Activation of Epidermal Growth Factor Receptor in Macrophages Mediates Feedback Inhibition of M2 Polarization and Gastrointestinal Tumor Cell Growth*

EGF receptor (EGFR) in tumor cells serves as a tumor promoter. However, information about EGFR activation in macrophages in regulating M2 polarization and tumor development is limited. This study aimed to investigate the effects of EGFR activation in macrophages on M2 polarization and development of gastrointestinal tumors. IL-4, a cytokine to elicit M2 polarization, stimulated release of an EGFR ligand, HB-EGF, and transactivation and down-regulation of EGFR in Raw 264.7 cells and peritoneal macrophages from WT mice. Knockdown of HB-EGF in macrophages inhibited EGFR transactivation by IL-4. IL-4-stimulated STAT6 activation, Arg1 and YM1 gene expression, and HB-EGF production were further enhanced by inhibition of EGFR activity in Raw 264.7 cells using an EGFR kinase inhibitor and in peritoneal macrophages from Egfrwa5 mice with kinase inactive EGFR and by knockdown of EGFR in peritoneal macrophages from Egfrfl/fl LysM-Cre mice with myeloid cell-specific EGFR deletion. Chitin induced a higher level of M2 polarization in peritoneal macrophages in Egfrfl/fl LysM-Cre mice than that in Egfrfl/fl mice. Accordingly, IL-4-conditioned medium stimulated growth and epithelial-to-mesenchymal transition in gastric epithelial and colonic tumor cells, which were suppressed by that from Raw 264.7 cells with HB-EGF knockdown but promoted by that from Egfrwa5 and Egfrfl/fl LysM-Cre peritoneal macrophages. Clinical assessment revealed that the number of macrophages with EGFR expression became less, indicating decreased inhibitory effects on M2 polarization, in late stage of human gastric cancers. Thus, IL-4-stimulated HB-EGF-dependent transactivation of EGFR in macrophages may mediate inhibitory feedback for M2 polarization and HB-EGF production, thereby inhibiting gastrointestinal tumor growth.

Tumor-associated macrophages (TAMs) in the tumor microenvironment usually exhibit M2 polarization in human and mouse models of cancers (5,6) and possess pro-tumorigenic activities in most cancers, including promoting tumor cell proliferation, migration, and invasion, increasing angiogenesis, and suppressing immunity through secretion of growth factors such as HB-EGF, EGF, CSF-1, VEGF, and PDGF, enzymes, including metalloproteinases (MMPs), and cytokines, such as IL-6, IL-10, and TNF (7). Clinical studies have revealed that there is a strong correlation between the abundance of TAMs and poor prognosis in most cancers. However, both positive and negative associations between TAMs and clinical outcomes have been reported in lung, gastric, prostate, and bone cancers, depending on tumor types and stages (8 -10).
It is known that cytokines, such as IL-4, IL-10, IL-13, IFN-␥, TNF, macrophage colony-stimulting factor (MCSF), TGF␤, and PGE2 produced by lymphocytes and tumor cells promote M2 polarization and functions of TAMs (11). However, the mechanism underlying the diverse functions of TAMs in different tumor microenvironment remains unclear.
EGF receptor (EGFR) belongs to the ErbB family of type 1 transmembrane receptor-tyrosine kinases. EGFR can be activated by direct ligand binding and transactivated by a wide variety of pharmacological and physiological stimuli, including TNF (12) and bacterial products, such as LPS (13) in intestinal epithelial cells, through stimulation of EGFR ligand release, such as EGF and HB-EGF. The cytoplasmic domain of EGFR contains the kinase domain as well as autophosphorylation sites at tyrosine residues. Ligation of EGFR leads to receptor dimerization, phosphorylation, and increased tyrosine kinase activity (14). These phosphorylated amino acids provide docking sites for a variety of signaling molecules that regulate intracellular signaling networks and ultimately define biological responses, such as proliferation, differentiation, migration, and survival (15,16). Activated EGFR is subsequently sorted by incompletely understood mechanisms for either recycling to the plasma membrane or destruction (17,18). Excessive functions of EGFR signaling are involved in initiation and progression of cancers in many human cancers of epithelial origin (19,20) and are implicated in numerous animal models of gastrointestinal tumorigenesis (21)(22)(23).
Studies regarding EGFR in macrophages have revealed that EGFR is phosphorylated at tyrosine and serine and threonine residues through stimulation of HB-EGF release and ligandindependent manner, respectively (24 -26). EGFR signaling has been shown to be involved in TLR3 (27)-and IFN-␥ (28)-dependent signaling pathways in macrophages. Our previous studies have demonstrated that EGFR is transactivated by LPS in macrophages (29). EGFR activation in macrophages suppresses both pro-and anti-inflammatory cytokines in response to inflammatory stimuli, and the decrease in the IL-10 plays a role in further up-regulating proinflammatory cytokine production, resulting in enhancing intestinal inflammation in dextran sulfate sodium-induced colitis (29). However, EGFR signaling in macrophages in directing M2 polarization for regulating gastrointestinal tumor development remains to be elucidated.
