Lung macrophage heterogeneity reflects anatomical niche occupancy and function

As tissue-specific macrophages, AMs display a unique phenotype and transcriptional signature, which is dictated by the lung environment (Gautier et al., 2012). AMs live on the luminal side of the alveolar niche, surrounded by type I and type II alveolar epithelial cells, capillary endothelial cells, and alveolar interstitial fibroblasts, that provide the cytokine-rich nurturing scaffold for AMs to thrive (Westphalen et al., 2014) (Figure 1). They are abundant in the lung as just the upper lobe of the human lung already contains 1.5 × 10⁹ AMs, the majority of which are localized to the diffusing area of the lung and a minority in the conducting small airways (Hume et al., 2020). Lung IMs mainly reside in the interstitial space around the bronchovascular bundle, though a few have been found within the airspace and alveolar interstitium (Hume et al., 2020). The IMs locate adjacent to many nonhematopoietic cell types such as neurons, lymphatic vessels, blood vessels, and hematopoietic cells (Chakarov et al., 2019; Gibbings et al., 2017; Keerthivasan et al., 2021; Schyns et al., 2019; Ural et al., 2020). In contrast to AMs, IMs are present throughout the body and display a similar transcriptional profile across multiple organs, suggesting they have a conserved, specialized function regardless of the organ they reside in (Dick et al., 2022; Gibbings et al., 2017).

Precise identification of lung macrophages by flow cytometry is complex. At steady state, distinguishing IMs from AMs, dendritic cells (DCs), and extravascular monocytes is still relatively straight-forward with sufficient markers (Gibbings and Jakubzick, 2018a, b; Gibbings et al., 2017; Leach et al., 2020) (Figure 1; Table 1). A reliable discrimination method to distinguish intravascular myeloid cells from tissue residents is by intravenous injection of CD45 antibodies 5 min prior to the sacrifice of the animal, which is a more pervasive method than perfusion alone (Desch et al., 2016; Gibbings et al., 2017). The differential expression of CD88 and CD26 helps distinguish monocytes and macrophages (both are CD88⁺) from CD26⁺ conventional DCs (Dutertre et al., 2019; Gibbings and Jakubzick, 2018a; Guilliams et al., 2016; Li et al., 2022; Nakano et al., 2015). C5AR1 (the gene encoding CD88) and DPP4 (CD26) may also be used at the transcriptional level. Exclusion markers are equally important to clean up the dataset within the myeloid gate, such as Ly6G for neutrophils in mice and CD15 in humans (Table 1).

In humans, AMs can be identified by flow cytometry and CITE-seq as large auto-fluorescent cells, expressing surface HLA-DR, CD43, CD88, CD169, CD206, CD10 (MME), CD11c, CD36, CD141, CD64, the scavenging antigen uptake receptors MARCO, and low amounts of CD14 (Bharat et al., 2015; Bosteels et al., 2020a; Desch et al., 2016; Leach et al., 2020; Misharin et al., 2013; Snelgrove et al., 2008; Vermaelen and Pauwels, 2004; Yu et al., 2015). In mice, CD11c, SiglecF, CD64, F4/80, and MerTK are useful phenotyping markers. In humans and mice, key AM genes include FABP4, MARCO, and PPARγ (Leach et al., 2020; Serezani et al., 2022). In mice, other robustly expressed genes are type 4 carbonic anhydrase (Car4), chitinase-like protein Ym1 (Chil3), CD2, several of the ribonucleases (Ear1), and keratins (Krt79) (Sajti et al., 2020; Schneider et al., 2014; van de Laar et al., 2016) (Figure 2; Table 1).

Open in a separate window

Figure 2.

Development of alveolar macrophages from fetal precursors

Around embryonic (E) day 18.5, fetal monocytes exit the lung vasculature and under the influence of autocrine TGF-β turn on a macrophage differentiation program characterized by expression of macrophage lineage-determining transcription factors. These cells next migrate to the alveolar niche where around the day of birth (DOB), they develop into immature TRAM, under the influence of granulocyte-macrophage colony-stimulating factor (GM-CSF) produced by type II alveolar epithelial cells and helped by macrophage-intrinsic transforming growth factor beta (TGF-β). This induces the expression of the transcription factor PPARγ that controls a large part of the AM-specific transcriptome. Mature TRAMs also contain transcription factors that allow for AM self-renewal. AM identity is instructed by other epithelial-cell-derived molecules like surfactant proteins A and D (SP-A and SP-D), TGF-β, and isthmin.

AMs have been previously considered as a relatively homogenous population of cells. Single-cell sequencing studies of healthy human AMs are quickly revealing their heterogeneity (Mould et al., 2021). At least 4 superclusters of AMs containing 14 subclusters can be discriminated across several individuals, with IFI27 and APOC2 as defining differentially expressed genes. One of the major determinants of heterogeneity appears to be the highly regulated production of certain combinations of chemokines, metallothioneins, interferon (IFN)-inducible genes, cholesterol-biosynthesis-related genes, and IGF1 by AM subsets. Moreover, analysis across several diseases including cystic fibrosis, chronic obstructive pulmonary disease (COPD), and COVID-19 demonstrated that these superclusters and their subclusters are universally observed (Li et al., 2022).

