Loss of FOXM1 in macrophages promotes pulmonary fibrosis by activating p38 MAPK signaling pathway.

Idiopathic pulmonary fibrosis (IPF) is a chronic disease with high mortality and is refractory to treatment. Pulmonary macrophages can both promote and repress fibrosis, however molecular mechanisms regulating macrophage functions during fibrosis remain poorly understood. FOXM1 is a transcription factor and is not expressed in quiescent lungs. Herein, we show that FOXM1 is highly expressed in pulmonary macrophages within fibrotic lungs of IPF patients and mouse fibrotic lungs. Macrophage-specific deletion of Foxm1 in mice (myFoxm1-/-) exacerbated pulmonary fibrosis. Inactivation of FOXM1 in vivo and in vitro increased p38 MAPK signaling in macrophages and decreased DUSP1, a negative regulator of p38 MAPK pathway. FOXM1 directly activated Dusp1 promoter. Overexpression of DUSP1 in FOXM1-deficient macrophages prevented activation of p38 MAPK pathway. Adoptive transfer of wild-type monocytes to myFoxm1-/- mice alleviated bleomycin-induced fibrosis. Altogether, contrary to known pro-fibrotic activities in lung epithelium and fibroblasts, FOXM1 has anti-fibrotic function in macrophages by regulating p38 MAPK.


Introduction
Idiopathic pulmonary fibrosis (IPF) is a chronic lung disease characterized by excessive accumulation of extracellular matrix, ultimately leading to disruption of gas exchange and respiratory functions. This disease remains poorly understood and is refractory to treatment. Inflammatory cells, including phagocytic cells like macrophages, have been shown to be intimately involved with all stages of lung injury and repair [1]. During acute wound healing, monocytes respond to a variety of chemo attractive cytokines and chemokines, are recruited to the lung and differentiate into macrophages [2]. Once present at the source of injury macrophages promote inflammation and wound healing to clear dead cells and to restore normal tissue homeostasis. Macrophage functions are diverse during fibrogenesis, consisting of both pro-and anti-fibrotic roles [1,3,4]. Macrophages produce anti-fibrotic mediators such as IL-13, ARG1 and FIZZ1 that repress inflammatory and fibrotic responses, and induce apoptosis of myofibroblasts [5][6][7][8]. Macrophages promote extracellular matrix (ECM) turnover by direct production of matrix metalloproteinases (MMPs), including MMP9 and MMP13 which degrade collagen [9,10]. In contrast to anti-fibrotic functions, macrophages can promote fibrosis through secretion of pro-fibrotic and pro-inflammatory mediators including TGF-β1, PDGF and IL-1β [11][12][13][14][15][16][17]. Activation of p38 MAPK signaling pathway in pulmonary macrophages promotes lung fibrosis through secretion of IL-6, IL-1β, and TNF-α, all of which are the direct targets of p38 MAPK pathway [18]. While pulmonary macrophages can both promote and inhibit fibrogenesis, transcriptional mechanisms regulating these dual functions remain poorly understood.
The Forkhead box M1 (FOXM1) is a member of the Forkhead box (FOX) family of transcription factors. FOXM1 is highly expressed during embryogenesis but nearly absent from adult respiratory cell types, including pulmonary macrophages [19]. FOXM1 is aberrantly induced in the lungs of IPF patients and in the mouse fibrotic lungs after radiation-or bleomycin-induced lung injury [20]. Genetic deletion of Foxm1 in alveolar epithelial type II cells (AECII) prevented both radiation-and bleomycin-induced lung fibrosis in mice, while overexpression of FOXM1 in AECII exacerbated fibrosis [20]. FOXM1 directly activated transcription of Snail, a critical regulator of epithelial to mesenchymal transition [20]. Furthermore, FOXM1 was found to be expressed in IPF fibroblasts [21,22]. FOXM1 induced TGFβ-mediated differentiation of fibroblasts into myofibroblasts, and protected them from apoptosis. Deletion of Foxm1 in myofibroblasts protected mice from bleomycin-induced pulmonary fibrosis and sensitized myofibroblasts to FasL-induced apoptosis [21]. In another study, FOXM1 increased expression of DNA repair proteins RAD51 and BRCA2 in IPF fibroblasts, and protected IPF fibroblasts from radiation-induced apoptosis [22]. All these published studies implicate FOXM1 in fibrotic responses and suggest that pharmacological inhibition of FOXM1 can be beneficial in IPF.
In the present study, we found a novel and unexpected role of FOXM1 in pulmonary fibrosis. Contrary to pro-fibrotic functions of FOXM1 in AECII and fibroblasts, FOXM1 has an anti-fibrotic role in macrophages. Mice lacking Foxm1 in macrophages developed exacerbated pulmonary fibrosis following repeated intra-tracheal administration of bleomycin. FOXM1 inhibited the p38 MAPK signaling pathway in pulmonary macrophages through transcriptional activation of DUSP1, a negative regulator of p38 MAPK pathway [23]. Adoptive transfer of FOXM1-expressing monocytes protected FOXM1-deficient mice from bleomycin-induced pulmonary fibrosis. Altogether, FOXM1 has anti-fibrotic function in macrophages, limiting potential clinical use of FOXM1 inhibitors in IPF.

