Purpose:

The different prognostic values of tumor-infiltrating neutrophils (TIN) in different tissue compartments are unknown. In this study, we investigated their different prognostic roles and the underlying mechanism.

Experimental Design: We evaluated CD66b+ neutrophils in primary tumors from 341 patients with breast cancer from Sun Yat-sen Memorial Hospital by IHC. The association between stromal and parenchymal neutrophil counts and clinical outcomes was assessed in a training set (170 samples), validated in an internal validation set (171 samples), and further confirmed in an external validation set (105 samples). In addition, we isolated TINs from clinical samples and screened the cytokine profile by antibody microarray. The interaction between neutrophils and tumor cells was investigated in transwell and 3D Matrigel coculture systems. The therapeutic potential of indicated cytokines was evaluated in tumor-bearing immunocompetent mice.

Results:

We observed that the neutrophils in tumor parenchyma, rather than those in stroma, were an independent poor prognostic factor in the training [HR = 5.00, 95% confidence interval (CI): 2.88–8.68, P < 0.001], internal validation (HR = 3.56, 95% CI: 2.07–6.14, P < 0.001), and external validation set (HR = 5.07, 95% CI: 2.27–11.33, P < 0.001). The mechanistic study revealed that neutrophils induced breast cancer epithelial–mesenchymal transition (EMT) via tissue inhibitor of matrix metalloprotease (TIMP-1). Reciprocally, breast cancer cells undergoing EMT enhanced neutrophils' TIMP-1 secretion by CD90 in a cell-contact manner. In vivo, TIMP-1 neutralization or CD90 blockade significantly reduced metastasis. More importantly, TIMP-1 and CD90 were positively correlated in breast cancer (r2 = 0.6079; P < 0.001) and associated with poor prognosis of patients.

Conclusions:

Our findings unravel a location-dictated interaction between tumor cells and neutrophils and provide a rationale for new antimetastasis treatments.

Translational Relevance

We found parenchymal neutrophils, but not the stromal ones, were an independent risk factor in breast cancer, especially the triple-negative subtype with poor prognosis due to the lack of targeted therapy. Furthermore, we uncovered that tumor-contacted neutrophils promote metastasis by a CD90-TIMP-1 juxtacrine–paracrine cycle. This study enhances the accurate prognosis prediction and provides new therapeutic strategies for patients with breast cancer.

Neutrophils are the most prevalent white blood cells in human blood and constitute a significant proportion of inflammatory cells in tumor microenvironment of various types of malignancies (1–3). In mouse models, their role in cancer metastasis is controversial. It has been reported that neutrophils mediated by G-CSF suppress antitumor immunity and promote tumor colonization of distant organs (4, 5). In contrast, it has been shown that tumor-entrained neutrophils inhibit lung metastasis (6) and mAb-induced tumor reduction is abolished in mice depleted of neutrophils (7). These contradictory observations highlight the diversity of tumor-associated neutrophils and a pressing need to further evaluate their function in a more patient-relevant scenario (8, 9).

In patients with breast cancer, elevated neutrophil-to-lymphocyte ratio (NLR) in peripheral blood has been associated with poor prognosis, particularly in triple-negative subtype, which is most difficult to treat due to the lack of endocrine and target therapies (10, 11). However, the clinical significance of neutrophils in tumors is less studied. Furthermore, the different prognostic values of tumor-infiltrating neutrophils (TIN) in different tissue compartments are largely unknown.

Epithelial–mesenchymal transition (EMT) plays a crucial role in tumor metastasis (12, 13). It has been well documented that EMT endows tumor cells with enhanced motility and invasiveness (14). In addition, we and others demonstrated that tumor cells undergoing EMT foster a permissive microenvironment, which reciprocally promotes EMT and metastasis by suppressing T-cell functions (15), recruiting monocytes (16), and skewing macrophage polarization to protumor phenotypes (17–19). Therefore, the interaction between EMT and tumor microenvironment exerts profound effects on tumor dissemination. Understanding its complex cellular and molecular mechanisms is important for designing more effective antimetastasis treatments (20, 21).

