Platelet-derived Growth Factor-C (PDGF-C) Induces Anti-apoptotic Effects on Macrophages through Akt and Bad Phosphorylation*

Background: Aggressive breast tumor cells secrete platelet-derived growth factor C (PDGF-C). Results: PDGF-C prevents staurosporine-induced macrophage apoptosis through PI3K/Akt activation and Bad phosphorylation, resulting in the inhibition of caspase activation and PARP cleavage. Conclusion: PDGF-C has an anti-apoptotic effect on macrophages. Significance: PDGF-C secreted from malignant tumor cells could affect the survival of tumor-associated macrophages. PDGF-C, which is abundant in the malignant breast tumor microenvironment, plays an important role in cell growth and survival. Because tumor-associated macrophages (TAMs) contribute to cancer malignancy, macrophage survival mechanisms are an attractive area of research into controlling tumor progression. In this study, we investigated PDGF-C-mediated signaling pathways involved in anti-apoptotic effects in macrophages. We found that the human malignant breast cancer cell line MDA-MB-231 produced high quantities of PDGF-C, whereas benign MCF-7 cells did not. Recombinant PDGF-C induced PDGF receptor α chain phosphorylation, followed by Akt and Bad phosphorylation in THP-1-derived macrophages. MDA-MB-231 culture supernatants also activated macrophage PDGF-Rα. PDGF-C prevented staurosporine-induced macrophage apoptosis by inhibiting the activation of caspase-3, -7, -8, and -9 and cleavage of poly(ADP-ribose) polymerase. Finally, TAMs isolated from the PDGF-C knockdown murine breast cancer cell line 4T1 and PDGF-C knockdown MDA-MB-231-derived tumor mass showed higher rates of apoptosis than the respective WT controls. Collectively, our results suggest that tumor cell-derived PDGF-C enhances TAM survival, promoting tumor malignancy.

greater number of malignant cells can produce chemoattractants as well as survival signals for monocytes/macrophages; these leukocytes increase the levels of factors that enhance the aggressiveness of tumor cells.
To identify proteins that could affect TAMs, we reanalyzed previously published Gene Expression Omnibus profiles of the malignant human breast cancer cell line MDA-MB-231 and the benign breast cancer cell line MCF-7. PDGF-C synthesis was Ͼ40-fold higher in MDA-MB-231 than in MCF-7 cells. In preclinical models of human lung tumors, it was confirmed that PDGF-C recruits PDGFR-␣-positive tumor fibroblasts, thereby promoting tumor growth (29). Similarly, PDGF-C autocrine signaling has also been observed in the initiation and progression of brain tumors such as glioblastoma and medulloblastoma (30,31). Although PDGF-C production in invasive breast tumor patients is up-regulated (3), to date, no specific evidence regarding the role of PDGF-C on TAM is available.
In an effort to extend our understanding of the role of tumorderived PDGF-C in the tumor microenvironment, and particularly its effects on macrophages, we assessed PDGF-C-mediated signaling pathways in THP-1 macrophages. Here, we report that PDGF-C mediates anti-apoptotic effects through Akt/Bad phosphorylation, underlining the importance of the survival mechanisms of macrophages resident in tumors.
Human Peripheral Blood Mononuclear Cell Isolation and Differentiation-Human blood acquisitions were approved by the Institutional Review Board of Seoul National University, Korea (no. 1308-002-508). Documented written informed consents provided by Ethics Committee were obtained from all participants in this study. Human peripheral blood mononuclear cells were isolated from the buffy coats of normal donors on a Ficoll-Paque PLUS (GE Healthcare) gradient in accordance with standard procedures. Monocytes were purified from peripheral blood mononuclear cells by magnetic bead sorting using a human monocyte isolation kit II (Miltenyi Biotec, Bergisch Gladbach, Germany). Monocytes (Ͼ90% CD14 ϩ cells) were cultured at a density of 10 6 cells/ml in RPMI medium supplemented with 10% FBS, 2 mM L-glutamine, and 1% penicillin/streptomycin. The peripheral blood mononuclear cellderived monocytes were differentiated using medium containing 50 ng/ml GM-CSF (eBioscience, San Diego, CA) or 50 ng/ml M-CSF (eBioscience) to generate mature macrophages.