FOXM1 expression is increased in pulmonary macrophages within human IPF and mouse fibrotic lungs
The macrophage-specific expression of FOXM1 in the lung tissue of IPF patients was visualized by a co-staining of lung sections for FOXM1 and the macrophage marker CD68. Fibrotic lesions in IPF lungs were identified by H&E staining and immunostaining for αSMA (S1 Fig).
In human donor lungs without IPF, less than 10% of macrophages were positive for FOXM1 ( Fig 1A, upper panels). In contrast, lungs of IPF patients showed abundant expression of FOXM1 in macrophages within fibrotic lesions as well as adjacent alveolar regions (Fig 1A, middle and bottom panels). Approximately 55% of the CD68 + cells were positive for FOXM1 in IPF lungs ( Fig 1B). Thus, FOXM1 expression is increased in macrophages within human IPF lungs.
Expression of FOXM1 in macrophages was examined in mouse lungs during progression of bleomycin-induced pulmonary fibrosis. Wild type mice were treated with bleomycin via intra-tracheal instillation once every week for four consecutive weeks and mice were sacrificed one week after the final bleomycin treatment. Saline-treated mice were used as controls. Immunostaining for FOXM1 showed the presence of only single FOXM1-positive cells in control mouse lungs, and a gradual increase in the number of FOXM1-positive cells after bleomycin treatments (Fig 1C). Expression of FOXM1 in macrophages was visualized by co-staining of lung sections for FOXM1 and MAC3. Similar to human donor lungs, less than 10% of macrophages within normal mouse lungs were positive for FOXM1 (Fig 1D and 1E). The percentage of macrophages expressing FOXM1 was significantly increased in bleomycin-treated mouse lungs (Fig 1D and 1E), a finding consistent with the results from the human IPF studies ( Fig 1B). In addition to increased FOXM1 staining, Foxm1 mRNA was increased in macrophages isolated from bronco-alveolar lavage fluid (BALF) of bleomycin-treated mice (Fig 1F). Foxm1 mRNA was also increased in neutrophils purified from bleomycin-treated lungs (S2 Fig). Altogether, FOXM1 mRNA and protein are increased in pulmonary macrophages within mouse and human fibrotic lungs.

Genetic deletion of Foxm1 in myeloid cells exacerbates bleomycin-induced pulmonary fibrosis
To determine FOXM1 requirements in macrophages during progression of pulmonary fibrosis, we utilized a previously generated myeloid-specific Foxm1 knockout mouse model (LysM-Cre tg/-;Foxm1 fl/fl ; abbreviated as myFoxm1 -/-) [24,25]. Following Cre-mediated recombination of the Foxm1-flox allele, exons 4-7 encoding the DNA binding and transcriptional activation domains of FOXM1 protein were deleted in myeloid cells, including monocytes, macrophages and neutrophils (Fig 2A) [25]. The myFoxm1 -/and control Foxm1 fl/fl mice were is not expressed in CD68 + macrophages (green) of control donor lungs (upper panels) but is upregulated in macrophages within fibrotic lesions (middle panels) and adjacent alveolar regions (bottom panels) of idiopathic pulmonary fibrosis (IPF) lungs. (B) Percent of FOXM1 + /CD68 + double treated with bleomycin via intra-tracheal instillation once every week for three consecutive weeks (Fig 2B). The lungs of uninjured control and myFoxm1 -/mice had normal lung architecture ( Fig 2C, top panels), indicating that FOXM1 expression in myeloid cell lineage is dispensable for lung development. Following chronic bleomycin injury, deletion of Foxm1 increased the severity of pulmonary fibrosis as shown by H&E staining (Fig 2C and S3 Fig) and Sircol assay measuring collagen content (Fig 2D). Masson's trichrome and Sirius red/ Fast green staining showed increased collagen depositions in bleomycin-treated myFoxm1 -/lungs compared to controls ( Fig 2E). We next assessed the presence of myofibroblasts using immunostaining for α-Smooth muscle actin (αSMA). Without bleomycin injury, no difference in αSMA staining was observed in control and myFoxm1 -/lungs ( Fig 1F, upper panels). Bleomycin treatment increased the number of αSMA + myofibroblasts in the lungs of myFoxm1 -/mice compared to bleomycin-treated controls (Fig 2F, middle, and bottom panels), a finding consistent with elevated collagen levels in the lung tissue (Fig 2D and 2E). Thus, deletion of Foxm1 from myeloid cells increased lung fibrosis in bleomycin-treated mice.