The purpose of this work is to determine the roles and mechanisms of EGFR activation in macrophages in regulating M2 polarization and gastric and colonic tumor growth. We demonstrated that IL-4 transactivated EGFR through stimulation of HB-EGF release in macrophages. Transactivation of EGFR by IL-4 in macrophages could serve as a negative feedback mechanism for M2 polarization and HB-EGF production, thereby inhibiting tumor growth. Our clinical assessment of human cancer samples revealed that EGFR was down-regulated in macrophages in the late stage of gastric cancers, which suggests that the less inhibitory effect of EGFR in macrophages may be associated with promoting cancer development by TAMs. These findings provide pivotal information for understanding the cell type-specific function of EGFR in tumorigenesis.

IL-4 Stimulates Transactivation of EGFR in Macrophages,
Which Requires HB-EGF Release-To determine whether EGFR signaling contributes to regulation of M2 polarization, we first studied the effects of IL-4 on EGFR activation in macrophages. Our data showed that IL-4 treatment stimulated transactivation of EGFR in Raw 264.7 cells (Fig. 1A) and in peritoneal macrophages isolated from WT mice (Fig. 1B). Degradation of EGFR occurred following its activation stimulated by IL-4 in macrophages (Fig. 1, A and B). As expected, an EGFR kinase inhibitor, AG1478, blocked IL-4-stimulated EGFR transactivation and degradation (Fig. 1A). Furthermore, IL-4 failed to stimulate EGFR activation and degradation in Egfr wa5 peritoneal macrophages, which have kinase-defective EGFR. These data indicate that EGFR degradation in IL-4-stimulated macrophages is a subsequent event after EGFR activation.
TAMs produced EGFR ligands such as EGF and HB-EGF to promote tumor cell growth (30 -32). We next tested whether EGFR ligand release mediated IL-4-induced EGFR transactivation in macrophages. Release of EGFR ligand by IL-4 treatment in macrophages was detected using ELISA assay. We found that IL-4 stimulated HB-EGF release in Raw 264.7 cells ( Fig. 2A). IL-4 treatment for 0.25-0.5 h induced EGFR transactivation (Fig. 1A), during which time points HB-EGF release was increased ( Fig. 2A). Thus, the time point for HB-EGF release was correlated with EGFR transactivation. No effect of IL-4 on EGF production in Raw 264.7 cells was identified (data not shown).
Next we examined the requirement of HB-EGF release for IL-4-stimulated EGFR transactivation in macrophages. MMPs Raw 264.7 mouse macrophages (A) and peritoneal macrophages isolated from WT and Egfr wa5 mice (B) were treated with IL-4 (10 ng/ml) for the indicated time periods in the presence and absence of 1-h pretreatment of an EGFR-tyrosine kinase inhibitor, AG1478 (450 nM). Cellular lysates were collected for Western blot analysis of EGFR expression and phosphorylation on tyrosine 1068 (EGFR-Tyr-1068). The ␤-actin blot was used as a protein loading control. In A, lanes were run on the same gel but were noncontiguous, as indicated by the empty spaces. Data are representative of at least three independent experiments. play a critical role in the release of EGFR ligands (33,34). We showed that EGFR transactivation and degradation by IL-4 were suppressed by broad-spectrum metalloproteinase inhibitors, GM6001 and TAPI-1 (Fig. 2B).
Furthermore, we applied the siRNA method to knock down HB-EGF expression in macrophages. The level of HB-EGF expression was decreased, and IL-4-stimulated EGFR transactivation was suppressed in Raw 264.7 cells transduced with HB-EGF siRNA but not with non-targeting RNA (Fig. 2C). Taken together, these results indicate that IL-4 stimulates MMPdependent HB-EGF release for transactivation of EGFR in macrophages.
IL-4-stimulated EGFR Transactivation Contributes to Inhibition of M2 Polarization in Macrophages-To test the roles of EGFR transactivation in IL-4-induced M2 polarization in macrophages, we used macrophages with inhibition of EGFR kinase activity and knockdown of EGFR expression. IL-4-stimulated STAT6 phosphorylation, which represents its activation status, was enhanced by inhibition of EGFR kinase activity by AG1478 and MMP inhibitors in Raw 264.7 cells (Fig. 3, A and  B). Inconsistent with these data, the activation level of STAT6 by IL-4 treatment in Egfr wa5 peritoneal macrophages was higher than that in WT peritoneal macrophages (Fig. 3C).