Unlike AMs, which can easily be isolated from the lung environment by bronchoalveolar lavage, IMs require a digestion and enrichment protocol from whole mouse or tissue excised human lungs to obtain high enough cell counts for transcriptional and functional analyses (Atif et al., 2018; Gibbings and Jakubzick, 2018a). IMs also rapidly die upon removal from their tissue niche. Therefore, to observe heterogeneity of IMs using single-cell RNA sequencing (scRNA-seq), in addition to rapid processing, it is important to exclude contaminating cell types, including AMs and recMacs, since both cell types are highly abundant and will dilute the granularity of an IM dataset. In mice, IMs universally express Pf4, C1q, and CD88. In humans, IMs selectively express LGMN and MARCKS. However, contrary to the mouse, human AMs also express C1Q, and therefore C1Q cannot be used as a distinguishing AM versus IM gene (Leach et al., 2020; Li et al., 2022). In addition to differentially expressed genes (DEGs) (Table 1), key reference genes help support the identification of a given scRNA-seq cluster. These genes may not be included in the top DEGs and might be shared with other cell types but are essential to validate the given classification of a cluster. For instance, macrophage identity genes include transcription factors, growth factors, and cell surface molecules such as Mafb, Csf1r, and Cd14. In a recent scRNA-seq study, three IM subtypes have been identified across multiple organs, including the lung, and based on the expression of Timd4, Lyve1, Folr2, Ccr2, and H2-Ab (Dick et al., 2022; Gibbings et al., 2017). Others have identified only two IM subtypes: CD206^hi and CD206^int IMs (Chakarov et al., 2019; Lim et al., 2018; Schyns et al., 2019; Ural et al., 2020) (Figure 1; Table 1). Even though these populations encompass a simplistic view, their identification in mouse tissue by flow and immunohistochemistry (IHC) can be performed by staining for C1q, in addition to MerTK, CD64, F4/80, or the CX₃CR1 reporter mouse (Hume et al., 2020; McCubbrey et al., 2017; Meghraoui-Kheddar et al., 2020; Ogawa et al., 2022).

Added complexity in identifying lung macrophages arises during the perturbation of homeostasis, such as under inflammation, sterile injury, or aging. Under these conditions, recMacs, which derive from circulating classical Ly6C⁺ monocytes, migrate into the lung and can be easily confused with TRMs. RecMacs upregulate all the surface molecules used to distinguish IMs from other mononuclear phagocytes in steady state (Bosteels et al., 2020b; Gibbings et al., 2015; Nakano et al., 2015) (Figure 1). RecMacs initially display a remarkably altered phenotype and responsiveness. Some recMacs can acquire overlapping features with TRMs, such as recruited alveolar macrophages (recAMs) (Aegerter et al., 2020; Arafa et al., 2022; Guillon et al., 2020; Misharin et al., 2017), while other recMacs are phenotypically distinct, providing functional roles in epithelial repair, induction of fibrosis, or enhanced inflammation (Grant et al., 2021; McCubbrey et al., 2018; Misharin et al., 2017; Mould et al., 2019; Ogawa et al., 2022). The phenotypic overlap between recMacs and TRMs, combined with the promiscuity of many myeloid markers in inflammatory contexts such as those expressed on inflammatory DCs, has resulted in descriptions in the literature of an overarching and generic functional response by “lung macrophages” that might be more accurately annotated to individual cell types in future research. Sequencing either at the transcriptomic or epigenetic level, or making use of a traceable monocyte-descent marker in adoptive transfer models, has been shown to provide a reliable way to distinguish recMacs from TRMs.

Lung macrophages develop from multiple precursors in homeostasis and inflammation

Over 50 years ago, van Furth proposed the mononuclear phagocyte system, in which the central theme was that all TRM are derived or replenished from adult circulating monocytes (van oud Alblas and van Furth, 1979). From the start, this model was contested, as radiation chimera studies in the rat, parabiotic studies in the mouse and monocyte depletion experiments demonstrated that resident lung macrophages can self-replenish without input from circulating monocytic progenitors (Sawyer et al., 1982; Tarling and Coggle, 1982; Volkman and Gowans, 1965). More recent research from several labs has shown that mouse and human AMs develop prenatally from circulating fetal progenitors, that occupy the alveolar niche during perinatal lung growth, and self-maintain later in life through local low-grade proliferation in the absence of inflammation, without input from circulating adult monocytes (Figure 2) (Evren et al., 2022; Gomez Perdiguero et al., 2015; Guilliams et al., 2013; Hoeffel et al., 2015; Jakubzick et al., 2013; Mass et al., 2016; Schulz et al., 2012; van de Laar et al., 2016; Yona et al., 2013). scRNA-seq analysis and Brdu pulse-chase experiments indeed show that at any given time, up to 5%–10% of AMs have Ki67 expression or active BrdU incorporation (Coggle and Tarling, 1984; Hashimoto et al., 2013; Li et al., 2022; Mould et al., 2021). With increasing age of mice, there seems to be a larger component of resident AM replaced from circulating monocytes (Liu et al., 2019); however, parabiotic mice suggest that replacement is minimal if the mouse experiences no major injury or insult (Hashimoto et al., 2013; Jakubzick et al., 2013). In human patients undergoing lung transplantation, macrophage origin, and replacement can also be followed, and reports supporting donor replacement by host recMacs (Byrne et al., 2020), as well as persistence of donor TRMs have been shown (Eguiluz-Gracia et al., 2016; Nayak et al., 2016). More research in humans and mice is warranted to determine what signals control TRM replacement.