Deletion of Foxm1 increases expression of pro-inflammatory cytokines in pulmonary macrophages
To examine the efficiency of Foxm1 deletion, BALF cells containing >90% of macrophages were isolated from myFoxm1 -/and control mice and used for qRT-PCR. Foxm1 mRNA was decreased by 94% in BALF cells from myFoxm1 -/mice, confirming efficient deletion of Foxm1 by the LyzM-cre transgene ( Fig 3A). We also assessed FOXM1 protein in pulmonary macrophages by co-staining lung sections for FOXM1 and MAC-3. After bleomycin treatment, FOXM1 protein was often found in the nuclei of macrophages in control Foxm1 fl/fl lungs ( Fig 3B, upper panels). In contrast, in bleomycin-treated myFoxm1 -/lungs, macrophages were negative for FOXM1 ( Fig 3B, bottom panels). The percentage of FOXM1-positive macrophages was significantly decreased in bleomycin-treated myFoxm1 -/lungs compared to controls ( Fig 3C). Thus, FOXM1 was efficiently deleted in macrophages in myFoxm1 -/lungs.
Since FOXM1 increased the recruitment of macrophages during lung carcinogenesis [24] and liver injury [26], we used flow cytometry to examine inflammatory response in bleomycin-treated mouse lungs (Fig 3D and S4A Fig). Surprisingly, deletion of Foxm1 did not change the percentages of monocytes, interstitial and alveolar macrophages in the bleomycin-injured lung tissue (Fig 3E). mRNAs of Cx3cr1 and Ccr2, the chemokine receptors critical for recruitment of macrophages to the lung tissue [24,27] were not changed in bleomycin-treated myFoxm1 -/lungs (S4B Fig). The percentages of T and B lymphocytes were also similar in bleomycin-treated control and myFoxm1 -/lungs ( Fig 3E and S4C Fig). Numbers of neutrophils were low in bleomycin-treated lungs ( Fig 3E). Thus, deletion of Foxm1 from myeloid cells did not change the percentages of inflammatory cells in fibrotic lungs. Since FOXM1 induces cellular proliferation in various cell types [19], co-localization of Ki-67 and MAC3 was performed to assess the number of proliferating lung macrophages. No difference in the number of Ki-67 + /Mac-3 + macrophages was found in bleomycin-treated myFoxm1 -/lungs positive cells were counted in 5 random fields and presented as mean ± SEM. Scale bar = 20μm. (C) FOXM1 (brown) is not expressed in control mouse lungs and gradually increased during the progression of bleomycin-induced pulmonary fibrosis in mice. Numbers of FOXM1 + cells were counted in 5 random fields and presented as mean ± SEM. N = 3 mice per group. Scale bar = 100μm. (D) FOXM1 (red) is not expressed in MAC3 + macrophages (green) of control mouse lungs (upper panels) but is upregulated in macrophages within fibrotic lesions and adjacent alveolar regions of bleomycin-injured fibrotic lungs (bottom panels). (E) Percent of FOXM1 + /MAC3 + double positive cells were counted in 5 random fields and presented as mean ± SEM. N = 3 mice per group. Scale bar = 20μm. (F) Foxm1 mRNA is increased in BALF cells from bleomycin-treated mice as determined by qRT-PCR. Actb mRNA was used for normalization. N = 3 mice per group. � = P < 0.05. ��� = P < 0.0001.
Interestingly, mRNAs of anti-fibrotic Arg1 and Mmp12 were decreased while mRNAs of pro-fibrotic Il1b and Tnf were increased in BALF macrophages from bleomycin-treated myFoxm1 -/mice ( Fig 3I). Since ARG1-positive macrophages were shown to protect liver from fibrosis [8], we examined ARG1 expression in fibrotic mouse lungs. Deletion of Foxm1 decreased the number of ARG1 + cells (S5A Fig). Agr1 mRNA was decreased in macrophages isolated from BALF and from lung tissue using FACS sorting (S5B Fig and S5C Fig). Interestingly, Mrc1 mRNA, encoding a type I membrane receptor that mediates the endocytosis of glycoproteins by macrophages [28], was decreased in lung interstitial and BALF macrophages isolated from bleomycin-treated myFoxm1 -/mice (S6A Fig and S6B Fig). While Tgfb1 mRNA was increased in the whole lung RNA from bleomycin-treated myFoxm1 -/mice (S6C Fig), Tgfb1 was unchanged in FOXM1-deficient macrophages (Fig 3I and S6D Fig). To compare the functional ability of macrophages to degrade collagen, we generated bone-marrow derived macrophages from control and myFoxm1 -/mice. Consistent with decreased Mrc1 expression, the collagen degradation ability of FOXM1-deficient macrophages was significantly decreased compared to WT macrophages as shown by collagen degradation assay (S6E Fig, S6F Fig and  S6G Fig).
Thus, genetic deletion of Foxm1 from myeloid cells did not change the percentage of inflammatory cells in the bleomycin-treated lungs, but altered expression of fibrosis-associated genes, and decreased the ability of macrophages to degrade collagens.

Cre-mediated deletion of Foxm1 from macrophages activates p38 MAPK pathway
Next, we examined gene expression in different subsets of lung macrophages. Interstitial (CD45 + CD11b + Ly6G -F4/80 hi CD68 low ) and alveolar (CD45 + CD11b + Ly6G -F4/80 hi CD68 hi ) macrophages were isolated from bleomycin-treated control and myFoxm1 -/lungs using FACS sorting (Fig 4A and 4B). While the efficiency of Foxm1 deletion was better in interstitial macrophages, Foxm1 mRNA was significantly decreased in both macrophage subsets (Fig 4A  and 4B). Consistent with the BALF data ( Fig 3I), mRNAs of Il1b, Il6 and Tnf were increased in both interstitial and alveolar macrophages isolated from myFoxm1 -/lungs (Fig 4A and 4B). Since these cytokines are known downstream targets of p38 MAPK pathway [29], we examined phosphorylation of p38 MAPK (Thr180/Tyr182), which activates p38 MAPK protein leading to increased expression of pro-inflammatory cytokines [30]. Phosphorylation of p38 (phospho-p38) was increased in pulmonary macrophages in bleomycin-treated myFoxm1 -/mice compared to controls ( Fig 4F). We next analyzed the expression of DUSP1, a dual specificity phosphatase that dephosphorylates p-p38 MAPK in mice. 8-10 weeks old myFoxm1 -/and control mice were treated with bleomycin intratracheally (1U/kg; once a week for three weeks). Mice were sacrificed, and lungs were collected one week after the final bleomycin treatment. (C) Deletion of Foxm1 from myeloid cells does not affect lung architecture in uninjured mice as shown by H&E staining (top panels). Increased pulmonary fibrosis was detected in bleomycin-treated myFoxm1 -/mice compared to control bleomycin-treated mice (bottom panels). Scale bar = 200μm. (D) Increased lung collagen content, as quantified by Sircol assay, was detected in bleomycin-treated myFoxm1 -/mice compared to control (n = 6 mice per group). (E) Increased lung collagen deposition in bleomycin-treated myFoxm1 -/mice was shown using Masson's trichrome and Sirius red fast green staining. Scale bar = 200μm. (F) Deletion of Foxm1 from myeloid cells increased number of αSMA + myofibroblasts in bleomycin-treated myFoxm1 -/mice. No changes in the number of αSMA + lung myofibroblasts were found in uninjured myFoxm1 -/mice compared to control lungs (n = 5 mice per group). Scale bar = 200μm. � = P < 0.05. https://doi.org/10.1371/journal.pgen.1008692.g002

