extraction kit according to the manufacturer's protocol (Thermo Scientific). Western blot was performed, and the membranes were incubated with the primary antibody for PPAR-γ, PU.1, H3K4me1, and H3K4me3 followed by anti-rabbit IgG and horseradish peroxidase (HRP)-linked antibody. The membranes were developed using a clarity ECL reagent on a UVP chemstudio 815 system. Stripping was performed and then membranes probed to detect β-actin or histone H3 levels. See further information in the detailed supplemental methods section.

Extracellular flux analysis
Real-time cellular metabolism of AML cells and MDM was determined by measuring oxygen consumption rate (OCR) (picomoles per minute) and extracellular acidification rate (ECAR) (mpH/min) using a Seahorse XFe96 Analyzer, according to the manufacturer's instructions (Agilent Technologies). Mito stress assay was performed after the sequential addition of 5 µM oligomycin, 4 µM carbonyl cyanide-p-trifluoromethoxyphenyl-hydrazon (FCCP), and 2 µM rotenone and antimycin A. For glycolysis stress analysis, AML and MDM cells were injected with 2 µM rotenone and 2 µM antimycin A followed by 100 mM 2-deoxyglucose to determine the glycolytic rate. See information in the detailed supplemental methods section.

Lactate release
MDM and AML cells were stimulated with or without lipopolysaccharide (LPS) (MDM: 10 ng/mL and AML: 100 ng/mL) for 24 hours. The supernatants (collected from three donors and stored at −80°C) were used for lactate measurements. Lactate was quan tified using a lactate colorimetric enzymatic assay kit according to the manufactur er's instructions (K627; BioVision, Milpitas, CA, USA). Data were expressed as per the manufacturer's instruction in nanomoles per microliter cell culture supernatant.

MitoSOX assay and cellular ROS detection
Mitochondrial and intracellular non-mitochondrial reactive oxygen species (ROS) levels in AML and MDM cells were measured by staining with mitochondrial superoxide (MitoSOX) (5 μM) and H 2 DCFDA (5 μM), respectively, for confocal or flow cytometry analysis. See further information in the detailed supplemental methods section.

EPR assay
Electron paramagnetic resonance (EPR)-based ROS detection for mtO 2 was performed using Mito-Tempo-H (100 µM). About 2 µM of rotenone and 2 µM of antimycin A mix were added to inhibit mitochondrial complexes. The EPR spectra were measured on the Bruker EMXnano ESR system. Expanded methods are described in the detailed supple mental methods section.

Phagocytosis assay for M.tb
Fixed HAM, AML, or MDM monolayers were incubated with either rabbit polyclonal anti-M.tb whole cell lysate antibody or an IgG rabbit isotype control antibody. Later, cells were incubated with an AlexaFluor 488 donkey anti-rabbit secondary antibody. Imaging was executed on a Zeiss LSM 800 microscope. The complete protocol and analysis are given in the detailed supplemental methods section.

Macrophage infection with M.tb
MDM, HAM, and AML cells were infected (multiplicity of infection, MOI 2) with virulent M.tb H 37 R v . The intracellular growth of M.tb at each post-infection time point (2,24,48, and 72 hours) was measured in cell lysates. CFUs were assessed after 3 and 4 weeks on 7H11 agar plates. Expanded information are given in the detailed supplemental methods section.

Generation of rSARS-CoV-2 expressing reporter genes
Recombinant SARS-CoV-2 expressing mCherry and nanoluciferase (Nluc) reporter genes (rSARS-CoV-2/mCherry-Nluc) was rescued in Vero E6 cells and viral stocks prepared (22). Viral titers in the stocks were determined and used for infection studies. Complete information is given in the detailed supplemental methods section.

Cytation 5 live cell imaging assay
Freshly obtained HAM, AML, MDM, and Vero E6 cells were infected with rSARS-CoV-2/ mCherry-Nluc virus (MOI: 1 and 10). Live cell imaging was performed using Cytation 5 paired with BioSpa (BioTek/Agilent). Data analysis was achieved with Gen5 software by calculating mCherry mean fluorescence intensity (MFI). Cells were counted after 120 hours by counterstaining with DAPI. Persistence of SARS-CoV-2-mCherry in cells was monitored in time-lapse videos (0/4-84-hour time period) using Gen5 software. The extended protocol is given in the detailed supplemental methods section.