Expression of M2 polarization markers was examined using real-time PCR analysis. Both AG1478 and MMP inhibitors significantly increased IL-4-induced Arg1 gene expression in Raw 264.7 cells (Fig. 3D). In addition, IL-4 stimulated significantly higher levels of Arg1 and YM1 in peritoneal macrophages from Egfr wa5 mice than those in WT peritoneal macrophages (Fig.  3E). These data suggest that inhibition of EGFR kinase activity in macrophages increases IL-4-stimulated STAT6 activation and M2 polarization.
We further examined the effects of knockdown of EGFR in macrophages on M2 polarization. Knockdown of EGFR expression in peritoneal macrophages from Egfr fl/fl LysM-Cre mice enhanced IL-4-stimulated STAT6 activation and Arg1 and YM1 gene expression (Fig. 4, A and B).
The in vivo effects of EGFR activation on M2 polarization were studied in Egfr fl/fl LysM-Cre mice and their littermate control, LysM-Cre mice with i.p. injection of chitin. Chitin, a biopolymer of N-acetylglucosamine from fungi, arthropods, and helminthes, recruits macrophages with the M2 polarization to the site of administration (35). The levels of YM1 and Fizz1 gene expression were significantly up-regulated in peritoneal macrophages from chitin-elicited Egfr fl/fl LysM-Cre mice as compared with those from Egfr fl/fl mice (Fig. 4C). These data indicate that EGFR activation in macrophages exerts a negative feedback for M2 polarization.
Furthermore, we investigated the role of EGFR activation in IL-4-stimulated HB-EGF production in macrophages. In agreement with the data that IL-4 increased the level of HB-EGF release in Raw 264.7 cells ( Fig. 2A), the level of HB-EGF production was up-regulated by IL-4 treatment in WT and Egfr fl/fl macrophages. IL-4-induced HB-EGF release was further increased in Egfr wa5 and Egfr fl/fl LysM-Cre macrophages (Fig. 5,  A and B).
We further studied the regulatory effects of EGFR activation on HB-EGF gene expression. IL-4 significantly up-regulated HB-EGF gene expression in Raw 264.7 cells (Fig. 5C). IL-4stimulated HB-EGF gene expression in peritoneal macrophages from Egfr wa5 and Egfr fl/fl LysM-Cre mice were significantly higher than that in WT and Egfr fl/fl macrophages, respectively (Fig. 5, D and E).
These results suggest that HB-EGF release is required for IL-4-stimulated EGFR transactivation. However, transactivation of EGFR has negative feedback for IL-4-induced M2 polarization and HB-EGF production.

EGFR Transactivation in Macrophages Exerts Inhibitory Effects on Gastrointestinal Intestinal Cell Growth and EMT-It
has been reported that HB-EGF produced by TAMs promotes tumor cell growth (30 -32). We tested if inhibition of EGFR kinase activity and knockdown of EGFR expression in macrophages could affect the effects of IL-4-conditioned media from macrophages on regulation of gastrointestinal epithelial cell growth and EMT.
We treated peritoneal macrophages isolated from WT and Egfr wa5 mice with IL-4, then conditioned media were collected for treating ImSt and IMCE ras cells (Fig. 6A). Conditioned medium from IL-4-stimulated Egfr wa5 peritoneal macrophages induced higher levels of cell growth in ImSt cells than those by conditioned medium from IL-4-stimulated WT peritoneal macrophages (Fig. 6B). In agreement with this finding, we found that conditioned media from IL-4-stimulated WT peritoneal macrophages promoted formation of colonies in IMCE ras cells, as compared with control, which was further increased by conditioned media from IL-4-stimulated Egfr wa5 peritoneal macrophages (Fig. 6C). Next we examined EMT in ImSt and IMCE ras cells treated by conditioned media from macrophages. IL-4-stimulated conditioned medium from WT peritoneal macrophages up-regulated Snail gene expression in ImSt and IMCE ras cells, which was further increased by conditioned medium from IL-4-stimulated Egfr wa5 macrophages (Fig. 6, D and E).
We next tested the effects of conditioned medium from macrophages with inhibition of EGFR expression on colonic tumor growth. IMCE ras cells were treated with conditioned media from IL-4-stimulated peritoneal macrophages from Egfr fl/fl and Egfr fl/fl LysM-Cre mice for examining EMT. IL-4stimulated conditioned medium from Egfr fl/fl peritoneal macrophages up-regulated Snail gene expression in IMCE ras cells, which was further increased by conditioned medium from IL-4-stimulated Egfr fl/fl LysM-Cre macrophages (Fig. 7A).
