RELMα-expressing macrophages protect against fatal lung damage and reduce parasite burden during helminth infection

See allHide authors and affiliations

Science Immunology  24 May 2019:
Vol. 4, Issue 35, eaau3814
DOI: 10.1126/sciimmunol.aau3814

In the RELMα realm

Alternatively activated macrophages (AAMs) are critical to many different immune responses. Krljanac et al. developed a tool to track and characterize AAMs based on expression of the immunomodulatory protein resistin-like molecule (RELM) α. They generated RELMα reporter/deleter mice and observed that RELMα+ macrophages are enriched in white adipose tissue, gut, and peritoneum at steady state. Primary infection with the helminth Nippostrongylus brasiliensis induces expansion of RELMα+ lung interstitial macrophages but not alveolar macrophages in a STAT6-dependent manner. The presence of RELMα+ macrophages was required for protection from fatal primary infection and resistance against secondary infection. Together, these data define RELMα as a marker of AAMs and reveal its role in defense against helminth infection in the lung.


Alternatively activated macrophages (AAMs) can contribute to wound healing, regulation of glucose and fat metabolism, resolution of inflammation, and protective immunity against helminths. Their differentiation, tissue distribution, and effector functions are incompletely understood. Murine AAMs express high levels of resistin-like molecule (RELM) α, an effector protein with potent immunomodulatory functions. To visualize RELMα+ macrophages (MΦs) in vivo and evaluate their role in defense against helminths, we generated RELMα reporter/deleter mice. Infection with the helminth Nippostrongylus brasiliensis induced expansion of RELMα+ lung interstitial but not alveolar MΦs in a STAT6-dependent manner. RELMα+ MΦs were required for prevention of fatal lung damage during primary infection. Furthermore, protective immunity was lost upon specific deletion of RELMα+ MΦs during secondary infection. Thus, RELMα reporter/deleter mice reveal compartmentalization of AAMs in different tissues and demonstrate their critical role in resolution of severe lung inflammation and protection against migrating helminths.


Macrophages (MΦs) adapt their effector functions in response to local tissue environments and recognition of pathogens, which results in a broad spectrum of polarization states. Classically activated MΦs (CAMs) are prevalent during infections involving intracellular pathogens, components of bacterial cell walls, or pro-inflammatory T helper 1 (TH1) cell products like interferon-γ (IFN-γ) or tumor necrosis factor–α. Alternatively activated MΦs (AAMs) are enriched in tissues containing cytokines like interleukin-4 (IL-4), IL-13, IL-10, tumor growth factor–β, products of fungi, apoptotic cells, or immune complexes (1). Important inducers of AAMs are also parasitic worms such as the gastrointestinal helminth Nippostrongylus brasiliensis. Infective third-stage (L3) larvae of N. brasiliensis enter the host through the skin and migrate to the lungs where they produce proteolytic enzymes to break into alveoli (2). Here, AAMs are an important source of matrix metalloproteases (MMPs) and inhibitors of MMPs that control extracellular matrix turnover, thereby mediating tissue regeneration and repair at the site of helminth-induced injury. From alveoli, parasite larvae crawl up the trachea, are coughed up and swallowed, and finally reach the small intestine by day 3 post infection (p.i.) (3). After primary infection of mice, AAMs block helminth-associated TH1-type responses that may cause pathogenesis. Polarized TH2-type responses are then able to expel parasites from the intestine by day 10 p.i. (2); however, inflammation in the lung continues to develop (4, 5). These studies point to AAMs as vital contributors to the anti-helminth response during the parasite’s life cycle. However, detection of AAMs and understanding their function are difficult owing to scarcity of available models. Yellow fluorescent protein–arginase 1 (YARG) reporter mice were used in the past (6); however, expression of arginase 1 is characteristic not only for AAMs but also for CAMs (7, 8), thus limiting its use to reliably identify AAMs in the presence of both TH1- and TH2-derived stimuli. Apart from arginase 1, another highly expressed product of AAMs is resistin-like molecule (RELM) α [also known as Fizz1 (found in inflammatory zone 1) or as HIMF (hypoxia-induced mitogenic factor)]. RELMα is encoded by the Retnla gene and was originally discovered by Holcomb et al. as a cysteine-rich secreted protein abundant in the bronchoalveolar lavage (BAL) fluid during ovalbumin-induced pulmonary inflammation (9). Studies of Retnla−/− mice gave insights into the importance of RELMα protein for protection against helminth-induced lung injury. Retnla−/− mice infected with N. brasiliensis (10), or after challenge with Schistosoma mansoni eggs (11), had intensified lung inflammation and pathology, accompanied by exacerbated TH2 adaptive immune response. During the transient lung stage of N. brasiliensis infection around days 2 and 3 p.i., RELMα protein was shown to protect mice from exacerbated infection-induced weight loss and, to some extent, from infection-induced mortality of female mice (12). In bleomycin-induced lung injury, RELMα promotes pulmonary fibrosis by inducing differentiation of myofibroblasts and exhibits chemoattractant activity for bone marrow (BM)–derived dendritic cells (DCs) (13). However, it is necessary to differentiate the effects of RELMα protein per se and RELMα-producing AAMs, which may also perform other critical functions during helminth infection, such as larval killing and repair of helminth-induced tissue damage.