Statistical analyses
Graphs were prepared and statistical comparisons were applied using GraphPad Prism version 9 (GraphPad). Statistical comparisons were performed by unpaired two-tailed Student's t-test. Ordinary one-way and two-way analysis of variance (ANOVA) with Sidak's multiple comparisons test for multiple testing (GraphPad Prism 9) was applied wherever applicable (indicated in the figure legends). For correlation analysis, Spearman's rank test was applied. Statistical differences between groups were reported as significant when the P-value was ≤0.05. The data are presented as mean ± SEM.

In vitro development and differentiation of human AML cells
We established a method for providing exogenous surfactant components (Infasurf ) and specific lung-associated cytokines (GM-CSF, TGF-β, IL-10), critical for AM differentiation, to cultured monocytes in PBMCs to determine whether the lung-associated components would drive monocyte differentiation into macrophages resembling a HAM phenotype (23). Importantly, we analyzed freshly obtained HAM within 6 hours of acquisition which best enables retention of the in vivo phenotype (17).
We isolated PBMCs from healthy adult human donors and first cultured them in increasing concentrations of GM-CSF, TGF-β, and IL-10 without Infasurf for 6 days, during which time monocytes differentiated into macrophages. We identified the optimal concentration of these cytokines to induce the expression of PPAR-γ and MRC1, two well-established AM markers ( Fig. S1A and B). Next, to understand the role of individual lung-associated cytokines and surfactant in generating AML cells, we treated PBMCs with GM-CSF, TGF-β, IL-10, and Infasurf individually or in combination and assessed the expression of a subset of genes that are differentially expressed in HAM relative to MDM. Treatment with all four components (termed "ALL cocktail") drove more robust gene expression changes than individual cytokines or Infasurf treatment alone ( Fig. S1C through K). ALL cocktail treatment did not affect the viability of AML cells (Fig. S1L). Next, we cultured PBMCs in the presence of ALL cocktail at the optimal concentration for 6 days (ALL cocktail) (Fig. 1A). To assess the cultured macrophages further, we identified a set of 30 genes that are differentially expressed in fresh HAM compared with bloodbased MDMs (Table 1). These genes were chosen carefully based on the literature and a previously generated AmpliSeq database from our laboratory comparing MDM and HAM transcriptomes (17). We assessed the gene expression pattern in the cultured macro phages from a randomly selected subset of these genes from Table 1 that resemble the HAM phenotype. Monocytes cultured in ALL cocktail developed into macrophages that exhibited expression patterns similar to HAM with significant increases in the expression of PPAR-γ, MRC1, MARCO, CES1, MCEMP1, myeloid cell leukemia-1 (MCL1), DUSP1, CXCL3, PU.1, CXCL5, CD170, and CCL18 and significant decreases in expression of MMP7, MMP9, CD36, CCL22 and CD84 when compared to monocytes that were cultured without lung components and thus differentiated into MDM (Fig. 1B through Q). We named the cells cultured in ALL cocktail AML cells. Increases in PPAR-γ transcript in AML cells versus MDM corresponded with an increase in PPAR-γ protein levels (Fig. 1R). The transcription factor PU.1 (SPI1) is induced by GM-CSF and is important for AM function (8). Like HAM, both AML and MDM expressed PU.1, although increased in AML cells ( Fig. 1J and R). Thus, the established culture conditions drive both PPAR-γ and PU.1 expression, which are critical transcriptional determinants of AML development.

AML cells undergo similar epigenetic changes as reported for HAM
During development, AMs undergo specific histone modifications, with higher levels of histone H3 lysine 4 mono-methylation (H3K4me1) and lower levels of H3K4me3 (24).
These epigenetic changes result in the recruitment of PU.1, which is essential for the maintenance of high H3K4me1 at macrophage-specific enhancers (25). As observed for HAM (24), AML cells also showed higher expression of H3K4me1 and lower expression of H3K4me3 when compared to MDM (Fig. 1R) (23). These data indicate that culturing human monocytes in lung components during differentiation drives them to an AML phenotype with characteristics similar to HAMs.

