The ability of AMs to prevent inappropriate immune responses has been inferred from depletion experiments prior to (30, 31) and after antigen sensitization (17, 31–33). Intratracheal instillation of clodronate-filled liposomes enables the specific depletion of phagocytic cells localized in the airway lumen, i.e. AMs in vast majority. The depletion is therefore transient and non-specific to AMs per se. More recently, AMs were also depleted using CD169-DTR mice (34). However, caution should be taken when interpreting the results obtained using these transgenic mice to target AMs as IMs also express CD169, even though less than AMs (35). In vivo elimination of AMs using clodronate-filled liposomes led to overt inflammatory reactions in sensitized mice to model antigens, such as trinitrophenyl-keyhole limpet hemocyanin (TNP-KLH) (30), ovalbumin (OVA) (17, 32, 33) or house dust mites (HDM) (31, 33). Increased IgE levels and development of mononuclear cell infiltrates in the lung were also observed following AMs depletion in OVA-sensitized rats (36). AMs exerted their immunosuppressive properties toward HDM no matter whether their depletion occurs prior the sensitization or challenge phase (31). Clodronate treatment during the sensitization phase resulted in a reduction of HDM-induced IL-27 concomitant with exacerbation of Th2 pathology, suggesting a role for IL-27 in regulating Th2 responses at mucosal surfaces (31). Surprisingly, AMs depletion alleviated the trimellitic anhydride (TMA)-induced drop in lung function parameters observed in TMA-sensitized rats (37). The levels of serum IgE were also decreased (37). In contrast, TMA-induced tissue damage and inflammation were augmented following AMs elimination (37). Indeed, AMs seemed to suppress non-specific inflammation caused by TMA conjugated to endogenous protein (TMA-BSA) challenge (38). In line with this study, transfer of naïve AMs to OVA-sensitized AMs-depleted mice resulted in decreased airway hyperreactivity and eosinophil counts in the bronchoalveolar lavage (BAL) fluid whereas no improvement was observed upon transfer of sensitized AMs (32). The same phenomenon was observed in rats (39), suggesting that allergen sensitization modulates AMs function. AMs phagocytosis was although diminished in sensitized AMs (39) underscoring the importance of AMs status for their control of the pulmonary response in a suppressive way.

Looking at the mechanisms, in vitro co-culture of rat AMs with antigen presenting cells (APCs) across a semipermeable membrane revealed an inhibition of APC maturation, amplified by TNF-α and abrogated via blockade of the nitric oxide synthase pathway (40). When mixed with T cells, AMs appear to allow T-cell activation and expression of T-cell effector function, while selectively inhibiting T-cell proliferation (41) ( Figure 2 ). This suppression involves a unique form of T-cell anergy, associated with inhibition of IL-2 receptor signal transduction (42). The induction of unresponsiveness was reversed upon removal of AMs from the T cell (42) or upon granulocyte-macrophage colony-stimulating factor (GM-CSF) treatment (43) in rodents and by the addition of CD28 costimulation or IL-2 in human (44). Rodent and human AMs also differ in the mechanisms employed to achieve this inhibition: rodent AMs appear to utilize reactive nitrogen intermediates, while this does not appear to be the case for human AMs (41). Nevertheless they both release prostaglandin and TGFβ (45, 46) suggesting that pulmonary macrophages use multiple mechanisms for locally suppressing lymphocyte activation. More recently, apoptotic cell uptake by lung AMs was shown to suppress HDM-driven allergic asthma while dampening AMs capacity to make inflammatory cytokine, increasing their responsiveness to adenosine (whose receptor limit allergic inflammation upon agonist treatment) and their retinoic acid (RA) production (47). In line with this study, mouse AMs were also shown to induce regulatory T cells (Tregs) in vitro through the release of RA and TGFβ (48, 49) ( Figure 2 ) even though IMs appear to be more potent in inducing the expression of the forkhead box P3 transcription factor, Foxp3, the master regulator of Tregs, in naïve T cells (50). These Treg-inducing AMs were nevertheless able to promote airway tolerance as their transfer into sensitized mouse airways prevented the development of asthmatic lung inflammation upon subsequent challenge with Ag (49). Several mediators from the lung microenvironment such as TGFβ production, SIRPα and CD200R stimulation and low doses of nutrients are also able to drive AMs toward tolerogenic function, preventing potentially detrimental lung inflammation (51).