To further assess the effects of EGFR activation in macrophages on colonic tumor growth, we used a xenograft tumor model. We inoculated Rag2 Ϫ/Ϫ mice with IMCE ras cells with co-treatment of conditioned media from peritoneal macrophages with and without IL-4 treatment. In the control media, IL-4 was added into the non-treated conditioned medium before inoculation to exclude the effects of IL-4 presenting in IL-4-conditioned media. Conditioned media from IL-4-stimulated Egfr fl/fl peritoneal macrophages significantly increased tumor volumes after 6 days of implantation, as compared with those with control media from Egfr fl/fl peritoneal macrophages and RPMI media (Fig. 7B). Conditioned media from IL-4-stimulated Egfr fl/fl LysM-Cre peritoneal macrophages significantly increased tumor volumes after 6 days of implantation, as compared with those by control media from Egfr fl/fl LysM-Cre macrophages and IL-4-stimulated media from Egfr fl/fl macrophages (Fig. 7B). In addition, significant increases in cell proliferation, detected by immunohistochemistry of a proliferation marker, Ki-67 (Fig. 7, C and D) and Snail expression by RT-PCR analysis in tumor cells (Fig. 7E), were observed in xenograft treated with IL-4-stimulated Egfr fl/fl LysM-Cre macrophageconditioned medium, as compared with those by IL-4-stimulated Egfr fl/fl macrophage-conditioned medium.
These results indicate that transactivation of EGFR in macrophages is capable of inhibiting colon cancer cell growth and EMT. It should be noted that IL-4 did not show any direct effects on growth and expression of Snail in ImSt cells (Fig. 8, B and C). Together, these data suggest that transactivation of EGFR in IL-4-stimulated macrophages suppresses the effects of conditioned medium from macrophages on growth and EMT in ImSt and IMCE ras cells.
HB-EGF Mediates IL-4-conditioned Medium-stimulated Tumor Growth and EMT-To determine whether increased production of HB-EGF by IL-4 treatment in macrophages serves as a functional factor for promoting tumor growth, we first evaluated the effects of IL-4-conditioned medium on EGFR activation, cell growth, and EMT in WT and Egfr Ϫ/Ϫ ImSt cells. Conditioned medium from IL-4-stimulated Raw 264.7 cells activated EGFR in WT ImSt cells (Fig. 8A). Compared with control-conditioned medium from untreated Raw 264.7 cells, conditioned medium from IL-4-stimulated Raw 264.7 cells stimulated higher levels of cell growth and Snail and Vimentin gene expression in WT but not Egfr Ϫ/Ϫ ImSt cells (Fig. 8, B and C). These data indicate that EGFR in ImSt mediates IL-4-conditioned medium regulation of growth and EMT. Furthermore, IL-4-conditioned media from Raw 264.7 cells transduced with siRNA HB-EGF failed to stimulate EGFR activation and Snail and Vimentin gene expression in WT ImSt cells (Fig. 8, A and D).
These results suggest that HB-EGF may contribute to promoting tumor cell growth and EMT by IL-4-conditioned media from macrophages. Because EGFR activation inhibits HB-EGF production, the effects of EGFR in macrophages on inhibiting tumor development may be through decreasing HB-EGF production.
Expression of EGFR in Macrophages Is Decreased in the Late Stage of Human Gastric Cancers-TAMs are associated with poor prognosis in gastric cancer (36). To provide some infor-mation regarding EGFR in macrophages in human cancers, we sought to study whether there is a correlation between the level of EGFR expression in macrophages in cancer tissues and tumor development.
The gastric tissue array, including normal and cancer tissues (Fig. 9A), was analyzed using immunohistochemistry to detect EGFR expression in CD68-positive macrophages (Fig. 9B). The immunohistochemistry data showed that the percentage of macrophages with EGFR expression in stage I gastric cancer tissues was similar to that in normal gastric tissues (p Ͼ 0.05); however, the percentages of macrophages with EGFR expression were significantly decreased in stage II-IV gastric cancer tissues, as compared with that in normal and stage 1 gastric cancer (p Ͻ 0.05) (Fig. 9C). These results indicate that the level of EGFR expression in macrophages may be associated with roles of TAMs in gastrointestinal tumor development.

Discussion
Diverse functions of TAMs have been reported in different types and stages of tumors (8 -10). Thus understanding of the mechanisms underlying these different functional programs in macrophages in response to signals in tumor microenvironment is important for cancer therapy. A finding from our stud- ies demonstrates that EGFR activation in macrophages exerts a negative feedback for M2 polarization. Therefore, EGFR signaling may be involved in the functional plasticity of macrophage development in the tumor microenvironment. However, the mechanisms underlying inhibition of M2 polarization by EGFR in macrophages is under investigation. It is known that in addition to activation of STAT6 pathway for M2 polarization, IL-4 induces negative feedback to inhibit STAT6 phosphorylation by up-regulating suppressor of cytokine signaling 1 (SOCS1) (1). In fact, inhibition of EGFR kinase activity in Egfr wa5 macrophages inhibits IL-4-stimulated SOCS1 production (supplemental Fig. 1). This evidence suggests that SOCS1 may serve as a target of EGFR activation for inhibition of M2 polarization.