Fig 3. Deletion of Foxm1 from macrophages increases expression of pro-fibrotic cytokines without affecting number of myeloid cells. (A)
Foxm1 mRNA levels were decreased in BALF cells isolated from bleomycin-treated myFoxm1 -/lungs as determined by qRT-PCR. Actb mRNA was and acts as a negative regulator of the p38 MAPK signaling pathway [23]. Dusp1 mRNA was decreased in both interstitial and alveolar macrophages isolated from bleomycin-treated myFoxm1 -/mice (Fig 4A and 4B). Immunostaining for DUSP1 and the number of DUSP1positive macrophages were also decreased in myFoxm1 -/lungs compared to controls (Fig 4D  and 4F). In addition, we found the increased number of IL-1β-positive macrophages in bleomycin-treated myFoxm1 -/lungs compared to control lungs ( Fig 4E and 4F). Thus, deletion of Foxm1 increased phosphorylation of p38 MAPK, decreased DUSP1 and increased expression of pro-inflammatory cytokines that are downstream targets of the p38 MAPK signaling pathway.

Inhibition of FOXM1 in vitro activates p38 MAPK pathway
The effect of Foxm1 deletion on p38 MAPK pathway was tested in vitro using RAW 264.7 macrophage cell line. Expression of Foxm1 in RAW 264.7 cells was inhibited using siRNA-mediated gene silencing (siFoxm1), which reduced Foxm1 mRNA ( Fig 5A) and FOXM1 protein ( Fig 5B). Consistent with our in vivo studies, knockdown of Foxm1 reduced Dusp1 and increased Tnf, Il6 and Il1b mRNAs ( Fig 5A). Furthermore, knockdown of Foxm1 increased protein levels of phosphorylated p38 MAPK without affecting total p38 MAPK protein as shown by Western blots (Fig 5B). The protein levels of TNF-α, IL-6, IL-1β were also increased after inhibition of FOXM1, consistent with the increased phospho-p38 MAPK (Fig 5B). Interestingly, supernatant from FOXM1-deficient macrophages increased the number of viable fibroblasts after 48 days in culture ( Fig 5C) and increased fibroblast activation as shown by increased expression of Acta2, Col1a1 and Ccnd1 (Fig 5D). Neutralizing antibodies against IL-1β or IL-6 decreased number of fibroblasts and their activation after treatment with supernatant derived from FOXM1-deficient macrophages (S7A Fig, S7B Fig and S7C Fig). Neutralizing antibodies against TNF-α decreased the number of fibroblasts (S7A Fig) but did not change activation of fibroblasts ( S7B Fig and S7C Fig). Thus, FOXM1-deficient macrophages directly increase fibroblasts numbers and activation, at least in part, through IL-1β, IL-6 and TNF-α. Altogether, inhibition of FOXM1 in vitro and in vivo reduced DUSP1 expression, increased p38 MAPK activation and increased expression of pro-fibrotic cytokines.
https://doi.org/10.1371/journal.pgen.1008692.g003  Fig 5F). This reporter was co-transfected with CMV-Empty or CMV-FOXM1 expression plasmids and transcriptional activation of the reporter was tested by measuring LUC activity. CMV-FOXM1 plasmid induced transcriptional activity of the Dusp1 promoter region (Fig 5F). Furthermore, chromatin immunoprecipitation (ChIP) was performed using RAW264.7 macrophages to identify functional FOXM1 binding sites in the Dusp1 promoter. FOXM1 specifically bound to the -152/-141 bp, but not to -1003/-992 bp region of the Dusp1 promoter ( Fig 5G). These data indicate that FOXM1 directly binds to and transcriptionally activates Dusp1 promoter in macrophages.

Overexpression of DUSP1 decreases activation of p38 MAPK pathway in FOXM1-deficient macrophages
To determine whether a decrease in DUSP1 contributes to the activation of p38 MAPK pathway in FOXM1-deficient macrophages, we overexpressed DUSP1 in siFoxm1-transfected RAW 264.7 cells. The efficient overexpression of DUSP1 is shown by increased Dusp1 mRNA (Fig 6A). Inhibition of FOXM1 increased the expression of p38 MAPK downstream targets Tnf, Il6 and Il1b in macrophages (Fig 6A), a finding consistent with the in vivo data (Fig 4A  and 4B). Overexpression of DUSP1 was sufficient to prevent the increase of Tnf, Il6 and Il1b mRNAs in FOXM1-deficient macrophages ( Fig 6A) and to decrease phospho-p38 MAPK protein in FOXM1-deficient macrophages (Fig 6B). While supernatant from FOXM1-deficient macrophages increased number of 3T3 fibroblasts and their activation, over-expression of DUSP1 attenuated these effects (Fig 6C and 6D), supporting a critical role of DUSP1 in profibrotic signaling from macrophages to fibroblasts. Altogether, these data indicate that FOXM1 inhibits activation of p38 MAPK pathway and expression of its targets through transcriptional activation of DUSP1.