Flow Cytometric Analysis-THP-1 macrophages and human macrophages were examined for PDGF-R expression level using FACS. Resting day 3 and 5 THP-1 macrophages and differentiation day 5 and 7 human macrophages were fixed with 4% paraformaldehyde and permeabilized with BD Perm/ Wash TM buffer (BD Biosciences) before antibody staining. Flow cytometry of fixed and permeabilized cells was performed using anti-PDGF-R␣ and anti-PDGF-R␤ (Cell Signaling) primary antibodies, and Alexa Fluor 488 Donkey anti-rabbit IgG (Invitrogen) was used as secondary antibody followed by flow cytometry. To detect macrophage marker expression, the fixed human macrophages were incubated with fluorescence-conjugated anti-human CD11b, CD14, CD68, and CD163 antibodies (eBioscience). All flow cytometry was performed using LSRII Green (BD Biosciences).
Apoptosis Induction and Detection-THP-1 macrophages were treated with 200 nM staurosporin (STS, Cayman Chemicals) and incubated for 3-8 h. Externalization of phosphatidylserine to the outer layer of the cell membrane was examined using the annexin V-FITC apoptosis detection kit I (BD Biosciences). Cells were washed, suspended in the Annexin V binding buffer, and stained with FITC-conjugated annexin V antibody and 7-aminoactinomycin D for 15 min at room temperature. Flow cytometry was performed using LSRII Green (BD Biosciences). Data were analyzed using the FlowJo software (Tree Star, Ashland, OR).
MTT Assay-Cell viability of THP-1 macrophages at resting day 7 was assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assay. THP-1 cells (1 ϫ 10 6 /ml) were differentiated in a 12-well plate with 100 ng of PMA stimulation for 3 days and a further 7 days of resting after PMA removal. 100 l of MTT solution (5 mg/ml) in culture medium was added. Cells were then incubated further 4 h at 37°C. Thereafter, medium was discarded, and cells were lysed in 1 ml of dimethyl sulfoxide. The absorbance of the resulting solutions was read at a wavelength of 560 nm in a VICTOR TM X3 microplate reader (PerkinElmer Life Science).
Animal Experiments-All animal procedures were performed according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the Institution of Animal Care and Use Committee of Seoul National University of Korea (approval no. SNU-130923-1). The 4T1 murine breast cancer cell line was obtained from ATCC and main-tained in RPMI complete medium. The 4T1_PDGFC knockdown cell line and MDAMB231_PDGFC knockdown cell line were generated through shRNA lentiviral particle (Santa Cruz Biotechnology) infection followed by selection in 4 g/ml puromycin for 2 weeks. Six-week-old female BALB/c mice were purchased from Orient Bio (Korea) and maintained in pathogen-free housing. NOD-SCID female mice were kindly provided by Dr. J. G. Park (Seoul National University, Seoul, Korea) at 6 -7 weeks of age. For orthotopic implantation of tumor cells, mice were anesthetized with zoletil (Virbac)/xylazine (Bayer), and a total of 10 5 4T1 cells or 2 ϫ 10 6 MDA-MB-231 cells (suspended in 100 l ice cold PBS) were injected into the right inguinal mammary fat pad of BALB/c or NOD-SCID mice, respectively. Tumor volumes were calculated as 0.5 ϫ length ϫ width 2 . Necropsy was performed on 16 (Balb/c) or 28 days (NOD-SCID) after tumor cell implantation. Resected tumor mass was dissected into thin slices and incubated for 30 min with 2 mg/ml collagenase IV (Roche Applied Science) and Dnase I (Sigma Aldrich) at 37°C. Single cell suspensions were washed three times and fixed with 3% formalin followed by methanol until examination. Tunnel assay was performed using FlowTACS flow cytometry apoptosis detection kit (R&D Systems) according to the manufacturer's instructions. TAMs were first gated with anti-mouse F4/80 PE (eBioscience) and anti-mouse CD45 Qdot650 (eBioscience) double positive population, and the FITCpositive apoptotic cells were gated and analyzed.