Although TH2 cytokines acting via the signal transducer and activator of transcription 6 (STAT6) pathway strongly induce expression of RELMα in MΦs during nematode infection (7, 14, 15), experimentally induced hypoxia can also stimulate RELMα expression in airway epithelial cells independently of IL-4 and STAT6 (16). However, it is less clear whether RELMα expression in MΦs in naïve mice is also controlled by an interplay of environmental factors, including the TH2 cytokine milieu and/or endogenous hypoxia in different tissues. Furthermore, in recall responses against N. brasiliensis helminths, AAMs and arginases play a role in blocking the egress of parasite larvae from the site of infection in the skin (17). AAMs can also impair parasite survival in the lungs (18), thus blocking parasite migration to the intestine. However, it is necessary to evaluate how bona fide RELMα-expressing MΦs contribute to the AAM-mediated anti-helminth response during primary and secondary N. brasiliensis infection. Also, little is known about the tissue distribution, development, and function of RELMα-producing MΦs at steady state. To gain more insight into these aspects of AAM-cell biology, we generated a RetnlaCre mouse model, which enables detection of AAMs via RetnlaCre-mediated expression of fluorescent reporters or AAM depletion via expression of cytotoxic mediators. We show that, at steady state, white adipose tissue (WAT), peritoneum, and gut are highly enriched with RELMα_tdTomato+-expressing MΦs, indicating their current or historic expression of the Retnla gene. During primary N. brasiliensis infection, RELMα-expressing lung interstitial but not alveolar MΦs (Alv-MΦs) expanded in a STAT6-dependent manner. Using a specific AAM-deleter mouse model, we found that RELMα-expressing MΦs were essential to prevent fatal lung damage during primary N. brasiliensis infection and to provide protective immunity against migrating parasites during secondary challenge.


Tissue distribution of RELMα-producing MΦs under steady-state conditions

RELMα is one of the most strongly induced proteins in AAMs in mice (14). To develop a mouse model for detection and deletion of RELMα-expressing MΦs, we generated bacterial artificial chromosome (BAC)–transgenic mice expressing Cre recombinase under the control of the promoter for the Retnla gene encoding RELMα protein (figs. S1 and S2). The Cre cassette was inserted at the start codon of the Retnla gene on the BAC vector, thereby destroying expression of the Retnla gene from this construct without affecting endogenous Retnla gene expression (fig. S2). RetnlaCre founder lines were then crossed to Rosa26tdTomato mice containing a Cre-inducible tdTomato fluorescent reporter in the ubiquitously expressed Rosa26 locus to generate RetnlaCre_R26tdTomato fate mapping mice (19). RELMα_tdTomato expression correlated with the RELMα mRNA expression in peritoneal MΦs (pMΦs) and blood neutrophils (fig. S3A), as well as with intracellular RELMα protein expression in pMΦs, lung interstitial MΦs (Int-MΦs), and DCs (fig. S3, B and C).

Next, we stimulated BM-derived MΦs (BMDMs) from five independent RetnlaCre founder lines with IL-4 and found that lines 7 and 13 exhibited similar RELMα_tdTomato to endogenous RELMα expression kinetics (fig. S3, D and E). As expected, the RetnlaCre BAC construct was specifically induced by IL-4 in a STAT6-dependent manner. IFN-γ and lipopolysaccharide (LPS), which are stimuli that induce CAMs, could not induce the RetnlaCre construct, thus confirming its IL-4–inducible pattern of activation (fig. S3F).

To trace MΦs and other cell lineages expressing RELMα, we analyzed organs of naïve RetnlaCre_R26tdTomato mice by flow cytometry and immunofluorescence histology. MΦs residing in the perigonadal fat, the peritoneum (pMΦs), and, to a lesser extent, the lung interstitium (Int-MΦs) almost entirely consisted of RELMα_tdTomato+ cells (Figs. 1, B and C, and 2). However, only a fraction of RELMα_tdTomato+ pMΦs and lung Int-MΦs/DCs were found positive for the RELMα protein by intracellular staining (Fig. 4C and fig. S3, B and C). This phenomenon could be explained by reduced sensitivity to detect intracellular RELMα protein from ex vivo isolated and resting cell populations (this is well known for detection of cytokines in nonrestimulated TH cell subsets) and/or by historic expression of RELMα protein in these cells because the RELMα_tdTomato signal continues to be expressed after cessation of Retnla-driven Cre recombinase expression.

Fig. 1 Tracing of RELMα expression in MΦs using RetnlaCre_R26tdTomato mice (see also fig. S6).

(A) BMDMs were generated from RetnlaCre_R26tdTomato mice and stimulated with recombinant murine IL-4 (20 ng/ml) in vitro. RELMα_tdTomato indicates the RetnlaCre-induced expression of the tdTomato fluorescent reporter from the Rosa26 locus and was analyzed by flow cytometry at the indicated hours (h) after stimulation. (B) Flow cytometry dot plots show the frequencies of RELMα_tdTomato+ MΦs and other cell types that had expressed RetnlaCre at some stage during their development. Parental gates are given in fig. S6. (C) Bar diagram shows the frequencies (mean + SD) of RELMα_tdTomato+ expressors among total cells of the respective type, isolated from at least five mice.

Fig. 2 Expression of RELMα_tdTomato in tissues of naïve RetnlaCre_R26tdTomato mice (see also fig. S4).

Indicated tissues isolated from naïve RetnlaCre_R26tdTomato mice were stained with BODIPY 493/503 dye for adipocytes, anti-CD68 for tissue MΦs, anti–SP-C for surfactant protein C-positive type 2 pneumocytes, anti–Pecam-1 for endothelial cells of blood vessels, and 4′,6-diamidino-2-phenylindole for nuclei. RELMα_tdTomato indicates endogenous RetnlaCre-induced expression of the tdTomato fluorescent reporter from the Rosa26 locus. Images on the far right show overlay of the respective channels. Scale bars, 50 μm.