Here, we investigated the clinical values of neutrophils as prognostic biomarkers in tumor parenchyma and stroma, respectively. In addition, we studied the functions and cytokine profiles of primary TINs isolated from clinical samples. Moreover, we explored the underlying mechanism in the Matrigel coculture system and in vivo models.

Patients and specimens

We obtained 446 tissue samples from patients with stage I–III primary breast carcinomas for this study. The patients were enrolled from two independent breast cancer centers. A total of 341 patients enrolled between January 21, 2003 and December 8, 2011 from Sun Yat-sen Memorial Hospital (Guangzhou, China) were randomized into the training cohort (n = 170) and internal validation cohort (n = 171). The independent cohort consisted of 105 patients enrolled between January 20, 2008 and July 3, 2012 from the First Affiliated Hospital of Shantou University (Shantou, China). The last follow-up time was February 10, 2017. This study was approved by the Clinical Research Ethics Committee of both hospitals. Written informed consents were obtained from all patients before recruitment. All procedures performed in this study were in accordance with the ethical standards of the institutional research committee and the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

IHC

Formalin-fixed and paraffin-embedded surgical specimens or xenografts, liver, and lung tissues from animal experiments were sectioned at 4 μm and used for IHC staining. Antigen retrieval was performed using a pressure cooker for 9 minutes in antigen unmasking solution (ETDA, pH 8.0), followed by incubated in 3% H2O2 for 10 minutes and blocked with PBS containing 5% BSA for 15 minutes. Afterward, the samples were incubated with primary antibodies against human CD66b (1:100, catalog no. 555723, BD Biosciences), mouse tissue inhibitor of matrix metalloprotease (TIMP-1; 10 μg/mL, catalog no. AF-980, R&D Systems), or mouse CD90 (1:50, catalog no.105312, BioLegend) overnight at 4°C, followed by a secondary antibody (HRP rabbit/mouse, GTVision III, GeneTech) for 1 hour at room temperature. Then, immunodetection was performed by using DAB (Dako) according to the manufacturer's instructions. Images were obtained by microscopy (BX51WI, Olympus).

Isolation and culture of human neutrophils

Primary TINs were isolated from fresh invasive ductal carcinoma samples obtained from surgery as described previously with slight modifications (17, 22). Briefly, fresh tissues were minced into small pieces on ice and digested by collagenase type I, collagenase type III, and hyaluronidase (1.5 mg/mL, Sigma Aldrich) at 37°C with agitation for 30 minutes in DMEM with 10% FBS. The cell suspension was collected by filtering through a 70 μm cell strainer and the isolation of neutrophils was performed by a magnetic-activated cell sorting using direct CD66b Isolation Kit (catalog no. 130-104-913, Miltenyi Biotec) according to the manufacturer's instructions (23). In addition, autologous neutrophils were isolated from EDTA-anticoagulated peripheral blood of the same patients using positive selection of CD66b+ cells with magnetic beads. Remaining erythrocytes were lysed with Aqua Braun (B. Braun). The isolated neutrophils were cultured in DMEM supplemented with 10% FBS at 37°C in 24-well culture plates (Costar, Corning). In some experiments, neutrophils were seeded at 2 × 105 cells/well. The condition media were collected after 24 hours and centrifuged at 3,000 rpm for 15 minutes at 4°C to remove the cell debris. All samples were collected from the donors with informed consent, and all related procedures were performed with the approval of the internal review and ethics boards of Sun Yat-sen Memorial Hospital.

Giemsa staining

The morphology of neutrophils was evaluated by Giemsa Staining. The isolated neutrophils were smeared onto a slide and stained in commercial Giemsa dye (catalog no. G4486, GBCBIO). After 30 minutes, the dye was poured and smear was washed with water. Images were obtained by microscopy (BX51WI, Olympus).