Adoptive transfer of wild type macrophages protects myFoxm1 -/mice from bleomycin-induced pulmonary fibrosis
Next, we tested whether adoptive transfer of wild type (WT) macrophages decreases the severe fibrotic burden in bleomycin-treated myFoxm1 -/mice. Control and myFoxm1 -/mice were treated twice with 1.0 U/Kg bleomycin via intra-tracheal instillation followed by adoptive transfer of bone marrow-derived WT monocytic progenitors (Fig 6E).  and collagen depositions were measured to assess fibrosis after the adoptive transfer. Adoptive transfer of WT monocytes into bleomycin-treated myFoxm1 -/mice significantly reduced the fibrotic load to collagen levels seen in control mice (Fig 6F and 6G). Thus, adoptive transfer of WT macrophages decreases pulmonary fibrosis in myFoxm1 -/mice. Altogether, FOXM1 expression in macrophages is critical to prevent pulmonary fibrosis.

Discussion
Pulmonary fibrosis results from deregulated repair of damaged tissue and is refractory to available treatment. Several different cell types are essential for the development of pulmonary fibrosis, including epithelial cells (AECII), fibroblasts and inflammatory cells [31]. The Forkhead box M1 (FOXM1) is a transcription factor that is highly expressed during embryogenesis, but is not expressed in quiescent adult lung [19]. FOXM1 stimulates cellular proliferation and its expression is induced in lung cancers [32,33]. High levels of FOXM1 were found in tumor cells and tumor-associated macrophages [24,25]. Genetic deletion of Foxm1 in macrophages inhibited lung cancer. FOXM1-deficient macrophages failed to migrate to the lung tissue resulting in reduced tumor-associated inflammation known to promote tumorigenesis [24]. Thus, FOXM1 has oncogenic and pro-inflammatory role in adult lungs during carcinogenesis. In addition to lung cancer, it was also shown that FOXM1 is highly expressed in several other chronic lung diseases, including COPD, asthma and IPF [20-22, 34, 35]. In human and mouse fibrotic lungs, increased levels of FOXM1 were found in AECII and fibroblasts [20,21]. Mouse genetic studies demonstrated that deletion of Foxm1 from either AECII [20] or fibroblasts [21] inhibits pulmonary fibrosis. Likewise, overexpression of FOXM1 in AECII exacerbated pulmonary fibrosis induced by thoracic irradiation or bleomycin [20], which is consistent with pro-fibrotic role of FOXM1. These studies suggest that inhibition of FOXM1 can be beneficial for the treatment of IPF and other fibrotic diseases. In the present manuscript, we found that in addition to epithelial cells and fibroblasts, FOXM1 is highly overexpressed in pulmonary macrophages and neutrophils of mouse and human fibrotic lungs. Surprisingly, deletion of Foxm1 from myeloid cells exacerbated bleomycin-induced pulmonary fibrosis, indicating anti-fibrotic role of FOXM1 in myeloid lineage. Thus, FOXM1 can induce or inhibit fibrosis depending on cell specificity and therefore, the use of FOXM1 inhibitors in IPF should be approached with caution. Interestingly, Mlyz-Cre deleted Foxm1 from all myeloid cells [34]. It is possible that the macrophage lineage is central to the fibrotic phenotype in myFoxm1 -/mice. Adoptive transfer of monocytic cells was sufficient to decrease fibrosis in myFoxm1 -/lungs, providing a direct support to this concept.
We have previously showed that FOXM1-deficient macrophages fail to migrate to the lung tissue during lung tumorigenesis [24]. After acute liver injury, deletion of Foxm1 from and Il1b mRNAs in siFoxm1 transfected macrophages as shown by qRT-PCR. Actb mRNA was used for normalization (n = 3). (B) Overexpression of Dusp1 in FOXM1-deficient macrophages decreases activation of p38 MAPK pathway. RAW264.7 macrophages were co-transfected with nontargeting siRNA (siControl) or siFoxm1, and pHIV-empty vector or pHIV-Dusp1 and protein levels were analyzed by Western blot. Expression of DUSP1 in Foxm1-depleted RAW264.7 macrophages decreased phospho-p38 MAPK protein levels. No changes were detected in total p38 MAPK protein levels. β-ACTIN was used as loading control. Graph shows densitometry quantification of Western blots. Protein levels were normalized to β-ACTIN. Protein expression is presented as fold change over pHIV empty + siControl transfected cells. Blots are representative of three independent experiments. (C-D) Supernatant from Dusp1 overexpressing Foxm1-deficient macrophages decreases fibroblast number and activation. 3T3 fibroblasts were cultured in supernatant obtained from RAW264.7 macrophages co-transfected with control siRNA or siFoxm1, and pHIV-empty vector or pHIV-Dusp1. 48 hours after transfection number of live cells were counted by trypan blue (n = 3). Expression of Acta2, Ccnd1, and Col1a1 mRNAs in fibroblasts was measured by qRT-PCR. Actb mRNA was used for normalization (n = 3). (E) Experimental design of adoptive transfer. (F) Representative H&E images (left) and trichrome images (right) show significant reduction in fibrosis in myFoxm1 -/mice upon adoptive transfer of FOXM1-expressing WT monocytes (n = 3 mice per group). Scale bar = 200μm. (G) Lung collagen content was measured using the Sircol assay and revealed that transfer of FOXM1-expressing WT monocytes into myFoxm1 -/mice significantly reduced bleomycininduced lung fibrosis. � = P < 0.05, �� = P <0.01, ��� = P < 0.001, by Student's t-test.
https://doi.org/10.1371/journal.pgen.1008692.g006 macrophages severely reduced macrophage recruitment to the injured liver and delayed liver repair [25]. FOXM1 regulated migration of macrophages through transcriptional activation of chemokine receptors Cx3cr1 and Ccr2 [24,25]. Surprisingly, in the present studies we found that deletion of Foxm1 did not affect the numbers of macrophages after bleomycin lung injury. Moreover, expression of Cx3cr1 and Ccr2 in macrophages were not changed after deletion of Foxm1. Thus, FOXM1 activates Cx3cr1 and Ccr2 in macrophages during lung cancer and acute liver injury/repair [24,25], but is not doing so during lung fibrosis. These surprising results indicate that functions of FOXM1 in macrophages are dependent on type of the injury and tissue/organ specificity. Being a transcription factor, FOXM1 physically binds to other coactivators/co-repressors. It is possible that specific combinations of these regulatory co-factors alter FOXM1 functions by activating different sets of transcription targets. In addition, transcription factors are often regulated by the epigenetic landscape. Thus, co-activators/co-repressors and epigenetic changes in macrophages could explain distinct functions of FOXM1 during lung cancer, acute injury and lung fibrosis.
Mitogen-activated protein kinase (MAPK) superfamily plays a crucial role in growth, differentiation, and response to stress [36]. The p38 MAPK pathway is activated by environmental stresses, and plays a central role in inflammatory response in the lung. Activation of the p38 MAPK signaling cascade induces secretion of IL-8 by bronchial epithelial cells [37], increases expression of ICAM-1 adhesion molecule on endothelial cells [38], stimulates secretion of TNF-α, IL-6 and IL-8 by neutrophils [39] and monocytes [40]. Activation of the p38 MAPK pathway directly correlates with severity of pulmonary fibrosis. In comparison to normal lung, p38 MAPK pathway is activated in epithelial, endothelial, smooth muscle cells, and fibroblasts in lung tissues from patients with IPF [41]. Activation of p38 MAPK pathway by TGF-β is critical for aberrant proliferation of pulmonary interstitial fibroblasts [42]. P38 MAPK activates MK2 and induces differentiation of fibroblasts into myofibroblasts [43]. On the other hand, inhibition of p38 MAPK signaling decreased expression of collagen I and fibronectin [44]. Inhibition of p38 MAPK signaling with specific inhibitor FR-167653 prevented bleomycininduced pulmonary fibrosis in mice [30], indicating an essential role of p38 MAPK pathway in pulmonary fibrotic responses. The important contribution of the current study is that FOXM1 inhibits p38 MAPK signaling pathway in macrophages through transcriptional activation of Dusp1, a negative regulator of p38 MAPK. In the absence of FOXM1, the p38 MAPK pathway is activated in macrophages leading to production of pro-fibrotic mediators IL-1β, IL-6 and TNF-α that directly stimulate fibroblast activation and survival, exacerbating pulmonary fibrosis. Given a robust pro-fibrotic activity of FOXM1 in macrophage lineage and critical role of macrophages in mouse models of pulmonary fibrosis [45], [46], [47,48], the use of FOXM1 inhibitors in IPF or other fibrotic diseases should be approached with caution.
In summary, macrophage-specific inactivation of FOXM1, exacerbates bleomycin-induced pulmonary fibrosis. FOXM1 inhibits the p38 MAPK signaling pathway and its downstream target genes, TNFα, IL-6 and IL-1β, through transcriptional activation of Dusp1. Our studies suggest an important limitation of using pharmacological FOXM1 inhibitors in IPF.