It is important to elucidate the mechanisms underlying regulation of EGFR activation in macrophages in the tumor microenvironment. Our data support IL-4-stimulated HB-EGF release for induction of EGFR transactivation in macrophages. HB-EGF activates EGFR in Raw 264.7 cells in the same timedependent manner as IL-4-conditioned media do (supplemental Fig. 2). We did not find EGF production in Raw 264.7 cells and in peritoneal macrophages from WT and Egfr wa5 mice with and without IL-4 treatment by using a Proteome Profiler Mouse Angiogenesis Array kit and ELISA (data not shown). Because regulation of proteolytic activity of ADAMs (a disintegrin and metalloproteinases), which are membrane-anchored proteases, to cleave the extracellular domains of membrane-bound proteins including EGFR ligands is still poorly defined (37, 38), more studies are needed for elucidating the mechanisms involved in specific release HB-EGF by IL-4 in macrophages. We cannot exclude the possibility that other cytokines produced by lymphocytes and tumor cells, such as IL-10 and TNF, which play a central role in directing polarization and functions of ATMs (11), stimulate release of EGF, leading to activation of EGFR in macrophages.
It should be noted that compared with DMEM and the control medium from Egfr fl/fl peritoneal macrophages, the control medium from Egfr fl/fl LysM-Cre macrophages induced higher levels of tumor growth, cell proliferation, and Snail expression in the IMCE ras xenograft model (Fig. 7). However, HB-EGF production in untreated Egfr fl/fl LysM-Cre peritoneal macro-

. Inhibition of EGFR kinase activity in macrophages enhances the effects of IL-4-conditioned medium from macrophages on growth and EMT in gastric and colonic epithelial cells.
Conditioned media were prepared from WT and Egfr wa5 peritoneal macrophages treated with IL-4 (10 ng/ml) for 1 h (A). IL-4 (10 ng/ml) was added to the media from untreated WT and Egfr wa5 macrophages as control media. ImSt and IMCE ras cells were treated with control and IL-4-conditioned media for 24 h (B, D, and E). Cell viability was tested (B). The cell number change (cell number at the end of treatment Ϫ cell number before treatment) in the control medium-treated ImSt cells was set up as 1, and the cell number changes in other groups were compared with it to obtain the -fold change. RNA was isolated for detecting the level of Snail gene expression using real-time PCR analysis (D and E). The expression level in cells treated with control media from WT macrophages was set as 100% for comparison with other groups. IMCE ras cells were plated in 12-well dish (1000 cells/well) and cultured in control and IL-4-conditioned medium for 14 days (C). IMCE ras cells were stained using the cell proliferation assay kit. Area covered by colonies with the size Ͼ0.50 mm is shown. *, p Ͻ 0.05 compared with cells treated with control media from WT macrophages. #, p Ͻ 0.05 compared with cells treated with IL-4-conditioned media from WT peritoneal macrophages. Data are quantified from at least three independent experiments. SEPTEMBER 23, 2016 • VOLUME 291 • NUMBER 39

JOURNAL OF BIOLOGICAL CHEMISTRY 20467
phages was similar to that by untreated Egfr fl/fl peritoneal macrophages (Fig. 5B). These results suggest that factors other than HB-EGF may be produced in an EGFR-dependent manner in macrophages without IL-4 stimulation, for promoting tumor development. For example, IL-10 produced by TAMs has been shown to promote tumor development (7). Thus, more studies are needed to characterize factors produced by macrophages in EGFR-dependent and independent manners for tumorigenesis.
Because both LPS (29) and IL-4 stimulate EGFR transactivation in macrophages, it is interesting to identify whether the functions of EGFR activation are specific in M1 and M2 polarized macrophages. Both LPS (supplemental Fig. 3) and IL-4 stimulate HB-EGF release for EGFR transactivation. However, IL-4, but not LPS, up-regulates HB-EGF gene expression in macrophages. More importantly, IL-4-stimulated EGFR transactivation mediates inhibitory feedback for HB-EGF production. However, LPS-stimulated EGFR transactivation does not affect HB-EGF production (supplemental Fig. 3). We have shown that LPS-stimulated EGFR transactivation mediates inhibition of LPS-stimulated NF-B activation and proinflammatory and anti-inflammatory cytokine production in Raw 264.7 macrophages (29). Thus, EGFR activation may exert different functions in macrophages upon M1 and M2 stimulation.
