Multinucleated Giant Cells in Adipose Tissue Are Specialized in Adipocyte Degradation

Obesity is a growing epidemic worldwide, leading to diabetes type 2, stroke, cardiovascular disease, and cancer (1). For this reason, obesity is one of the main threats to global human health and life expectancy (2). To date, the only effective long-term treatment for obesity is invasive bariatric surgery; there is an urgent need for new therapies. Therefore, it is essential to broaden our limited understanding of the cellular mechanisms underlying adipose tissue (AT) dysfunction.

Obesity is associated with chronic low-grade inflammation of visceral AT closely linked to insulin resistance (3). In obese individuals, a high caloric diet leads to massive lipid uptake in adipocytes, resulting in adipocyte hypertrophy. In addition, adipocyte death attracts macrophages, which is indicative of AT inflammation, resulting in an accumulation of AT macrophages (ATMs), so-called crown-like structures (CLSs), surrounding dying adipocytes (4). Additionally, there is a switch in ATM immune phenotype from an anti-inflammatory (M2) population in lean individuals to a more proinflammatory (M1) population in obesity (5,6).

Interestingly, in histological sections of obese AT, multinucleated giant cells (MGCs) are frequently observed (4,7). Before now, little was known about these macrophage-like cells, which presumably form through cell-to-cell fusion. In general, the function of MGCs has been debated and is probably context specific. In tuberculosis, schistosomiasis, and other granulomatous diseases, MGCs are disease hallmarks; foreign-body giant cells attack foreign material, and osteoclasts (endogenous MGCs of the bone) are important in bone homeostasis (8). In tuberculosis, MGCs seem to restrict the spread of mycobacteria, whereas in HIV infection, virus particles seem to spread using cell-to-cell fusion, including MGCs (9,10).

Importantly, MGCs in obese AT have never been studied in detail before. Therefore, the aim of this study was to characterize MGCs with regard to the following: pro- and anti-inflammatory properties, uptake of lipids from dying adipocytes, mechanisms involved in MGC formation, and growth in MGC number. We tested the effect of different cytokines on MGC formation in living AT ex vivo using an established AT explant model and found that interleukin (IL)-4 treatment was sufficient to enhance the generation of MGCs in AT. Moreover, MGCs expressed not only pro- but also anti-inflammatory marker proteins. Most importantly, MGCs exhibited significantly higher lipid content in vivo and were able to phagocytize larger particles compared with regular, mononucleated ATMs in vitro. Therefore, our data suggest that MGCs are specialized for clearance of dead adipocytes in obesity.

Mice strains were maintained in the local animal facility in 12-h light/dark cycles and with free access to food and water. For diet-induced obesity, male C57BL/6J, LysMCre^+/− × IL-4Rα^flx/− (IL-4Rα^Δmyel), or MacGreen (CSF1R-eGFP^+/−) mice on a C57BL/6 background were fed a high-fat diet (HFD; 60% kcal fat; Ssniff-Spezialdiäten GmbH, Soest, Germany) starting at 6–8 weeks of age. Control littermates were kept on a regular chow diet (chow; 9% kcal fat; Ssniff-Spezialdiäten). Leptin receptor–deficient mice (The Jackson Laboratory, Bar Harbor, ME) were analyzed at 12 weeks of age. All experiments were performed in accordance with the National Institutes of Health guidelines and approved by the local authorities (Regierungspräsidium Leipzig).

Twenty-five obese patients who underwent laparoscopic Roux-en-Y gastric bypass surgery were recruited at the Leipzig University Medical Center (Leipzig, Germany). The human study was approved by the ethics committee of the University of Leipzig (approval number: 017–12–23012012), and all participants provided written informed consent before taking part in the study. Exclusion criteria included chronic or acute inflammatory disease as defined by white cell counts and/or hs-CRP along with clinical symptoms, antibiotic treatment within 2 months before surgery, pregnancy, and/or nursing. All individuals underwent routine clinical phenotyping as listed in Table 1. Abdominal subcutaneous AT samples were acquired during surgery, paraffin embedded, sectioned, and stained as described previously (11). After staining, two sections per individual were evaluated with regard to occurrence of MGCs within the total sections in a blinded fashion. MGCs were defined as Mac-2⁺ cells (1:1,000; Cedarlane, Burlington, ON, Canada) with at least three nuclei according to previous reports (12). Finally, patients with detectable MGCs and without detectable MGCs were statistically analyzed.

