The lung environment controls alveolar macrophage metabolism and responsiveness in type 2 inflammation

Abstract

Fine control of macrophage activation is needed to prevent inflammatory disease, particularly at barrier sites such as the lungs. However, the dominant mechanisms that regulate the activation of pulmonary macrophages during inflammation are poorly understood. We found that alveolar macrophages (AlvMs) were much less able to respond to the canonical type 2 cytokine IL-4, which underpins allergic disease and parasitic worm infections, than macrophages from lung tissue or the peritoneal cavity. We found that the hyporesponsiveness of AlvMs to IL-4 depended upon the lung environment but was independent of the host microbiota or the lung extracellular matrix components surfactant protein D (SP-D) and mucin 5b (Muc5b). AlvMs showed severely dysregulated metabolism relative to that of cavity macrophages. After removal from the lungs, AlvMs regained responsiveness to IL-4 in a glycolysis-dependent manner. Thus, impaired glycolysis in the pulmonary niche regulates AlvM responsiveness during type 2 inflammation.

The specialized mucosal environment of the lung enables breathing during continuous exposure to debris and micro-organisms, and offers diverse mechanisms to restrict disease caused by overexuberant inflammatory responses¹. Lung macrophages have been proposed to be central to mediating and regulating type 2 inflammation against allergens and parasitic worms, which together affect billions of people worldwide². Against these challenges, macrophages can expand in situ to type 2 cytokines such as IL-4 that trigger ‘alternative’ M(IL-4) activation, linked to wound repair and type 2 pathology^3–5. Although pulmonary macrophage subpopulations inhabit different anatomical sites, such as the airways and tissue parenchyma, it is not known how location influences their ability to respond to type 2 inflammation. Reports of M(IL-4) marker expression on lung macrophages during type 2 inflammation^6–9 have involved experimental approaches that may not clearly distinguish macrophages from other myeloid cells, indicating that functional differences in key macrophage subpopulations have been inadvertently overlooked.

As the predominant macrophage subpopulation in airways, AlvMs are vital for maintaining lung health and function by clearing debris, surfactant and apoptotic cells¹⁰. In the absence of AlvMs, fluid buildup leads to primary pulmonary alveolar proteinosis, severe lung dysfunction and respiratory failure¹¹. Most AlvMs are thought to derive from embryonic precursors that seed the lung tissue before birth¹², with recent evidence suggesting that cytokines GM-CSF and TGF-β induce PPAR-γ, a crucial transcription factor for AlvM development^11,13,14. During inflammation, AlvMs mediate bacterial clearance and initiate neutrophil recruitment¹⁵, functions that can be regulated by cytokines such as IL-10 or TGF-β and/or the engagement of cell surface receptors such as SIRPα or CD200 (ref.¹⁶). Because clear discrimination between AlvMs and other lung macrophage subpopulations is technically challenging¹⁷, less is known about the function and origin of tissue-residing interstitial macrophages (IntMs). Although IntMs may comprise up to three subpopulations¹⁸, earlier work may have mistakenly identified them as AlvMs, monocytes or dendritic cells.

Mucosal environments such as the lung play a principal role in determining both development and function of macrophages¹⁹, although many of the factors that shape such processes remain unclear, particularly in type 2 inflammation. Lung macrophage upregulation of M(IL-4) markers during parasite-mediated type 2 responses is promoted by environmental factors such as surfactant protein A (SP-A) and engagement of TAM receptors during clearance of apoptotic cells^20,21. We show that lung macrophage subsets, particularly AlvMs, were considerably less responsive to type 2 inflammation than macrophages from other tissues. We demonstrate that this muted phenotype was conferred by the lung environment, and was independent of potential negative regulators such as CD200–CD200R, SP-D, Muc5b or the host microbiota. Hyporesponsive AlvMs had an altered metabolic profile compared to IL-4-responsive peritoneal exudate cell macrophages (PECMs), and were unable to upregulate glycolysis in situ. After removal from the lung, AlvMs recovered their IL-4 responsiveness in a glycolysis-dependent manner. Thus, the pulmonary environment controlled AlvM responsiveness during type 2 inflammation via modulation of their metabolic activity.

Results

AlvMs are hyporesponsive to IL-4 in vivo.

To better understand how pulmonary macrophages respond during type 2 inflammation, we used MerTK, CD64, Siglec-F and CD11b as markers that distinguish AlvMs from IntMs^11,17,18. Most lung tissue and bronchoalveolar lavage (BAL) macrophages were MerTK⁺CD64⁺CD11b⁻Siglec-F⁺ AlvMs (>89%), alongside fewer (<10%) MerTK⁺CD64⁺CD11b⁺Siglec-F⁻ IntMs (Fig. 1a,b and Supplementary Fig. 1a). Analysis of additional macrophage markers showed that, while both AlvMs and IntMs expressed F4/80 and CD11c, AlvMs were also Ym1^hi, a feature of M(IL-4) (Fig. 1c). Further, only IntMs expressed CX3CR1 (Fig. 1c), supporting the idea that IntMs are derived from monocytes, while AlvMs at steady state are resident cells^12,18. To verify that AlvMs reside in airways and IntMs in lung tissue, we administered CD45-PE antibodies intranasally (i.n.) and CD45-FITC antibodies intravenously (i.v.) before lung processing, to discriminate CD45-PE⁺ airway macrophages from CD45-FITC⁺ blood monocytes and CD45-PE⁻FITC⁻ tissue macrophages²². This approach indicated that AlvMs (defined throughout this study as MerTK⁺CD64⁺CD11b⁻Siglec-F⁺) were mainly found in the airways and IntMs (defined throughout this study as MerTK⁺CD64⁺CD11b⁺Siglec-F⁻) in the lung tissue (Fig. 1d), demonstrating that refined flow cytometry could discriminate between AlvM and IntM subsets.