Continuous supplementation of the lung cocktail retains the AM phenotype after differentiation of AML cells
To determine whether continuous supplementation of the ALL cocktail during monocyte differentiation is necessary to drive monocytes to AML cells, we treated PBMCs with one dose of ALL cocktail (day 0) versus multiple doses (days 0, 2, 4). Monocytes treated with ALL cocktail on alternative days showed changes in gene expression more akin to HAM when compared to one dose only (day 0). We observed a stronger increase in PPAR-γ, MRC1, MARCO, CES1, PU.1, and MCEMP1 gene and protein expression when cells were treated with multiple cocktail doses ( Fig. 2A through G). Together, the results indicate a These genes were carefully chosen based on their abundancy in HAM relative to MDM, from the literature and a previously generated AmpliSeq database from our laboratory (17,20). Transcriptomic analysis was assessed from PBMC-derived differentiated MDM and fresh HAM after 2-hour adherence in culture. "RPM" means "reads per million, " that is, the count of how many reads map to the gene, divided by the total number of aligned reads and multiplied by one million. Lightface shows higher expression in HAM, and boldface shows higher expression in MDM. Research Article mBio that monocytes must be continuously supplemented with ALL cocktail during differentiation to drive the monocytes to AML cells.
To determine if the lymphocytes present in PBMCs aid in the differentiation of monocytes into AML cells, we assessed the expression of select genes for AML cells generated from PBMCs versus purified monocytes during cultivation. We found that lymphocytes are not required for AML cell development. AML cells generated from PBMCs and isolated monocytes showed similar increases in the expression of PPAR-γ, MRC1, and MARCO, and reduced expression of MMP9 (Fig. 2H through K). We also observed significant differential expression of TLR genes in AML cells versus MDM (Fig. 2L through O). These data indicate that AML cells can be developed from monocytes in the absence or presence of other cell types present in PBMCs. For ease, unless indicated otherwise we cultured PBMCs with three doses of ALL cocktail to generate the AML cells described below.
We have previously determined [Papp et al. (17), Table S1] that AMs rapidly lose their phenotype on isolation from the lung and time in culture (17). Our data demonstrate that treatment with multiple doses of ALL cocktail (days 0, 2, 4) is optimal for AML development ( Fig. 2A through G). To investigate if continuous supplementation of the cocktail is required to retain the AM phenotype after differentiation of AML cells, we generated AML cells, plated them, and subsequently incubated them with or without ALL cocktail for 24, 48, and 72 hours. We observed that additional supplementation of the cocktail after adherence enables maintenance of the AML phenotype with higher expression of PPAR-γ, MRC1, and MARCO compared to cells that are not treated after adherence ( Fig. S2A through C). This phenotype can be maintained for a longer period of time with ALL cocktail added ( Fig. S3A through C), which is beneficial for longer-term studies.

AML cells have similar morphological features and limited self-proliferation capacity to those reported for HAM
HAMs have a unique morphology (26). We assessed the morphology of MDM and AML cells by light microscopy and TEM. By light microscopy, AML cells were more rounded and had long pseudopodia closely resembling HAMs, as opposed to MDMs which are flatter and more irregularly elongated cells ( Fig. 3A and B). Similarly, by TEM, AML cells appeared rounded and had a similar morphology to what is reported for HAM (26). Like HAM, the cytoplasm of AML cells contained various structures which vary in appearance and number (Fig. 3C). AML cells contained prominent onion-shaped phagolysosomes with phospholipid-rich surfactant stored in lipid inclusion bodies, resembling lamellar bodies, composite bodies, and large and small floccular or reticular inclusions, some showing fusion (26). We also observed coated vesicles, some large heterophagic vacuoles (HVs), Palade granules (PGs), and very dense granules interpreted as ferritin (F) (26) (Fig. 3C). Double-membrane autophagosome (DMA) structures were also visible as were several round/irregular or elongated mitochondria (M) and various elements of the endoplasmic reticulum (ER). In contrast to AML cells or HAM, MDM were flat, large, and irregularly shaped with an eccentrically placed nucleus, numerous vesicles and vacuoles, a ruffled surface, and contained free or membrane-bound lysosomal inclusions in the vacuole. Round or ovoid electron-dense bodies were more prevalent in MDM than AML cells. Round or elongated mitochondrial structures and ER were also visible in MDM (Fig.  3D). Overall, the micrographs showed that AML cells have a morphology similar to that described for HAM, which represents morphologically distinct macrophages (26). In addition, PBMCs were exposed to ALL cocktail for 6 days on alternative days (AML-PBMCs) or left untreated (MDM-PBMCs), then macrophages were purified by adherence. The cells were collected for qRT-PCR analysis of selected HAM signature genes (17). Gene expression was compared within the groups: MDM and AML cells that were matured from purified monocytes (Continued on next page)

Research Article mBio
Adult AMs have been described as long-lived terminally differentiated lung resident cells (27). In the murine model, AMs originate from either peripheral blood monocytes and/or fetal monocytes and undergo cell proliferation for self-renewal and maintenance in the steady state (4,5,28,29). In contrast, much less is known about the ontogeny and cell proliferation of healthy adult human AMs in the normal steady-state condition. Some evidence suggests a low-grade proliferation capacity of adult human AMs in disease states such as respiratory infection or inflammatory/autoimmune diseases but not in healthy humans (30,31). To ascertain the proliferative capacity of AML cells and MDM, cells were stained using Ki67 and analyzed by confocal microscopy and flow cytometry ( Fig. 3E through H). AML cells and MDM demonstrated a very limited proliferation capacity. The THP-1 monocytic cell line was used as a positive control.