Within the first week of murine life, alveolar epithelial-cell-derived granulocyte-macrophage colony-stimulating factor (GM-CSF, encoded by Csf2) provides the instructive cytokine signal that programs fetal monocytes to become AMs in the alveolar niche, along a gradual differentiation scheme (Figure 2) (Gschwend et al., 2021; Guilliams et al., 2013; Schneider et al., 2014; Suzuki et al., 2008; Tazawa et al., 2019). While M-CSF (Csf1) deficient mice have AMs, GM-CSF (Csf2)- or GM-CSFR (Csf2r)-deficient mice lack AMs from birth. This instructive process is enforced by autocrine transforming growth factor-β (TGF-β) signaling and lung epithelial α_νβ₆ integrin-mediated presentation of TGF-β to AMs (Branchett et al., 2021; Koth et al., 2007; Meliopoulos et al., 2016; Yu et al., 2017). Triggering of these cytokine receptors leads to stable expression of the transcription factor (TF) PPARγ. To be able to respond to GM-CSF, the progenitors need to reach the alveolar space, and the actin binding protein L-Plastin (encoded by Lpl) is crucial for reaching the alveolar niche and upregulation of PPARγ (Todd et al., 2016). PPARγ is well known for regulating several lipid handling transporters and enzymes involved in surfactant turnover, and in the absence of this TF, AMs are absent and replaced by pro-inflammatory macrophages as pulmonary alveolar proteinosis develops (Schneider et al., 2014). AMs have also been shown to rely on Egr2, Bhlhe40, Bhlhe41, Bach2, and Klf4 for development and self-maintenance. These TFs instruct and maintain the unique identity and function of AMs by binding to enhancer regions of AM specific genes, on top of the generic macrophage transcriptional program that is controlled by pioneer or lineage-determining TFs such as C/EBPβ and PU.1 that appear earlier during AM development (Figure 2) (Daniel et al., 2020; Gorki et al., 2021; Mass et al., 2016; Rauschmeier et al., 2019; Sajti et al., 2020; Schneider et al., 2014; Yu et al., 2017). Mouse AM also have a great capacity for proliferation and stem cell-like replenishment, partially imparted by the low expression of the transcription factors MafB and cMaf, and high expression of Bhlhe40 and Bhlhe41, which control self-renewal genes (Busch et al., 2019; Rauschmeier et al., 2019; Soucie et al., 2016; Subramanian et al., 2022). Strikingly, VitD3 is required for restricting proliferation because accumulation of AMs is seen in Vdr^−/− mice (Hu et al., 2019).

Steady-state IMs also originate from the yolk sac and are then replaced by fetal liver monocytes. IMs then either self-replenish without further input from circulating monocytes or are gradually replaced by monocytes at different rates and to a different extent according to the subtype of IM (Chakarov et al., 2019; Dick et al., 2022; Gibbings et al., 2017). Whereas AMs require GM-CSF, IMs are dependent on M-CSF for development and maintenance in the lung, as exemplified by the fact that Op/Op mice lack IMs due to deficiency in M-CSF and inhibitors of the kinase activity of CSF-1R or blocking CSF-1R antibodies deplete IMs (Lim et al., 2018; Ural et al., 2020). More research is needed on additional niche-instructive signals and precise TFs that govern the differentiation program of IMs, as this could also open possibilities for selective targeting of these cells. Using epigenetics analysis and TF binding motif enrichment on gene expression differences between mouse strains, important TFs have already been predicted, but remain to be validated (Sajti et al., 2020).

During lung infection or other inflammatory insults, the diversity of macrophage populations expands, and populations of recruited macrophages, recAMs and recMacs, replace or join the pool of resident AMs and IMs (Figure 1). These cells develop from CCR2-positive Ly6C^hi monocytic precursors, express CD115, and depend on the cytokine M-CSF for their development in lung tissues and alveolar space. This pathway of lung inflammatory macrophage development is the reflection of the original mononuclear phagocyte system concept in which the circulating monocyte is the precursor for tissue macrophages (Volkman et al., 1983).

AM function is instructed by homeostatic and inflammatory environmental conditions

AMs adhere to and often crawl tightly attached to the luminal side of alveolar epithelial cells. This important surveillance function is mediated by the actin cytoskeleton and dysregulated in the absence of surfactant-protein A (Meliopoulos et al., 2016; Phelps et al., 2011). Only one in every three alveoli has an AM at any given time point (Westphalen et al., 2014), but migration of AMs from one alveolus to another by squeezing through the connecting pores of Kohn can occur, under the integrin-dependent guidance of the complement system component C5a (Neupane et al., 2020). In their exposed position, AMs continuously capture, phagocytose, conceal, and neutralize a large cargo of inhaled pathogens and particles (Westphalen et al., 2014). This is done in an immunologically “silent” way, i.e., without triggering neutrophil influx and causing excessive inflammation, which could damage the delicate alveoli. It has been estimated that the pool of AMs of the rat can even conceal up to 10⁹ inhaled heat killed bacteria before the adaptive immune system is set in motion (MacLean et al., 1996). By expressing scavenging and apoptotic cell receptors like MARCO and Axl, AMs also avidly efferocytose apoptotic granulocytes and epithelial cells and phagocytose cellular and extracellular matrix debris that deposits on the alveolar wall (Fujimori et al., 2015; Roberts et al., 2017). During efferocytosis, AMs produce epithelial repair cytokines, like trefoil factor 2 (Hung et al., 2019). In addition to their phagocytic function in the innate immune system, AMs handle and recycle surfactant, a phospholipid detergent that keeps alveoli in an open inflated state, crucial for gas exchange. Genetic or acquired deficiency of AMs causes surfactant accumulation and leads to pulmonary alveolar proteinosis in humans and mice (Dranoff et al., 1994).

When their phagocytic function is surpassed, AMs initiate inflammation via the production of chemokines and cytokines (such as type I IFNs, TNF-α, and IL-1β) that recruit and activate neutrophils, monocytes, and DCs (Goritzka et al., 2015; Mould et al., 2019). The inflammatory response is not evenly distributed among all AMs recovered from the lungs, and chemokine production appears in a highly coordinated manner (Li et al., 2022; Xu-Vanpala et al., 2020), suggesting the existence of a specific transcriptional program that dictates a specialized function, tailored to the need of the stimulus, or the specialized location in the lung. As well as their ability to initiate inflammation, AMs are also responsible for resolving inflammation through the secretion of immunoregulatory cytokines like TGF-β, IL-1RA, and prostaglandins (Branchett et al., 2021; Jardine et al., 2019; Thepen et al., 1991).