Fig. 1 |. AlvMs are unresponsive to IL-4.

a, Flow cytometry plots identifying AlvMs and IntMs from BAL fluid or lung tissue of naïve mice. Data representative of eight independent experiments. b, Imaging cytometry of AlvMs and IntMs from lung tissue of naïve mice (scale bar, 10 μm). c, Histograms of expression of F4/80, CD11c, Ym1 and eGFP by Cx3cr1^eGFP/+ mice by AlvMs and IntMs from lung tissue of naïve mice. d, Flow cytometry plots of AlvMs, IntMs and monocytes from lung tissue of naïve mice following CD45 i.n. and i.v. administration. b–d, Representative data from three independent experiments. e, Numbers of lung tissue AlvMs and IntMs, or PECMs, on day 4 following i.p. PBS or IL-4c administration on day 0 and day 2. Data representative of seven to nine independent experiments, n = 26 (AlvM PBS, AlvM IL-4c, IntM PBS, IntM IL-4c), n = 18 (PECM PBS), n = 13 (PECM IL-4c) mice per group. f, RELMα and Ki67 expression, or EdU incorporation, by lung tissue AlvMs and IntMs, or PECMs, on day 4 following i.p. PBS or IL-4c administration on day 0 and day 2, and EdU injection i.p. 3 h before tissue collection. Graphs show individual replicate mice, data pooled from five to nine independent experiments. RELMα: n = 29 (AlvM PBS, AlvM IL-4c, IntM PBS, IntM IL-4c), n = 24 (PECM PBS), n = 23 (PECM IL-4c) mice per group. Ki67: n = 24 (AlvM PBS, IntM PBS), n = 23 (AlvM IL-4c, IntM IL-4c), n = 20 (PECM PBS, PECM IL-4c) mice per group. EdU: n = 22 (AlvM PBS, AlvM IL-4c, IntM PBS), n = 23 (IntM IL-4c), n = 14 (PECM PBS), n = 17 (PECM IL-4c) mice per group. g, Percentage of RELMα⁺ PECMs, PLECMs, Kupffer cells, IntMs and AlvMs on day 4 following i.p. PBS or IL-4c administration on day 0 and day 2. Data representative of two to five independent experiments, n = 5 (PBS PECMs, Kupffer cells, Colon Ms, IntMs and AlvMs), n = 4 (IL-4c PLECMs), n = 3 (PBS PLECMs, colon Ms and IL-4c PECMs, Kupffer cells, colon Ms, IntMs, AlvMs) mice per group. e–g, Data analyzed by two-way analysis of variance (ANOVA) with Tukey’s post-test for multiple comparisons, displayed as mean ± s.e.m., *P <0.05, ***P <0.001 and ****P <0.0001.

We next investigated whether lung AlvMs and IntMs were functionally similar to macrophages in other tissues following systemic (intraperitoneal, i.p.) administration of recombinant IL-4 complexed with mAb to IL-4 (IL-4c), which extends the bioactive half-life of the cytokine and induces type 2 inflammation in C57BL/6 and BALB/c mice^4,5. PECMs underwent rapid expansion by day 4 after i.p. IL-4c injection on day 0 and day 2 (Fig. 1e)^4,5, while AlvMs and IntMs were markedly less responsive to IL-4c, with no measurable increase in numbers of either population (Fig. 1e). Additionally, PECMs from IL-4c injected mice had elevated expression of markers of M(IL-4) activation (RELMα) and proliferation (Ki67 and EdU) (Fig. 1f and Supplementary Fig. 1b)⁴. AlvMs did not upregulate RELMα, Ki67 or EdU in response to IL-4c, whilst IntMs expressed intermediate amounts of RELMα and Ki67 in comparison to PECMs (Fig. 1f and Supplementary Fig. 1b). Similar observations were made in BALB/c mice (data not shown). AlvMs and IntMs had lower responsiveness to systemic IL-4c compared to MerTK⁺CD64⁺CD11b⁺ liver, colon or pleural cavity (PLEC) macrophages, which responded similarly to PECMs (Fig. 1g and Supplementary Fig. 1c). Together, this indicated that hyporesponsiveness to IL-4c was a feature of lung macrophages, and was particularly evident in AlvMs.

AlvMs express functional IL-4 receptor.

We next assessed whether AlvMs had reduced expression of IL-4 receptor (IL-4R) compared to IntMs and PECMs⁴. AlvM IL-4Rα expression was similar to that of IntMs and PECMs, and was not significantly affected by i.p. IL-4c (Fig. 2a). In addition, IL-4 was detected in BAL fluid with similar dynamics as in peritoneal washes (Fig. 2b), indicating that i.p. injected IL-4c could reach the airways. To further address whether AlvM responsiveness to IL-4c depended on route of administration, we administered a range of concentrations of IL-4c i.n. (0.05 μg, 0.5 μg or 5 μg), with the lowest dose typical of that detected in airways during type 2 inflammation. Although IntMs significantly upregulated RELMα in response to i.n. IL-4c compared to PBS (Fig. 2c), AlvMs did not do so, even at the highest IL-4c dose (Fig. 2c). These observations indicated that lack of M(IL-4) activation was characteristic of AlvMs, irrespective of IL-4c delivery route.

Fig. 2 |. AlvMs are less responsive than IntMs to IL-4c administered directly into the airways.

a, IL-4Rα expression by AlvMs, IntMs or PECMs from mice injected with PBS or IL-4c i.p. on day 0 and day 2, and lung tissue and PEC collected on day 4. Histograms representative of two independent experiments. b, ELISA of IL-4 amounts in BAL or PEC fluids 6, 12, 24 or 48 h after i.p. injection of PBS or IL-4c. Data representative of two independent experiments, n = 2 (PBS, 24 and 48 h), n = 3 (6 and 12 h) mice per group. c, Flow cytometry plots of RELMα expression in lung tissue AlvMs and IntMs on day 4 following i.n. PBS or IL-4c administration on day 0 and day 2 (left) and quantification of the percentage of RELMα⁺ cells (right). Data representative of three independent experiments, n = 2 (5 μg AlvM), n = 3 (AlvM: PBS, 0.05 and 0.5 μg. IntM: PBS, 0.05, 0.5 and 5 μg) mice per group. d, Histograms of pSTAT6 amounts in lung tissue AlvMs or PECMs 15 min after PBS or rIL-4 administered i.n. or i.p. to wild-type (WT) or Il4ra^−/− mice (left), and quantification of AlvM and PECM pSTAT6 expression (right). Data representative of two independent experiments, n = 3 mice per group. Data analyzed by one-way ANOVA with Tukey’s post-test for multiple comparisons, displayed as mean ± s.e.m., *P <0.05, ***P <0.001 and ****P <0.0001.