MΦs in lamina propria of the small intestine contained ~22% RELMα_tdTomato+ cells, and RELMα_tdTomato+ MΦs were also located largely at the base of colonic crypts (Figs. 1, B and C, and 2). However, small and large intestines also contained other non-MΦ RELMα_tdTomato+ cell types (Fig. 2). Splenic DCs and MΦs in the thymus, spleen, and liver contained only ~2% of RELMα_tdTomato+ cells (Fig. 1, B and C, and figs. S4A and S5, A and B), indicating that these organs are essentially void of RELMα-rich MΦ niches. Similarly, total brain MΦs also contained only a fraction of ~1% of RELMα_tdTomato+ cells, residing both in the parenchyma and in meninges (Fig. 1B and fig. S4A), whereas subgating for MHC-IIhigh brain MΦs revealed that 55% of these cells expressed the RELMα_tdTomato reporter (fig. S6). Kidney, stomach, and epididymis also contained RELMα-expressing cells (fig. S5A). In the lung, most of the RELMα_tdTomato-expressing cells were alveolar epithelial type 2 (AE2) cells residing in the lung parenchyma and airways (Figs. 1B, 2, and 3A and fig. S4B). In contrast, only ~3% of all Alv-MΦs expressed RELMα_tdTomato under steady-state conditions (Fig. 1B). Expression of the endogenous RELMα protein largely correlated with the RELMα_tdTomato signal in AE2 cells located within airways, as shown by immunofluorescence (fig. S4B). Eosinophil granulocytes were also prominent RELMα expressors with 76, 17, and 13% of RELMα_tdTomato+ eosinophils found in the peritoneum, lamina propria of small intestine, and blood, respectively (Fig. 1B). Also, ~12% of blood neutrophils expressed RELMα_tdTomato, but no RELMα_tdTomato+ cells were detected among basophil granulocytes, monocytes, and T and B lymphocytes in the blood (summarized in Fig. 1C). These data show that cells exhibiting substantial RELMα expression under steady-state conditions include MΦs enriched in the peritoneum, gut, and WAT. In addition, RELMα expression was also evident in AE2 cells, as well as in subsets of eosinophils and neutrophils.

Fig. 3 RELMα expression in alveolar versus peritoneal MΦs.

(A) Immunofluorescence of fixed lung tissue (top) from RetnlaCre_R26tdTomato mice shows strong endogenous RELMα_tdTomato signal in the airways and lung parenchyma (red), but not in lung MΦs stained with anti-CD11c (green) and anti-CD68 (violet), as seen on the overlay image of the respective channels on the far right (merge). Bottom shows immunofluorescence of unfixed BAL cells, revealing bright RELMα_tdTomato+ Alv-MΦs stained with anti-CD11c (green) and anti–Siglec-F (violet). Scale bars, 50 μm. (B) The upper dot plots show gating for Alv-MΦs in Ly5.1 B6 → Ly5.2 RetnlaCre_R26tdTomato BM chimeras. Similar gating was applied for Ly5.2 RetnlaCre_R26tdTomato → Ly5.1 B6 chimeras. The lower dot plots show percentages of RELMα_tdTomato+ Alv-MΦs gated as above and isolated from the following mice (from left to right): RetnlaCre_R26tdTomato mouse lacking the RetnlaCre transgene as a negative control (R26tdTom), RetnlaCre_R26tdTomato mouse carrying the RetnlaCre transgene as a positive control, Ly5.1 B6 → Ly5.2 RetnlaCre_R26tdTomato chimeras, and Ly5.2 RetnlaCre_R26tdTomato →Ly5.1 B6 chimeras. Results are representative of three to five individual mice per group (see also fig. S7). (C) RELMα_tdTomato Alv-MΦs were sorted from BAL of RetnlaCre_R26tdTomato mice, labeled with CellTrace, and cultured for 48 hours in vitro with or without IL-4. Numbers indicate the frequency of RELMα_tdTomato+ cells among total Alv-MΦs. The lower histograms show dilution of the CellTrace proliferation dye after 48 hours of culture in the presence or absence of IL-4. (D) Relative mRNA expression of the indicated genes normalized to Hprt mRNA in sorted RELMα_tdTomato+ (black bars) and RELMα_tdTomato (white bars) MΦs from the peritoneum (pMΦs, top) or BAL fluid (Alv-MΦs, bottom). Data are obtained from >10 mice for isolating pMΦs and >30 mice for isolating Alv-MΦs from BAL of naïve RetnlaCre_R26tdTomato mice.

RELMα expression in alveolar versus peritoneal MΦs

Immunofluorescence of BAL cells from RetnlaCre_R26tdTomato mice confirmed that only few Alv-MΦs had high levels of RELMα_tdTomato expression (Fig. 3A). Flow cytometry revealed that an additional 10% of Alv-MΦs expressed intermediate tdTomato fluorescence, 10 to 100 times higher than autofluorescence (phycoerythrin channel) of Alv-MΦs isolated from control mice not expressing the transgene (Fig. 3B and fig. S7A). After transferring Ly5.1+ B6 BM into irradiated Ly5.2+ RetnlaCre_R26tdTomato recipients, 99% of Alv-MΦs were of donor Ly5.1+ B6 origin, whereas the RELMα_tdTomatohigh Alv-MΦs were lost, indicating that these cells are genuine RELMα expressors (Fig. 3B and fig. S7B). Donor-derived Ly5.1 B6 Alv-MΦs still formed the RELMα_tdTomatolow population, which probably acquired their fluorescence from radio-resistant AE2 cells, the only major tdTomato-expressing population in the lung of these chimeras. Also, in chimeras having wild-type (WT) nonfluorescent pneumocytes, the RELMα_tdTomatolow population was absent, and donor-derived Alv-MΦs of RetnlaCre_R26tdTomato origin gave rise to bona fide RELMα_tdTomatohigh expressors (Fig. 3B and fig. S7C). These data indicate that, at steady state, roughly 10% of Alv-MΦs may directly or indirectly acquire cytosolic proteins (e.g., tdTomato) from adjacent AE2 pneumocytes in lung alveoli. The prolonged life span of Alv-MΦs (20) in vivo may enable cytosolic accumulation of the fluorescent reporter, thus explaining the high RELMα_tdTomato signal in these cells.