HAM and AML cells share similar transcriptome profiles
To further compare AML cells to HAM and MDM, we performed transcriptomic analysis following RNA-seq of freshly collected HAM, AML cells, and MDM. Principal component analysis (PCA) showed a high degree of similarity between biological replicates within each group (Fig. 4A). Volcano plots depicting the false discovery rate (FDR) relative to the magnitude of change in gene expression highlighted that the majority of the AML cell transcriptomes resemble HAM, with 899 genes (of 14,097 expressed genes, 6.4%) significantly upregulated at least two-fold, and 102 genes (0.7%) significantly downregu lated at least two-fold in AML cells relative to HAM (Fig. 4B and D). In contrast, when comparing MDM and freshly isolated HAM, we found a significant difference with 1,516 upregulated and 1,319 downregulated genes in MDM versus HAM (Fig. 4C; 20.1% genes were differentially expressed). We also found differential gene expression with 744 upregulated and 438 downregulated genes in AML cells when compared to the MDM transcriptome (Fig. S4A). Results from the RNA-seq data validated the 30 gene signatures comparing fresh HAM to MDM (Table 1; Fig. 1; Fig. 4E and F).
Next, we assessed transcription factors that are important for AM development and function. Of the PPAR family, PPAR-γ is critical for AM development (7) and was significantly upregulated in AML cells and HAM (Fig. 4G), which corresponds with increased protein levels in AML cells (Fig. 1R). RXRB associates with PPAR-γ; regulates cell differentiation, lipid metabolism, and immune function; and was also highly expressed in HAM. The transcription factor KLF4, along with PPAR-γ, upregulates Myeloid cell leukemia-1 (MCL-1) expression, and both were increased in HAM and AML cells. Downstream signaling of GM-CSF (CSF2) and PU.1 (SPI1) is important for AM function and their expression was increased in HAM and AML cells, both at the gene ( Fig. 1J; Fig. 4G) and protein levels (Fig. 1R). CSF2RB is the receptor for GM-CSF, an important transcription factor for macrophage development and surfactant catabolism (32). CCAAT enhancer binding protein beta (CEBPB), runt-related transcription factor 1 (RUNX-1), transforming growth factor-beta (TGF-β) receptor type 2 (TGFBR2), interferon-regulatory factor (IRF) 6, and IRF7 are important transcription factor-associated genes and showed similar expression patterns in AML cells and HAM relative to MDM (Fig. 4G). They were highly expressed in HAM and associated with their development and function. We further performed STRING protein analysis on PPAR-γ, TGFB1, and GM-CSF (CSF2), which mediate key signaling pathways in HAM. Most of the interacting proteins in these pathways showed similar expression in HAM and AML cells, indicating that these cells activate similar signaling networks (Fig. 4H through J). Thus, AML cells showed a similar transcriptomic landscape to HAM. Although MDM showed some similarity in the gene profile with HAM, especially in regard to common inflammatory and immune function-related pathways, important differences in canoni cal pathways exist (17). In addition, the expression level of major transcription factors important for AM development differed between MDM and AML cells and HAM (Fig. 4G). Ingenuity pathway analysis (IPA) showed the involvement of the RXRA transcription factor with the upregulation of MARCO, collectin subfamily member 12 (COLEC12), heparin-binding epidermal growth factor (HBEGF), insulin-like growth factor (IGF) 1, S100A4, and versican (VCAN) in AML compared to the MDM transcriptome (Fig. S4B). Similar to a previous finding in HAM (17), the TREM1 signaling network was also upregu lated in the AML transcriptome when compared to MDM (Fig. S4C). IPA also identified the inflammatory response network as being distinct between AML cells and MDM, with involvement of PPAR-γ and downregulation of CD36 in AML cells compared to MDM, a profile more consistent with HAM (Fig. S4D). IPA network analysis identified network 1 (immune cell trafficking, cellular movement, cell-to-cell signaling and interaction), network 2 (cellular movement, immune cell trafficking, inflammatory response), and network 3 (immune cell trafficking, cellular movement, hematological system develop ment and function) as being distinct between AML cells and MDM (Fig. S4E through G). In summary, although MDM showed some similarity in the gene profile with HAM, AML cells have a transcriptomic profile more closely aligned with HAM.