Statistical Analysis-Statistical analysis was performed using the unpaired Student's t test with GraphPad Prism (version 5). Data are presented as means Ϯ S.D.

Reanalysis of Published MDA-MB-231 versus MCF-7
Microarray Results-To identify differences in the regulation of macrophages between malignant and benign tumors, we selected the two human breast cancer cell lines MDA-MB-231 (human malignant breast adenocarcinoma) and MCF-7 (human benign breast adenocarcinoma). We reanalyzed previously published microarray data of MDA-MB-231 and MCF-7 (GSM253207, GSM253208, GSM307014, GSM307015, GSM388187, GSM388191) in terms of differences in gene expression profiles between MDA-MB-231 and MCF-7. We identified 16 genes whose expression was more than 25-fold higher in MDA-MB-231 cells ( Table 1). Given that several proteins can influence macrophage behavior, we selected PDGF-C based on its known paracrine effects. Next, we determined whether various cancer cell lines, including MDA-MB-231, expressed PDGF-C.
PDGF-C Expression in Cancer Cell Lines-To assess the production and secretion of PDGF-C by various cancer cell lines, we collected culture supernatants from MDA-MB-231, MCF-7, MeWo, A549, and HT-29 cells after 48 h of incubation in serum-free medium. The PDGF-C contents of the supernatants were assayed by Western blotting under reducing and nonreducing conditions (Fig. 1). Reducing proteins of both the fulllength monomer of PDGF-C at 48 kDa and the processed GFD monomer (GFD-M) expression at 17 kDa were detected in only the MDA-MB-231 culture supernatant (Fig. 1A). Western blotting under non-reducing conditions confirmed the full-length dimer of PDGF-C (85 kDa), full-length monomer (48 kDa), and the GFD dimer (GFD-D) (26 kDa) (Fig. 1B). These results indicate that MDA-MB-231 breast cancer cells process PDGF-C into the biologically active GFD dimer form under these culture conditions. MCF-7, and the other cell lines secreted a very low level of the full-length dimer (Fig. 1B), highlighting that the metastatic breast cancer cell line MDA231 produces PDGF-C. The PDGF-C gene expression levels in the cancer cell lines were also compared using real-time PCR analysis. As expected, the PDGF-C gene expression levels in MDA-MB-231 were higher than those in the other cancer cell lines used (Fig. 1C). Furthermore, concentrated MDA-MB-231 culture medium induced PDGF-R␣ phosphorylation in THP-1 macrophages (Fig. 1D). Collectively, these data suggest that the malignant breast cancer cell line MDA-MB-231

Reanalysis of MDA-MB-231 vs. MCF-7 breast cancer cell line microarray data
Data uploaded from PubMed (GEO GSM253207, GSM307014, GSM388187, GSM253208, GSM307015, GSM388191) were used in this analysis. The HG U113 Plus 2.0 array (Affymetrix, Inc.) was used as the microarray platform. Expression levels of these 16 genes were Ͼ25-fold higher in the MDA-MB-231 cell line than in the MCF-7 cell line.

Gene
Gene  produced biologically active PDGF-C that could activate PDGF-R␣ on THP-1 macrophages.
Macrophage PDGF Receptor Expression-Before investigating PDGF-C-mediated signaling in macrophages, we examined macrophage surface PDGF receptor ␣ and ␤ expression using FACS (Fig. 2). THP-1 cells were differentiated with PMA (100 ng/ml) for 3 days and then were rested for a further 5 days in fresh medium without PMA ( Fig. 2A). Increased PDGF receptor ␣ expression along with macrophage differentiation was observed; similar results were also seen in the human peripheral blood mononuclear cell-derived macrophage differentiation process (Fig. 2B). These results indicated that mature human macrophages express surface PDGF-R and, thus, could respond to tumor-derived PDGF-C.