To understand the kinetics of RELMα expression in Alv-MΦs, we sorted BAL-derived tdTomato Alv-MΦs and stimulated them with IL-4 in vitro. Alv-MΦs spontaneously proliferated in vitro, but in contrast to BMDMs (Fig. 1A), we detected no significant induction of RELMα expression with or without IL-4 (Fig. 3C). RELMα_tdTomato+ Alv-MΦs sorted and pooled from more than 30 mice did not show a bona fide profile of AAMs, with only a 2.5-fold increase of Retnla or Arg1, and no significant up-regulation of Mrc1, Chil3, or Pparg mRNA compared to RELMα_tdTomato cells (Fig. 3D). Conversely, pMΦs expressing RELMα_tdTomato had a robust AAM profile with 40- to 45-fold higher levels of Retnla and Mrc1 mRNA, two- to threefold higher levels of Pparg and Chil3, but twofold lower Arg1 levels compared to RELMα_tdTomato pMΦs. Direct comparison of RELMα_tdTomato+ populations revealed that pMΦs contained 1000- and 4000-fold higher levels of Arg1 and Retnla mRNA compared to Alv-MΦs, respectively, whereas Mrc1, Pparg, and Chil3 were 5, 100, and 1000 times higher in Alv-MΦs compared to pMΦs, respectively. These data show that, under steady-state conditions in vivo, a minor subset of Alv-MΦs produces RELMα at a low but consistent rate, whereas pMΦs show robust RELMα production.

Requirement of STAT6 for expression of RELMα in MΦs, eosinophils, and neutrophils

The transcription factor STAT6 acting downstream of the IL-4 receptor appears to be required for RELMα expression in BMDMs (fig. S3E); thus, we investigated whether this requirement can be observed in vivo. For this, we examined organs of RetnlaCre_R26tdTomato mice lacking STAT6 (STAT6ko) or lacking IL-4 and IL-13 cytokine production (4-13ko). Unexpectedly, STAT6 and IL-4/IL-13 were not essential for RELMα expression in blood neutrophils, as well as in pMΦs (Fig. 4, A and B). However, STAT6- or IL-4/IL-13–deficient RetnlaCre_R26tdTomato mice had two- and fourfold lower frequencies of RELMα_tdTomato+ blood and peritoneal eosinophils, respectively, and a marked loss of RELMα_tdTomato+ Alv-MΦs (Fig. 4, A and B). At steady state, both WT and STAT6ko pMΦs expressed RELMα protein (Fig. 4C) and STAT6 deficiency did not significantly alter the levels of RELMα in serum, peritoneum, or BAL fluid (Fig. 4D).

Fig. 4 STAT6 regulates expression of RELMα in a cell-specific manner.

(A) Dot plots show percentages of RELMα_tdTomato+ cells among neutrophils (Ly6G+, Ly6C+), blood eosinophils (Siglec-F+, SSC-Hhi), pMΦs (CD11b+, F4/80+), and BAL Alv-MΦs (Siglec-F+, CD11c+) isolated from naïve RetnlaCre_R26tdTomato mice generated on a C57BL/6 (WT), STAT6ko, or IL-4/IL-13ko (4-13ko) background. (B) Bar diagram summarizes data in (A). Bars show the frequencies (mean + SD) of five individual mice per group and are representative of three independent experiments. **P < 0.01 by two-tailed Student’s t test. (C) Histograms show percentages of intracellular RELMα protein-expressing pMΦs [pregated as in (A) among cells positive for the RELMα_tdTomato reporter], isolated from WT and STAT6ko RetnlaCre_R26tdTomato mice. Data are representative of at least three to four mice per group. (D) Bar diagrams show concentrations of RELMα protein in serum, peritoneal, and BAL fluids harvested from WT and STAT6ko RetnlaCre_R26tdTomato mice and analyzed by enzyme-linked immunosorbent assay (ELISA). Data are pooled from at least three to four mice per group. (E) RELMα_tdTomato monocytes were sorted from BM of Ly5.2 RetnlaCre_R26tdTomato mice, labeled with CellTrace, and injected intraperitoneally into Ly5.1 B6 recipients (WT → WT), or from STAT6ko RetnlaCre_R26tdTomato mice into recipients lacking STAT6 (STAT6ko → STAT6ko), and from IL-4/IL-13ko RetnlaCre_R26tdTomato donors into recipients lacking STAT6 (4-13ko → STAT6ko). The upper dot plots show gating strategy for transferred cells, and the lower dot plots indicate percentages of RELMα_tdTomato+ monocyte-derived pMΦs harvested from the peritoneal cavity of recipient mice at day 6 after transfer. Bar diagram summarizes the cell frequencies (mean + SD) from four to five mice per group (pooled from two independent experiments) and points indicate individual mice (see also fig. S8B). *P < 0.05; **P < 0.01 by two-tailed Student’s t test.

Although not essential, STAT6 could still be important for promoting rapid RELMα expression in pMΦs. To address this, we studied the conversion of RELMα_tdTomato monocytes (Fig. 1B) into RELMα_tdTomato+ pMΦs in vivo. Monocytes from RetnlaCre_R26tdTomato mice (fig. S8A) adoptively transferred into WT RetnlaCre recipients developed into CD11b+ F4/80+ pMΦs, and ~70% of them expressed RELMα_tdTomato at day 6 after transfer (Fig. 4E and fig. S8B). In contrast, monocytes from STAT6-deficient RetnlaCre_R26tdTomato donor mice also differentiated into RELMα+ pMΦs but with a two- to threefold lower rate of conversion. Conversion of IL-4/IL-13–deficient but STAT6-sufficient monocytes was also reduced in peritoneum containing STAT6-deficient cells, demonstrating that STAT6-dependent mechanisms in the recipient mice were important to induce RELMα in transferred monocytes. These experiments indicate that, in contrast to eosinophils and Alv-MΦs, neutrophils and pMΦs express RELMα in a STAT6-independent manner; however, STAT6 does enhance conversion of monocytes into RELMα+ pMΦs.