Culture of cancer cells

Human breast cancer cell lines MCF-7, MDA-MB-231, and BT-474 and mouse breast cancer cell line 4T1 were obtained from ATCC and cultured according to standard protocols. For the induction of EMT of breast cancer cells, MCF-7 cells were cultured alone, treated with 30% condition media of primary TINs or 10 ng/mL rhTIMP-1 (catalog no. 410-01-10, PeproTech) for five days. The culture media were changed every two days. In some experiments, condition media of TINs were pretreated with 10 μg/mL neutralizing antibodies against TIMP-1 (catalog no. AF-970, R&D Systems), GM-CSF (catalog no. MAB215, R&D Systems), angiogenin (catalog no. AB-265-NA, R&D Systems), or CCL2 (catalog no. MAB679, R&D Systems) for 30 minutes at 37°C.

Coculture of cancer cells and neutrophils

Noncontact transwell system.

A total of 2 × 104 breast cancer cells were seeded in the top chamber and 1 × 105 neutrophils were seeded in the bottom chamber of 24-well Boyden chambers with 0.4-μm pore size (Corning Incorporated). After 24 hours, the top chambers were removed. The neutrophils in the bottom chambers were washed by PBS and cultured in fresh complete media for another 12 hours. Afterward, the culture media of neutrophils were collected for subsequent ELISA assay.

Mixed coculture system.

A total of 2 × 104 breast cancer cells and 1 × 105 neutrophils were mixed together and seeded in 24-well culture plate. Neutralizing antibodies (20 μg/mL) against Frizzled2 [catalog no. MOB-2608z-S(P), Creative Biolabs], TMPRSS4 (catalog no. 33R-5428, Fitzgerald Industries International), CD90 (catalog no. GTX88569-PEP, GENETEX), EphA4 (catalog no. LS-E9103, LifeSpan BioSciences), and Mac-1 (catalog no.101201, BioLegend) were added in the coculture. After 24 hours, neutrophils were isolated by CD66b magnetic-activated cell sorting (catalog no. 130-104-913, Miltenyi Biotec) and cultured in fresh complete media for another 12 hours. Afterward, the culture media of neutrophils were collected for subsequent ELISA assay.

Three-dimensional coculture.

Cells were seeded within the 8-well chamber slides that were precoated with growth factor–reduced Matrigel (catalog no. 354230, BD Biosciences). A total of 2 × 103 MCF-7 cells and neutrophils were mixed in a ratio of 1:5 and suspended in the 400 μL DMEM + 10% FBS + 2% Matrigel. Afterward, the mixed cells were seeded in the chamber slides and cultured for 2 days.

Flow cytometry

Neutrophils were resuspended in PBS containing 1% FBS and stained with fluorescent-conjugated antibodies against CD66b (catalog no. 130-104-395, Miltenyi Biotec), CD11b (catalog no. 11-0118-41, eBioscience), CD90 (catalog no. 328108, BioLegend) for 30 minutes at 4°C or propidium iodide solution (catalog no. BMS500PI, eBioscience) according to the manufacturer's instructions. During the analysis of cytometry data, cells were first gated on the basis of forward (FSC-A) and side (SSC-A) scatters to exclude cell debris.

Immunofluorescent staining

Paraffin-embedded samples were sectioned at 4-μm thickness. Antigen retrieval was performed by a pressure cooker for 15 minutes in 0.01 mol/L citrate buffer (pH 6.0). Then, sections were blocked in PBS containing 10% donkey serum or 2% BSA for 1 hour at room temperature. Cells for immunofluorescent staining were fixed by 4% paraformaldehyde for 15 minutes at room temperature, washed with PBS, and permeabilized with 0.2% Triton X-100 in PBS for 15 minutes. Thereafter, cells were blocked in PBS with 2% BSA for 1 hour at room temperature. After blocking, samples were incubated with primary antibodies against E-cadherin (1:100, catalog no. Sc-7870, Santa Cruz Biotechnology), Vimentin (1:100, catalog no. Sc-66002, Santa Cruz Biotechnology), CD90 (1:50, catalog no. 328108, BioLegend), cytokeratin (1:100, catalog no. ZM-0069, Zsbio), CD66b (1:100, catalog no.555723, BD Biosciences), and TIMP-1(10 μg/mL, catalog no. ZAa0429, Zsbio) overnight at 4°C. The slides were then incubated in Alexa Fluor–conjugated secondary antibodies (Invitrogen) for 1 hour at room temperature. DAPI was used for counterstaining the nuclei. The images were obtained by laser scanning confocal microscopy (LSM780, Zeiss).