To investigate whether the lack of IL-4c responsiveness in AlvMs was due to impaired signaling, we measured expression of phosphorylated STAT6 (pSTAT6), a key transcription factor downstream of IL-4Rα engagement². Both AlvMs after IL-4 i.n., and PECMs after IL-4 i.p., had increased pSTAT6 expression compared to PBS controls which was not evident in Il4ra^−/− mice (Fig. 2d). In addition, AlvMs had high basal expression of pSTAT6, Ym1, pAkt T308 (mTORC1) and pAkt S473 (mTORC2) compared to PECMs, which was also evident in Il4ra^−/− mice (Fig. 2d and Supplementary Fig. 2). This showed that AlvMs displayed IL-4Rα-independent ‘tonic’ STAT6 and mTORC signaling in the steady state and could respond to i.n. IL-4 through STAT6 phosphorylation.

AlvMs show limited M(IL-4) activation during helminth infection.

To assess whether differential activation of AlvMs and IntMs was apparent in settings other than IL-4c injection, we infected C57BL/6 mice subcutaneously (s.c.) with the parasite Nippostrongylus brasiliensis, against which a type 2 response is essential for tissue repair as larvae migrate through the lung, and for clearance of adult worms from the intestines²³. As expected, a type 2 response, with eosinophilia and increased amounts of RELMα in BAL fluid, was detected after infection, compared to naïve mice (Fig. 3a,b). As infection progressed from day 2 to day 7, IntMs increased in numbers (Fig. 3c, d) and upregulated the M(IL-4) markers RELMα, Arg1 and Ym1 markedly more than AlvMs (Fig. 3e and Supplementary Fig. 3b,c). IntMs expressed higher amounts of Ki67 than AlvMs by day 7 after infection (Fig. 3e and Supplementary Fig. 3b), indicating that AlvMs did not acquire a clear M(IL-4) phenotype during infection. These observations contradict previous reports of AlvM M(IL-4) activation during type 2 inflammation^6–9,23. However, these previous studies have generally defined AlvMs as CD11c⁺Siglec-F⁺ (Supplementary Fig. 4a). Reliance on CD11c and Siglec-F to identify AlvMs could result in the inclusion of RELMα⁺ IntMs and eosinophils, particularly in inflamed mice (Supplementary Fig. 4a–d). Furthermore, use of scatter parameters in flow cytometry to exclude eosinophils could remove macrophages with similar granularity and Siglec-F, CD11b or CD11c expression (Supplementary Fig. 4e,f). Using refined flow cytometry, we have demonstrated that M(IL-4) activation of AlvMs was impaired compared to IntMs during N. brasiliensis infection.

Fig. 3 |. AlvMs are less responsive than IntMs during helminth infection.

a, Eosinophil numbers from lung tissue of naïve mice or on day 2, day 4 and day 7 following infection s.c. with 500 L3 N. brasiliensis larvae. Graphs show individual replicate mice, data pooled from four independent experiments, n = 18 (naïve), n = 17 (day 2), n = 9 (day 4), n = 5 (day 7) mice per group. b, ELISA of RELMα amounts in BAL fluid from naïve or infected mice. Data pooled from two independent experiments, n = 8 (naïve), n = 7 (day 2), n = 4 (day 4), n = 3 (day 7) mice per group. c, Numbers of lung tissue AlvMs and IntMs from naïve or infected mice. Data pooled from four independent experiments, n = 18 (naïve), n = 17 (day 2), n = 9 (day 4), n = 8 (day 7) mice per group. d, Flow cytometry plots identifying lung tissue AlvMs and IntMs from naïve or infected mice. Data representative of four independent experiments. e, Quantification of the percentage of RELMα⁺ and Ki67⁺ lung tissue AlvMs and IntMs from naïve or infected mice. Data pooled from four independent experiments, n = 18 (Naïve RELMα), n = 17 (day 2 RELMα), n = 9 (day 4 RELMα), n = 8 (day 7 RELMα), n = 13 (Naïve Ki67), n = 12 (day 2 Ki67), n = 5 (day 4 Ki67), n = 3 (day 7 Ki67) mice per group. Data analyzed by one-wayANOVA with Tukey’s post-test for multiple comparisons, displayed as mean ± s.e.m., *P <0.05, **P <0.01, ***P <0.001 and ****P <0.0001.

The pulmonary niche regulates AlvM responsiveness to IL-4.

The lung environment is a unique site that shapes macrophage development²⁴, with environmental signals vital for directing this process¹⁹. Further, upon removal from tissues, macrophages in culture display fundamentally altered gene expression^25,26. Consistent with reports that AlvMs from mice and humans can respond to IL-4 in vitro², AlvMs isolated from the lungs of C57BL/6 mice significantly upregulated expression of Retnla (encoding RELMα) and Arg1 after 48 h in culture with IL-4, while expression of Chil3 (encoding Ym1) was elevated after 12 h compared to media controls (Fig. 4a). We next addressed whether the airway environment limited the ability of AlvMs to undergo M(IL-4) polarization. We transferred CD45.2⁺ PECMs, which responded strongly to IL-4 in vivo (Fig. 1), i.n. into naïve CD45.1⁺ mice, followed by administration of IL-4c i.p. Donor CD45.2⁺ PECMs were detected in the lungs of recipient mice at day 5 after transfer (Fig. 4b). However, PECMs transferred i.n. displayed an activation profile similar to that of resident AlvMs, failing to upregulate RELMα and Ki67 in response to i.p. IL-4c administration compared to recipient PECMs (Fig. 4c). IL-4Rα expression on transferred CD45.2⁺ PECMs was similar to recipient PECMs (Fig. 4d), indicating that the impaired response of i.n. PECMs to IL-4 was not due to altered IL-4Rα expression.