The unique environment occupied by AMs and signals provided by the surrounding stroma and epithelium has been shown to confer many functional properties of these cells, which is well revealed following their removal from the lung (Bleriot et al., 2020; Janssen et al., 2020; Roquilly et al., 2020; Subramanian et al., 2022; Svedberg et al., 2019). Following ex vivo culture, AMs lose expression of genes involved in lipid metabolism, oxygen response, TGF-β signaling, and adhesion molecules. They also respond differently in the absence of lung-derived restraints, becoming more responsive to the cytokine IL-4 ex vivo (Svedberg et al., 2019) and more phagocytic and efferocytic (Janssen et al., 2008). Importantly, the transfer to plastic is in itself a plastic phenotype, as a bona fide in vivo AM profile is restored once ex vivo cultured cells are re-established in the lung (Subramanian et al., 2022). The environmental signals that constrain AM efferocytosis include surfactant proteins A and D that signal through SIRPα (Janssen et al., 2008). The epithelium and stroma in general are likely responsible for a multitude of AM phenotypes, but this has not yet been dissected in full detail. However, with acute and chronic inflammation and aging, the alteration of the AM niche is reflected in their functional outcomes (McQuattie-Pimentel et al., 2021).

Cellular metabolism dictates unique AM functions

The environment of the lung has a direct influence on cell metabolism that is tightly linked to function, particularly with respect to differentiation and inflammation (Mould et al., 2017; Svedberg et al., 2019). The environment of the lung is well supplied with oxygen and lipids, yet actively depleted of glucose to prevent bacterial outgrowth (Baker and Baines, 2018). This is reflected in a low ability of luminal AMs to use glucose (Mould et al., 2017; Svedberg et al., 2019), a constraint imposed by the lung environment, as AMs become more glucose responsive outside of the lung environment; upregulating the glycolysis as well as PI3K-Akt, mTOR, and C/EBPβ pathways (Subramanian et al., 2022; Svedberg et al., 2019). Interestingly, AMs derived from different regions of the human lung appear to utilize different metabolic pathways, again reflecting tissue dictated heterogeneity (Lavrich et al., 2018). Macrophages with an “alternatively activated” phenotype use the tricarboxylic acid (TCA) cycle to provide ATP via oxidative phosphorylation, whereas a “classically activated” macrophage will divert citrate from the TCA cycle for fatty acid oxidation, and preferentially utilize glycolysis (O’Neill et al., 2016). The nuclear receptor PPARγ directly regulates genes involved in lipid metabolism and fatty acid β-oxidation to enable macrophage function (Schneider et al., 2014) (Figure 3). However, just as the M1/M2 dichotomy of macrophages has proved reductive, the metabolism of a cell within a tissue cannot be easily simplified. The many metabolic pathways that supplement and regulate each other within a cell are interconnected to allow rapid environmental adaptation and response. Upon infection, the lung is liable to fast and extreme changes in glucose concentration (Baker and Baines, 2018). Increased glucose provides a benefit for cellular growth and can enable proliferation. Additionally, the generation of NADPH by the pentose phosphate pathway during glycolysis is required to produce reactive oxygen species and anti-bacterial defense. The metabolic profile of recAMs has been shown to differ, and even one-month post-influenza infection recAMs have not yet upregulated key AM genes involved in lipid- or fatty acid metabolism (Aegerter et al., 2020) Figure 3), and may have a reliance on creatine metabolism (Guillon et al., 2020).

Open in a separate window

Figure 3.

Metabolism of alveolar macrophages in homeostasis and infection

Steady-state AMs exist in an environment of low glucose and other environmental signals that instruct their metabolism and inflammatory output. PPARγ is a key AM transcription factor induced by GM-CSF, which regulates lipid metabolism, a key feature of AM functionality. In contrast, recAMs develop in an environment of higher glucose accessibility, and due to an altered origin or altered environmental signals, remain in an altered metabolic state characterized by high HIF1a that results in increased of inflammatory cytokines and nitric oxide.

Inflammation and infection results in a mixed population of AMs: Recruited monocytes contribute to an altered responsiveness of the lung

During acute lung infection, or inflammation, a reduction in the resident AM population is often seen and referred to as the “macrophage disappearance reaction”, and this is certainly the case in SARS-CoV2 infection (Grant et al., 2021; Liao et al., 2020). It is still unclear whether tissue resident alveolar macrophages (TRAMs) are directly infected and die, or whether TRAMs disappear by loss of a survival signal due to the infection of airway or alveolar epithelial cells. The efferocytic role of AMs promotes a pro-repair, anti-inflammatory phenotype, which may be counterproductive during acute inflammation, and therefore their temporary disappearance may prove beneficial to license effective inflammation (Herold et al., 2011; Miki et al., 2021). The disappearance of AMs could also be caused by their detachment from epithelial cells and subsequent disposal by mucociliary transport. Whether AMs die or migrate during inflammation, the lungs must urgently fill the alveolar niche to preserve lung function and gas exchange. The signals that govern this are still unknown. Repopulation may occur either by local proliferation of the remaining resident, terminally differentiated AMs (Jenkins et al., 2011; Minutti et al., 2017; Mould et al., 2019), or differentiation of recruited Ly6C^hi monocyte precursors (recAMs) (Aegerter et al., 2020; Gschwend et al., 2021; Machiels et al., 2017; Mould et al., 2017). RecAMs have a different transcriptional, epigenetic, and functional phenotype than embryonically derived AMs, despite a similar expression of CD11c and CD64 and slightly lower SiglecF. However, recAMs do not comprise the entire AM population. TRMs are capable of simultaneous expansion and often seen as impervious to the viral insult, retaining a naive AM phenotype (Aegerter et al., 2020). In contrast, bone marrow-derived monocytes, as precursors of multiple populations, have considerable potential for education. As monocytes differentiate into recAMs during acute or chronic inflammation, the local signals that instruct AMs during homeostatic development may be lost or altered (Larsen et al., 2021). This may result in an interrupted or alternative differentiation of recAMs, who retain open chromatin regions relative to their monocyte origin for an extended period. This epigenetic legacy, as a natural result of differentiation (Aegerter et al., 2020; Chen et al., 2022), is distinct from “trained immunity". Trained immunity is defined operationally as a process by which an initial pathogenic insult leads to a lasting epigenetic imprint of innate immune cells, such that cellular responses become faster and greater upon repeated contacts with homologous or heterologous stimuli (Divangahi et al., 2021). While the “imprinting” of resident AMs has been reported, this is always in the apparent absence of recAMs, outlining two possibilities. Either monocytes are recruited but not sufficiently detected and recAMs are responsible for altered population reactivity, as previously shown (Aegerter et al., 2020; Chen et al., 2022; Machiels et al., 2017; Misharin et al., 2017). Or, in the absence of monocyte recruitment, the AMs gain plasticity. It has been demonstrated that AM phenotype is not always a result of intrinsic alterations to the cell but rather of extrinsic signals provided by the niche (Roquilly et al., 2020).