Transgenic mice and bleomycin-induced fibrosis model
Foxm1 was deleted from macrophages by breeding Foxm1 fl/fl mice [25] with LysM-Cre tg/transgenic mice (The Jackson Laboratory, Bar Harbor, ME, USA) to produce LysM-Cre tg/-: Foxm1 fl/fl mice (myFoxm1 -/-) as previously described [24,25]. Single transgenic Foxm1 fl/fl littermates were used as controls. Animal studies were reviewed and approved by the Animal Care and Use Committee of Cincinnati Children's Hospital Research Foundation. Pulmonary fibrosis was induced with bleomycin sulfate (EMD Millipore, Billerica, MA) [20]. 1.0 U / Kg of bleomycin was administered intratracheally once per week for three consecutive weeks. For the time-course experiment, CD57Bl/6 mice were administered 1.0 U / Kg of bleomycin intratracheally once per week for 4 weeks. Mice were sacrificed one week after final bleomycin treatment. Quantification of fibrosis was determined by measuring lung collagen content using the Sircol collagen assay (Biocolor, Carrickfergus, UK). Total RNA was prepared from BALF cells, and flow-sorted alveolar and interstitial macrophages from bleomycin-treated myFoxm1 -/or control mice. mRNA levels of specific genes were measured by qRT-PCR using TaqMan probes (S1 Table) and the StepOnePlus Real-Time PCR system (Applied Biosystems) as previously described [49].

Human Lung samples
Lung samples were obtained at the University of Vienna and the University of Giessen Lung Center from patients with IPF (n = 9) and control donors (n = 6) anonymously following lung transplantation. Diagnosis was done according to the American Thoracic Society/European Respiratory Society (ATS/ERS) criteria for IPF. Utilization of human lung samples was reviewed and approved by the Ethics Committee of Justus-Liebig-University of Giessen. Fibrotic lesions were identified based on the histology of the lung tissue sections [50,51].