Cultivation of AT explants from chow-fed C57BL6 and IL-4Rα^−/− mice or HFD-fed MacGreen or IL-4Rα^Δmyel mice was performed as described previously (13). In some experiments, AT explants were treated with indicated cytokines (50 ng/mL; PeproTech, Hamburg, Germany, and Sigma-Aldrich).

Staining of whole mounted AT was performed as previously described (14). Images were taken using an Olympus FV1000 confocal laser scanning microscope (Olympus Deutschland GmbH, Hamburg, Germany). Live imaging was performed as described previously with a special focus on putative MGCs using an Olympus FV300 confocal laser scanning microscope (Olympus) (15).

To determine the formation of MGCs, the stroma vascular fraction (SVF) of 100 mg freshly dissected epididymal AT or 10 cultured AT explants was isolated using collagenase type II (Worthington, Lakewood, NJ). Subsequently, the SVF cell suspension was filtered through 75-µm mesh and seeded onto a poly-L-lysine–coated (Sigma-Aldrich) coverslip in RPMI medium containing 10% FCS (Gibco) and antibiotics (1% penicillin/streptomycin; Gibco). After 2 h of attachment, cells were counterstained with Hoechst dye 33342 (1:10,000; Life Technologies) and subsequently fixed for 5 min using zinc formalin (Polysciences, Hirschberg, Germany).

For quantification, 10 randomly chosen images were taken per condition with a 10× objective using an Olympus BX51 epifluorescence microscope (Olympus), and GFP⁺ cells in MacGreen mice or Mac-2⁺ cells in nonreporter mice were semiautomatically counted using cellSens software (Olympus). MGCs were defined as macrophages with at least three nuclei according to previous reports (12).

For the phagocytosis assay, SVF cells were incubated for 48 h with serum-opsonized and BODIPY-stained 45-µm polystyrene beads (30,000 beads per well; Polysciences). Representative images were acquired using a confocal FV1000 microscope (Olympus) to ensure full engulfment of the phagocytized beads.

To perform flow cytometric analysis, freshly dissected AT or cultured AT explants were digested as described above. After digestion, the cell suspension was filtered through 75-µm mesh, and Fc receptors were blocked by using anti-CD16/32 (1:100; eBioscience, Frankfurt, Germany) treatment for 10 min on ice. Subsequently, cells were stained for 20 min on ice with anti-CD45-FITC (30-F11), anti-F/480-PE-Cy7 (BM8), anti-CD11b (M1/70), anti-CD11c-PE (N418, all 1:100; all from eBioscience), and anti-CD36-PE (HM36, 1:100; Biolegend, San Diego, CA) for standard flow cytometry. After extracellular live staining, cells were fixed using the Fixation/Permeabilization Solution Kit and stained with 7-AAD (both BD Pharmingen, Heidelberg, Germany) for 5 min at room temperature for nuclear counterstain. For imaging flow cytometry (ImageStream), cells were stained using anti-CD80-AlexaFluor647, anti-CD86-AlexaFluor647, anti-MHC-II-AlexaFluor647, anti-CD11c-AlexFluor647 (all 1:100; eBioscience), and anti-CD206-AlexaFluor647 (ER-MP23; 1:100; AbD Serotec, Kidlington, U.K.) for 20 min on ice, and Hoechst dye 33342 (1:10,000) was applied for 30 min on ice for nuclear counterstaining followed by fixation.

For analysis of flow cytometric data, single cells were gated for 7-AAD⁺ cells. Subsequently, CD45⁺ and F4/80⁺ cells were defined as ATMs (Supplementary Fig. 1). Analysis was performed using an LSR II (BD Pharmingen) equipped with FACS Diva software 8.0. Quantification was performed using FlowJo software 10.0.5 (Tree Star, Ashland, OR). For imaging flow cytometry, the Amnis ImageStream^X Mark II Imaging Flow Cytometer (Luminex, Austin, TX) equipped with INSPIRE software (Luminex) was used. Data analysis was performed using IDEAS software (Luminex).

3T3-L1 adipocytes were differentiated and cultured as previously described (16,17). Afterward, adipocyte death was induced using 25 nmol/L tumor necrosis factor-α (TNFα) for 24 h following established protocols (18). After incubation, adipocytes were cocultured with freshly isolated SVF cells of epididymal AT from HFD-fed MacGreen mice for 24 h.