STAT6-dependent accumulation of RELMα+ Int-MΦs and eosinophils in lungs after N. brasiliensis infection

After investigating its role in MΦs at steady state, we next asked whether STAT6 is required for RELMα expression and relative proliferation of tissue-resident Alv-MΦs versus lung Int-MΦs during N. brasiliensis infection. For this, we infected RetnlaCre_R26tdTomato mice (WT) and STAT6-deficient RetnlaCre_R26tdTomato mice (STAT6ko) with N. brasiliensis and analyzed lungs on days 11 and 20 p.i. by flow cytometry. For clarity, Alv-MΦs were gated as CD68+ Siglec-F+ and Int-MΦs as CD68+ Siglec-F cells (Fig. 5A, day 20 p.i.). MΦs were also subgated to confirm their identity as MHC-II+ CD11b+ CD11c+ (Int-MΦs) or MHC-IIdim/− CD11c+ (Alv-MΦs), and eosinophils were identified as Siglec-F+ CD68 MHC-II cells (Fig. 5A and fig. S9A). The frequencies of total lung Int-MΦs were significantly increased at 11 and 20 days after N. brasiliensis infection in both WT and STAT6ko mice as compared to uninfected control mice (Fig. 5D, left bar diagrams). In mice infected with N. brasiliensis versus naïve controls, WT RELMα_tdTomato-expressing MΦs were detected as aggregates around lung airways together with an enlarged pool of IL-4–producing cells as shown by immunofluorescence (fig. S9B).

Fig. 5 RELMα_tdTomato+ Int-MΦs increase whereas Alv-MΦs decline after N. brasiliensis infection.

(A) WT and STAT6ko RetnlaCre_R26tdTomato mice were infected with N. brasiliensis, and lungs were analyzed at 11 and 20 days p.i. Dot plots show simplified gating of total CD68+ Siglec-F Int-MΦs, CD68+ Siglec-F+ Alv-MΦs (gate R1), and CD68 Siglec-F+ lung eosinophils (gate R2). (B and C) Dot plots show percentages of RELMα_tdTomato+ cells among Int-MΦs and Alv-MΦs (B) or eosinophils (C) in WT versus STAT6ko RetnlaCre_R26tdTomato mice at day 20 p.i. (D and E) Bar diagrams indicate percentages (mean + SD) from total cells (D) or percentages of RELMα_tdTomato-expressing cells (E) among the indicated populations of WT and STAT6ko mice at day 11 (top) and day 20 p.i. (bottom). Note the requirement of STAT6 for induction of the RELMα_tdTomato+ Int-MΦs and eosinophils after infection (E). Data are representative of four independent experiments with at least three to four mice per group. *P < 0.05; **P < 0.01 by two-tailed Student’s t test. ns, not significant. (F) RELMα_tdTomato monocytes of Ly5.2 RetnlaCre_R26tdTomato mice were transferred intravenously into congenic Ly5.1 B6 mice, naïve or infected with N. brasiliensis 3 days before. Lungs of recipient mice were analyzed on day 37 post transfer (p.t.). Representative dot plots of two independent experiments show gating strategy of the donor-derived Ly5.2+ RELMα_tdTomato+ cells, which differentiated into MHC-II+ Siglec-F interstitial and Siglec-F+ MHC-II Alv-MΦs. Bar diagram summarizes the cell frequencies (mean + SD) from four mice per group, and points indicate individual mice (see also fig. S8C). *P < 0.05 by two-tailed Student’s t test.

The absence of STAT6 abolished the recruitment of eosinophils to the lung at the early stage, yet these cells accumulated at later stages of infection (Fig. 5D, right bar diagrams). Induction of the RELMα_tdTomato-expressing Int-MΦs and eosinophils was abolished in lungs of mice lacking STAT6 (Fig. 5, B, C, and E). We sorted WT RELMα_tdTomato+ Int-MΦsat day 8 p.i. and confirmed their AAM-polarized state of differentiation (fig. S9C). In contrast to Int-MΦs, the relative proportion of Alv-MΦs declined in the lungs of WT mice after N. brasiliensis infection. In mice lacking STAT6, the population of Alv-MΦs was stable at day 11; however, it also declined at the later stage of infection (Fig. 5D, middle bar diagrams).

Next, we investigated whether infiltrating monocytes contribute to the accumulation of RELMα_tdTomato+ Int-MΦs in the lung after N. brasiliensis infection. For this, we adoptively transferred RELMα_tdTomato monocytes (Fig. 1B and fig. S8A) from RetnlaCre_R26tdTomato Ly5.2 mice into congenic Ly5.1 mice infected or not with N. brasiliensis. The infection provided an inflammatory niche in the lung where transferred monocytes were still detected at 37 days p.i. Subgating confirmed that two-thirds of the transferred cells had differentiated into RELMα_tdTomato+ MΦs of the CD68+ MHC-II+ Siglec-F interstitial phenotype, and one-third of them acquired the CD68+ Siglec-F+ MHC-II phenotype of Alv-MΦs (Fig. 5F and fig. S8C). Together, we found that Alv-MΦs decline after N. brasiliensis infection, whereas infiltrating monocytes give rise to RELMα-expressing Int-MΦs, which increase in numbers in the lung along with RELMα_tdTomato+ eosinophils in a STAT6-dependent manner.

Control of parasite burden in secondary infection and protect from fatal primary N. brasiliensis infection by RELMα-expressing MΦs

Because RELMα dampens inflammation and promotes healing (10), the absence of RELMα-expressing MΦs may compromise the resolution of inflammation. To test this hypothesis, we used R26iDTR mice that contain a Cre-inducible allele of diphtheria toxin receptor (DTR) in the ubiquitously expressed Rosa26 locus (19), and after crossing them to RetnlaCre_R26tdTomato mice, we were able to efficiently deplete RELMα-expressing cells. After short-term exposure to diphtheria toxin (DT), we observed 93% depletion efficiency of RELMα_tdTomato-expressing cells among pMΦs and, to a lesser extent, among skin and lung Int-MΦs (Fig. 6, A and B, and fig. S10, A and B). Depletion of RELMα_tdTomato-expressing peritoneal eosinophils was also less efficient (about 40%), whereas no significant decrease was observed for RELMα_tdTomato-expressing lung AE2 cells and neutrophils in blood (Fig. 6, A and B, and fig. S10C), probably because of rapid de novo production of these cell types.

Fig. 6 Requirement of RELMα-expressing cells for prevention of fatal primary and protection during secondary N. brasiliensis infection.