NO₂ exposure induces the infiltration of an AM subpopulation in rodent BAL fluid that may exert anti-inflammatory functions by the production of high amounts of the immunosuppressive cytokine IL-10 (52). In the same way, delivery of low-dose LPS in mice to prime the lung was shown to augment AMs production of IL-10 in the BAL fluid and enhance resolution of lung inflammation induced by a lethal dose of LPS or by Pseudomonas bacterial pneumonia (53) ( Figure 2 ). On the other hand, a study revealed that mouse AMs do not produce IL-10 upon LPS stimulation in vitro (54). These results are in accordance with others, showing that IL-10 production by mouse AMs is quiet low in comparison to mouse IMs, using IL-10-β-lactamase reporter (ITIB) mice (35, 55) and in vitro culture with LPS-containing OVA stimulation (56). Caution should be taken when considering all BAL fluid cells as AMs since the presence of IMs was shown in mouse BAL fluid upon stimulation with CpG-DNA (35). Moreover, inflammatory stimuli induce phenotypical changes among AM and IM populations making their discrimination more complex. Recently, two subpopulations of macrophages have been described in the human BAL fluid based on their degree of autofluorescence and their ability to secrete IL-10 (57) suggesting that a fraction of human IMs could be present in the airway lumen.

Besides their localization in the lung interstitium, IMs have been defined as regulatory macrophages. Indeed, since the first description of their regulatory function in 2009 (56), many studies confirmed that IMs produce IL-10 in the steady state (35, 58–60) and can in this way contribute to lung homeostasis.

Upon LPS-containing OVA exposure, ex vivo cultured IMs were found to impair the ability of co-cultured bone marrow-derived dendritic cells (BMDCs) to maturate, migrate to the draining lymph node and induce features of Th2-mediated airway allergy once reinjected into the trachea of recipient mice (56). This effect was mediated through IL-10 production by IMs since Il10^-/- IMs failed to do so (56) ( Figure 2 ). In vivo, mouse IMs were shown to be localized in close vicinity of lung dendritic cells (DCs), which are endowed with the ability to trigger allergen-specific Th2 responses (56), and IMs-depleted mice developed airway allergy following exposure to low doses of allergens and LPS (56). The immunosuppressive potential of IMs is nevertheless surpassed by the high dose of the allergens and LPS, given that 100% of WT mice develop asthma upon exposure to high amounts of HDM extracts, a phenomenon that requires Toll-like receptor 4 (TLR4) activation by HDM-borne LPS (61). IMs also regulate Th17-mediated inflammatory response since their transfer to HDM-exposed Il10^-/- mice reduce the Th17-related neutrophilic inflammation (59).

IL-10 production by IMs is increased upon inflammatory stimuli such as HDM, LPS, CpG-DNA, Flagellin and FSL-1, a synthetic lipoprotein derived from Mycoplasms salivarium (35, 56, 58, 59). CpG-DNA is by far the most potent stimulator of IL-10 in mouse IMs and it also induces a dramatic expansion of IL-10-producing IMs (35). These CpG-induced IMs were able, by producing IL-10, to confer protection against allergic inflammation even when mice were sensitized and challenged with high doses of HDM (35) ( Figure 2 ). Intranasal delivery of mesenchymal stem cell-derived exosomes was also able to substantially expand lung IL-10-producing IMs and thus contributed to protection against allergic asthma in mice (62).

Signaling pathways that promote IL-10 expression in IMs have been little studied but the constitutive production of IL-10 in IMs was shown to be mediated through activation of the TLR4/myeloid differentiation factor 88 (MyD88) pathway in a microbiota-independent manner (59). In allergenic contexts, MyD88-dependent upregulation of the transcription factor Hypoxia-inducible factor 1-alpha (Hif1α) boosts the expression of IL-10 by lung IMs (58). Recently, semaphoring 3E/plexinD1 signaling in IMs was shown to be a critical pathway for their immunoregulatory activity as IMs genetic deficiency in plexinD1 impaired IL-10 production leading to airway allergy in mice exposed to HDM (63) ( Figure 2 ).

While IL-10 production is the most studied immunoregulatory mechanism used by IMs, it is not the only one. The interaction of repulsive guidance molecule b (RGMb), expressed on IMs, with programmed death ligand 2 (PD-L2), expressed on DCs, appear to be essential for respiratory tolerance. Indeed, blockade of the RGMb-PD-L2 interaction markedly impaired the development of respiratory tolerance in a mouse model of tolerance to OVA (64). Lung IMs were also shown to induce the proliferation and differentiation of Treg cells (50) ( Figure 2 ).