Migration and invasion assay

Migration and invasion assay were performed in 8-μm 24-well Boyden chambers (Corning). For invasion assay, Matrigel (catalog no. 354230, BD Biosciences) diluted with DMEM in 1:3 ratio was precoated in the chambers and solidified for 30 minutes in 37°C. A total of 5 × 104 breast cancer cells suspended in 100 μL 0.2% BSA DMEM were added in the top chamber, whereas 2 × 105 primary TINs or autologous peripheral blood neutrophils in 600 μL 5% FBS DMEM were added to the bottom chambers. The cancer cells migrated and invaded for 10 hours and 24 hours, respectively. Afterward, cancer cells in the top chamber were removed by swab and the cells attached in the underside of chambers were fixed by 4% paraformaldehyde for 15 minutes and then stained by 0.5% crystal violet. Quantification was performed by the mean number of cells in five microscopic fields per chamber. To offset the influence of proliferation, we pretreated tumor cells with 10 μg/mL mitomycin C (catalog no. S8146, Selleck Chemicals) to inhibit cell proliferation. In some experiments, TINs in the bottom chamber were pretreated with 10 μg/mL neutralizing antibody against TIMP-1 (catalog no. AF-970, R&D Systems), GM-CSF (catalog no. MAB215, R&D Systems), angiogenin (catalog no. AB-265-NA, R&D Systems), or CCL2 (catalog no. MAB-679, R&D Systems) for 30 minutes at 37°C.

Western blot analysis

Protein was extracted from the cells using RIPA buffer, resolved by SDS–polyacrylamide gels and then transferred to polyvinylidene difluoride membranes. Primary antibodies against E-cadherin (catalog no. Sc-7870, Santa Cruz Biotechnology), Vimentin (cat no. Sc-66002, Santa Cruz Biotechnology), Snail (Cell Signaling Technology, catalog no. 9782T), Twist (Abcam, catalog no. ab49254), Slug (Cell Signaling Technology, catalog no. 9782T), ZEB-1 (Cell Signaling Technology, catalog no. 9782T), CD90 (Abcam, catalog no. ab133350), and GAPDH (ProteinTech, catalog no. HRP-60004) were used. Peroxidase-conjugated secondary antibodies (Cell Signaling Technology) were used and the antigen–antibody reaction was visualized by enhanced chemiluminescence assay (ECL, Thermo Fisher Scientific).

Animal experiments

4T1 cells-luc (5 × 105 cells/mouse) were inoculated into the mammary fat pads of 6-week-old female BALB/c mice. For neutrophil depletion, rat anti-Ly6G antibody (12.5 μg/mouse, catalog no. BE0075-1, BioXcell) or rat IgG isotype control (12.5 μg/mouse, catalog no. BE0089, BioXcell) was administered daily via intraperitoneal injection after the grafts were palpable. For neutralization of TIMP-1, TIMP-1 mAb (0.4 mg/kg, catalog no. AF-980, R&D Systems) or the IgG control (0.4 mg/kg, catalog no. ab37373, Abcam) was administrated via intraperitoneal injection once a week after the grafts were palpable. For neutralization of CD90, CD90 mAb (250 μg/mouse, catalog no. BE0066, BioXCell) or the IgG control (250 μg/mouse, catalog no. BE0089, BioXcell) was administrated every 3 days via intraperitoneal injection after the grafts were palpable. We examined the metastasis using PET/CT imaging (Siemens) and IVIS Lumina Imaging System (Xenogen). The tumors, lung, and liver tissues were harvested for hematoxylin and eosin (H&E) and IHC staining when the tumors reached 1.5 cm in diameter.

