Acute respiratory distress syndrome (ARDS) is defined by the presence of lung inflammation and hypoxemia, leading to tissue hypoxia and the need for supplemental oxygen1. It can have many different etiologies and is typically associated with a mortality rate of around 35%, although mortality rates are higher in patients with COVID-19-associated ARDS2. Despite being a hallmark of ARDS, whether hypoxemia and the ensuing tissue hypoxia regulate inflammation remains unclear. In this issue of Nature Immunology, a tour-de-force study from Mirchandani et al.3 demonstrates that both hypoxemia and hypoxia are associated with blood monocytopenia in patients and mice. In a mouse model of ARDS, this was also associated with the impaired development of monocyte-derived pulmonary macrophages and persistent inflammation. The administration of colony-stimulating factor (CSF1) fused to the Fc region of IgG1a (CSF1–Fc) was sufficient to rescue macrophage numbers while also altering their phenotypes, leading to the resolution of inflammation and hence the identification of a potential therapeutic target for ARDS3 (Fig. 1).

Fig. 1: Treatment with CSF1–Fc alters monocyte and macrophage phenotypes resolving inflammation in the hypoxic lung.
figure 1

a, ARDS of various etiologies is associated with an initial reduction in circulating monocytes in mice and humans and prolonged alterations in monocyte phenotypes. b, Deficient monopoiesis and increased erythropoiesis in the bone marrow contribute to the reduction in circulating monocytes in a mouse model of ARDS. Equal number of monocytes are recruited to the lung after acute lung injury in hypoxic or normoxic conditions, but their differentiation into MDMs and eventually CD64+Ly6C−SiglecF− macrophages is disrupted under hypoxic settings. c, Administration of CSF1–Fc to ARDS mice increases the numbers of monocytes in the blood and lung and reverts them to a non-hypoxic phenotype. This in turn results in increased numbers of MDMs and CD64+Ly6C−SiglecF− macrophages, reduced neutrophilia and less inflammation. A MHCII−LYVE-1+IL-10+ subset of CD64+Ly6C−SiglecF− macrophages uniquely accumulates in the setting of CSF1–Fc administration. AM, alveolar macrophage.

Full size image

With the considerable recent research effort into SARS-CoV-2, several studies have reported alterations in populations of circulating monocytes in patients with ARDS2. However, the role of hypoxemia in this has not been investigated. Mirchandani et al.3 observed significant monocytopenia in the blood of patients diagnosed with ARDS of various etiologies less than 48 h before analysis compared with healthy controls, despite the presence of increased numbers of total leucocytes. The remaining monocytes were shown to have a distinct phenotype, proteome and transcriptome that persisted even after the monocytopenia was resolved. To gain further mechanistic insight into how hypoxemia or hypoxia can shape the immune landscape in ARDS, the authors3 combined lipopolysaccharide (LPS)-induced acute lung injury with reduced oxygen levels to develop a mouse model that recapitulates the inflammation and hypoxia observed in ARDS. Consistent with the patient studies, ARDS mice exhibited monocytopenia 24 h after inhalation of LPS, although unlike in the patients, this was coupled to leucopenia. Similar results were also observed when LPS inhalation was replaced by intratracheal infection with Streptococcus pneumoniae. As observed in patients, the remaining circulating monocytes had an altered phenotype, but, in contrast to patients, the remaining mouse monocytes were biased towards the Ly6Chi classical phenotype. These data indicated that, irrespective of etiology, ARDS was associated with changes in circulating monocytes (Fig. 1a) and highlighted a role for hypoxia in this phenotype.

Studying the mechanisms involved, Mirchandani et al.3 observed decreased BrdU incorporation in circulating monocytes in ARDS mice after a 12-h pulse-chase experiment, which suggested decreased monocyte output from the bone marrow. A similar decrease was observed in naive hypoxic mice, indicating that hypoxia alters monocyte generation in general. Erythropoiesis is one mechanism by which oxygen levels can be regulated in hypoxia4. As erythrocytes and monocytes share a common precursor (common myeloid progenitor, CMP)5, the authors investigated whether ARDS skews hematopoiesis towards erythropoiesis and away from monopoiesis. Consistent with this idea, an increase in erythropoietin and a decrease in interferon-α (IFNα), which are drivers of erythropoiesis and emergency monopoeisis, respectively, was observed in the serum of ARDS mice. Confirming a role for reduced IFN signaling, Ifnar−/− mice, which are deficient in the type I IFN receptor, had increased numbers of erythroid progenitors and decreased numbers of monocytes 5 days after LPS inhalation under normoxia. A reduction in Ifnar expression was also observed in hematopoietic stem cells (LSKs) in the bone marrow of ARDS mice, providing another explanation for the insufficient type I IFN signaling in ARDS. Looking into the distinct bone marrow progenitors, consistent with a skewing toward erythropoiesis, the direct precursors of erythrocytes were increased in hypoxic mice; however, no differences were noted in the numbers of upstream CMPs (Fig. 1b). It could be argued that, given the increase in erythroid progenitors, CMPs preferentially gave rise to erythrocytes under hypoxia. However, to support this conclusion, the number of downstream progenitors that can give rise to monocytes but not erythrocytes — such as monocyte–dendritic cell progenitors (MDPs) and common monocyte progenitors (c-MOPs) — and the number of bone marrow monocytes should also be investigated. Whether these changes occur consistently in patients with ARDS, in which monocytopenia is associated with an increased leucocyte pool, also remains to be determined.