Functional studies on IMs have been mainly conducted in mice, probably due the difficulties faced to isolate macrophages from healthy human tissue. However their existence have been described in human and non-human primates (60, 65). Human IMs produce IL-10 more potently than human AMs, mainly upon LPS stimulation but also at steady state (60) and would be functionally impaired in asthmatic patients (66). Recently, a population of macrophages with properties similar to IMs was described in the human BAL fluid (57). These cells, identified as autofluorescent^low (AF^low) AMs in comparison to classical AF^high AMs, expressed a unique transcriptional signature associated with specific immunoregulatory functions, including the ability to secrete IL-10 (57) ( Figure 2 ). Such signature, along with their small size is reminiscent of what is described for IMs in the murine lung indicating that some humans “IMs” might be present in the airway lumen. This would facilitate investigations on human IMs.

Recent studies have begun to reveal heterogeneity among IMs compartment (50, 55, 67) with at least two main populations with different phenotypes and localizations in mice (50, 55). Mouse CD206⁺ Lyve1^hi MHCII^lo IMs are mainly located in the bronchial interstitium (55) and associated to blood vessels (50) while mouse CD206^- Lyve1^lo MHCII^hi IMs are mainly located in the alveolar interstitium (55), and associated to nerve bundles (50). Regarding their function, CD206⁺ IMs produce more IL-10 than CD206^- IMs (50, 55). On another hand, CD206^- IMs were shown to induce Treg more potently than CD206⁺ IMs, in accordance with their high expression of MHCII (50). A study showed that MHCII^hi IMs can be further divided on two subsets based on CD11c expression (67). However, differences in function between these two subsets has not been investigated. A population of mouse CD169⁺ IMs, named nerve and airway-associated macrophage and described as a new subset of IMs, was reported recently (68). However, their phenotypic analysis reveals that these cells largely overlap with CD206^- IMs identified previously. In humans, two subsets transcriptionally similar to murine subsets were also described (50). However, unlike mouse IMs, MHCII (HLA-DR) cannot be used to identify human IM subsets as this marker is expressed at higher levels in CD206⁺ Lyve1⁺ IMs (50).

Two main populations of monocytes have been described in the mouse lung: the Ly6C^hi GR-1^hi classical monocytes and the Ly6C^lo GR-1^lo patrolling monocytes (69) with their human counterparts consisting of CD14⁺ CD16^- and CD14^lo CD16⁺ monocytes respectively (70). However, lung monocytes are mainly located in the blood vessels associated to the lung (35). Only a fraction of Ly6C^hi monocytes, called Ly6C^hi lung monocytes (35), and a discrete population of CD64⁺ CD16.2⁺ N4RA1-dependent (patrolling monocytes key transcription factor) monocytes (55) are truly located in the mouse lung tissue. Monocytes can extravasate into the lung tissue where they can differentiate into tissue macrophage or DC or recirculate to lymph nodes without any differentiation (71). Most of the monocytes and monocyte-derived cells harbor proinflammatory properties and contribute to the development of immune response. Nevertheless, some regulatory functions have been reported.

First of all, a fraction of lung resident CD64⁺ CD16.2⁺ monocytes were shown to express IL-10 (55), suggesting regulatory properties. Regarding Ly6C^hi lung monocytes, CpG-DNA exposure induced their differentiation into hypersuppressive CpG-induced IMs (35). Splenic monocytes were also recruited to the lung to constitute the pool of CpG-induced IMs (35). Infection with the murid herpesvirus 4 (MuHV-4) was shown to inhibit the development of HDM-induced experimental asthma by modulating lung innate immune cells (72). This immunosuppressive effect was attributed to monocytes that replenished resident AMs upon MuHV-4 infection (72). These monocyte-derived AMs displayed regulatory properties, including IL-10 production, and blocked the ability of DCs to trigger a HDM-specific response by Th2 cells in mice (72) ( Figure 2 ). MuHV-4-imprinted monocytes are also able to recruit CD4 T cells to the airways and trigger immunosuppressive signaling pathways through the PD-L1/PD-1 axis, thereby dampening the deleterious activation of cytotoxic CD4 T cells (73). Monocytes can also act as suppressor cells that promote Treg development (74). Indeed, adoptive transfer of GR-1⁺ monocytes in tumor-bearing mice revealed the differentiation of such monocytes into tolerogenic DCs that produce IL-10 and potently induce Treg response and expansion (74). During gut infection, monocytes can acquire regulatory properties in the bone marrow thanks to a priming by natural killer (NK) cells (75). This process could potentially occur in other mucosa like the lung. A population of GR-1⁺ cells was observed in ozone-exposed mice (76). These cells were diminished in the absence of CX3CR1 and appeared to protect the host from the biological response to ozone (76). Indeed, CX3CR1-null mice exhibited enhanced responses to ozone, including increased airway hyperresponsiveness, exacerbated neutrophil influx, accumulation of 8-isoprostanes and protein carbonyls and increased expression of cytokines (76). This population was identified by the authors as a novel macrophage subset, distinct from AMs. Despite their expression of F4/80, a major macrophage marker (77), these cells do not look like any population of lung macrophage already described and highly express GR-1 (76), a monocytic marker. As mentioned earlier, inflammation makes discrimination between cell population harder due to phenotypic changes and overlapping marker expressions, so it is possible that these cells are stuck at an intermediate state between classical monocytes and macrophages.

