CXCL17-derived CD11b⁺Gr-1⁺ myeloid-derived suppressor cells contribute to lung metastasis of breast cancer through platelet-derived growth factor-BB

All mice procedures were conducted and approved in accordance with the Institutional Animal Care and Use Committee at Kaohsiung Medical University. BALB/c mice (female, 5-week-old) were treated with rmCXCL17 for 14 days (1 μg/mouse, 2 times/week, n = 6 per group) by intra-tracheal route, and 4T1 cells (500,000 cells per fat pad) were implanted into the mammary fat pads and allowed to spontaneously metastasize to the lung for 24 days. For lung metastasis of a human breast cancer model, athymic nude mice (6-week-old females, n = 6 per group) were treated with rmCXCL17 protein (1 μg/mouse, 2 times/week) for 14 days, then human MDA-MB-231 or MDA-MB-231-RFP-Luc cells were injected via the tail vein for the indicated times (90 days for lung metastasis and 48 h for extravasation). Tumors in primary sites and the lungs were harvested after the indicated times, fixed and embedded in paraffin, sectioned, and immunostained with antibodies against CD31 antibodies (Catalog #ab28364, dilution 1:100). All lung nodules in both lung lobes were calculated and tumor nodules occurring in the same site of sequential sections were only calculated once. Quantitative studies of stained sections were performed independently by three researchers in blinded fashion. The total number of tumor nodules per whole lung lobe was counted and averaged among the animals of each group. Quantitative analyses of CD31 staining were performed with the image analysis program ImageJ. Assays were carried out three times, and three random fields per sample were analyzed in five high-power fields (100 magnification [10 objective and 10 ocular]). For L4T1 generation, 4T1 cells were transplanted into BALB/c mice from mammary fat pads. The animals were then sacrificed on day 24, and the 4T1 cells in the lungs were isolated by mincing, collagenase type I digestion, and filtering. 4T1 cells were cultured in a growth medium containing 10% fetal bovine serum and 1% penicillin–streptomycin and expanded for second-round transplantation. The 4T1 sub-line of transplantation was designated as L4T1. For MDSC depletion studies, mice were treated with isotype or anti-Gr-1 antibody (Bio X Cell, West Lebanon, NH) at 0.25 mg/mouse intraperitoneal injections every 4th day, and 4T1 cells were injected via tail vein into the mice (six per group). The control group received intraperitoneal injections of purified rat immunoglobulins. For PDGF receptor inhibition studies, mice were treated with imatinib mesylate (Sigma-Aldrich) (100 mg/kg) by oral gavage three times weekly, and 4T1 cells were transplanted into mice from mammary fat pads (n = 6 per group). The control group received oral gavage of vehicle control (10% ethanol and 90% corn oil).

Murine lung tissue was dispersed using 2% collagenase A and 0.75% DNase I (Roche Diagnostics GmbH, Germany) in RPMI1640 medium supplemented with 10% FBS at 20 °C for 1 h. CD11b⁺ F4/80⁺ cells were isolated from the lungs of mice by magnetic cell sorting using mouse CD11b and F4/80 antibodies conjugated with magnetic beads. CD11b⁺Gr-1⁺ and CD11b⁺Gr-1⁻ cells were isolated using MDSC Isolation Kit (MACS, Miltenyi Biotec) according to the manufacturer’s instructions.

Microarray experiment procedures were carried out following the manufacturer’s protocols. Total RNA was amplified by an Agilent Quick Amp Labeling Kit (Agilent Technologies, USA). Cy-labeled cRNA (0.3 μg) was cleaved to an average size of about 50–100 nucleotides by incubation with fragmentation buffer (Agilent Technologies) at 60 °C for 30 min. Equal Cy-labeled cRNA was pooled and hybridized to Agilent SurePrint G3 Mouse GE 8x60K Microarray (Agilent Technologies, USA), then scanned by an Agilent microarray scanner (Agilent) at 535 nm for Cy3 and 625 nm for Cy5. Scanned images were analyzed by Feature Extraction software 10.5 (Agilent Technologies), and image analysis and normalization software was used to quantify signal and background intensity for each feature, which substantially normalized the data using the rank-consistency-filtering LOWESS method.

The levels of angiogenic factors were determined by Luminex Assays (R&D Systems). CXCL17 levels were assessed by human or mouse CXCL17 ELISA kits (Cusabio Biotech, Wuhan, China).

C166 (4 × 10⁵ cells/well) were then plated on reduced growth factor Matrigel (200 μL/well) (BD Biosciences, San Josè, CA) in 48-well plates and treated with conditioned medium (CM) (50%) of CD11b⁺Gr-1⁺ MDSCs isolated from normal mice for 3 h. Photographs were taken using a Nikon fluorescence microscope. The total tube area was quantified as mean pixel density obtained from image analysis of three random microscopic fields using ImageJ software.