PDGF-C Phosphorylates PDGF-R␣, Akt, and Bad-The PDGF receptor is a trans-membrane tyrosine kinase receptor, the tyrosine kinase activity of which is encoded within the intracellular domain of PDGFRs (␣ and ␤). Proteolytically activated PDGF-C stimulates dimerization and phosphorylation of PDGFR-␣ but also activates PDGFR-␤ by inducing heterodimerization of PDGFR-␣␤ (10, 13). As expected, recombinant PDGF-C induced phosphorylation of PDGFR-␣, starting at 5 min after treatment in THP-1 macrophages (Fig. 3A). We also examined PDGFR-␤; however, this receptor showed little phosphorylation after PDGF-C treatment (data not shown). Next, we examined the known PDGFR-mediated PI3K/Akt signaling pathway. PDGF-C phosphorylated Akt at serine 473 at 5 min after treatment (Fig. 3B), and phosphorylation was completely inhibited by the PI3K inhibitor LY294002, indicating PI3K-dependent Akt activation (Fig. 3D). Among the downstream effects of PI3K/Akt, we detected Bad phosphorylation at Ser 112 from 10 min after PDGF-C stimulation (Fig. 3C).
LY294002 consistently inhibited PDGF-C-mediated Bad phosphorylation (Fig. 3E), suggesting that Bad phosphorylation is dependent on PI3K. Because growth factor activation of the PI3K/Akt signaling pathway culminates in the phosphorylation of Bad and promotes cell survival (32,33), these results prompted us to evaluate the anti-apoptotic effects of PDGF-C in macrophages.
Anti-apoptotic Effects of PDGF-C on Human Macrophages-Next, we determined whether PDGF-C could protect THP-1 macrophages from apoptosis, thus providing some evidence to explain the increased TAM density in breast cancer patients with a poor prognosis. We evaluated the ability of PDGF-C to protect THP-1 macrophages as well as primary human macrophages from STS-induced apoptosis. Macrophages were treated with 200 nM STS for 3 h with or without PDGF-C (100 ng/ml) pretreatment for 24 h (Fig. 4). Macrophage apoptosis was determined by surface annexin V staining using FACS. As expected, PDGF-C protected both THP-1 macrophages and human primary macrophages against STS-induced apoptosis (Fig. 4, A and B). The annexin V-FITC mean fluorescence intensity in the PDGF-C-pretreated cells before STS challenge was almost half that of the STS-only treated cells. Because PDGF-C also reduced the basal apoptotic rate of THP-1 macrophages, we investigated the influence of PDGF-C on the natural death of THP-1 macrophages. After PMA (100 ng/ml) stimulation for 3 days, THP-1 macrophages were further rested for an additional 7 days. The culture media were changed with fresh media with or without PDGF-C every other day. As shown in Fig. 4C, PDGF-C treatment increased the attached cell density at resting day 7. Incubation of THP-1 macrophages with PDGF-C led to a significant increase in cell viability as determined by the MTT assay (Fig. 4D). These results confirmed the  FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9

JOURNAL OF BIOLOGICAL CHEMISTRY 6229
protective effect of PDGF-C on THP-1 macrophages from apoptosis, resulting in enhanced survival.
PDGF-C Inhibits Caspase Activation in THP-1 Macrophages during STS-induced Apoptosis-Caspase activation and cleavage of PARP-1 play critical roles in the apoptosis process (34,35).
Collectively, these results show that PDGF-C enhances macrophage survival by inhibiting early apoptotic events.
Neutralization of PDGF-C Inhibits the Anti-apoptotic Effects of MDA-MB-231-conditioned Medium-To determine whether PDGF-C secreted from MDA-MB-231 cells can inhibit apoptosis, we investigated the anti-apoptotic effects of MDA-MB-231conditioned medium with or without PDGF-C neutralizing antibody on THP-1 macrophages by annexin V staining. As expected, apoptosis induced by STS was prevented by preincubation with MDA-MB-231-conditioned medium for 24 h (Fig.  6, A and B). PDGF-C neutralizing antibody again inhibited the anti-apoptotic effects of MDA-MB-231-conditioned medium, as evidenced by the increased annexin V-FITC mean fluorescence intensity values (Fig. 6B). Thus, our results provide evidence for the anti-apoptotic effects of MDA-MB-231-conditioned medium mediated by PDGF-C.