We have reported that deletion of EGFR in macrophages leads to increase in the IL-10 level in response to inflammatory stimuli, such as LPS, which plays a role in suppressing proinflammatory cytokine production, resulting in protection of mice from intestinal inflammation colitis (29). Current studies show that blocking EGFR in macrophages up-regulates M2 polarization. Therefore, it should be noted that more studies are needed to elucidate the roles of EGFR activation in macrophages in inflammation-associated tumorigenesis. In fact, EGFR in macrophages in liver has shown tumor-promoting effects on hepatocellular carcinoma formation induced by diethylnitrosamine/phenobarbital (39). In this model, IL-1␤, released by DEN-damaged hepatocytes, stimulates Kupffer cells to produce IL-6. IL-6 is required for hepatocellular carcinoma formation. IL-1␤-induced IL-6 production in macrophages requires EGFR transactivation. Therefore, the roles of EGFR in macrophages in cancer development may be affected by complex interactions existing in the tumor microenvironment.
Because excessive EGFR signaling in tumor cells is known to promote tumorigenesis (19,20), EGFR is one of the key targets of the therapeutic strategy designed to treat cancers. Currently, two anti-EGFR monoclonal antibodies have been approved for the treatment of metastatic colorectal cancer (cetuximab and panitumumab) (40). However, therapies to inhibit the EGFR activity are not universally efficacious, suggesting possible resistance of EGFR inhibition in tumor cells or divergent roles of EGFR in other cell types involved in tumor development. Accordingly, we demonstrated that transactivation of EGFR in macrophages inhibits M2 polarization and exerts an anti-tumor effect. The decreased level of EGFR expression in macrophages might be associated with pro-tumor effects of TAMs in late stage of cancers. Our results are supported by the reported study that cetuximab increases activities of tumor-promoting M2 macrophages in the colorectal tumor microenvironment (41). Thus, our studies broaden the understanding of the mechanisms of EGFR signaling in tumor development by supporting the concept of cell type-specific (epithelial cell versus macrophage) EGFR activation in regulation of tumor establishment and progression. This concept represents a new direction to elucidate the mechanisms of EGFR signaling in tumorigenesis and assess therapeutic application of EGFR inhibitors.
In summary, these findings revealed previously unrecognized roles of EGFR signaling in regulation of M2 polarization in macrophages and demonstrate the association between the EGFR expression level in macrophages and the roles of TAMs in cancer development. These results should be taken into account for the application of anti-EGFR antibodies for cancer treatment.

FIGURE 7. IL-4-conditioned medium from macrophages with inhibition of EGFR expression exerts a higher level of effects on colonic tumor development.
Peritoneal macrophages isolated from Egfr fl/fl and Egfr fl/fl LysM Cre mice were treated with IL-4 (10 ng/ml) for 1 h to prepare conditioned media. IL-4 (10 ng/ml) was added to the media from untreated Egfr fl/fl and Egfr fl/fl LysM Cre macrophages as control media. IMCE ras cells were treated with IL-4conditioned and control media for 24 h (A). RNA was isolated for detecting the level of Snail expression using real-time PCR analysis. The expression level in IMCE ras cells treated with control media from Egfr fl/fl peritoneal macrophages was set as 100% for comparison with other groups. Rag2 Ϫ/Ϫ mice were inoculated with IMCE ras cells in the indicated conditioned media and DMEM medium. Tumor volumes were measured at the indicated days post implant (B). Tumor tissues were prepared for immunohistochemistry of Ki-67 (C). Brown nuclei represent Ki-67-positive staining. The percentage of Ki-67-positive staining cells was quantitatively analyzed (D). RNA was isolated from tumor tissues. The level of Snail gene expression was quantified using realtime PCR analysis. The average of the expression levels in the DMEM-treated group was set as 100% for comparison with other groups. *, p Ͻ 0.05 compared with WT control medium-treated and the DMEM-treated groups. #, p Ͻ 0.05, compared with the group treated with IL-4-conditioned media from Egfr fl/fl macrophages. In B, comparison was performed for data from the same inoculation day. Data are quantified from at least three independent experiments in A. n ϭ 3-4 in each group in B-E.

Experimental Procedures
Human Tissue Microarrays-A human gastric tissue microarray was accessed from the Human Tissue Acquisition and Pathology Shared Resources at Vanderbilt University Medical Center. There were 78 paraffin-embedded stomach tissue samples (8 normal mucosae and 70 tumors) in the gastric tissue microarray cores. All tissue samples were coded and de-identified in accordance with Institutional Review Board approved protocols. Clinical staging was evaluated according to the American Joint Committee on Cancer criteria. The adenocarcinomas were analyzed ranging from well to poorly differentiation (stages I to IV). Gastric tumors included a mix of intestinal-and diffuse-type tumors. The histology of all tissue samples was verified using H&E staining. The tissue microarray was used for immunohistochemistry to analyze EGFR expression in CD68-positive macrophages.