Cytokine antibody arrays

We used Human Cytokine Antibody Arrays V Kit (catalog no. AAH-CYT-5-4, RayBiotech) to evaluate the cytokine profiles. The arrays were blocked, incubated with 1 mL of supernatants from neutrophils overnight at 4°C, and then probed with biotin-conjugated antibodies (1/250) for 2 hours. Afterward, the array membranes were incubated with horseradish peroxidase–linked secondary antibody (1/1,000) for 1 hour at room temperature. The membranes were then incubated with chemiluminescent substrate and exposed to X-ray films. The signals as relative units of spot color density were evaluated by ImageJ software. The Excel-based analysis software tools provided by RayBiotech company (https://www.raybiotech.com/analysis-tools) were used for the automatic data sorting, averaging, background subtraction, positive control normalization, and data comparison. The formula used for data normalization is as follows: X(nY) = X(Y) x P1/P(Y), where P1 = the average signal density of the positive control spots on the reference array, P(Y) = the average signal density of the positive control spots on Array Y, X(Y) = the signal density for a particular spot on Array for sample, and “Y”, X(nY) = the normalized value for that particular spot "X" on Array for sample “Y".

ELISA

TIMP-1 (catalog no. E-EL-H0184), GM-CSF (catalog no. E-CL-H0081), angiogenin (catalog no. E-CL-H0006), CCL2 (catalog no. E-EL-H0020), CCL5 (catalog no. E-EL-H0023), VEGF (catalog no. E-EL-H0111), TNFα (catalog no. 88-7346-88), and TGF-β1 (catalog no. 88-8350-22) ELISA kits were purchased from Elabscience or Thermo Fisher Scientific. All experiments were performed according to the manufacturer's instructions.

siRNA transfection

To knock down specific target genes, cells were plated at 5 × 105 cells/mL and transfected with specific siRNA duplexes using Lipofectamine 3000 Transfection Reagent (Invitrogen, catalog no. L3000015) according to the manufacturer's instructions. siRNAs were provided by GenePharma Inc. Oligonucleotide sequence of siRNAs are as following: Twist siRNA-1, 5′-GCAAGAUUCAGACCCUCAATT-3′ (sense), Twist siRNA-2, 5′-CCUGAGCAACAGCGAGGAATT-3′ (sense).

Statistical analysis

We identified the optimum cutoff for high and low neutrophils' infiltration, TIMP1+ neutrophils, or CD90+ tumor cells in tumors using X-tile plots based on the association with the patients' survival by X-tile software version 3.6.1 (Yale University School of Medicine, New Haven, CT) as we reported previously (24, 25). Statistical analyses were performed using SPSS version 19.0 (SPSS Inc). Association between TINs and clinicopathologic features was assessed by χ2 test. DFS and OS was estimated using the Kaplan–Meier method and log-rank test. The univariate and multivariate Cox proportional hazards models were used to identify independent prognostic factors. We used receiver operating characteristics (ROC) curves to investigate the prognostic sensitivity and specificity of TINs. All experiments for cell cultures were performed independently at least three times and in triplicate for each time. P values less than 0.05 were considered statistically significant.

Neutrophils are present in both tumor parenchyma and stroma of breast cancer

To investigate the presence of TINs in breast cancer, we obtained clinical breast cancer samples from two independent institutes and performed IHC staining for CD66b as a marker of neutrophils (26). A total of 341 patients from Sun Yat-sen Memorial Hospital were randomized into the training set (170) and internal validation set (171). The independent set comprising 105 patients was enrolled from the First Affiliated Hospital of Shantou University (Shantou, China). Patient characteristics are listed in Supplementary Table S1.

CD66b+ neutrophils were absent in normal breast tissues. In contrast, neutrophils can be observed in both parenchyma and stroma of breast cancer tissues (Fig. 1A). Among the training, validation and independent sets, TINs were identified in 78 of 170 (45.88%), 74 of 171(43.3%), and 51 of 105 (48.57%) cases in tumor parenchyma, respectively, and 60 of 170 (35.29%), 63 of 171 (36.84%), and 39 of 105 (37.14%) cases in tumor stroma, respectively.