The altered reactivity of infection-derived recAMs have beneficial roles in some contexts. Compared with TRMs, they produce higher amounts of IL-6 to protect against Streptococcus pneumoniae and Mtb infection (Aegerter et al., 2020; Mata et al., 2021), express higher Arginase-1 which results in resistance to helminth infection (Chen et al., 2022), produce amphiregulin in response to cell injury signals to repair vascular injury in helminth infection (Minutti et al., 2019), and restrict type-2 inflammation in house-dust mite induced allergic asthma (Machiels et al., 2017). These phenotypes are surprisingly long term and have persisted over a month following recruitment. In other instances, like in severe COVID-19, the acute replacement of homeostatic AMs by pro-inflammatory and profibrotic monocyte-derived cells is detrimental. The considerable cytokine (IL-1, IL-6, TNF-α) and chemokine production by RecAMs can contribute to lung damage and induce a systemic syndrome resembling cytokine release syndrome (Grant et al., 2021; Liao et al., 2020). .

AMs are generally described as having a less mature and more inflammatory phenotype during chronic inflammation or aging. Multiple causes can induce such a profile; altered instructions from the niche, accumulation of epigenetic changes in resident cells, or replacement of embryonic-derived AMs by recAMs. Equally, all three factors may play a role. AMs from older mice are more prone to replacement by monocyte-derived cells, which have down-regulated cell cycle genes, less Ki67 staining, and increased production of the transcription factors MafB and Mafc, as well as increased CD11b (Angelidis et al., 2019; McQuattie-Pimentel et al., 2021; Wong et al., 2017). Bulk sequencing and proteomics of the aged mouse lung demonstrate an inflammatory signature of increased IL-6, IL-1β, and TNF-α coinciding with the downregulation of PPARγ (Angelidis et al., 2019). These classic indications of monocyte origin have been confirmed by fate-mapping studies in the Ms4a3 mouse (Liu et al., 2019). Functionally, AMs from older mice have a lower phagocytic and efferoctyic capacity (Wong et al., 2017). However, these macrophages do not age in isolation. Deposition of matrix, and changes in the ECM have been identified in the aging lung. Old mice downregulate collagen XIV that may disrupt the ability of decorin to regulate TGF-β, a key signaling factor for AM maturation (Angelidis et al., 2019). Conversely, collagen IV is upregulated, and increased expression of matrix genes and growth factors (Areg, Ntf3, Vegfa, Igf1) are seen. Indeed, the cellular niche has been shown to impose an aged phenotype on AMs because old AMs transferred into young mice were able to restore their “young phenotype” and vice versa (McQuattie-Pimentel et al., 2021). The transfer of young macrophages to old mice is able to provide protection during influenza (Wong et al., 2017). While conducting these studies in humans is technically difficult, parallels can be drawn, as a more inflammatory profile is also seen in AMs from older humans, with a lower expression of cell adhesion and proliferation markers (McQuattie-Pimentel et al., 2021). However, a caveat to many murine aging and inflammation studies is the relatively sterile environment in which they are conducted. This contrasts with humans, who experience considerably more exposure to ambient pathogens, pollutants, and infections throughout their life. Thus, the situation in the human lung may be considerably different to that of the mouse, given the repetitive infections and insults which come with advancing age.

How exactly an altered lung environment changes AM polarization and function is highly complex and multifactorial. MicroRNAs (miRNAs) can regulate the expression of many genes simultaneously, and exosome-bound miRNAs could be important in the communication between niche cells and AMs in health and disease states, such communication heavily influenced by disease risk factors like smoking, alcohol consumption, and prior microbial exposure (Sturrock et al., 2014). For instance, miR-155 has been found to be induced in AMs of humans and mice exposed to lipopolysaccharide (LPS)-induced lung injury and cigarette smoke, and promote the expression of several pro-inflammatory genes, such that miR-155 deficient mice resist cigarette smoke-induced lung injury (De Smet et al., 2020); (Wang et al., 2016). MiR-155 is also involved in controlling the function of post-viral recAM, and in the absence of miR-155, post-influenza recAMs exposed to bacteria produced more IL-17 and IL-23 and mice are better protected against secondary bacterial infection (Podsiad et al., 2016). MiR-146a is an important miRNA induced not only by microbial exposure, but also by mechanostretch in AMs, e.g., induced by injurious mechanical ventilation. This miRNA dampens the activation of AMs and miR-146a overexpression reduces mechanical ventilation induced lung injury (Bobba et al., 2021).