Immunohistochemistry and immunofluorescence
Mouse lungs were inflated, fixed and embedded in paraffin. Sections (5 μm) were stained with hematoxylin and eosin (H&E) to assess gross lung morphology. Immunohistochemical staining and immunofluorescence were performed as described previously [52]. The list of antibodies is presented in S2 Table. Antibody-antigen complexes were detected using biotinylated secondary antibodies followed by avidin-biotin-horseradish peroxidase complex and 3,3'-diaminobenzidine substrate (Vector Labs), as previously described [53], [54]. Sections were counterstained with nuclear fast red (Vector Labs). For immunofluorescence imaging, secondary antibodies conjugated with Alexa Fluor 488 or Alexa Fluor 594 (Invitrogen/Molecular Probes) were used. Nuclei were counterstained with DAPI. Fibrosis was visualized using Masson's trichrome staining (Siemens) and Sirius Red Fast Green staining (Chondrex). Images were obtained using a Zeiss AxioPlan2 microscope.

Cell culture, qRT-PCR and Western blot
The Foxm1 siRNA (Dharmacon) was used to transiently knockdown Foxm1 in mouse macrophage cell line RAW 264.7 (ATCC) using Dharmafect Duo transfection reagent (Dharmacon) according to the manufacturer's instructions. Non-targeting siRNA was used as control. Cells were collected for RNA and protein extraction 48 hours after transfection. Culture supernatant was collected for 3T3 proliferation assay. For in vitro cytokine-neutralization assay, supernatant from RAW 264.7 cells was incubated with neutralizing antibodies against IL-6 (0.03μg/ mL), IL-1β (0.25μg/mL), or TNF-α (0.1μg/mL) at 37˚C for 2 hours. In vitro rescue experiments were performed by dual transfection of siFoxm1 and pHIVDusp1 overexpression plasmids using Dharmafect Duo transfection reagent (Dharmacon). Empty plasmid and non-targeting siRNA were used as control. Cells were collected for RNA and protein extraction 48 hours after transfection. Culture supernatant from macrophages was collected for 3T3 proliferation assay. 3T3 cells were cultured in the collected supernatant for 48 hours and number of live cells were counted using trypan blue by automated cell counter Countess TM II FL (Life Technologies). Protein and mRNA from 3T3 cells were used for qRT-PCR and Western blot as previously described [33]. TaqMan probes are provided in S1 Table. The  Chromatin immunoprecipitation assay RAW264.7 mouse macrophages were cultured in vitro according to the ATCC protocol. Cells were cross-linked by addition of formaldehyde, sonicated, and used for immunoprecipitation using a FOXM1 rabbit polyclonal antibody (C-20; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Reverse cross-linked chromatin immune-precipitation DNA samples were subjected to qRT-PCR using EvaGreen dye and the following PCR primers (5' to 3'): mDusp1 -1kb Forward GTTAGTAGGAGTCCGGGTCCA + Reverse GCCAACAAGCTCTTCCGTGC; mDusp1 -100bp: Forward ATCTGGGCTCAGCTGTTCTG + Reverse TTCACTCTGGCCCATGA AGC. Binding of FOXM1 was normalized to the DNA of the samples immunoprecipitated with isotype control IgG.

Cloning of the mouse Dusp1 promoter region and luciferase assay.
The mouse Dusp1 promoter corresponding to the -1033bp to -17bp was PCR amplified from C57/B6 gDNA using the following primers (5' to 3'): Fwd. GAG GTAC CAG GAG GAT ATA GAA AGC GGC; Rev. ATCTCGAGTTCGGGGAGATGATACTCC. The 1.0kb promoter region was cloned into the KpnI and XhoI sites of the pGL3 luciferase reporter plasmid (Promega). Dual luciferase reporter assay was performed using the Dusp1 luciferase reporter with CMV-empty or CMV-FOXM1 overexpression plasmids.

Isolation and adoptive transfer of monocytes during bleomycin-induced lung injury
For adoptive transfer experiments, recipient Foxm1 fl/fl and myFoxm1 -/mice were treated with 1.0 U / Kg bleomycin twice via intratracheal administrations. Monocytes were isolated from the bone marrow of donor Foxm1 fl/fl or myFoxm1 -/mice as described [25,57]. Cells were incubated with a biotin-conjugated anti-CD115 (M-CSF receptor) antibody (eBioscience) followed by anti-biotin MACS beads (Miltenyi Biotech, Auburn, CA, USA). A positive cell fraction was obtained through magnetic separation and labeled with CFSE as described [57]. The percentage of monocytes in the positive fraction was 85% as determined by flow cytometry defining the monocytic phenotype as CD115 + /CD11b + /Ly-6G − /Ly-6C hi /F4/80 lo cells. One week after treatment with bleomycin, 5 × 10 5 labeled cells were injected into tail veins of myFoxm1 -/or control Foxm1 fl/fl mice. Recipient mice were harvested one week after adoptive transfer and fibrotic load was assessed.