Fig. 4 |. The pulmonary niche regulates AlvM responsiveness to IL-4 independently of host commensals.

a, mRNA expression by quantitative PCR (qPCR) of lung tissue AlvMs from naïve mice following culture for 12, 24 or 48 h in media alone or with rIL-4 (20 ng ml⁻¹). AU, arbitrary units. Data representative of four independent experiments, n = 2 (media), n = 3 (rIL-4) wells per group, each group pooled cells from eight mice. b–e, Donor (CD45.2⁺) and host (CD45.1⁺) macrophage populations identified by flow cytometry in lung tissue or PEC from host mice on day 5 after i.n. PBS or donor PECM transfer on day 0, then injection with IL-4c i.p. on day 1 and day 3. b, Flow cytometry plots identifying donor PECMs in host lung tissue. Data representative of four independent experiments. c, Quantification of the percentage of RELMα⁺ and Ki67⁺ host and donor macrophages isolated from host lung tissue or PEC. Data representative of four independent experiments, n = 6 mice per group. d, IL-4Rα expression by donor PECMs isolated from host lung tissue. Histogram representative of two independent experiments. e, RELMα and Ki67 expression by lung tissue AlvMs and IntMs, or PECMs, from specific pathogen-free (SPF) or germ-free (GF) mice on day 4 following i.p. PBS or IL-4c administration on day 0 and day 2. Data representative of three independent experiments, n = 3 mice per group. Data analyzed by one-way ANOVA with Tukey’s post-test for multiple comparisons, displayed as mean ± s.e.m., **P <0.01, ***P <0.001 and ****P <0.0001.

Interaction between the inhibitory receptor CD200R and its ligand CD200 has been described as a dominant negative regulator of AlvM activation in non-type 2 settings¹⁶. Although expression of CD200R was highest on AlvMs compared to IntMs (Supplementary Fig. 5a), we observed no significant difference in numbers of AlvMs or IntMs, or their expression of RELMα or Ki67, following i.p. IL-4c injection of Cd200r1^−/− mice (Supplementary Fig. 5b), indicating that AlvM hyporesponsiveness to IL-4 was independent of regulatory CD200–CD200R interactions.

In addition to immune mechanisms, macrophage responses at barrier sites may be modulated by airway components such as surfactant or mucus. SP-A and SP-D are abundant in the lower airways²⁷ and have been implicated in promotion of type 2 inflammation and M(IL-4) activation of AlvMs during helminth infection^21,28, while mucus is a major regulator of responses in lung and airway macrophages²⁹. IL-4c increased expression of the dominant pulmonary mucin Muc5b in airway epithelial cells compared to PBS-treated mice (Supplementary Fig. 5c). However, IntMs and AlvMs in Muc5b^−/− and Sfptd^−/− mice responded to IL-4c similarly to wild-type mice (Supplementary Fig. 5d,e), indicating that neither Muc5b nor SP-D were dominant factors in limiting the ability of lung AlvMs to undergo M(IL-4) polarization.

The airways host a wide diversity of commensals that could influence macrophage responses and are proposed to be key in regulating pulmonary allergic inflammation³⁰. Further, gut microbe-derived short-chain fatty acids are able to systemically regulate type 2 responses in the lung³¹. To test the involvement of commensals in regulating IL-4 responsiveness of AlvMs, we compared expression of RELMα and Ki67 in gnotobiotic (germ-free) mice and conventionally-housed (specific pathogen-free) mice following i.p. IL-4c administration. RELMα and Ki67 expression on AlvMs, IntMs or PECMs was similar in IL-4c-treated germ-free and specific pathogen-free mice (Fig. 4e), indicating that neither commensals nor their metabolites were involved in regulation of IL-4 responsiveness in any of these macrophage types. Together, these data indicated that the lung environment controlled AlvM responsiveness to IL-4 but this was independent of the microbiota, Muc5b or SP-D.

AlvMs and PECMs have distinct metabolic gene profiles.

To address which factors might determine the lack of AlvM responsiveness to IL-4 in vivo, we performed genome-wide messenger RNA profiling of AlvMs, IntMs and PECMs isolated from C57BL/6 mice injected i.p. with IL-4c or PBS. IL-4c induced a marked alteration of PECM gene expression, with 2,074 transcripts significantly upregulated or downregulated compared to PECMs from PBS-injected mice, including upregulation of core M(IL-4) genes such as Chil3, Retnla, Arg1 and Mrc1 (encoding mannose receptor) (Fig. 5a,b and Supplementary Tables 1 and 2). IntMs in IL-4c-treated mice significantly upregulated or downregulated 107 transcripts relative to IntMs from PBS-injected mice, including upregulation of Chil3 and Retnla but not Arg1 or Mrc1 (Fig. 5a,b and Supplementary Tables 3 and 4), while IL-4c did not significantly upregulate any of the core transcripts previously associated with M(IL-4) responsiveness in AlvMs, having almost no measurable effect on mRNA expression, with only two genes significantly downregulated compared to AlvMs from PBS-treated mice: Mipol1, a putative tumor suppressor³² and Gnpat, involved in lipid metabolism³³ (Fig. 5a,b and Supplementary Table 5). This indicated that AlvMs were broadly unresponsive to IL-4 in vivo. Further, AlvMs in PBS-treated mice had high basal expression of Chil3 and Mrc1 mRNA (Fig. 5b), consistent with high expression of Ym1 protein in steady state AlvMs (Fig. 1c) and indicating that these markers are not suitable for M(IL-4) assessment in AlvMs. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed that the principal pathways altered in PECMs by IL-4c included those involved in proliferation and metabolic processes (Supplementary Fig. 6a). However, few of these pathways were altered in response to IL-4c in AlvMs or IntMs (Supplementary Fig. 6b,c).