Cell migration assays were conducted using the 3- or 8-μm inserts (EMD Millipore). PKH26-labeled CD11b⁺Gr-1⁺ or 4T1 cells (1 × 10⁵ cells/well) were seeded onto inserts, and CXCL17 (1 or 10 ng/ml in RPMI1640 medium containing 1% FBS) or CMs of CD11b⁺Gr-1⁺ MDSCs were added to the bottom wells for 24 h as chemoattractant. Migratory cells were counted using a fluorescence microscope. For transendothelial migration, C166 were seeded onto inserts with polyester membranes of 3 (for MDSCs) or 8 μm (for 4T1 cells) pore size (EMD Millipore) and cultured for 2 days to allow a confluent monolayer to form. PKH26-labled CD11b⁺Gr-1⁺ MDSCs or 4T1 cells were seeded onto C166-coated inserts for 24 h. CMs of CD11b⁺Gr-1⁺ MDSCs isolated from normal or 4T1-bearing mice were placed at the bottom wells to be acted as chemoattractant, and the migratory cells were made visible using a fluorescence microscope. For colony formation, 4T1 cells (1 × 10³) were seeded into a 60-mm dish and then treated with CM of CD11b⁺Gr-1⁺ MDSCs isolated from normal or 4T1-bearing mice (50%). After 7 days, the colonies were fixed by formalin (1%), then stained by crystal violet.

The SurvExpress contain 189 (Sotiriou Van de Vijver Breast GSE2990) and 327 (Kao Hung Breasr GSE20685) breast cancer patients with follow-up time intervals [20]. The data were used to estimate the prognostic significance of the CXCL17 transcript for metastasis curve, while Kaplan-Meier plotter database was used to evaluate distant metastasis-free survival [21].

Data were expressed as mean ± standard deviation (SD) or standard error of mean (SEM). Two treatment groups were compared by Student’s t test. Multiple group comparisons were performed by two-way analysis of variance with Tukey’s post hoc test. Metastatic quantifications were assessed with a Mann-Whitney U test. GraphPad Prism version 7.04 (GraphPad Inc., La Jolla, CA) was used for statistical analyses. Results were considered statistically significant when P < 0.05.

To identify specific candidate secretory factors that mediate breast cancer metastasis by shaping lung niches, we found upregulated genes only in lung metastatic 4T1 cells, but not in lymph, intestinal, and liver 4T1 cells (Fig. 1a, b). Thirty-three secretory factors were found in lung metastatic 4T1cells (Table 1). Among these genes, we focused on CXCL17, which is upregulated at the highest fold (24.092-fold) of lung metastatic 4T1 cells, compared to the primary 4T1 in the fat pads of mice. The ELISA also confirmed that CXCL17 was enhanced at protein level in lung metastatic 4T1 (L4T1) cells, compared to 4T1 cells in tumors of primary sites (Fig. 1c). We also assessed the association between CXCL17 expression and metastasis in breast cancer via the online databases SurvExpress and KM plotter (PMID: 24066126 and PMID: 20020197). In two independent datasets from SurvExpress (PMID: 21501481 and PMID: 11823860), distant metastasis-free survival revealed that the high-risk group expressed high levels of CXCL17 and had shorter time to metastasis in both datasets (Fig. 1d, e). Similar results were shown in Kaplan-Meier survival analysis (Fig. 1f). The results suggest that CXCL17 might be associated with metastasis in breast cancer.

To assess whether CXCL17 promotes lung metastasis through metastatic niche establishment, we treated mice with rm (recombinant mouse protein) CXCL17 via intra-tracheal administration and then implanted breast cancer cells by orthotropic graft or tail vein injection. Compared to control mice, rmCXCL17 treatment increased the numbers of tumor nodules in the lungs of mice that had received tumor implantation in an orthotropic model (Fig. 2a, b). In a second approach, we implanted human breast MAD-MB-231 cells into mice by tail vein injection after rmCXCL17 treatment for 2 weeks. Compared to control mice, more and larger tumor nodules were observed in the lungs of mice pre-treated with rmCXCL17 (Fig. 2c, d). Histological analysis revealed that mice treated with rmCXCL17 had a higher number of metastatic foci, and with larger sizes, compared with mice injected with PBS (Fig. 2d). Moreover, knockdown of CXCL17 decreased the spontaneous metastatic ability of L4T1 cells (Fig. 2e, f). However, CXCL17 did not change cell proliferation, colony formation, and migration of 4T1 breast cancer cells in vitro (Additional file 1: Figure S1A to S1C). These data indicate that CXCL17 might contribute to the supportive metastatic niche formation rather than cancer cell growth or mobility regulation, for breast cancer spreading.

Given that soluble factor CXCL17 is mainly responsible for recruiting immune cells in the lung [22], a critical step for the formation of metastatic niche favored by the cancer [5, 23], so we speculated that CXCL17 might function through remodeling of the lung microenvironment in a paracrine manner. Intra-tracheal administration of rmCXCL17 increased the infiltration of CD11b⁺Gr-1⁺ MDSCs in the lungs of mice, but CD11b⁺Gr-1⁻ MDSCs or macrophages (CD11b⁺F4/80⁺) did not (Fig. 3a–c). CXCL17 also enhanced basal and transendothelial migration of CD11b⁺Gr-1⁺ MDSCs isolated from mice in vitro (Fig. 3d, e). The inhibitor of GPR35 (G Protein-Coupled Receptor 35), a receptor of CXCL17, prevented the stimulatory effect of CXCL17 on the enhancement of CD11b⁺Gr-1⁺ MDSCs basal and transendothelial migration (Fig. 3f, g), indicating that CXCL17 might functionally mediate the inhibition of anti-cancer immunity of the lungs in mice via a GPR35-dependent manner.