PDGF-C Knockdown in Breast Cancer Cells Could Not Prevent Macrophage Apoptosis in the Tumor Microenvironment-
Finally, we examined the effects of tumor cell-derived PDGF-C on macrophage apoptosis in the tumor environment using a mouse breast cancer model. We selected a 4T1 orthotopic syngeneic mouse model, because it can provide the most natural tumor immunological microenvironment. We established stable PDGF-C knockdown in 4T1 and MDA-MB-231 cell lines using shRNA viral vector and confirmed down-regulation of PDGF-C mRNA expression compared with expression in WT cells using real-time PCR (Fig. 7A). Cells were implanted into the mammary fat pad of female Balb/c mice or NOD-SCID mice, respectively. Surprisingly, the tumor mass size was persistently down-regulated in the 4T1 knockdown group (Fig.   FIGURE 5. PDGF-C treatment renders THP-1 macrophage cells resistant to STS-induced apoptosis by inhibiting caspase activation. THP-1 macrophages were pretreated with PDGF-C (100 ng/ml) for 6 and 24 h, respectively; apoptosis was then induced by 200 nM STS. Caspase-3, -7, -8, and -9 activation and PARP cleavage were analyzed by Western blotting.  FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9 7B), although no detectable difference in tumor cell growth rate was found compared with WT cells in vitro (data not shown). To examine apoptosis in TAMs, TUNEL staining was performed in single-cell isolates from the tumor mass, and additional F4/80 and CD45 staining was conducted to gate the macrophage population (Fig. 7C). Consistent with our in vitro data, macrophages from PDGF-C knockdown cell-implanted mouse tumors showed higher apoptotic rates compared with the control (Fig. 7D). We also confirmed similar results in the MDA-MB-231 xenograft model using NOD-SCID mice (Fig.  7D). Collectively, we showed that aggressive breast cancer cells prevent apoptosis of macrophages through PDGF-C secretion, and this interaction could affect tumor progression.

DISCUSSION
Although PDGF-C is known to be an important regulator of cell growth and apoptosis (36 -39), the present study is the first to demonstrate its anti-apoptotic effects on macrophages and the underlying mechanisms of action. Because macrophages constitute the major cell population present in tumors and contribute to disease progression (16,20,22), our findings provide important insights into malignant cell-derived PDGF-C production and macrophage survival in the tumor microenvironment.
TAMs produce a unique tumor microenvironment that can modify the neoplastic properties of both tumor and surrounding stromal cells. Thus, the relationship between TAMs and tumors has been extensively researched in the past. Based on accumulated data, it is generally thought that TAMs function mostly to promote tumor growth, metastasis, and angiogenesis (16,23,40). Along with its supposed tumor-promoting roles, the extent of TAM infiltration has been studied as a prognostic factor (40 -42). For example, in some human tumors, i.e., breast, thyroid, and esophageal cancers, the extent of macrophage infiltration was correlated with clinical aggressiveness (43)(44)(45). These findings have led to an evaluation of macrophage targeting in a tumor setting (46 -48). The fundamental principles of TAM-targeted anti-tumor approaches are based on inhibiting macrophage recruitment (49 -51), suppressing TAM survival (52)(53)(54), enhancing the anti-tumor activity of TAMs (55)(56)(57), and blocking their protumor activity (58 -60). Targeting TAMs is thus an important and efficient strategy for cancer therapy. However, the effectiveness of these approaches as a potential therapy should be further explored, as protumoral TAMs could be depleted, whereas anti-tumoral TAMs may not be affected.
Cytokines, chemokines, growth factors, and proteolytic enzymes from tumor cells and stromal cells are known to facilitate the recruitment of macrophages to tumor tissues (40,42,61). For example, overexpression of the chemoattractants CCL2, CCL5, IL-6, and VEGF was correlated with increased macrophage infiltration and poor prognosis in human cancers (62)(63)(64)(65); this finding has been a focus of clinical trials of antitumor drugs that have the potential to target TAMs (46). Actually, we found that up to 10ϳ20% of cells in the tumor mass comprised mature macrophages in mouse breast cancer models (Fig. 7). Intriguingly, there were fewer infiltrated macrophages in the PDGF-C knockdown cell-derived tumor mass (data not shown), a finding that is consistent with a reduced tumor growth rate as well as an up-regulated macrophage apoptotic rate. Because the tumor microenvironment might be insufficient for appropriate macrophage survival caused by hypoxia or nutrient competition with actively proliferating cells, it can be assumed that infiltrated macrophages easily undergo apoptosis. However, in the current study, we revealed the possible survival mechanisms of macrophages in relation to abundant PDGF-C from aggressive breast cancer cells.