AML cells have upregulated lipid uptake genes and a drive toward oxidative phosphorylation
An important function of AMs is to regulate lipid metabolism, including the degradation of lipid-rich surfactants to maintain proper lung function (33,34). The lung alveolar space is rich in surfactant proteins and lipids and has low levels of glucose (35,36), an environ ment that may be conducive to the known low immunoreactivity of AMs, representing an adaptation to their specific environment. AMs engage in OxPhos over glycolysis as their core source of ATP (37), a finding in both human and murine cellular studies. Patients suffering from sepsis endure a shift from OxPhos to aerobic glycolysis, which is reversed on patient recovery (38). In vitro studies established that LPS-stimulated inflammatory macrophages typically depend on glycolysis, and alternatively activated M2 macrophages use OxPhos to generate energy (39). It is also well-known that IL-10 suppresses glycolysis in LPS-activated wild-type bone-marrow-derived macrophages (BMDMs) (40). Conversion from OxPhos to glycolysis in macrophages is generally important in host defense (41).
We assessed the metabolic status of AML cells relative to HAM and MDM. RNA-seq data demonstrated that the expression of OxPhos-related genes is upregulated in AML cells and HAM relative to MDM (Fig. 5A). Cholesterol and triglyceride metabolism-related genes are also upregulated in AML cells and HAMs (Fig. 5B). Further, like reported for HAM (36,37), we observed that AML cells exhibited much higher basal and maximal OCR (42) compared with MDM, signifying the engagement of OxPhos and mitochondrial activities (Fig. 5C). Notably, AML cells also had a higher basal, maximal, and spare respiratory capacity (Fig. 5D through F). Proton leak and non-mitochondrial OCRs were Research Article mBio elevated in AML cells compared to MDM (Fig. 5G through I). These data support our RNAseq analysis, where OxPhos is upregulated in HAM and AML cells (Fig. 5A and B). Interestingly, the opposite trends were noted for ECAR in AML cells, which indicate that the glycolytic rate (glycoPER), basal proton efflux rate, and compensatory glycolysis are Research Article mBio much lower in AML cells as compared to MDM (Fig. 5J through M). OCR can be increased for the production of ATP through either glycolysis or OxPhos (43). The ATP rate assay indicated that the production of ATP is increased in AML cells by OxPhos relative to MDM (Fig. 5H). MDM is more prone toward glycolytic-linked ATP production than AML cells or HAM (Fig. 5J). We further assessed the metabolic response of cells after LPS stimulation, which was correlated with the glycolytic response. We measured extracellular lactate release in the supernatant after 24 hours of treatment in AML cells and MDM. In contrast to MDM, AML cells did not respond to LPS to induce glycolysis (Fig. 5N), similar to a previous finding reported in human AMs (37). Mitochondrial mass, mitochondrial membrane potential, the rate of proton leak, oxygen consumption, and ATP synthesis dynamically influence mitochondrial ROS production (44). We observed that mitochon drial ROS (mt-ROS) and non-mitochondrial ROS are increased in AML cells, as demonstra ted by confocal microscopy and flow cytometry (Fig. 5O through R). The data were further validated by measuring EPR-based mt-ROS detection, which demonstrated an increase in the Mito-TEMPO signal intensity in AML cell lysate ( Fig. 5S and T). Overall, these data provide evidence that AML cells are driven toward OxPhos rather than glycolysis and are consistent with the RNA-seq data indicating that fatty acid metabolism is more active in HAM and AML cells when compared to MDM.