Data are presented as means ± SD or as box plots (whiskers represent minima and maxima) for at least three animals evaluated by the Student t, Mann-Whitney U, or Wilcoxon test, according to data distribution. Identification of outliers (Grubbs test) and assessment of normality of data distribution (Shapiro-Wilk test) were performed using GraphPad Prism 8.0 (GraphPad Software, Inc., La Jolla, CA). The Pearson correlation coefficient was calculated using GraphPad Prism (GraphPad). A P value <0.05 was considered statistically significant.

All data generated or analyzed during this study are included in the published article (and its Supplementary Material). The mouse models analyzed during the current study are available from the corresponding author upon reasonable request.

Previous studies focused on the role of ATMs in obesity and AT inflammation in several mouse models. Extraordinarily large cells with multiple nuclei, often appearing in a circular orientation, are regularly found in obese mice and humans (Fig. 1A–C). Of note, MGCs were defined as macrophages with at least three nuclei according to previous reports (12). In obese humans undergoing bariatric surgery, 28% (7 of 25) exhibited MGCs within subcutaneous AT. Interestingly, when comparing the characteristics of patients with MGCs with those of patients without detectable MGCs, blood leukocytes were significantly elevated in the MGC group. Furthermore, there was a trend toward higher values for fasting plasma insulin, fasting plasma glucose, and HOMA indices in the MGC group, suggesting pathophysiological relevance (Table 1). However, these MGCs were preferentially observed near dead adipocytes in CLSs. Expression of the pan-macrophage marker F4/80 and GFP-labeled macrophages (MacGreen mouse line) in obese animals further validated their macrophage origin (Fig. 1A). Additionally, by long-term live imaging of AT explants, we could detect MGCs, which exhibited a remarkable motility (Fig. 1D and Supplementary Movie 1). Therefore, the aim of this study was to characterize MGCs and their function in AT in more detail.

Here, we established a model of AT inflammation in MacGreen mice using an HFD. After 12 weeks of HFD feeding, mice exhibited a significantly higher body weight (Fig. 2A), and the weight of the epididymal AT also increased significantly compared with AT in their lean littermate controls (data not shown). Furthermore, the frequency of GFP⁺ cells among SVF cells, corresponding to ATMs in MacGreen (15), increased after 12 and 24 weeks of an HFD compared with that in their lean littermates (Fig. 2B), and the ratio of pro- (M1) to anti-inflammatory (M2) ATMs shifted toward a proinflammatory (M1) phenotype (Fig. 2C and D). In line with these findings, CLS density increased starting from 12 weeks of an HFD, indicating adipocyte death (Fig. 2E and F). Furthermore, semiautomatic quantification of ATM size by GFP⁺ area revealed a significantly enlarged cell size of ATMs after 12 and 24 weeks of an HFD compared with that in lean littermates (Fig. 2G). MGCs in vivo were not detectable in mice after 4 or 12 weeks of an HFD. However, after 24 weeks of an HFD, MGCs appeared in the SVF cells, representing ∼3% of all ATMs (Fig. 2H and I). Because MGCs were defined as ATMs with at least three nuclei, they represented up to ∼10% of ATM-associated nuclei in obese AT.

Formation of MGCs in obese AT occurred late in the course of AT inflammation, and their appearance was linked to inflammatory signs, such as CLS occurrence (Fig. 1A–C), a rising number of ATMs (Fig. 2B), and a phenotypic switch toward proinflammatory M1 macrophages (Fig. 2C and D). Therefore, we next tested whether adipocyte death per se could stimulate MGC formation ex vivo (Fig. 3). First, we performed long-term cultivation of AT explants from lean mice with almost no endogenous CLSs (<0.1% of adipocytes; Fig. 2E) or MGCs (Fig. 2H) and studied the potential formation of CLSs and MGCs as a result of adipocyte death ex vivo. Interestingly, CLSs and MGCs both formed in parallel starting from 7 days of long-term cultivation (Fig. 3A–D), indicating a strong association between both processes. For a more controlled experimental condition, TNFα overstimulation of adipocytes was successfully established as model to kill 3T3-L1 adipocytes in vitro as described previously (18) (data not shown). Interestingly, cocultivation of dead adipocytes with SVF cells of obese AT led to a significantly higher number of MGCs compared with the number in the PBS-treated control (Fig. 3E and F), indicating that adipocyte death increases MGC formation.