Fig. 5 |. AlvMs and PECMs display different metabolic gene profiles.

a, mRNA expression profiles (volcano plots) as determined by RNA-seq of PECMs, IntMs or AlvMs isolated from PEC or lung tissue by flow cytometry on day 4 following i.p. PBS or IL-4c administration on day 0 and day 2. Dashed lines represent P <0.01, and about two-fold change, IL-4c relative to PBS. b, Heatmaps of selected mRNA transcripts of genes that have been previously described as M(IL-4) markers, *indicates significance between IL-4c vs PBS of at least P <0.01.c, mRNA expression profile (volcano plot) of AlvMs versus PECMs isolated from IL-4c injected mice. Dashed lines represent P <0.01, and about two-fold change. d, Selected pathways from KEGG analysis (Supplementary Fig. 6) of significantly altered mRNA transcripts from c, black lines represent P <0.05. e, Relative transcript expression by AlvMs versus PECMs from IL-4c injected mice that were significantly altered (P <0.01, log2 normalized intensity), as identified from the glycolysis pathway by network analysis (several genes displayed more than one altered transcript variant), n = 2 (PECM PBS, PECM IL-4c, AlvM PBS, IntM PBS, IntM IL-4c), n = 3 (AlvM IL-4c) separate biological replicates, each replicate pooled cells from three to five mice.

We next directly compared transcript expression in AlvMs and PECMs from IL-4c-treated mice. This analysis indicated different gene expression profiles between the two macrophage populations (Fig. 5c and Supplementary Tables 6 and 7). In particular, pathways associated with glycolysis were impaired, while those associated with lipid metabolism and differentiation, such as PPAR and TGF-β, were elevated in AlvMs compared to PECMs (Fig. 5d,e and Supplementary Fig. 6d). This indicated that AlvM hyporesponsiveness to IL-4 may be due to impaired glycolysis and confirmed previous reports that lung macrophages have a distinctive metabolic state compared to macrophages in other tissues¹³.

Impaired glycolysis limits AlvM IL-4 responsiveness in vivo.

To investigate whether AlvMs had reduced glycolytic ability compared to PECMs, we analyzed changes in extracellular acidification rates (ECAR), a measure of glycolytic activity through detection of lactic acid as an end-product of glucose metabolism³⁴, in AlvMs and PECMs from naïve C57BL/6 mice. AlvMs exhibited significantly impaired glycolysis (reduced ECAR following glucose addition) and glycolytic reserve and capacity (defined as the ability to upregulate aerobic glycolysis) compared to PECMs (Fig. 6a). Analysis of oxygen consumption rates (OCR) showed that AlvMs also displayed reduced respiratory capacity (oxidative phosphorylation, OXPHOS) compared to PECMs (Supplementary Fig. 7a). Culture of freshly isolated AlvMs or PECMs with the glucose analog 2-NBDG, to measure uptake potential and glycolytic activity, showed that AlvMs acquired less 2-NBDG than PECMs, even when cocultured at a 1:1 ratio with PECMs (Fig. 6b). Further, both CD45.2⁺ PECMs transferred i.n. into CD45.1⁺ mice and resident CD45.1⁺ AlvMs had a reduced ability to acquire 2-NBDG in vivo following i.p. IL-4c, when compared to resident CD45.1⁺ PECMs (Fig. 6c). These observations indicated that the lung environment impaired the ability of AlvMs to both take up and utilize glucose.

Fig. 6 |. Impaired uptake and use of glucose renders AlvMs unresponsive to IL-4.

a, ECAR of AlvMs and PECMs isolated from lung tissue or PEC of naïve mice by flow cytometry, at baseline and after sequential treatment (vertical lines) with glucose, oligomycin (Oligo) or 2-Deoxy-d-glucose (2-DG) to measure glycolysis, glycolytic reserve and glycolytic capacity. Data representative of four independent experiments, n = 6 (AlvM), n = 10 (PECM) glycolytic stress test profile, n = 6 (AlvM) glycolysis, glycolytic capacity and glycolytic reserve, n = 10 (PECM) glycolysis, glycolytic capacity, n = 9 (PECM) glycolytic reserve, wells per group, each group pooled cells from eight mice. b, Flow cytometry plots of 2-NBDG uptake in vitro by BAL AlvMs, or PECMs, from naïve mice cultured separately or at a 50:50 mix for 20 min with fluorescently labeled 2-NBDG. c, 2-NBDG uptake in vivo by donor (CD45.2⁺) and host (CD45.1⁺) macrophage populations identified by flow cytometry in lung tissue or PEC from host mice on day 5 after i.n. PBS or donor PECM transfer on day 0, administration of IL-4c i.p. on day 1 and day 3, and i.p. injection of fluorescently labeled 2-NBDG 20 min before lung tissue and PEC collection. Data representative of two independent experiments, n = 6 mice per group. d, mRNA expression by qPCR of lung tissue AlvMs from naïve mice cultured for 0, 12, 24 or 48 h in media alone. Data representative of six (Eno1) or five (Slc2a6) independent experiments, n = 3 (Eno1), n = 2 (Slc2a6) wells per group, each group pooled cells from six to eight mice. e, mRNA expression by qPCR of lung tissue AlvMs from naïve mice cultured for 48 h in media alone, or with rIL-4 ± 2-DG. Data representative of three independent experiments, n = 2 (media), n = 3 (rIL-4), n = 3 (rIL-4 + 2-DG) wells per group, each group pooled cells from six to eight mice. Data analyzed using unpaired t-test (a) or a one-way ANOVA with Tukey’s post-test for multiple comparisons as indicated (b, c, e) or compared to 0 h (d), displayed as mean ± s.e.m., *P <0.05, **P <0.01, ***P <0.001 and ****P <0.0001.

However, AlvMs isolated from the lung and cultured for 48 h in vitro showed increased expression of Slc2a6 and Eno1, genes involved in glucose uptake and glycolysis (Fig. 6d), indicating that ex vivo culture of AlvMs enhanced their glycolytic ability. Next, we addressed whether glucose or fatty acid utilization was required for AlvMs to regain IL-4 responsiveness in vitro. The ability of cultured AlvMs to upregulate Retnla, Arg1 and Chil3 in vitro in response to IL-4 was markedly inhibited by 2-deoxyglucose (2-DG), a competitive glucose inhibitor, compared to culture with IL-4 alone (Fig. 6e) but not significantly affected by addition of etomoxir, an inhibitor of fatty acid oxidation (FAO) (Supplementary Fig. 7b). Similarly, even though AlvMs had high expression of genes associated with the TGF-β pathway (Fig. 5d) and addition of TGF-β reduced expression of IL-4-induced Retnla in cultured AlvMs (Supplementary Fig. 7c), it had no significant effect on Chil3 expression, and increased Arg1 expression, indicating that TGF-β was not vital in limiting AlvM IL-4 responsiveness. Together, these data indicated that the lung environment regulated AlvM M(IL-4) activation through modulation of their metabolism.