Interstitial macrophages perform homeostatic and inflammatory functions

There are still few data on the functional role of IM subtypes apart from common macrophage features. Due to the proximity of IMs to non-hematopoietic cells, it is logical to assume that IMs communicate with adjacent cell types and that this crosstalk promotes the survival of both macrophages and adjacent cell types during the development and regeneration of the lung, although these mediators are still largely unknown. Lyve-1⁺ macrophages in the aorta communicate with hyaluronic acid on vascular smooth muscle cells to regulate collagen synthesis in the latter and maintain vascular tone by preventing arterial stiffness (Lim et al., 2018). In the lung, loss of Lyve-1⁺ IMs using a genetic targeting strategy leads to increased vascular leakage (Chakarov et al., 2019). However, in the absence of IMs in Csf1^op/op mice, lung function is not affected (Ural et al., 2020). Perhaps tissue-resident IMs are therefore not solely required for normal maintenance of lung nonhematopoietic cells, and instead, the crosstalk between a tissue-specific macrophage and its adjacent cell may have other functions. It has indeed been shown that CCR2-dependent cells and IL-4Rα signaling are required for lung regeneration following partial pneumonectomy (Lechner et al., 2017). Monocytes enter tissue continuously, surveying the environment during steady-state conditions and abundantly during inflammation (Jakubzick et al., 2013; Rodero et al., 2015). Recruited monocytes have a naive M0-like phenotype; a plastic cell capable of differentiation into a spectrum of phenotypes, including into an “alternatively activated” macrophage, which can secrete numerous growth factors required for structural cell development and survival (Mould et al., 2019). Thus, perhaps a combination of surveying recruited monocytes and TRMs will provide the cellular crosstalk with nonhematopoietic cells currently theorized.

Some IM functions have been suggested based on their transcriptional profile and anatomical location but have not yet been adequately investigated. IMs might have a role as neuroimmune communicators, leukocyte chemoattractants, and secrete immune modulators (Fu et al., 2022b; Ural et al., 2020). The production of immunoregulatory cytokines like IL-10, TGF-β, IGF-1, and reactive oxygen scavengers such as SOD2 suggest that IMs might have an important immunoregulatory role (Bedoret et al., 2009; Kawano et al., 2016; Martinez et al., 1997; Ural et al., 2020). In the induced absence of IMs (in Cd169Cre × Cx₃CR1^LStopL–DTR mice), sublethal influenza virus infection runs a more severe course, with heightened production of chemokines and cytokines (Ural et al., 2020). In a model of systemic LPS-induced endothelial injury, lung endothelium-derived Rspondin3 profoundly expands and alters the IM metabolism and epigenome to suppress lung inflammation (Zhou et al., 2020). In a model of hypoxia-induced acute respiratory distress syndrome, M-CSF is able to promote the differentiation of monocytes into tissue injury restoring macrophages (Mirchandani et al., 2022). The precise functional role of major histocompatibility complex (MHC) class II on lung IMs is unclear but could be important for providing a long-term residence niche for tissue resident CD4⁺ T cells, that derive tonic survival signals from this interaction (Snyder et al., 2021; Tang et al., 2022). This subset of MHCII⁺CX₃CR1⁺ IM has also been shown to be recruited to sites of emboli that clog up lung capillaries in the lung vascular filter, which contribute to the formation of T-cell rich inflammatory lesions around intravascular embolic antigens (Willart et al., 2009). All in all, there is much to explore of IMs function and tools to selectively investigate their functional role in health and disease are just beginning to unfold.

Macrophages demonstrate an altered phenotype and function in chronic lung diseases

In the previous sections, we discussed the many alterations in resident AMs and IMs, particularly the confounding effects of recruited macrophages that dominate the macrophage population in acute inflammatory and infectious settings. However, certain chronic lung diseases are also prone to altering lung macrophage biology. In some, these cells are even seen as important culprits in the disease process.

A major disease in which lung macrophages are of principal importance is tuberculosis, caused by the bacterium Mycobacterium tuberculosis (Mtb), responsible for over 1.5 million annual deaths. AMs are the primary cells to be infected by Mtb and ultimately act as a bacterial reservoir. AMs have a gene profile associated with easy access to iron (Pisu et al., 2020), and their slow cytokine and chemokine production in response to infection permits bacterial replication (Papp et al., 2018; Rothchild et al., 2019), which is even more prominent in AMs from infants (Goenka et al., 2020). Experimental infection of AMs in vivo but not in vitro caused upregulation of NRF-2 associated genes, which prevent host cell death and therefore permits bacterial replication (Rothchild et al., 2019), demonstrating niche-derived signals that promote infection. Elegant intravital imaging, combined with intravascular labeling, and monocyte-deficient mice demonstrated how resident luminal AMs actively migrate into the interstitium to transfer bacteria to IMs (Cohen et al., 2018; Huang et al., 2018; Mata et al., 2021). This is likely an anti-bacterial host response, as IMs exert high stress, low iron condition on Mtb (Huang et al., 2018; Pisu et al., 2020). RecAMs also represent an ample host for infection (Antonelli et al., 2010; Cohen et al., 2018), and monocyte-derived AMs derived from old mice harbor more bacteria (Lafuse et al., 2019). Strikingly, pulmonary-delivered Bacille Calmette-Guérin (BCG) has been shown to be contained within AMs for up to five months post-vaccination, demonstrating not only their longevity, but profound ability to act as self-sacrificing protectants of the lung (Mata et al., 2021).