To further characterize MGCs in AT, we established flow cytometric imaging of SVF cells from epididymal AT of obese MacGreen mice (Fig. 4A). As expected, MGCs were threefold bigger in size than mononucleated ATMs (Fig. 4B). Moreover, MGCs did show a significantly higher fluorescence intensity for nuclear counterstaining (Hoechst related to the cell size) compared with regular ATMs as a sign of polyploidy (Fig. 4D). Furthermore, the intensity of the nuclear staining was significantly correlated with cell area (Fig. 4C), validating our approach of detecting MGCs by imaging flow cytometry. More importantly, MGCs expressed both pro- and anti-inflammatory marker proteins, such as CD11c, CD206, CD80, CD86, and MHC-II (Fig. 4A–E), with a substantial variety of expression intensity (Fig. 4G). Of note, ∼90% of all MGCs expressed GFP and F4/80 as well as the most common proinflammatory (M1) marker, CD11c (Fig. 4A–E), which has been validated by whole mount staining of obese AT in situ (Fig. 4F). Most interestingly, more anti-inflammatory markers, such as CD206 and CD80, were expressed on ∼50% of all MGCs, indicating a possible fusion of pro- and anti-inflammatory ATMs or a de novo expression of M2 markers by MGCs (Fig. 4A–E and Supplementary Fig. 2). Furthermore, expression of MHC-II as well as T-cell costimulatory markers CD80 and CD86 indicated a role in antigen presentation of MGCs (Fig. 4A–E). In comparison, MGCs expressed higher levels of GFP and F4/80 as mononucleated ATMs (Fig. 4G).

Next, we specifically focused on MGCs in living AT by long-term live imaging. Putative MGCs were identified by an increased cell size, which was several-fold larger than the size of regular ATMs also expressing GFP (Fig. 5A). Interestingly, regular ATMs struggled with the uptake of large lipid remnants (∼25 µm), resulting in several failed attempts at phagocytosis by regular ATMs (t = 3–6 h). After 13 h of observation, an MGC formed and rapidly phagocytized the respective lipid remnant, with subsequent intracellular degradation (13–30 h; Fig. 5A and Supplementary Movie 2). Thus, live imaging indicated that MGCs could take up lipid remnants, which are indigestible for regular ATMs. To quantify the phagocytosis of large particles, acutely isolated SVF cells of obese MacGreen mice were incubated with 45-µm beads, which are too big for phagocytosis by regular ATMs (cell size ∼20 µm) (19). Interestingly, MGCs were responsible for all observed events of successful bead uptake (14 of 14); none were attributable to mononucleated ATMs (Fig. 5B and C).

Next, we aimed at estimating the phagocytic capacity of MGCs in vivo. Therefore, we used flow cytometric imaging to compare the lipid content and cell granularity of ATMs and MGCs (Fig. 5D–I). Interestingly, MGCs exhibited a significantly higher lipid content (related to cell area) than mononucleated ATMs measured by BODIPY staining (Fig. 5D and E). In addition, BODIPY fluorescence was strongly correlated with cell size (Fig. 5F), indicating higher lipid uptake. Of note, binucleated ATMs exhibited an intermediate phenotype between mono- and multinucleated ATMs with regard to cell area, Hoechst intensity, and lipid content (Supplementary Fig. 3). Next, MGCs showed a significantly higher side scatter signal related to cell area, indicating greater cell granularity (Fig. 5G and H), as a result of more intracellular vesicles, which was also positively correlated with respective cell size (Fig. 5I). Of note, granules did not colocalize with intracellular lipid droplets (Supplementary Fig. 4).