(A and B) To test the efficiency of DT-mediated depletion of RELMα+ cells, naïve RetnlaCre_R26iDTR mice were either left untreated or injected intraperitoneally with DT at day 0 and day 1, and blood, peritoneal exudates, skin, and lung samples were harvested on day 2 p.i. Representative dot plots (A and fig. S10B) and bar diagrams (B) show percentages + SD of RELMα_tdTomato+ cells among the indicated cell populations from three to four untreated (gray bars) and DT-treated mice (black bars). Data are representative of two independent experiments. *P < 0.05; **P < 0.01 by two-tailed Student’s t test. (C) RetnlaCre+_R26iDTR mice were treated with DT daily from day 0 until day 3 and concomitantly subcutaneously infected with N. brasiliensis (+DT +Nb, n = 7) or left uninfected (+DT, n = 6). Alterations to the well-being of the mice were monitored daily through day 11 after primary infection, and mice exhibiting clinical signs of debilitating phenotype (heavy or slow breathing, severely reduced movement or body weight, and low body temperature) were euthanized and removed from the study. **P = 0.0005 by log-rank test. (D) RetnlaCre+_R26iDTR mice were subcutaneously infected with N. brasiliensis, challenged for the second time with 500 larvae per mouse 30 days after primary infection, and daily treated (+DT) or not (−DT) with DT at days 0 until day 4 during secondary infection. Numbers of retained larvae in the skin at the site of infection (left plot) or numbers of larvae in the lung (middle plot) were analyzed at days 5 and 2 after secondary challenge, respectively. Plot on the right shows analysis of N. brasiliensis worm burden in the small intestine at day 5 after secondary infection. Lines indicate the means, and points indicate individual mice pooled from two independent experiments. *P < 0.05; **P < 0.01 by two-tailed Student’s t test. (E) Image on the right shows severe lung damage in RetnlaCre+_R26iDTR mice challenged for the second time with N. brasiliensis and treated with DT as in (D). DT injection in infected mice lacking the RetnlaCre transgene (left image), as well as secondary N. brasiliensis infection per se (middle image), caused no significant lung damage.

Given the importance of RELMα for resolution of tissue inflammation, depletion of RELMα-expressing cells may increase vulnerability of mice to lung-damaging larvae of N. brasiliensis during primary infection. N. brasiliensis infection of RetnlaCre_R26iDTR mice followed by daily injections of DT led to significantly increased mortality by 5 days p.i., whereas toxin treatment itself did not show effects on survival (Fig. 6C). Next, we asked whether depletion of RELMα+ cells affects protection from secondary infection with N. brasiliensis where migrating larvae are efficiently trapped in skin and lung so that only few worms mature to adults in the small intestine (17). We injected DT daily from day 0 to 4 after secondary infection and analyzed parasite burden in the skin, lung, and small intestine on day 5. Compared to controls, mice treated with DT had reduced numbers of N. brasiliensis larvae in the skin at the site of injection (Fig. 6D, left panel). Consequently, the larvae burden in the lungs at day 2 (Fig. 6D, middle panel), lung tissue damage (Fig. 6E), and the number of adult parasites in the small intestine at day 5 (Fig. 6D, right panel) were higher in mice concomitantly treated with N. brasiliensis and DT. These data demonstrate that RELMα-expressing cells are necessary to promote larval trapping in the skin and prevent the subsequent high burden of adult parasites during secondary infection.

To further dissect a specific role of RELMα-expressing MΦs in protection from N. brasiliensis, we used Csf1rLsL-DTR mice (hereafter named CD115iDTR) that Cre-inducibly express DTR-mCherry fusion protein in CD115+ cells, including MΦs and monocytes (21). Because we observed no RELMα_tdTomato expression in monocytes (Fig. 1B), RetnlaCre_CD115iDTR mice enabled us to specifically delete RELMα+ MΦs.

Our data with RetnlaCre_R26iDTR mice indicated that, in contrast to secondary infection, where more than 80% of injected larvae are trapped in the skin, this level of protection does not exist during primary infection, and RELMα-expressing cells help to prevent massive tissue damage in the lung between days 3 and 5 p.i. To avoid a fatal course of infection, RELMα-expressing MΦs might play an important role in lung tissue repair. To test this hypothesis, we treated RetnlaCre_CD115iDTR mice and Cre-negative control mice with DT on days 4, 5, and 7 after primary N. brasiliensis infection. DT treatment notably reduced survival of RetnlaCre_CD115iDTR mice compared to Cre-negative controls (Fig. 7A). These data indicate that RELMα+ MΦs are essential for resolution of inflammation in the lung and prevention of a fatal course of primary N. brasiliensis infection. Furthermore, after rechallenging RetnlaCre_CD115iDTR mice with N. brasiliensis as indicated for RetnlaCre_R26iDTR mice (Fig. 6D), DT treatment efficiently depleted lung CD11b+ CD11c+ Int-MΦs and led to massive lung tissue damage (Fig. 7, B and C). This was accompanied by reduced expression of characteristic AAM markers in the lung at day 5 after secondary infection, indicating a loss of AAMs (Fig. 7D). Infected and DT-treated RetnlaCre_CD115iDTR mice had higher parasite burden in the small intestine compared to infected and DT-treated Cre-negative controls and a tendency of higher larval load in the lungs, although no significant alteration of larval retention in the skin was observed (Fig. 7E). These results demonstrate that RELMα+ MΦs play a critical role in controlling protective immunity during secondary N. brasiliensis infection.

Fig. 7 Depletion of RELMα-expressing MΦs compromises survival during primary and control of lung tissue repair and intestinal parasite burden in the secondary N. brasiliensis infection.