Animals and Treatment-All animal experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee at Vanderbilt University Medical Center (Nashville, TN). WT C57BL/6j and Rag2 ؊/؊ on C57BL/6j mice were from The Jackson Laboratory (Bar Harbor, ME). Egfr wa5 mice on a C57BL/6 background with a dominantnegative mutation in EGFR receptor kinase domain were obtained from Dr. David Threadgill (University of North Carolina, Chapel Hill). PCR primers specific for the EGFR sequence containing the relevant point mutation were used for genotyping.
Egfr fl/fl LysM-Cre mice were generated by crossing Egfr fl/fl mice, which harbor a floxed allele of EGFR on a C57BL/6j back-ground with homozygous LysM-Cre mice on a mixed C57BL/6J and C57BL/6N background in our laboratory (29). Gene mutation was confirmed by genotyping. EGFR expression in macrophages was tested using Western blot analysis.
Chitin (Sigma) was solubilized in PBS, as described before (4). Egfr fl/fl LysM-Cre and Egfr fl/fl mice were treated with chitin (800 ng in 100 l of PBS) intraperitoneally. Peritoneal macrophages were isolated 48 h after chitin administration.
Tumor Xenograft-1 ϫ 10 6 IMCE ras cells in conditioned medium and in DMEM medium containing IL-4 (10 ng/ml) were subcutaneously inoculated into the flank of 6 -8-week-old Rag2 Ϫ/Ϫ mice. Conditioned media were prepared from peritoneal macrophages isolated from the Egfr fl/fl LysM-Cre and Egfr fl/fl mice with and without IL-4 (10 ng/ml) treatment for 1 h. To exclude the potential effect of IL-4 in IL-4-conditioned medium on xenograft growth, IL-4 (10 ng/ml) was added to the conditional medium from untreated macrophages when inoculation as a control. Conditioned medium was injected into the xenograft again at 7 days after inoculation. Mice were euthanized 14 days after inoculation. Tumor volume was calculated by using the formula of tumor volume ϭ larger diameter ϫ (smaller diameter) 2 ϫ 0.5.
Cell Culture and Treatment-The mouse Raw 264.7 (ATCC TIB-71 TM ) monocyte/macrophage cell line and peritoneal macrophages from mice were cultured in DMEM medium containing 10% FBS, 1% glutamine, 100,000 IU/liter penicillin, and 100 mg/liter streptomycin at 37°C with 5% CO 2 . Peritoneal macrophages were isolated from mice and plated on cell culture dishes for 3 h. Unattached cells were removed by changing medium. For each experiment, peritoneal macrophages isolated from 4 -5 mice were mixed for treatments.
Macrophages were treated with IL-4 at 10 ng/ml (R&D Systems, Inc., Minneapolis, MN) in the presence or absence of a 1-h pretreatment of an EGFR-tyrosine kinase inhibitor, AG1478, at 450 nM (Calbiochem-EMD Millipore Corp., Billerica, MA), and the broad-spectrum MMP inhibitors GM6001 at 2.5 M (EMD Millipore) and TAPI-1 at 10 M (Enzo Life Sciences, Farmingdale, NY). Cells were collected for isolation of total cellular lysate and RNA. Culture supernatants were collected for ELISA and as conditioned media for treatment of gastric and colonic epithelial cells and xenograft. We adjusted the ratio of cell-to-medium at 5 ϫ 10 5 cells per 1 ml of medium.
Mouse conditionally immortalized stomach epithelial cells (ImSt) were isolated from the gastric epithelium of transgenic mice with a temperature-sensitive mutation of the simian virus 40 (SV40) large tumor antigen gene (tsA58) fused to the pro-moter of the mouse H-2K b class I gene (H-2K b -tsA58 mice) (42). The Egfr Ϫ/Ϫ ImSt cell line was generated from the stomach epithelium of EGFR-null mice crossed to the Immortomouse (43).
The Immorto-Min colonic epithelial (IMCE) cell line was generated from the colonic epithelium of F1 Immorto-Apc min/ϩ mouse hybrid (44). Thus, IMCE cells carry both the mutant Apc min gene and a temperature-sensitive mutant of the SV40 large T gene. The IMCE ras cell line was generated by overexpression of v-Ha-ras gene in IMCE cells (45).
Transient Transfection of HB-EGF siRNA-Raw 264.7 cells were transiently transfected with either 20 nM non-targeting siRNA or 20 nM mouse HB-EGF siRNA (Santa Cruz Biotechnology, Santa Cruz, CA) at 80% confluence using Lipofectamine 2000 (Invitrogen) for 6 h according to the manufacturer's instructions. Cells were then cultured for 48 h before treatment.