Phenotypic and functional characterization of AML cells
HAM differentially expresses several cell surface receptors, such as CD206, CD64, CD11c, CD163, CD170, MARCO, human leukocyte antigen-DR isotype (HLA-DR), CD11b, and CD36, that can be used to distinguish HAM from other cell types (45)(46)(47). We character ized AML cell surface receptors by flow cytometry and confocal microscopy. We found that the AML cells exhibit increased expression of CD64, CD206, MARCO, CD163, CD11c, MerTK, and CD170, and reduced expression of CD11b, CD36, and HLA-DR compared to MDM (Fig. 6A through J; Fig. S5A). These data were verified by confocal fluorescence microscopy, which revealed higher expression of CD200R, CD11c, CD206, CD163, MARCO, MerTK, CD170, CD68, and CD64 in AML cells and downregulation of CD36 and CD11b ( Fig. 6K through P; Fig. S5B and C). These changes in protein levels corresponded with the qRT-PCR data, where MRC1 and MARCO were highly expressed in AML cells versus MDM, and CD36 was highly expressed in MDM versus AML cells (Fig. 1C, D, and M). MARCO is a scavenger receptor that mediates binding and ingestion of unopsonized environmental particles (47). MARCO is highly expressed in AML cells compared to MDM ( Fig. 1D; Fig. 6C, O, and P), and AML cells have higher capacity to bind unopsonized fluorescent beads compared to MDM ( Fig. S5D and E).

AML cells secrete inflammation-related proteins
We assessed the amounts of secreted inflammation-related proteins from AML cells compared to MDM. Several secreted proteins reported for HAM were present in higher amounts (CD163, CXCL18, IL-13, and IL-4) in AML cells compared to MDM (Fig. 7A  through D). Also similar to the profile reported for HAM, the levels of MMP7, MMP9, CCL22, TNFα, and IFN-γ were decreased significantly in AML cell supernatants compared to MDM (Fig. 7E through I), which correlates with the RNA-seq data (Fig. 4F). Soluble ICAM-1, M-CSF, IFN-α, RAGE, and IL-1β levels in AML cell supernatants were similar to MDM (Fig. 7J through N). We found GM-CSF and IL-10 in the cell supernatants as would be expected ( Fig. 7O and P). Similar profiles were observed in AML cells differentiated from purified monocytes (Fig. 7Q through V).

HAM and AML cells demonstrate increased uptake and intracellular growth of M.tb and increased persistence of SARS-CoV-2 compared to MDM
AMs are the first myeloid cells to phagocytose airborne M.tb and allow for M.tb growth (48)(49)(50). To investigate the phagocytic capacity of HAM, AML, and MDM, we infected each cell type with mCherry-H 37 R v M.tb. We found that HAM and AML cells have similar increased phagocytic capacity compared to MDM as demonstrated by confocal micro scopy ( Fig. 8A and B). In addition to calculating mean bacteria per cell, we found that 62.6% ± 1.5 HAM and 59.8% ± 1.5 of AML cells contained one or more bacterium, compared to 41.6% ± 8.4 of MDM (mean ± SEM, n = 3-4). We also found increased intracellular M.tb growth in HAM and AML cells over time (Fig. 8C). Similarly, murine AMs are more permissive to M.tb infection and growth than BMDMs (16). We next explored the cellular response to SARS-CoV-2 infection. The Vero E6 kidney cell line is extensively used in COVID-19 research for viral propagation, passaging and stock preparation, and antiviral assays (51). These cells highly express the angiotensin converting enzyme 2 (ACE2) receptor for SARS-CoV-2 attachment but lack the coreceptor transmembrane serine protease 2 (TMPRSS2) protease that also participates in entry into human cells (52). Viral entry into Vero E6 cells is reported to be cathepsinmediated but may not mimic viral infection of human cells (53). Cells expressing both ACE2 and TMPRSS2 are highly permissive to infection. We confirmed that Vero E6 cells express ACE2 receptor but not TMPRSS2. ACE2 basal level receptor expression was higher in AML cells than in MDM but lower than in Vero E6 cells (Fig. S6A). TMPRSS2 basal level receptor expression was higher in AML cells than MDM. Vero E6 cells did not express TMPRSS2 (Fig. S6B). Basigin (CD147) is another reported route of cellular entry for SARS-CoV-2 (54). We found that BSG/CD147 expression is similar in AML cells, MDM, and Vero E6 cells (Fig. S6C).
The cellular response to SARS-CoV-2 in primary human macrophages, particularly HAM, has not been explored. We infected HAM, AML cells, MDM, and VeroE6 cells using our previously described replication-competent recombinant SARS-CoV-2 expressing bioluminescent luciferase (Nluc) and fluorescent mCherry reporter genes (rSARS-CoV-2/ mCherry-Nluc) (22) and continuously monitored viral infection over time by the Cytation 5 fluorescence live cell imaging system ( Fig. S6D; Movies S1 through S4). These data demonstrated that similar to HAM, AML cells have rapid, increased SARS-CoV-2 uptake and persistence over time without replication in contrast to Vero cells, where SARS-CoV-2 entry is initially lower but propagation is higher (Fig. 8D and E; Movies S1 through S4). Finally, we assessed the SARS-CoV-2 persistence after DAPI counterstaining in HAM and AML cells at day 5 post-infection (Fig. 8F). We observed lower infection and persistence of SARS-CoV-2 in MDM than in HAM or AML cells.