Discussion

We have shown that AlvMs were hyporesponsive to type 2 inflammation mediated by IL-4c injection or helminth infection. This lack of responsiveness was conferred by the lung environment and affected AlvM metabolic activity and ability to both take up and metabolize glucose. Removal of AlvMs from the lung reversed this metabolic constraint, enabling M(IL-4) activation.

Although numerous studies have reported that pulmonary macrophages upregulate M(IL-4) markers, they either did not unequivocally distinguish between AlvMs and IntMs in their analyses, relied on IL-4 stimulation of macrophages ex vivo, or used M(IL-4) markers that are already highly expressed by AlvMs at steady state^6–9,20,21. Our results suggest that such work may require reassessment to precisely identify which macrophage populations respond to IL-4 in vivo. Our data indicate that IntMs will be the major macrophage subpopulation to respond in pulmonary type 2 inflammatory settings. This distinction is likely to be important for accurate understanding of the pathogenesis of pulmonary type 2 disease, given that M(IL-4) macrophages have been implicated in wound repair during type 2 inflammation^23,35. We would speculate that IntMs play a more important role than AlvMs in processes such as resolving tissue damage in the lung, due to their greater ability to respond to IL-4.

Although negative regulation of macrophage activation is a well-described feature of the lung and is thought to be vital to restrict overexuberant responses against inhaled material, viral or bacterial infection¹, how pulmonary M(IL-4) responses are regulated is currently poorly understood. While we have not identified which specific components of the pulmonary environment restricted AlvM activation by IL-4, we have shown that this was independent of Muc5b, SP-D and commensals or their metabolite products, all of which are features of the lung that have previously been implicated in modulating pulmonary macrophage responses to bacteria, helminth infection and allergic airway inflammation^28,29,31.

Metabolism is a key determinant of immune cell function and is central in governing how macrophages respond to a variety of signals, including type 1 and type 2 cytokines^34,36. Most studies so far have profiled metabolic responses in bone marrow-derived macrophages in vitro³⁷ and have not addressed how tissue environments alter macrophage metabolism and function in vivo. From such work, it has been proposed that type 2 cytokines promote amino acid and lipid metabolism (including FAO) feeding into OXPHOS, whereas glycolysis is more associated with type 1 macrophage polarization^34,37–39. We found that AlvMs had a distinctive metabolic state compared to PECMs, with elevated expression of genes associated with PPAR-γ and lipid metabolism, a profile that would be expected to enhance FAO, OXPHOS and M(IL-4) activation^34,37,38. However, defective glycolytic ability rendered AlvMs hyporesponsive to IL-4, consistent with recent observations that glycolysis can mediate macrophage responses to IL-4^40,41 and with studies linking altered metabolic state with AlvM ability to respond to Mycobacterium tuberculosis⁴². Our demonstration that the lung environment controls macrophage metabolism during type 2 inflammation, together with recent evidence that metabolism also regulates dentritic cell control of allergic airway inflammation⁴³, suggests caution in interpreting metabolic data generated from model macrophages or dentritic cells in culture. Our data also imply that the distinctive metabolic profile of AlvMs may be directly linked to negative regulation of their activation and function at steady state and during inflammation¹.

One factor to consider in how the lung may affect AlvM activation is amounts of metabolic substrates, including glucose, present in airways. Glucose levels in air surface liquid, which covers the airway epithelium, are 12.5-fold lower than in blood⁴⁴. Such low glucose concentrations, maintained through highly effective epithelial cell glucose transport⁴⁵, appear vital to prevent bacterial outgrowth in airways^44,46. Elevated glucose is found in patient sputum during chronic obstructive pulmonary disease⁴⁷, while glucose levels and glucose metabolism rise in the lung during asthma^48,49. Together with our data, this leads to the intriguing hypothesis that glucose availability and/or use could be exploited to therapeutically modify pulmonary disease.

We showed that AlvMs removed from the airways regained ability to respond to IL-4 in vitro, while PECMs transferred into the airways lost IL-4 responsiveness. The stark difference between AlvM ability to respond to IL-4 in vitro and in vivo resonates with the reported transformation of microglial transcriptional identity when removed from the brain^25,26 and cautions against reliance on AlvMs in vitro for functional studies. This may be particularly relevant for human AlvMs, given current experimental dependence on their culture ex vivo, and highlights the need for development of innovative approaches to better assess human AlvM function in vitro. Similarly, identification of markers for human macrophage subpopulations and their M(IL-4) activation is urgently needed. The current revolution in single-cell sequencing for definition of cellular networks indicates that this approach applied to human AlvMs should be illuminating. In both human and murine type 2 inflammation, it will also be important to understand how monocytes recruited to the lung differentiate and influence airway or tissue macrophages, as resident AlvMs can be replaced by regulatory monocytes during viral infection⁵⁰. Our data indicate that the airway environment will play a key role in influencing activation and function of AlvMs during type 2 inflammation, irrespective of their origin.

More broadly, this work illustrates the pivotal role of the tissue environment in the regulation of macrophage metabolic activity and ability to respond to type 2 cytokines, a principle that will likely be relevant in diverse tissue and disease settings. Local differences in substrate availability, and alterations of such during inflammation, may provide a metabolic mechanism to modulate the activation and function of macrophages and other immune cells in a tissue-specific manner in health and in disease.

Methods

Experimental animals.