(A) RetnlaCre_CD115iDTR mice (Cre+, n = 5) and Cre-negative controls (Cre, n = 8) were treated with DT on days 4, 5, and 7 after primary N. brasiliensis infection. Alterations to the well-being of the mice were monitored as in Fig. 6C. **P = 0.0026 by log-rank test. (B to E) Lungs of RetnlaCre_CD115iDTR mice were analyzed at day 5 after secondary N. brasiliensis infection and DT treatment as in Fig. 6D. (B) Dot plots and bar diagram indicate percentages of CD11b+ CD11c+ Int-MΦs from total lung cells in naïve, N. brasiliensis-infected RetnlaCre, and RetnlaCre+_CD115iDTR mice. Bars indicate means + SD of four individual mice per group and are representative of two independent experiments. (C) Image on the right displays aggravated lung damage in RetnlaCre+_CD115iDTR mice versus Cre-negative controls (left image), both challenged for the second time with N. brasiliensis and treated with DT. (D) Expression of genes that characterize AAMs in lungs of DT-treated mice (+DT) compared to untreated controls (−DT). (E) Parasite burden in the indicated organs of RetnlaCre+_CD115iDTR mice (Cre+) and Cre-negative controls (Cre), both treated with DT. Data are representative of two independent experiments. *P < 0.05; **P < 0.01 by two-tailed Student’s t test.


RELMα-expressing AAMs are found in IL-4/IL-13–rich inflammatory niches during tissue repair or helminth infection. Here, we developed a RetnlaCre fate mapping mouse model to show that the main sites of MΦ-derived RELMα production include adipose tissue, peritoneum, and gut, whereas few Alv-MΦs expressed RELMα in a STAT6- and IL-4/IL13–dependent manner. In the course of primary N. brasiliensis infection, STAT6 promoted the increase of lung Int-MΦs and eosinophils expressing RELMα. During secondary infection, specific removal of RELMα expressors impaired protection from N. brasiliensis. Furthermore, specific deletion of CD115+ RELMα+ MΦs severely impaired survival of mice during primary infection and led to an increased parasite burden during secondary N. brasiliensis infection.

Early studies have shown that signaling downstream of the IL-4 receptor was crucial for up-regulating RELMα production in MΦs (14). Correspondingly, we demonstrated that STAT6-dependent mechanisms promoted RELMα induction in monocytes upon their adoptive transfer into the peritoneal cavity. In contrast, we observed that naïve mice lacking cytokines IL-4 and IL-13 or STAT6 transcription factor exhibited similar frequencies of RELMα_tdTomato+ pMΦs compared to WT controls. Also, concentrations of RELMα protein in the peritoneum, lung, and blood of both naïve WT and STAT6ko mice were comparable. It has been reported that RELMα can also be up-regulated under hypoxic conditions as shown in the lung (22, 23), as well as by other MΦ-specific factors, including CCAAT box enhancer binding proteins (24) or regulator of G protein signaling (25). Furthermore, the chitinase-like protein Ym1 was recently shown to induce RELMα expression in the lung of IL-4Rα–deficient mice (26). On the other hand, absence of STAT6 or IL-4 and IL-13 cytokines strongly abolished generation of RELMα-expressing Alv-MΦs under basal conditions. This could be due to the specificities of lung environment that is, under naïve conditions, characterized by high oxygen tension (27) and thus may lack hypoxic and/or other STAT6-independent stimuli, which promote RELMα expression.

STAT6 was required for expansion of RELMα-expressing Int-MΦs in the lung of N. brasiliensis–infected mice. Furthermore, we demonstrated that RELMα-expressing Int-MΦs can originate from monocytes infiltrating the infected lung tissue. In parallel, we observed no increase of resident Alv-MΦs whose absolute numbers actually declined after infection, compared to naïve controls. In situ proliferation of resident MΦs was shown to be important for parasite sequestration or wound repair during filarial nematode infection (28). However, in the course of bleomycin-induced lung fibrosis, tissue-resident Alv-MΦs die from apoptosis (29), are depleted in the acute phase (30), and are replaced with monocyte-derived MΦs that eventually differentiate into resident Alv-MΦs (31). Similarly, depletion of resident Alv-MΦs followed by replacement with recruited mononuclear phagocytes were also observed during Yersinia pestis–, LPS-, or Streptococcus pneumoniae–induced lung infection (3234). We detected Int-MΦs as well as Alv-MΦs that originated from adoptively transferred monocytes as late as day 40 after N. brasiliensis infection, indicating that monocyte-derived MΦs remained in the tissue long after the initial inflammation had resolved.

We demonstrated that removal of RELMα-expressing cells impaired protection from secondary N. brasiliensis as evidenced by larger parasite burden in lungs and small intestine. Obata-Ninomiya et al. (17) indicated that blocking arginases, enriched not only in AAMs but also in other cell types, reduces trapping of N. brasiliensis larvae in the skin. Chen et al. (35) demonstrated a protective function of lung MΦ isolated from N. brasiliensis–infected mice and adoptively transferred into naïve mice. By specifically depleting CD115+ RELMα+ MΦs, we strongly abolished the protective population of lung Int-MΦs and demonstrated a subsequent increase in parasite burden in the small intestine after secondary infection. On the other hand, during primary infection, MΦs and eosinophils were implicated in contributing the wound healing effect of the type 2 immune response in the lung. Namely, depletion of CD11b+ cells up to day 3 after primary infection resulted in increased hemorrhaging and inflammation accompanied by decreased expression of factors important in wound healing (36). However, it was unclear whether bona fide AAMs alone mediate the protective wound healing response. Here, we demonstrated that removal of RELMα-expressing MΦs reduced survival of mice compared to controls. This observation is in line with a recent report demonstrating that tissue-damage-induced apoptotic cells in combination with IL-4/IL-13 in the lung elicit RELMα expression, which is required for the tissue repair response (37).

Overall, the RetnlaCre mice serve as a valuable tool to study the development and localization of AAMs and other RELMα-expressing cell types in vivo. However, one has to bear in mind that RetnlaCre_R26tdTomato fate mapping mice report not only current but also historic expression or Retnla gene in a particular cell and its offspring. The obvious limitation is that the tdTomato signal may not always correlate with current RELMα protein expression. On the other hand, it allows studying functional plasticity of AAMs in vivo. RetlnaCre mice further offer the opportunity to specifically delete conditional alleles in AAMs to investigate their role in homeostasis and effector function mediating the protection against helminths, tissue repair, and other processes including regulation of fat and glucose metabolism, tumor growth, and organogenesis.