DISCUSSION
We developed an AML cell model using blood-derived monocytes that recapitulates unique features of AMs, which require TGF-β and GM-CSF for development along with IL-10 for maintenance and have a critical role in catabolizing lipid surfactant. We show that culturing human blood-derived monocytes, either purified or with other PBMCs, in a cocktail containing surfactant, IL-10, TGF-β, and GM-CSF allows monocytes to differentiate into AML cells and maintain this phenotype over time in culture. Using multiple complementary approaches, we demonstrate that these AML cells mimic AMs in that they have (i) similar morphology containing lipid bodies; (ii) similar gene expression patterns with only 6.4% of detectable genes showing a significant change in expression between AML cells and HAM; (iii) increases in gene and protein levels of CD206, PPAR-γ, MARCO, and other key markers for AMs; (iv) similar expression of genes in key pathways (Continued on next page) Research Article mBio been described for healthy adult HAMs (30,31). In contrast to MDM, AML cells did not respond to LPS to induce glycolysis, as reported for HAM (37). However, previous data suggest that virulent M.tb can drive the shift toward aerobic glycolysis in HAM (42). These findings suggest the differential metabolic activity of HAM in response to a pathogen. The study of HAM has been hindered by the invasive and costly nature of BALs, which require extensive prescreening tests and experienced pulmonologists. In contrast, venipuncture is significantly cheaper, less invasive, and requires less specialized training than needed for the BAL procedure. People are generally more willing to undergo venipuncture than BAL, and venipuncture can occur more frequently, thus making the recruitment of donors easier for studying blood-derived cells. Our AML model is based on culturing blood monocytes with commercially available products, and AML cells can be purified from the lymphocytes through a simple and inexpensive adherence step, thus providing a HAM model that is much more readily available to a range of labs.
A second challenge in working with HAM is that each BAL yields approximately 2-4 × 10 6 HAM per person, thus restricting studies to smaller experiments than what is feasible when working with cell lines. In contrast, from a full blood draw, approximately 50 × 10 6 monocytes (and thus potentially AML cells) can be recovered per donor, allowing for much larger studies. Cell number is a particular problem when working with mice because BAL results in only approximately 1-2 × 10 5 AMs per mouse. Thus, many investigators pool cells from multiple mice to have sufficient cells for one experiment. This need makes studying heterogeneity in the population challenging. The ability to work with blood-derived AML cells obviates the need to pool samples, allowing for studies that interrogate donor heterogeneity. Finally, some important inflammatory pathways differ in humans and mice (55). Thus, working with primary human cells has a distinct advantage. Studies using non-human primate (NHP) BAL cells to obtain AMs are another option. However, there is more limited access to NHPs, studies are expensive, and there is strict IRB regulation similar to humans. It will be interesting to adopt our AML model to NHP monocytes for greater accessibility and application.
An alternative approach to studying AMs is to digest animal, typically murine lungs, and study total lung macrophages. This results in the recovery of many more mac rophages than BAL. However, the lung contains a range of macrophages, including alveolar, interstitial, and intravascular. AMs constitute about 10% of lung macrophages and have a unique phenotype relative to interstitial and intravascular macrophages; thus, the study of lung macrophages is not specific to AMs and the majority of recovered macrophages from lung tissue are actually IMs. AMs can be isolated from digested lung tissues based on cell surface receptor expression, but the cell yield is still low.
A third challenge with studying AMs is that their phenotype rapidly changes after removal from the lung, which makes long-term studies and mechanistic work challeng ing. Importantly, we show that culturing AML cells in Infasurf, IL-10, TGF-β, and GM-CSF allows them to better retain a HAM phenotype once adhered. This is expected to allow investigators to conduct longer mechanistic studies than what is currently feasible in vitro and so represents a large step forward for the field.