Cx3cr1^eGFP/+, Cd200r1^−/−, CD45.1⁺, Il4ra^−/−, Muc5b^−/− and Sfptd^−/− were generated as described previously^16,28,29. All were on a C57BL/6 background except Il4ra^−/− which were BALB/c. C57BL/6 or BALB/c mice were purchased from Envigo. Mice were bred and maintained under specific pathogen-free conditions at The University of Manchester. Germ-free mice were from the University of Manchester Gnotobiotic Facility. All experiments were approved under a project license granted by the Home Office UK, and by the University of Manchester Animal Welfare and Ethical Review Body, and performed in accordance with the United Kingdom Animals (Scientific Procedures) Act of 1986.

In vivo mouse models.

IL-4 complex delivery in vivo.

Long-acting IL-4 complexes (IL-4c: IL-4/mAb to IL-4) were prepared and used as previously described^4,5. Recombinant murine IL-4 (BioLegend) was combined with rat IgG1 mAb to IL-4, 11B11 (BioXcell), at a 1:5 molecular weight ratio⁵¹. Mice were injected i.p. with 5 μg of IL-4 (complexed to 11B11) or Dulbeccos PBS (PBS, Sigma) on day 0 and day 2. Alternatively, 50 μl PBS or varying doses of IL-4c (5–0.05 μg) was administered directly i.n. on day 0 and day 2. Tissues were collected on day 4 after initial injection.

N. brasiliensis infection.

Wild-type mice were infected s.c. with 500 N. brasiliensis third-stage larvae and tissues collected days 2, 4 and 7 after infection.

In both models, to assess cell proliferation, mice were injected i.p. with 0.5 mg 5-ethynyl-2′-deoxyuridine (EdU; ThermoFisher) in 200μl PBS 3 h before harvest to label cells in S-phase of cell cycle as has previously been described³. This short window was chosen to provide an accurate readout of in situ cell proliferation at the tissue of interest and avoid detection of cells that had recently proliferated elsewhere before recruitment.

Isolation of immune cells from the peritoneal cavity, BAL, lung, intestine and liver.

Following killing, PEC or BAL cells were obtained by washing of the peritoneal cavity or lungs with PBS containing 2% FBS and 2 mM EDTA (Sigma). Lungs were processed as previously described⁵², incubated at 37 °C for 40 min with 0.8 U ml⁻¹ Liberase TL and 80 U ml⁻¹ DNase I type IV in HBSS (all Sigma). The digestion was stopped with PBS containing 2% FBS (Sigma) and 2 mM EDTA (Sigma), with the resulting suspension then passed through a 70 μm cell strainer. In some cases, before collection, i.v. or i.n. instillation of fluorescently labeled anti-CD45 (clone: 30-F11) was used to distinguish between blood circulating (i.v. CD45-FITC⁺), airway resident (i.n. CD45-PE⁺) and tissue resident leukocytes (CD45-FITC⁻PE⁻), as described previously^22,53. Mononuclear cells from the intestine and liver were isolated as previously described^54,55. Erythrocytes were lysed using RBC lysis buffer (Sigma) and cells counted and processed for flow cytometry.

Flow cytometry and cell sorting.

Equal numbers of cells were stained for each sample, washed with ice-cold PBS and stained with Zombie UV dye (1:2,000, BioLegend) for 10 min at room temperature. All samples were then blocked with 5 μg ml⁻¹ αCD16/CD32 (2.4G2; BioLegend) in FACS buffer (PBS containing 2% FBS and 2 mM EDTA) before staining for specified surface markers at 4 °C for 25 min. For detection of intracellular molecules, following surface staining, cells were fixed with 1% paraformaldehyde in PBS for 10 min at room temperature, and permeabilized with the Transcription factor staining kit (eBioscience) then stained with the relevant antibodes. If mice had been treated with EdU, cells were stained using the Click-iT Plus EdU Alexa Fluor 647 Flow Cytometry Assay Kit (Molecular Probes) using an adapted protocol for a final staining volume of 50 μl. Samples were acquired using a 5 laser Fortessa with BD FACSDiva software and analyzed with FlowJo software (versions 9 and 10, Tree Star). Sorting of macrophage populations from the PEC (based on DAPI⁻F4/80⁺) and lung (DAPI⁻, CD45⁺MerTK⁺CD64⁺ and CD11b⁺ IntMs or Siglec-F⁺ AlvMs) was performed using an Influx (BD Biosciences) using the 140 μm nozzle and 7.5 psi pressure, to a purity of ~ 95–99%. In some cases, myeloid cells were enriched before sorting by removal of lymphoid cells using Dynabeads (ThermoFisher; biotinylated anti-CD3, CD19, B220, Ly6G, NK1.1, Ter119 and streptavidin-Dynabeads) according to the manufacturers instructions.

pSTAT6 and pAkt intracellular staining.

To assess pSTAT6 and pAkt activation, 5 μg rIL-4 was administered i.n. or i.p. 15 min before tissue collection. Cells from PEC or BAL washes were directly incubated with an equal volume of formalin (final concentration 2% formalin) for at least 10 min at room temperature, resuspended in 500 μl ice-cold methanol at 4 °C for 10 min, washed twice with FACS buffer, then stained and acquired (as described above).

Intranasal transfer of PEC macrophages into the airways.

PECMs were sorted as described above from CD45.2 mice. PBS, or 1 × 10⁶ donor PECMs in PBS, were instilled into the airways of CD45.1 recipient mice via i.n. transfer. Mice were treated with IL-4c (as described above) and cells from the lungs were isolated and processed as described above. In some experiments, 100 μg FITC labeled 2-NBDG (Sigma), internalization of which measures glucose uptake potential and glycolytic activity⁵⁶, was injected (i.p.) 20 min before tissue collection.

In vitro culture of macrophages.

AlvMs or PECMs FACS isolated from naïve mice (as described above) were cultured in RPMI 1640 (containing 10% FBS,1% PenStrep, 1% l-glutamine, all Sigma) for up to 48 h at 37 °C. In some experiments, they were incubated with 50 μg ml⁻¹ FITC labeled 2-NBDG (Sigma) for 20 min (either separately or a 50:50 mix of the two) or in the presence of rIL-4 (20 ng ml⁻¹) ± 1 mM 2-DG (Sigma), 200 μM etomoxir (Sigma) or recombinant human TGF-β (10 ng ml⁻¹) (Peprotech).

Imaging cytometry.