Lung fibrosis can result from defective lung repair following acute or chronic lung injury, a manifestation of autoimmune collagen vascular disorders, or an idiopathic process, often with a very poor patient clinical outcome (Figure 4). In a vast body of literature, it was mainly proposed that lung fibrosis is associated with the differentiation of macrophages to a profibrotic “alternatively activated” phenotype expressing the mannose receptor CD206, producing profibrotic chemokines (like CCL17, CCL18, and CCL23) and several matrix remodeling enzymes and components like fibronectin. Indeed, macrophage-mediated fibrosis can be blocked by interfering with IL-4Rα signaling (Jakubzick et al., 2003; Kropski and Blackwell, 2019; McCubbrey et al., 2018; Ogawa et al., 2022; Prasse et al., 2006). In animal models of bleomycin-induced lung fibrosis, recAMs are the preeminent lung macrophage driving the profibrotic process (Gibbons et al., 2011; McCubbrey et al., 2018). Some recruited bone marrow-derived monocytes are already poised for an alternatively activated signature because they express the marker Chil3 or contain a segregated nucleus and atypical morphology (SAT-M monocytes), but it is also possible that pro-repair genes are instructed by environmental cues of the fibrotic niche (Fukushima et al., 2020; Ikeda et al., 2018; Satoh et al., 2017). Macrophage-driven fibrosis results most likely from an initially appropriate response triggered by the ingestion of dead cells or damage-associated molecular patterns (DAMPS) and acts to stimulate epithelial and vascular repair, through the release of growth factors, like VEGF-A, amphiregulin, platelet-derived growth factor (PDGF-A and B), and TGF-β (Minutti et al., 2019). However, when out of balance or prolonged, an epithelial-endothelial-fibroblast-macrophage fibrotic niche can occur around an injured area. Monocyte-derived recAMs recruited by CXCL12 initially depend on M-CSF rather than GM-CSF and produce excessive amounts of growth factors that stimulate the differentiation of myofibroblasts. These myofibroblasts, in turn, synthesize collagen, a process that can be maintained because recAMs have a long-life span (Aran et al., 2019; Fukushima et al., 2020; Greiffo et al., 2020; Joshi et al., 2020; Misharin et al., 2017; Satoh et al., 2017). Alternatively, in a silica-induced fibrosis model Lyve1^neg MHCII⁺ IMs that locally divide or recMacs produce many profibrotic cytokines like PDGF-A and PDGF-B, TGF-β, IGF-1, TGF-β receptors, and matrix remodeling enzymes. These macrophages express the complement component C1q, and it has been elegantly shown that C1qa-DTR mediated depletion of these cells abrogates fibrosis but enhances inflammation (Ogawa et al., 2022). In mice, IMs are also a source of IL-10, and triggering of IL-10R on recMacs stimulates a PDGF-A-dependent profibrotic program in a bleomycin model, suggesting that anti-inflammatory pathways might be intrinsically associated with fibrosis when unabated (Bhattacharyya et al., 2022).

Open in a separate window

Figure 4.

Involvement of lung macrophages in development of lung fibrosis

In homeostasis, TRAMs clear out apoptotic cells and other debris and, in this process, produce pro-repair genes like amphiregulin and TGF-β that is necessary for normal tissue homeostasis and repair. In fibrosis developing in response to injury, or due to an idiopathic epithelial injury, however monocytes are recruited and as these develop into recMacs or recAMs, they become programmed to contribute to fibrosis by producing profibrotic growth factors and chemokines, matrix metalloproteinases, matrix components, and other mediators. These cause activation of type 2 epithelial cells, which become hyperplastic, and activation of myofibroblasts that produce too much collagens. However, both epithelial cells and myofibroblasts then produce cytokines that can activate the recMacs to contribute further to fibrosis and can activate other niche cells to produce more collagens. This becomes a self-sustaining functional unit in progressive fibrosis.

In humans with various profibrotic lung diseases, a subset of AMs expressing SPP1 (osteopontin), CHI3L1 (YKL-40), MARCKS, IL1RN, PLA2G7, and MMP9 has been described (Reyfman et al., 2019). YKL-40 is a known severity marker in idiopathic lung fibrosis but is also able to promote fibrosis by further programming of recruited monocytes to become M2-like profibrotic fibroblasts by triggering the CRTH2 chemokine receptor (Cao et al., 2022). The chemokine CCL18 is strongly associated with lung fibrosis and is produced by an AM subset that clusters separately from other AMs, suggesting that only some AMs might be profibrotic (Li et al., 2022; Prasse et al., 2006; Serezani et al., 2022). A profibrotic population of macrophages has recently been found in the BAL fluid of COVID-19 patients with severe acute lung injury, yet co-expressed high amounts of IM-markers CD163, Lyve-1, and C1q and fibrosis promoting genes like SPP1 and legumain, suggesting that profibrotic AMs might differentiate from monocytes over a phenotype resembling IMs, or IMs enter the airspace and contribute to fibrosis (Bosteels et al., 2020a; Wendisch et al., 2021; Li et al., 2022). Specifically, in idiopathic pulmonary fibrosis, scRNA-seq on AMs identified genes upregulated by IFN-γ, cautioning that the binary M1/M2 macrophage view of fibrosis being associated with M2 has been an oversimplification (Serezani et al., 2022).

These new insights in macrophage-associated lung fibrosis are slowly leading to new therapeutic avenues, such as targeting TGF-β (Singh et al., 2022; Xu et al., 2022) and M-CSF (Joshi et al., 2020). With increasing age, a known risk factor for lung fibrosis, recMacs occupy the AM niche and contribute to fibrosis, and this has been linked to aging-associated loss of GM-CSF responsiveness (McQuattie-Pimentel et al., 2021). Therefore, an alternative therapeutic option might be to promote differentiation of homeostatic AMs, through inhalation of GM-CSF, a strategy that has been tested in several profibrotic lung diseases like pulmonary alveolar proteinosis, and even COVID-19-associated acute lung injury (Bosteels et al., 2020a; Tazawa et al., 2019). Also, the complement protein and chitinase-like protein family are largely unexplored yet highly interesting targets for immune intervention in fibrotic diseases.