PDGFs are classified as members of the superfamily of growth factors (10,39,66,67). Because PDGFs show high sequence identity with VEGFs, they are often referred to collectively as the PDGF/VEGF family (68). PDGFs are important in connective tissue growth and survival (10,12,69). Recently, the role of PDGFs in lymphocyte survival, via an autocrine regulating pathway, was investigated (70). Comparison of VEGFs and PDGFs and their receptor properties suggests that PDGFs may have an impact on TAM apoptosis. Based on this possibility, we hypothesized that PDGF-C could have anti-apoptotic effects on macrophages. Interestingly, the microarray data from up-regulated PDGF-C in MDA-MB-231 compared with MCF-7 were correlated with the tissue microarray analysis results of 216 patients with invasive malignant breast cancer (3). However, no experimental evidence to date has demonstrated a direct effect of PDGF-C on macrophages. In the present study, we demonstrated that the MDA-MB-231 human malignant breast adenocarcinoma cell line produced biologically active PDGF-C that could activate PDGFR-␣ on THP-1 macrophages. Concentrated MDA-MB-231 culture medium failed to inhibit STS-induced apoptosis in the presence of PDGF-C neutralizing antibodies (Fig. 6). Therefore, PDGF-C might represent a potential anti-cancer therapy target to modulate macrophage survival rates.
The PI3K/Akt signaling pathway plays a critical role in various cellular events such as apoptosis, cell cycle progression, and transcriptional regulation (71)(72)(73). The ability of Akt to prevent apoptosis in cells through phosphorylation and inhibition of proapoptotic mediators such as Bad and caspase-9 has been well documented (74). Hu et al. (75) demonstrated that activation of the PI3K pathway by PDGFRs promotes actin reorganization, cell movement, cell growth, and inhibition of apoptosis. PI3K signaling by PDGFs both directly and indirectly regulates the apoptotic pathway through the action of effectors, including serine/threonine kinases such as Akt/PKB (76,77). Akt activation is important for cell survival through regulation of multiple target pathways by phosphorylation of critical proteins. For example, phosphorylation of Bad at Ser 112 by Akt induces a conformational change that blocks the ability of Bad to interact with anti-apoptotic Bcl-2 proteins (78). These free anti-apoptotic Bcl-2 proteins could then inhibit Bax-triggered apoptosis by maintaining the integrity of the outer mitochondrial membrane (78 -80). Mitochondria play important roles in the regulation and transmission of apoptotic signals, which are regulated by the balance among Bcl-2 family members; therefore, phosphorylation of Bad at Ser 112 promotes cell survival (32,74,81). In the present study, we showed that the PDGF-C-mediated anti-apoptotic signaling pathway was mediated via PI3K/ Akt activation and proapoptotic Bad phosphorylation at Ser 112 , culminating in Bad inactivation of THP-1 macrophages. Subsequently, phosphorylated Bad inhibited caspase activation and PARP cleavage (Fig. 4). However, Weisser et al. (82) reported that PI3K could activate the M2 skewing of macrophages through increasing STAT6 activity. Therefore, PI3K activation by PDGF-C led to an increase in immune-suppressive TAM via an anti-apoptotic mechanism.
Taken together, our results demonstrate the anti-apoptotic effect of PDGF-C on macrophages and clarify the signaling pathway of PDGF-C-mediated receptor ␣ phosphorylation, followed by PI3K/Akt activation and Bad inactivation. These events, in turn, lead to inhibition of caspase activation and PARP cleavage (Fig. 8). These findings support a therapeutic role for tumor-derived PDGF-C that is mediated by its promotion of TAM survival.