Finally, we tested the effect of different cytokines on their ability to force MGC formation ex vivo. Therefore, we used an obese AT explant model treated for 48 h with several cytokines. Interestingly, stimulation with granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-4, and IL-13 led to a significantly enhanced frequency of MGCs in the SVF by approximately twofold (Fig. 6A and Supplementary Fig. 5A). In contrast, other factors well known to induce formation of MGCs in vitro by fusion of bone marrow–derived macrophages, such as Rankl, had no significant effect on MGC generation in AT (Fig. 6A). Additionally, quantification of the cell size of ATMs did verify a general increase in mean ATM size after GM-CSF and IL-4 stimulation (Fig. 6B), as seen for ATMs after 12 and 24 weeks of an HFD in vivo. Furthermore, we studied whether IL-4 directly stimulated MGC generation or functioned through other cell types in AT. Therefore, the effect of a myeloid cell–specific knockout of IL-4Rα, the mandatory signaling component of the IL-4 receptor, was tested ex vivo. In control mice (IL-4Rα^−/fl), the accumulation of MGCs as a result of IL-4 stimulation was unaffected, whereas in myeloid-specific knockout mice (IL-4Rα^∆myel), IL-4 was unable to generate MGCs (Fig. 6D). Next, we aimed at testing whether IL-4Rα signaling was also a prerequisite for MGCs in obese animals in vivo. Surprisingly, IL-4Rα^∆myel did exhibit a trend toward more MGCs within the AT of obese mice after 24 weeks of an HFD, which could have been related to an altered immune phenotype, body composition, or CLS density in these animals (data not shown). Therefore, we used our established model of MGC induction produced by long-term cultivation of lean AT explants to ensure similar conditions (Fig. 3A–D). However, the number of MGCs was not affected by knockout of IL-4Rα after 10 days of long-term cultivation, suggesting that although IL-4 can be used to increase MGCs in AT, other signals dominate physiological MGC formation in vivo.

The link between obesity and associated diseases is AT dysfunction caused by chronic low-grade inflammation (1). Accordingly, AT inflammation can be observed before obesity-associated diseases manifest, and immune cell infiltration is the strongest predictor of insulin resistance in obese patients (20). AT inflammation is characterized by an increase of immune cells, mostly resulting from an increase in ATMs, an increasing number of CLS, and a shift in ATM polarization toward a more proinflammatory (M1) phenotype (5,21). Mechanistically, this proinflammatory phenotype of ATMs is closely linked to adipocyte death and the occurrence of CLS (22). Of note, up to 90% of ATMs are found in CLS in AT of obese mice (4,23). Although susceptibility of AT depots to obesity-induced adipocyte death is different, adipocyte death is increased by 10- to 100-fold as a result of obesity (24). However, the clearing of dead adipocytes seems challenging for resident ATMs, because hypertrophic adipocytes are fivefold larger than regular phagocytes (25). This seems to result in extraordinarily long clearing periods of adipocyte remnants. For instance, after global induction of adipocyte death, the formation of CLSs occurs several days after induced cell death (26). In contrast, apoptotic cells in other tissues are cleaved in <24 h (27–29). More importantly, after switching back from a CLS-inducing HFD to a chow diet in mice, increased CLS density was detectable for up to 6 months, indicating that adipocyte degradation takes several months (30). This implies that local ATM proliferation and activation toward a M1 phenotype continue for this prolonged period. In line with these findings, several studies have shown that induction of AT inflammation as a result of HFD feeding is not (or only partially) reversible after switching back to regular chow (30,31).

In general, the ineffective clearing of dead cells increases the release of proinflammatory cytokines (32–34), leading to chronic inflammation and autoimmune diseases (33). This failure may also involve the dysregulation of receptors mediating contact between MGCs/macrophages and dying adipocytes or lipid remnants. This could include BAI-1 or TIM-4, which are generally involved in efferocytosis (35), or TREM-2, which was recently shown to mediate macrophage-adipocyte interaction (36,37).

Importantly, one way that ATMs digest large cells, such as adipocytes, was recently described and includes the extracellular degradation of lipid remnants as a result of lysosomal exocytosis. Here, CLS-associated ATMs secrete lysosomal enzymes to the adipocyte-macrophage interface and subsequently take up liberated lipids (18,25). In this report, we describe a potential second alternative of the immune system to improve large-cell efferocytosis through the generation of MGCs, possibly as a result of macrophage fusion. Direct macrophage fusion was recently confirmed to be a mechanism for the maintenance of osteoclasts, which are physiologically occurring MGCs of the bone (38). Of note, fusion of other myeloid cell types, such as dendritic cells, cannot be excluded, because of the lack of exclusive markers as discussed by others (39). However, endoreplication (nucleus replication without cell division) as an alternative origin of MGCs may also play a role in AT (40). IL-4 and GM-CSF have convincingly been demonstrated to induce macrophage fusion in vitro (8,41), and IL-4 signaling (IL-13–, IL-4Rα–, and IL-4–stimulated STAT6 phosphorylation) is also increased in AT from obese mice as shown previously (13). Although our data on obese IL-4Rα knockout mice suggest that signaling other than IL-4Rα signaling dominates the formation of MGCs in vivo, the increase in the number of MGCs in AT via IL-4 stimulation seems possible and may have therapeutic potential.