Obstructive airway diseases like asthma and COPD have also been studied extensively. In allergic asthma, many rodent studies have suggested that AMs have an important dampening role in disease inception by capturing allergens in an immunologically silent way while suppressing the function of DCs and antigen-specific T cells (Lauzon-Joset et al., 2013; Soroosh et al., 2013; Strickland et al., 1993; Thepen et al., 1989, 1992). Other studies have suggested that AMs are more pro-inflammatory, an observation that may be confounded because of the accumulation of recAMs (Kurowska-Stolarska et al., 2009; Lee et al., 2015; Naessens et al., 2012). In humans with asthma, AMs show increased antigen-presenting capacity, costimulatory molecule, and adhesion molecule expression (Balbo et al., 2001; Chanez et al., 1993; Tang et al., 1998; Vieira Braga et al., 2019; Viksman et al., 2002). The function of IMs in asthma is emerging. Interleukin-10 derived from IMs has a dampening effect on the activation of lung DCs and may prevent overt immune reactions to allergens (Bedoret et al., 2009), although the IL-10 production is not extensive (Sabatel et al., 2017), and recMacs may be a more prominent source of this cytokine (Tewari et al., 2021). One subset of lung IMs are phenotypically similar to monocyte-derived DCs (moDCs), which are known to be a copious source of chemokines that recruit effector and regulatory T cells, as well as innate effector cells like eosinophils and neutrophils to the lung (Fu et al., 2022b; Medoff et al., 2009; Plantinga et al., 2013). Recently, IL-9, mainly produced by ILC2 and T helper 9 (Th9) cells in asthma, has been shown to promote activation of IL-9R-positive lung IMs, leading to the acquisition of Arg1 expression and production of chemokines that regulate allergic airway inflammation, suggesting that IMs, or recMacs resembling IMs, could also be proinflammatory (Fu et al., 2022b). Following Sendai virus infection, IMs produce increased amounts of triggering receptor expressed on myeloid cells (TREM-2). Strikingly, TREM-2 is cleaved into sTREM-2 and prevents apoptosis of IMs, leading to a prolonged M2-like state, characterized by overproduction of IL-13 and expression of Chil3 and Arg1, and intense collaboration with ILC2 (Wu et al., 2015, 2020). Clearly, more work and better tools are necessary before we can fully understand the role of macrophages in asthma.

In COPD, AMs have long been known to be in an activated state, acting even as orchestrators of the disease process, producing increased amounts of cytokines, chemokines, ROS, and matrix-degrading enzymes that could contribute to the recruitment of CD8⁺ T cells and neutrophils and to the process of alveolar destruction that characterizes lung emphysema (Barnes, 2004). The metabolic alterations associated with chronic smoke inhalation in AMs profoundly affect the phagocytic capacity of AMs, explaining how COPD patients might be at risk for bacterial and viral infections, often the cause of exacerbation of disease (Belchamber et al., 2019; Kono et al., 2021; Wrench et al., 2018). New scRNA-seq of COPD patients has revealed the accumulation of AMs expressing a broad range of metal-binding metallothioneins (Sauler et al., 2022). Deficiency in VitD3 or genetic polymorphisms in VitD pathway genes is a risk factor for COPD development. In Vdr^−/− mice, a GM-CSF mediated expansion of ROS-producing AMs leads to progressive emphysema of the lung (Hu et al., 2019). Other pathways that control AM numbers can also lead to emphysema when disrupted. Isthmin-1 is produced by lung structural cells and promotes AM apoptosis by triggering the cell surface Grp78 receptor. Ism1^−/− mice develop progressive emphysema, characterized by increased MMP12 production from AMs (Lam et al., 2022). These studies illustrate how AM activation needs to be tightly balanced to avoid lung structural damage to the alveolus. In this regard, it is well-known that interrupting endothelial or epithelial function alters AM activation and vice versa and can lead to emphysema, suggesting equilibrium in the alveolar niche is crucial (Jakkula et al., 2000).

Specific macrophage subsets participate in the tumor microenvironment

Lung cancer is very prominent particularly in smokers, although non-small-cell lung cancer (NSCLC) is also increasingly seen in non-smokers. Due to the highly vascularized nature of the gas-exchange apparatus and blood-filtering capacity, the lung is also a very frequent site of metastasis of other primary cancers. A detailed description of the biology of lung macrophages in cancer is beyond the scope of this paper, and the reader is referred to other detailed review articles (Hegde et al., 2021; Ma et al., 2022). Several scRNA-seq studies have unraveled the extraordinary complexity of macrophages in the lung tumor microenvironment (TME), and the dynamic nature of macrophage behavior as the tumor or the metastatic lesion sets up symbiosis in the host lung (Casanova-Acebes et al., 2021; Huggins et al., 2021; Lavin et al., 2017). These studies revealed there to be at least 7 different types of lung macrophages, including lipid-associated macrophages and specific tumor-associated macrophages, some of which are pro-angiogenic and immunosuppressive. Macrophages shape the TME, tumor immunity, and response to immunotherapy, which makes them an important target for cancer treatment. Using lineage tracing, it has been shown that developing NSCLC lesions in the mouse initially accumulate TRMs close to tumor cells to promote epithelial-mesenchymal transition and invasiveness in tumor cells, and these TRMs also induced a potent regulatory T cell response that protects tumor cells from adaptive immunity (Casanova-Acebes et al., 2021; Loyher et al., 2018). Depleting TRMs diminishes and alters the phenotype of regulatory T cells, promoting the accumulation of CD8⁺ T cells and decreasing tumor invasiveness and growth. During tumor growth, TRMs become redistributed at the periphery of the TME, which gradually become dominated by recMac in both mouse and human NSCLC (Casanova-Acebes et al., 2021). The precise signals that alter lung macrophages in the TME are still unclear. CD4⁺ T cell-derived IL-9 promotes the expansion of both IMs in lung tumor models, and IL-9 is overexpressed in NSCLC lesions. Mechanistically, the IL-9 macrophage axis required arginase 1 to mediate tumor growth, and adoptive transfer of Arg1⁺ lung macrophage to Il9r^−/− mice promoted tumor growth (Fu et al., 2022a). Modulating macrophages in cancer is still difficult, as we still lack a complete understanding of the molecular and functional diversity of the tumor macrophage compartment before this can lead to effective new therapies.

B.N.L. is supported by a FWO Methusalem grant, and by an FWO Excellence of Science grant. He has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 789384). H.A. is supported by an FWO-VLAIO Fellowship. C.V.J. is supported by National Institutes of Health grant R35 HL155458.