Colony Formation Assay-IMCE ras cells were plated in 12-well dishes (1000 cells/well) in 500 l of macrophage-conditioned medium with 5 units/ml of murine IFN-␥ and cultured at 33°C with 5% CO 2 for 14 days. Conditioned medium was changed every 4 days. Cells were stained at the end of the experiment using the CellTiterH120 AQueousOne Solution Cell Proliferation Assay (Promega, Madison, WI) according to the manufacturer's instructions. The area covered by colonies of Ͼ0.50 mm was measured (46).
ELISA-Raw 264.7 macrophages and peritoneal macrophages were treated with mouse IL-4 (10 ng/ml) in the presence or absence EGFR inhibitor AG1478 (450 nM) in 1 ml of medium for 1, 2, 8, and 24 h. The ratio of cell:medium was 5 ϫ 10 5 cells/1 ml of medium. Cell culture media were collected for determining the levels of HB-EGF using mouse HB-EGF ELISA kits (DuoSet ELISA Development System, R&D Systems, Inc.), according to the manufacturer's instructions. The HB-EGF concentration in the cell culture medium was calculated as pg/ml.
Cell Viability Assay-WT and Egfr Ϫ/Ϫ ImSt cells were plated in 96-well plates (5000 cells/well) and cultured overnight under permissive condition followed by culture in the starved medium under non-permissive condition for 8 -12 h. Then cells were treated with macrophage-conditioned media at 37°C for 24 h. Cell viability was assessed using the CellTiter120 AQueousOne Solution Cell Proliferation Assay (Promega) according to the manufacturer's instructions. The cell number standard curve was generated to determine the cell number in each experimental group.
Cellular Lysate Preparation and Western Blot Analysis-Total cellular lysates were prepared by solubilizing cells using cell lysis buffer containing 1% Triton X-100, 10 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, and protease and FIGURE 9. EGFR expression in macrophages is decreased in advanced stages of gastric cancer. The human gastric tissue array, including normal and stage I, II, III, and IV tumors, was prepared for H&E staining (A) and immunohistochemistry to detect macrophages using anti-CD68 and FITC-conjugated secondary (green) and EGFR expression using anti-EGFR and Cy3-conjugated secondary (red) (B). Nuclei were stained using DAPI (blue). Merged images are shown. Yellow and green arrows indicate macrophages with and without EGFR expression, respectively. The percentage of macrophages with EGFR expression was determined by counting the number of EGFR-expressing cells in CD68 positive cells (C). *, p Ͻ 0.05 compared with the normal and stage I groups.
Real-time PCR Analysis-Total RNA was isolated from cultured cells and xenograft tissues using a RNA isolation kit (Qiagen, Valencia, CA) and treated with RNase-free DNase. Reverse transcription was performed using the High Capacity cDNA Reverse Transcription kit and the 7300 Real Time PCR System (Applied Biosystems, Foster City, CA). The data were analyzed using Sequence Detection System V1.4.0 software. All primers were purchased from Applied Biosystems, including Arg1 (Mm00475988_ml), YM1 (Mm00657889_ml), FIZZ (Mm00445109_ml), Snail (Mm00441533_gl), vimentin (Mm01333430_ml), and HB-EGF (Mm00439306_ml). The relative abundance of GAPDH mRNA was used to normalize the levels of the mRNAs of interest. All cDNA samples were analyzed in triplicate.
Immunohistochemistry-Tissues sections were deparaffinized and rehydrated. Sections on human gastric tissue microarray were unmasked using EDTA solution and costained using mouse anti-human CD68 (Dako, Carpinteria, CA) and rabbit anti-human EGFR antibodies (RayBiotech, Inc). The sections were incubated sequentially with FITC-labeled anti-rabbit IgG followed by Cy3-labeled anti-mouse IgG (Jackson ImmunoResearch Laboratories, Inc. West Grove, PA) antibodies. Sections were then mounted using mounting medium containing DAPI (Vector laboratories, Inc. Burlingame, CA) for nuclear staining. Slides were observed under fluorescence microscopy. FITC, Cy3, and DAPI images were taken from the same field.
For Ki-67 staining, antigen retrieval was carried out by using antigen unmasking solution (Vector Laboratories). The sections were then incubated with a rabbit anti-Ki-67 monoclonal antibody (Biocare Medical, Concord, CA) at 4°C overnight followed incubation with a goat anti-rabbit polymer-HRP secondary antibody (Biocare Medical) for 1 h at room temperature. The sections were developed using the ImmPACT TM DAB substrate (Vector Laboratories). Sections were counterstained using hematoxylin and the observed light microscopy.
Statistical Analysis-Statistical significance was determined by one-way analysis of variance followed by Newman-Keuls analysis using Prism 5.0 (GraphPad Software, Inc. San Diego, CA) for multiple comparisons and t test for paired samples. A p value Ͻ0.05 was defined as statistically significant. Data are presented as the mean Ϯ S.E.