Several studies have highlighted the unique susceptibility of HAM to airborne bacterial and viral infections (12,48,49,56). In an effort to begin to demonstrate the applicability of AML cells for infection studies, we analyzed the uptake and intracellular growth of M.tb and SAR-CoV-2 and determined that these parameters were similar in AML cells and HAM. For example, AML cells had increased phagocytosis and intracellular growth of M.tb (50). Regarding SARS-CoV-2, we detected moderate expression of ACE2 and TMPRSS2 in AML cells. It is of interest that AML cells demonstrated rapid uptake of SARS-CoV-2 and subsequent persistence without detectable growth. Viral persistence Research Article mBio of SARS-CoV-2 is reported in research and clinical settings (57,58). Infection of NHPs showed that rhesus macaques and baboons develop moderate SARS-CoV-2 viremia with COVID-19-related pathology and some degree of viral persistence (59). Analysis of human PBMCs showed viral persistence in the form of fragmented SARS-CoV-2 RNA and the presence of S1 viral proteins in the post-acute sequelae of COVID-19 patients for up to 15 months post-acute infection (60). Finally, a humanized mouse model identified that SARS-CoV-2-infected human lung resident macrophages contribute to hyperinflammation of the lung by upregulation and release of IL-1 and IL-18 and activation of the inflammasome (61). AML cells hold promise for further investigation of the molecular and cellular events enabling SARS-CoV-2 uptake and persistence in human cells.
Prior studies have generated macrophages from primary human cells using specific factors such as GM-CSF or MCSF to generate either M1 or M2 types of macrophages. HAM shares both M1 and M2 characteristics and thus cannot be characterized as such (45). Our approach was to optimize dosage by combining critical factors into a cocktail (Infasurf, GM-CSF, TGF-β, IL-10) to better recapitulate the alveolar environment of AMs. We based the concentration range of factors to study on prior in vitro studies. It is difficult to measure the true concentration of these factors in the alveolar hypophase. We did not include all known soluble factors of the alveolar hypophase (e.g., eicosanoids) in our model but found that factors previously found to be critical for AM development recapitulated the AM phenotype well. In future studies, it will be interesting to explore the role of other factors such as eicosanoids (especially, prostaglandin E2, PGE 2 ) in AML longevity (29).
As noted above, previous studies, including our own, have determined that AMs will change their phenotype when removed from the lung (16,17). Thus, isolation and handling of AMs could affect their transcriptome and raise the concern that the ex vivo-studied AMs have deviated from the in vivo phenotype. Studying human AMs in vivo is not possible. We contend that isolated HAM analyzed soon after harvest (within 6 hours) without manipulation most closely approximates the in vivo pheno type (17). Including the cocktail during in vitro cultivation maintains this phenotype. The inclusion of these factors needs to be considered when studying these cells in a variety of lung inflammation contexts. Finally, although we demonstrate that cultivating human monocytes in the cocktail enables cells to differentiate in a manner that more closely recapitulates HAM, the use of MDM in culture has generated very significant and useful data over the years in delineating human macrophage innate immune responses confirmed in vivo and humans. Thus, the success of this and other new models does not necessarily diminish the importance of data from older models.
A bottleneck in studying respiratory biology has been the ability to study AMs in vitro. The lungs are a unique organ that must balance pro-and anti-inflammatory responses and much of this is mediated through AMs, which have a unique biology relative to other tissue macrophages. We expect that this AML model will significantly advance respiratory biology research, including for inflammatory diseases like chronic obstructive pulmonary disease, asthma, and cystic fibrosis, and infectious diseases, including COVID-19 and tuberculosis. This model should also aid in assessing the impact of aging on AM biology. Finally, this model will aid in translational human studies, that is, therapies and vaccines. Thus, we expect this AML model will help in identifying key pathways/responses to interrogate in the context of the lung alveoli and can be further validated in less readily available primary HAM as necessary.

ACKNOWLEDGMENTS
This work was supported by National Institutes of Health (NIH) award (AI136831) (to L.S.S.), Texas Biomed Cowles and Forum Postdoctoral Fellowship (to S.P.), and NIH awards (1R01AI161175-01A1) and (1R01AI161363-01) (to L.M.-S.). Research reported in this publication was supported by the NIH-National Institute of Allergy and Infectious Diseases under award no. P30AI168439. Research reported in this publication was also supported by the Office of the Director, NIH award (S10OD028653). RNA sequencing

ETHICS APPROVAL
PBMCs were isolated from human peripheral blood collected from healthy adult donors, following a Texas Biomed-approved institutional review board (IRB) protocol (20170315HU). HAMs were isolated from BAL of healthy adult human donors, follow ing a Texas Biomed-approved IRB protocol (20170667HU). All donors for these studies provided informed, written consent.

ADDITIONAL FILES
The following material is available online.