Cells were stained and fixed (as described above) in ImageStream buffer (PBS containing 1% FBS and 2 mM EDTA). Data acquisition was performed on ImageStreamX (Amnis/EMD Millipore) equipped with 405, 488, 561 and 642 nm lasers. Single cells were discriminated from cell aggregates on the basis of area and aspect ratio. In focus cells were selected on the basis of high-gradient RMS of the bright field image. Images of cells were acquired with a ×40 objective including bright field images (Channels 1 and 9; 420–480 nm and 570–595 nm), CD11b (Channel 2; 480–560 nm), MerTK (Channel 3; 560–595 nm), Siglec-F (Channel 4; 595–660 nm), CD64 (Channel 6; 740–800 nm), Zombie UV (Channel 7; 420–505 nm) and CD45 (Channel 8; 505–570 nm). All data analysis was performed using the IDEAS software version 6.

Histology.

Histological sections were prepared from lungs perfused with freshly prepared Carnoy’s solution (60% absolute methanol, 30% choroform, 10% acetic acid) and embedded in paraffin. Sections of 5 μM were subjected to immunohistochemical analysis for Muc5b (custom polyclonal antisera)⁵⁷. Bound primary antibody was detected with goat anti-rabbit Alexa fluor 488. Images were captured using an Olympus BX51 upright microscope using a ×20/0.5 EC Plan-neofluar objective and captured using a Coolsnap ES camera (Photometrics) through MetaVue Software (Molecular Devices). Images were then processed and analyzed using ImageJ (ref.⁵⁸).

Enzyme-linked immunosorbent assays.

ELISAs to detect RELMα (PeproTech) and IL-4 (BioLegend) were performed on BAL or PEC fluid, as per manufacturers instructions.

RNA isolation, library construction and analysis.

To generate RNA libraries of sorted macrophage populations, mice were exposed to PBS or IL-4c and two separate pooled biological replicates were generated for PECM PBS, PECM IL-4c, AlvM PBS, IntM PBS and IntM IL-4c groups while three separate pooled replicates were collected for the AlvM IL-4c group. Each pooled biological replicate was generated from cells isolated from 3–5 mice. After FACS sorting, each sample was lysed with RLT buffer (Qiagen) and RNA isolated with RNeasy microkits (Qiagen) according to the manufacturers instructions. Sample RNA integrity was confirmed using TapeStation (Agilent), with all samples showing RNA integrity numbers of ~8.8–10. RNA quality was assessed by Fragment Analyzer (Advanced Analytical Technologies) and 20 ng total RNA was used for each library. RNA samples were processed with an Illumina TruSeq RNA Access Library prep kit, following the manufacturers instructions. Libraries were quantified with Qubit HS (ThermoFisher) and Fragment Analyzer (Advanced Analytical Technologies). Indexed libraries were pooled and sequenced on an Illumina NextSeq 500 using paired-end chemistry with 75 base pair (bp) read length.

For analysis, the raw RNA sequences were quality assessed using FASTQC and no further trimming was performed. The latest mouse transcript set (release 87, ‘REL87’) was obtained by ftp from ensembl (ftp://ftp.ensembl.org/pub/release-87/fasta/mus_musculus/) and annotation acquired using BioMart. Transcripts for both complementary DNA and non-coding RNA were used. Alignments (–end-to-end,–very-sensitive -p 30–no-unal–no-discordant settings) to the REL87 reference set were performed using bowtie2 (version 2.2.7). Alignments were stored in indexed BAM files. Normalized data provided the input for statistical hypothesis testing, in which we sought to identify loci that were significantly different between sample groups. We were also interested in the degree of difference, that is the fold-change. In the outputs, the fold-changes (logFC) are given as log2 values, with a positive logFC representing upregulation and a negative logFC indicating downregulation. For each comparison, the first group (A) is the numerator, while the second group (B) is the denominator. A positive logFC for the comparison ‘A–B’ indicates upregulation in A relative to B. Comparisons, manually chosen to explore the data, were undertaken using linear modeling. Subsequently, empirical Bayesian analysis was applied (including vertical, in a given comparison, P value adjustment for multiple testing, which controls for false discovery rate). For each comparison, the null hypothesis was that there was no difference between the groups being compared. The Bioconductor package limma was used and an overview of the underlying biological changes occurring in each comparison obtained by functional enrichment analysis from KEGG pathway membership. The significance threshold for functional analysis was manually chosen to be P <0.01

Quantitative PCR.

After culture macrophages were lysed in the plate using RLT lysis buffer and RNA was isolated with RNeasy microkits (Qiagen) according to the manufacturers instructions. Complementary DNA was generated from extracted RNA using SuperScript-III and Oligo-dT (ThermoFisher). Relative quantification of genes of interest was performed by qPCR analysis using QuantStudio 12 K Flex system and SYBR Green master mix (ThermoFisher), compared with a serially diluted standard of pooled cDNA. Expression was normalized to β-actin (primers as in Supplementary Table 8).

Seahorse extracellular flux analysis.

FACS isolated PECMs or AlvMs from a pool of eight mice were plated at 150,000 cells per well and allowed to adhere for at least 1 h. ECAR and OCR were measured in XF media (modified DMEM containing 2 mM L-glutamine) under basal conditions, in response to 25 mM glucose, 20 μM oligomycin, 100 mM 2-DG (ECAR) or 20 μM Oligomycin, 15 μM FCCP, 10 μM Antimycin A, 1 μM Rotenone (OCR) (Sigma) using a 96-well extracellular flux analyzer XFe-96 (Seahorse Bioscience).

Statistical analysis.

Data are shown as mean values ± s.e.m. Where applicable, data were analyzed by unpaired t-test, one-way or two-way analysis of variance (ANOVA) with Tukey’s post-test as appropriate. Significant differences were defined at P <0.05. Statistical analysis was performed using GraphPad PRISM version 7.

Reporting Summary.

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Code availability

Bioinformatics analyses were performed with publicly available code from bioconductor.org.

Data availability

The data that support the findings of this study are available from the corresponding author upon request. RNA-seq data were deposited at Gene Expression Omnibus, with the following accession code: GSE126309.