Increased angiogenesis in the metastatic niche is considered a crucial event for dissemination of cancer cells invading distant organs [23, 24], and MDSCs have been implicated in orchestrating aberrant angiogenesis in metastatic niches of various cancers [25]. IHC staining of lungs of CXCL17-treated mice revealed increased CD31⁺ cells in the lungs of mice (Fig. 4a). Tube formation analysis shows that the conditioned medium (CM) of CD11b⁺Gr-1⁺ MDSCs isolated from the lungs of CXCL17-treated mice enhanced tube formation in mouse endothelial C166 cells compared to the CM of CD11b⁺Gr-1⁺ MDSCs isolated from the lungs of control mice (Fig. 4b). High-throughput screening by a Luminex system identified increased expressions of PDGF-BB expression in CD11b⁺Gr-1⁺ MDSCs isolated from lungs of CXCL17-treated mice in vivo, compared to the CD11b⁺Gr-1⁺ MDSCs isolated from the lungs of control mice. There were increased trends in the expressions of PDGF-AA, VEGF-A, and EGF basic, although they did not reach statistical significance (Fig. 4c–f). rmCXCL17 increased the expression of PDGF-BB in CD11b⁺Gr-1⁺ MDSCs isolated from lungs of normal mice in situ (Fig. 4g). Inhibitor of PDGFR-β, a specific receptor for PDGF-BB, partially decreased the stimulatory effects of CXCL17-treated CD11b⁺Gr-1⁺ MDSC’s CM in tube formation of C166 cells, revealing that MDSC-derived PDGF-BB is the mediator of angiogenesis in lung metastatic niches (Fig. 4h).

Next, we determined if CXCL17 would also facilitate breast cancer cell extravasation and long-term lung colonization. MDA-MB-231 cells expressing RFP and luciferase were injected into the tail veins of mice after CXCL17 treatment. A significant increase of MDA-MB-231 cells was observed in the lungs of mice 48 h after implantation, showing extravasation of cancer cells into the lung (Fig. 5a). Consistently, more tumor nodules and increased size of tumors were found in the lungs of CXCL17 pre-treated mice compared with those of control mice (Fig. 2c, d). These results indicate that CXCL17 provides metastatic supporting niches for the distant spread of breast cancer cells.

Given the prominent role of CD11b⁺Gr-1⁺ MDSCs in breast cancer metastasizing to the lungs, we next examined whether depletion of CD11b⁺Gr-1⁺ MDSCs could decrease lung metastasis of breast cancer induced by CXCL17. Mouse anti-Gr-1 antibody (0.25 mg/mouse, MDSC depletion) was administrated by IP every 4 days during CXCL17 pretreatment in mice prior to PKH26-labeled 4T1 cancer cell implantation by tail vein injection (Fig. 5b). A significant decrease of 4T1 cancer cells was observed in the lungs of mice after 48-h implantation, showing that extravasation of the 4T1 cells into the lungs was reduced by MDSC depletion (Fig. 5c). Consistently, there were fewer metastatic foci and decreased tumor sizes in the lungs of anti-Gr-1 antibody treated, when compared with isotype treated, CXCL17 pre-treated mice (Fig. 5d, e).

Because PDGF-BB has been implicated as being an oncogenic factor for metastasis in various cancers [26, 27], and CXCL17 provokes CD11b⁺Gr-1⁺ MDSCs to express high levels of PDGF-BB, we investigated the impact of CD11b⁺Gr-1⁺ MDSC-derived PDGF-BB in cancer extravasation, as determined by transendothelial migration analysis. We found that the CM of CD11b⁺Gr-1⁺ MDSC isolated from the lungs of CXCL17-treated mice increased the transendothelial migration of 4T1 cells (Fig. 6a). In addition, elevated colony formation of 4T1 cultured in CM of CD11b⁺Gr-1⁺ MDSCs isolated from CXCL17-treated mice further supported that CXCL17-conducted CD11b⁺Gr-1⁺ MDSC increased 4T1 cell survival (Fig. 6b). Treatment of 4T1 cells with rmPDGR-BB enhanced transendothelial migration and colony formation (Fig. 6c, d). PDGFR inhibitor prevented transendothelial migration and colony formation of 4T cells induced by CM of CD11b⁺Gr-1⁺ MDSC (Fig. 6e, f), indicating that CD11b⁺Gr-1⁺ MDSC may support breast cancer extravasation and survival by releasing PDGF-BB. Furthermore, PDGFR inhibitor imatinib decreased metastatic foci and decreased tumor sizes in the lungs of mice, when compared with isotype-treated and CXCL17 pre-treated mice (Fig. 6g, h).