Study design

This research was designed to investigate the differentiation, tissue distribution, and host-protective effector functions of RELMα-expressing MΦs in mice. Whenever possible, negative littermate mice were used as controls. For statistical analysis of most experiments, at least three to five samples per group were analyzed by two-tailed Student’s t test as indicated in the figure legends. The statistical analysis of the survival curves in Figs. 6C and 7A was analyzed by log-rank test.


RetnlaCre mice were generated as described in Supplementary Methods. STAT6-deficient mice (38), IL-4/IL-13–deficient mice (39), Rosa26iDTR mice (40), Csf1rLsL-DTR mice (21), and Rosa26tdTomato mice (41) were described previously. Mice were housed under specific pathogen–free conditions and used at 8 to 12 weeks of age. All experiments were performed in accordance with German Animal Protection Law and European Union guidelines 86/809 and were approved by the Federal Government of Lower Franconia.

N. brasiliensis infection and determination of worm burden

N. brasiliensis L3 were recovered from feces of infected rats, mixed with activated charcoal, and cultured in humidified chambers at room temperature. After extensive washing in sterile 0.9% saline (37°C), 500 larvae were injected into mice subcutaneously in 200 μl of saline. Mice were provided water containing antibiotics [neomycin sulfate (2 g/liter) and polymyxin B sulfate (100 mg/liter); Sigma-Aldrich] for the first 7 days p.i. For recall responses, mice were reinfected 4 weeks after the primary infection. Worm burden was determined by incubating pieces of small intestine cut longitudinally or 1- to 2-mm lung pieces for 2 hours at 37°C in RPMI 1640 and counting worms or larvae under a dissecting microscope. Worm burden in the skin was determined by measuring N. brasiliensis actin mRNA by quantitative reverse transcription polymerase chain reaction (RT-PCR), and the corresponding number of larvae was obtained from a standard curve displaying N. brasiliensis actin Ct values versus serial dilutions of larvae.

RELMα quantification by ELISA

Peritoneal or BAL fluids were harvested in 5 or 1 ml of Dulbecco’s minimum essential medium supplemented with 10% fetal calf serum, respectively. ELISA plates were coated overnight at 4°C with rabbit anti-murine (1 μg/ml) RELMα, and RELMα was detected in serum (1:50 diluted) or lavage fluids (undiluted) by biotinylated rabbit anti-murine RELMα (0.125 μg/ml) (all antibodies were obtained from PeproTech) and streptavidin-coupled alkaline phosphatase (1:3000 diluted, SouthernBiotech).

Flow cytometry

See Supplementary Methods.


See Supplementary Methods.

Quantitative RT-PCR

See Supplementary Methods.

Statistical analysis

Two-tailed Student’s t test and graphs were done using GraphPad Prism 4 (GraphPad Software, La Jolla, CA) and SigmaPlot 12.3 (Systat Software, San Jose, CA). P ≤ 0.05 was considered statistically significant (*P < 0.05; **P < 0.01). The survival curves in Figs. 6 and 7 were analyzed by log-rank test.


Raw data


Fig. S1. Generation of RetnlaCre mice.

Fig. S2. Structure and sequence information on the recombined BAC vector and its copy number in the genome.

Fig. S3. RELMα_tdTomato expression correlates with Retnla mRNA and RELMα protein expression and can be induced in vitro by IL-4.

Fig. S4. Immunofluorescence analysis of RELMα_tdTomato expression in selected organs of RetnlaCre_R26tdTomato mice.

Fig. S5. Ex vivo RELMα_tdTomato fluorescence imaging of selected organs of naïve RetnlaCre_R26tdTomato mice.

Fig. S6. Flow cytometry gating strategies for the indicated cell types isolated from naïve RetnlaCre_R26tdTomato mice.

Fig. S7. Analysis of RELMα_tdTomato-expressing Alv-MΦs in RetnlaCre_R26tdTomato mice.

Fig. S8. RELMα_tdTomato expression in MΦs derived from adoptively transferred monocytes.

Fig. S9. Analysis of MΦs in the lungs of RetnlaCre_R26tdTomato mice infected with N. brasiliensis.

Fig. S10. Depletion efficiency of RELMα_tdTomato+ cells after DT treatment of RetnlaCre_R26tdTomato/iDTR mice.

Table S1. List of antibodies used for flow cytometry and immunofluorescence.

Table S2. Sequences of primers used for quantitative RT-PCR.

References (4245)


Acknowledgments: We thank K. Castiglione, A. Matthies, and D. Döhler for technical assistance; M. Kirsch, L. Gundel, and C. Dietz for animal husbandry; G. Krönke and S. Culemann for providing the Rosa26iDTR and Csf1rLsL-DTR mice; S. Wirtz for providing the Rosa26tdTomato mice; D. Weiss for pulsed-field gel electrophoresis equipment; R. Salvi for help with neural tissue isolation; M. Kindermann and T. Fraass for tdTomato imaging; the Optical Imaging Center Erlangen (OICE) for help with microscopy; and members of the Voehringer lab for helpful comments. Funding: This work was supported by grants from the Deutsche Forschungsgemeinschaft (CRC1181_A02 to D.V., CRC1181_A03 to G.K., and CRC1181_A08 to S.W.). Author contributions: Conceptualization: B.K. and D.V.; methodology: R.N., S.C., and S.W.; investigation: B.K. and C.S.; statistical analyses: B.K.; resources: G.K.; writing—original draft: B.K. and D.V.; writing—review and editing: B.K. and D.V.; funding acquisition: D.V. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data supporting the findings of this study are available within the article. RetnlaCre mice are available upon reasonable request to D.V. and dependent on a mutually signed material transfer agreement.

Stay Connected to Science Immunology

Navigate This Article