Human CD14⁺ monocytes are potent activators of TGFβ, via expression of the integrin αvβ8 and matrix metalloproteinase 14, which dampens their production of TNFα in response to LPS (78). In the healthy human intestine, a mucosa which, like the lung, have to deal with foreign compounds, integrin αvβ8 is highly expressed on mature tissue macrophages, with these cells and their integrin expression being significantly reduced in active inflammatory bowel disease (78). This suggests a key role of integrin αvβ8-mediated TGFβ activation in the regulation of inflammatory responses and mucosal homeostasis by monocytes and macrophages.

Accumulating evidence indicates that, besides their pro-inflammatory roles in Th2 responses associated with helminth infections or allergic diseases, eosinophils also regulate homeostatic processes at steady state and exhibit protective role under certain conditions (113, 114).

At steady state, the mouse lung contains resident eosinophils (rEos) which display unique morphological and phenotypical features that unambiguously distinguish them from the inflammatory eosinophils (iEos) that are recruited to the lung during HDM-induced allergic airway inflammation (115). CD101 is the main characteristic that enable to distinguish rEos from iEos: rEos do not express CD101 while iEos are CD101⁺ (115). These rEos were shown to inhibit the maturation, and therefore the pro-Th2 function, of allergen-loaded DCs and correspondingly, mice lacking lung rEos showed an increase in Th2 cell response to inhaled allergens (115) ( Figure 2 ). In human, the parenchymal rEos identified in non-asthmatic lungs were phenotypically distinct from the iEos isolated from the sputa of eosinophilic asthmatic patient, suggesting that the findings in mice are relevant to humans (115). Mouse lung eosinophils play also a crucial role in lung allograft acceptance. While associated with rejection of other solid organs, local nitric oxide (NO) generation is critical for lung allograft acceptance (116). Eosinophils were shown to be the dominant inducible NO synthase (iNOS)-expressing cells in the lung allograft and their depletion reduced NO levels to that of recipient mice and led to allograft rejection (117). NO production by eosinophils depends on stimulation by IFN-γ and TNF-α since neutralization of such mediators in graft recipients abrogates eosinophil suppressive capacity (117). The iNOS⁺ lung eosinophils were phenotypically similar to the previously described lung rEos, indicating that rEos may display several immunoregulatory functions.

In guinea pigs, ozone exposure induced eosinophil hematopoiesis which limit ozone-induced airway hyperreactivity since depletion of these newly recruited eosinophils worsened airway hyperreactivity (118). This ozone-induced hematopoiesis of beneficial eosinophils was blocked by TNF-α antagonist or by prior allergen sensitization, suggesting that atopic individuals might have worsened airway hyperreactivity following ozone exposure or delayed resolution of symptoms because of a lack of bone marrow response (119).

Like for eosinophils, emerging evidences point out regulatory functions for neutrophils (120). Indeed, it was shown that neutrophils can decrease DCs function (121–123), protect host from LPS-induced lethal inflammation (124) and produce anti-inflammatory molecules such as IL-10 and act as T-cell suppressors in different contexts (125–128). However, these functions have been mainly attributed to circulating neutrophils and immunoregulatory properties of lung neutrophils have been poorly investigated. So far, a study revealed that mycobacteria-infected DCs attract neutrophils that produce IL-10 and specifically shut down the otherwise exuberant Th17 response in the mouse lung (127).

Besides their well-known roles in allergy and innate immunity, mast cells have also the potential to turn immune responses off (129, 130). Although an immunosuppressive role has not been uncover yet in the lung, several protective properties have been reported in the skin where mast cells are important to suppress UVB-induced contact hypersensitivity (131), limit leukocyte infiltration in contact dermatitis (132), impair the development of Ag-specific T cell response following Anopheles mosquitoes bites (133) and induce an optimal tolerance to skin allograft through Foxp3⁺ Treg cells (134).