Generation of Bone-marrow derived macrophages and Collagen degradation assay
Bone marrow derived macrophages were generated as previously described [58]. Bone marrow cells were obtained from 8-10 weeks-old Foxm1 fl/fl , or myFoxm1 -/mice. Mononuclear cells were isolated and cultured overnight. The following day, non-adherent cells were recovered, transferred to a new dish and cultured in in DMEM containing 10% heat-inactivated FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.25 μg/ml Amphotericin B, 10ng/ml GM-CSF and 5 ng/ml M-CSF (both from R&D Systems) to allow differentiation and expansion of macrophages. After 48 hours, the culture medium was changed, and adherent bone marrow-derived macrophages were isolated by trypsin-EDTA 5 days after seeding. Cells were plated at a concentration of 50,000 cells per well in black 96-well plates in triplicate in DMEM containing 10% heat-inactivated FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.25 μg/ml Amphotericin B, and 10ng/mL LPS and cultured overnight at 37˚C in a humidified environment containing 5% CO 2 . Following day, cells were washed with PBS and incubated in PBS with 1μg/ml of FITC-conjugated type I collagen (Invitrogen) in a total volume of 100 μl. After 4 hours, fluorescence from cleaved collagen was quantified using a spectrofluorometer (SpectraMax Minimax 300) at 499nm Absorption maximum and 525nm Emission maximum. The fluorescent signal from control wells containing only collagen was subtracted from that from wells containing macrophages and collagen.

Statistical analysis
Statistical significance differences in measured variables between experimental and control groups were assessed by Student's t-test (2-tailed) P-values <0.05 were considered significant. Values for all measurements were expressed as the mean ± standard deviation or as the mean ± standard error of mean. Statistical analysis was performed, and data were graphically displayed using GraphPad Prism v.5.0 for Windows (GraphPad Software, Inc.).

S5 Fig. FOXM1-deficient macrophages express decreased anti-fibrotic marker ARG1. (A)
Decreased expression of ARG1 is found in bleomycin-treated lungs of myFoxm1 -/mice compared to controls. Number of ARG1 + cells were counted in 5 random fields and presented as mean ± SEM (n = 3 mice per group). Scale bar = 200μm. (B) Arg1 mRNA is decreased in BALF cells isolated from bleomycin-treated myFoxm1 -/mice compared to controls as determined by qRT-PCR. Actb mRNA was used for normalization (n = 3 mice per group). (C) Arg1 mRNA is decreased in interstitial macrophages FACS-sorted from bleomycin-treated myFoxm1 -/mice compared to controls as determined by qRT-PCR. Actb mRNA was used for normalization (n = 3 mice per group). � = P > 0.05; �� = P > 0.01, ��� = P < 0.001, by Student's t-test. (TIF) S6 Fig. FOXM1-deficient macrophages have decreased expression of Mrc1 and inefficiently degrade collagen. (A) Mrc1 mRNA is decreased in interstitial macrophages FACS-sorted from bleomycin-treated myFoxm1 -/mice compared to controls as determined by qRT-PCR. Actb mRNA was used for normalization. N = 3 mice per group. (B) Mrc1 mRNA is decreased in BALF cells isolated from bleomycin-treated myFoxm1 -/mice compared to controls as determined by qRT-PCR. Actb mRNA was used for normalization. N = 3 mice per group. (C) Tgfb1 mRNA is increased in total lung RNA from bleomycin-treated myFoxm1 -/mice as determined by qRT-PCR. Actb mRNA was used for normalization (n = 5 mice per group). (D) Depletion of Foxm1 in mouse macrophages did not change Tgfb1 mRNA as shown by qRT-PCR. RAW264.7 cells were transfected with control siRNA or siFoxm1 and mRNA levels were measured by qRT-PCR. Actb mRNA was used for normalization (n = 3). (E) Foxm1 mRNA is decreased in bone-marrow derived macrophages from myFoxm1 -/mice compared to controls as determined by qRT-PCR. Actb mRNA was used for normalization. N = 3 mice per group. (F) Csfr1, Csf3, and Cxcr4 mRNA levels are unchanged in bone-marrow derived macrophages from treated myFoxm1 -/mice as compared to control mice as determined by qRT-PCR. Actb mRNA was used for normalization (n = 3 mice per group). (G) Collagen degradation is decreased in bone-marrow derived macrophages from myFoxm1 -/mice compared to controls. The fluorescent signal from control wells containing only collagen was subtracted from that from wells containing macrophages and collagen. Assay was done in triplicates using bone-marrow derived macrophages from 3 individual mice per group. � = P > 0.05; �� = P > 0.01, ��� = P < 0.001, by Student's t-test. (TIF) S7 Fig. (A) Supernatant from Foxm1-deficient macrophages decreases fibroblast survival through IL-6, IL-1β, and TNF-α. The NIH-3T3 fibroblasts were cultured in supernatant obtained from RAW264.7 macrophages transfected with control siRNA or siFoxm1. IL-6, IL-1β, and TNFα were inhibited in the cell culture using neutralizing Abs. 48 hours after addition of supernatant, the numbers of live fibroblasts were counted using trypan blue (n = 3). Neutralizing antibody against IgG was used as control. (B-C) Supernatant from Foxm1-deficient macrophages decreases Acta2 and Col1a1 mRNAs in fibroblasts through IL-6 and IL-1β as shown by qRT-PCR. mRNA levels in fibroblasts were analyzed by qRT-PCR. Actb mRNA was used for normalization (n = 3). Immunostaining shows FOXM1-positive macrophages in bleomycin-treated control lungs (top panels) and myFoxm1 -/lungs (bottom panels) after adoptive transfer of WT monocytes. No FOXM1-positive macrophages are found in bleomycin-treated myFoxm1 -/lungs without adoptive transfer (middle panels). Control and myFoxm1 -/mice were treated twice weekly with bleomycin (1.0 U / Kg). One week after treatment, bone marrow-derived monocytes were injected (I.V.) into both groups of mice. Mice were sacrificed one week after the adoptive transfer. Numbers of FOXM1 + cells were counted in 5 random fields and presented as mean ± SEM. N = 3 mice per group. (TIF) S1