The authors also examined the consequences of the changes in the circulating monocytes for lung myeloid cells3. Consistent with the presence of inflammation, both normoxic and hypoxic mice challenged with LPS 24 h earlier had an increased population of neutrophils in whole lung digests. However, although total levels of CD64hi macrophages expanded in normoxic LPS-treated mice, this was not observed in hypoxic mice. Similar results were observed in the S. pneumoniae model of ARDS and in Ifnar−/− LPS-challenged normoxic mice. The reduction in this cell population did not result from the monocytopenia observed in the blood, as the numbers of Ly6ChiCD64int monocytes recruited to the lung were not altered compared with controls. Although the precise reasons for this are unclear, the remaining circulating monocytes showed increased expression of the chemokine receptor CCR2, which is known to be important for the recruitment of monocytes to inflamed tissues6. Moreover, the expression of CCL2 was also increased in ARDS mice, suggesting that an interaction between CCR2 and CCL2 may be one mechanism that prevents pulmonary monocytopenia. Further investigation suggested that the reduced numbers of CD64+ macrophages resulted from impaired differentiation of monocytes into Ly6C+MHCII+/−CD64hiSiglecF-CD11c+/− cells, defined by the authors as monocyte-derived macrophages (MDMs)3. On day 5 after LPS challenge, total CD64+SiglecF− macrophages (including Ly6Chi MDMs and Ly6C− macrophages) were still reduced in the ARDS mice, highlighting a prolonged block in their differentiation, correlating with persistent inflammation not observed in normoxic controls (Fig. 1b). This suggests that the differentiation of MDMs and subsequently Ly6C− macrophages may be crucial for the resolution of inflammation.

The precise nature of MDMs in ARDS mice remains to be defined. Given the expression of Ly6C and the presence in the lung just 24 h after LPS administration, MDMs probably represent recently recruited monocytes that are in the process of transitioning to macrophages, and as such may be a cell population often referred to as transitioning monocytes, rather than bona fide macrophages. Notably, as CD64 expression can also be upregulated in dendritic cells via IFN and LPS signaling7, it would be interesting to examine whether the CD64hiLy6ChiMHCII− cells identified within the MDM population may be genuine Ly6Chi monocytes. If this is the case, the reduction in MDMs may be due to reduced recruitment alongside impaired differentiation. Lung-shielded bone marrow chimeras suggested that at least some of the MDMs could give rise to Ly6C−CD64hiSiglecF− macrophages at day 5 after LPS treatment. Fitting with reduced differentiation, the chimerism in Ly6C−CD64hiSiglecF− macrophages was lower in ARDS mice than in normoxic controls. It will be interesting to determine whether this stems from an altered pulmonary macrophage niche8 in hypoxia. Notably, the fact that MDMs were found in the bronchoalveolar lavage and lung tissue may suggest some heterogeneity within this population, and hence in their niches, and suggests that some of these cells may contribute to both the alveolar and interstitial macrophage pools. In addition, as monocyte engraftment in an empty macrophage niche can be coupled to proliferation of the engrafting monocytes9, it will be interesting to examine whether impaired proliferation also contributes to the reduced number of MDMs and Ly6C−CD64hiSiglecF− macrophages in ARDS mice.

Mirchandani et al.3 next queried whether increasing the numbers of monocytes and macrophages by treatment with CSF1–Fc, a growth factor important for monocytes and macrophages but not alveolar macrophages10, could be an effective therapeutic strategy in ARDS. CSF1–Fc administration for 4 days after LPS inhalation increased the numbers of circulating monocytes in ARDS mice and also reverted the hypoxia-induced phenotype (Fig. 1c), including the restoration of the expression of type I IFN-related genes. In the lung, CSF1–Fc treatment increased the numbers of infiltrating monocytes and MDMs, and restored the levels of chimerism within CD64hiSiglecF−Ly6C− macrophages (Fig. 1c). Although, whether this was due to increased differentiation or to the increased number of monocytes remains unclear. Crucially, the increase was accompanied by ameliorated inflammation, which suggests that CSF1–Fc could be an effective therapeutic strategy. Phenotypical analysis indicated that the MHC class II-negative (MHCII−) subset of CSF1–Fc-induced CD64hiLy6C−SiglecF− macrophages expressed the CD44 homolog LYVE-1 and interleukin-10 (IL-10), and thus resembled repair-associated macrophages11. This phenotype was not observed in normoxic controls, indicating it was an ARDS-specific signature. However, it did not seem to be due to the increased accumulation of a population of macrophages already present in ARDS mice in the absence of CSF1–Fc treatment, nor was it induced by CSF1–Fc, as it was not observed in normoxic controls treated with CSF1–Fc. It is tempting to speculate that the monocytes and macrophages that accumulate in the lung after CSF1–Fc administration may communicate differently with the hypoxic niche, leading to the provision of altered signals and ultimately to an altered macrophage phenotype. Indeed, macrophage-niche cell interactions have been shown in both directions, and population densities seem to be an important aspect of this communication12.

Together, Mirchandani et al.3 have uncovered a previously unappreciated role for hypoxemia and hypoxia in modulating the immune landscape and hence tissue inflammation. They demonstrate a crucial role for macrophages in resolving hypoxia-induced inflammation and with their CSF1–Fc studies, they provide a potential therapeutic strategy to target these cells to improve patient outcomes.