Importantly, we demonstrated that MGCs isolated from the AT of obese mice had a higher phagocytic capacity ex vivo, because only these cells were able to take up large particles, such as 25-µm lipid remnants or 45-µm polystyrene beads, which are otherwise indigestible for regular, mononucleated ATMs. Of note, the specialization of in vitro–generated, bone barrow–derived MGCs for large particle uptake has been previously demonstrated (19). To estimate the phagocytic activity in adipocyte degradation in vivo, we measured the lipid content of MGCs, which was threefold higher than that in mononucleated ATMs of the respective cell size, indicating a higher density of lipid droplets as a result of increased lipid uptake. AT foam cells (lipid laden macrophages) are associated with an unhealthy obese phenotype and AT inflammation (42–44), similar to MGCs, but they form earlier in obesity (45). However, the influence of polyploidy on the lipid content of ATMs has not been studied before. Of note, binucleated ATMs exhibit an intermediate phenotype between mono- and multinucleated ATMs, and some of them may represent precursors of MGCs. However, uptake of liberated fatty acids after extracellular fat mobilization resulting from lysosomal exocytosis could also explain the higher lipid content of mononucleated ATMs and lead to a metabolically activated immune phenotype (25). Either way, our data show for the first time that MGCs found in CLSs are specialized in adipocyte degradation. Our findings also show that provoked adipocyte death induces MGC formation in vitro.

Whether the formation of MGCs is protective of or detrimental to AT inflammation requires further study. Of note, patients with detectable MGCs within AT sections had significantly elevated blood leukocytes as an indicator of a more proinflammatory AT phenotype. However, we also showed that ∼50% of MGCs also expressed anti-inflammatory marker proteins, such as CD206. This could have been due to endogenous de novo expression of M2 markers by MGCs or due to fusion of proinflammatory (M1) with anti-inflammatory (M2) macrophages. Thus, a reprogramming of proinflammatory macrophages toward a tissue-remodeling M2 phenotype in MGCs would fit into a model of resolution of inflammation after tissue clearing by MGCs. Although this is a tempting speculation, more data on the function and extensive transcriptional profiling of AT-derived MGCs are needed to prove this concept.

In summary, for the first time, we characterized naturally occurring MGCs in AT of obese mice in detail. We showed that these unique cells represented a significant number of ATMs in obesity, making up to 3% of ATMs or ∼10% ATM-associated nuclei. Of note, because the criteria for detecting MGCs seemed to fit with the exclusion criteria from other laboratories, this cell population may be underrepresented in other studies. However, our data indicate that adipocyte death is a trigger for MGC formation. Most importantly, MGCs seem to be specialized for large cell destruction, such as adipocytes, and express pro- and anti-inflammatory marker proteins. However, more data are needed to unravel the function of these unique cells in the course of AT inflammation as well as in the resolution of obesity-associated tissue damage in AT.

Acknowledgments. The authors are grateful for the technical assistance of Christine Fröhlich and Michaela Kirstein (Institute of Anatomy and Cell Biology, Martin-Luther-University Halle-Wittenberg). The authors also thank Kathrin Jäger and Andreas Lösche from the FACS core unit, University of Leipzig, as well as Alexander Navarrete-Santos from the Flow Cytometry core unit, Martin-Luther-University Halle-Wittenberg, and Claudia Müller from the Fraunhofer Institute Leipzig for providing flow imaging cytometers. Furthermore, the authors thank Frank Brombacher from the Health Sciences Faculty, University of Cape Town, Cape Town, South Africa, and David Hume from the Roslin Institute and Royal School of Veterinary Studies, Roslin, U.K., for kindly providing IL-4Rα^−/−, LysMCre^+/− × IL-4Rα^flx/− (IL-4Rα^Δmyel), and MacGreen (CSF1R-eGFP^+/−) mice, respectively.

Funding. This work was funded by Deutsche Forschungsgemeinschaft project number 209933838–SFB 1052 (project B09 to M.G. and I.B.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.B. designed, performed, and analyzed gene expression, flow cytometry, imaging flow cytometry, and microscopic imaging data and wrote the manuscript. A.L. performed the beat assay. J.F. performed and analyzed murine explants experiments. C.H. performed murine cell culture experiments. P.K. and M.B. analyzed human data. J.E. and I.B. performed live imaging and edited the manuscript. M.G. designed the study and wrote the paper. M.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.