Figure 1.

TINs in parenchyma, but not the ones in stroma, are associated with poor prognosis. A, Representative images for IHC staining of CD66b in normal breast tissue, breast cancer tissue, and appendicitis tissue. The neutrophils in appendicitis tissue served as positive controls for CD66b staining. Boxes indicate the area with higher magnification. Scale bars, 50 μm. B and C, Kaplan–Meier survival curves for patients with breast cancer with high and low infiltration of CD66b+ TINs in tumor parenchyma in the training (n = 170), internal (n = 171), and independent (n = 105) cohorts, respectively. The optimal survival cutoff point for the CD66b+ TINs infiltration was determined by X-Tile based on the association with disease-free survival (B) or overall survival (C) in the training cohort and applied to the other two cohorts. D, ROC analysis and AUC were used to assess the prognostic capacity of DFS according to the parenchymal CD66b+ TINs.

Figure 1.

TINs in parenchyma, but not the ones in stroma, are associated with poor prognosis. A, Representative images for IHC staining of CD66b in normal breast tissue, breast cancer tissue, and appendicitis tissue. The neutrophils in appendicitis tissue served as positive controls for CD66b staining. Boxes indicate the area with higher magnification. Scale bars, 50 μm. B and C, Kaplan–Meier survival curves for patients with breast cancer with high and low infiltration of CD66b+ TINs in tumor parenchyma in the training (n = 170), internal (n = 171), and independent (n = 105) cohorts, respectively. The optimal survival cutoff point for the CD66b+ TINs infiltration was determined by X-Tile based on the association with disease-free survival (B) or overall survival (C) in the training cohort and applied to the other two cohorts. D, ROC analysis and AUC were used to assess the prognostic capacity of DFS according to the parenchymal CD66b+ TINs.

Close modal

Parenchymal TINs, but not stromal ones, are associated with poor prognosis

To investigate the clinical significance of TINs in breast cancer, we used X-tile plots to generate the optimum cut-off value (11.0 cells/HPF) for the CD66b+ TINs in tumor parenchyma based on association with disease-free survival (DFS) in the training set (Supplementary Fig. S1A). The univariate analysis showed significant association between TINs in tumor parenchyma and DFS (Fig. 1B; Supplementary Table S2). More importantly, the patients with high TINs in parenchyma exhibited worse overall survival (OS) compared with the ones with low TINs in tumor parenchyma (Fig. 1C; Supplementary Table S3). Cox multivariate analysis demonstrated that TINs in parenchyma was an independent prognostic factor for both DFS (Table 1) and OS (Supplementary Table S4). In addition, high TINs in tumor parenchyma was significantly associated with advanced histologic grade, tumor size, lymph nodes metastasis, high TNM stage, triple-negative breast cancer subtype, and distant metastasis (Supplementary Table S5). To confirm these findings, we applied the same cut-off setting to the validation group and independent one. In both cohorts, TINs in parenchyma was an independent prognostic factor (Table 1; Supplementary Table S4) and correlated with similar clinicopathologic features (Supplementary Table S5).

Table 1.

Multivariate Cox regression analyses of DFS in the training, validation, and independent cohorts

DFS
Training cohort (n = 170)Validation cohort (n = 171)Independent cohort (n = 105)
Factor HR (95% CI) P HR (95% CI) P HR (95% CI) P 
Tumor grade (III vs. I–II) 1.282 (0.718–2.289) 0.401 1.123 (0.644–1.959) 0.682 2.228 (0.935–5.307) 0.070 
Tumor size (>2 cm) 2.168 (1.152–4.079) 0.016* 1.976 (1.054–3.704) 0.034* 4.424 (1.274–15.368) 0.019* 
LN (positive vs. negative) 3.098 (1.535–6.251) 0.002** 2.970 (1.525–5.782) 0.001** 4.947 (1.619–15.116) 0.005** 
ER (positive vs. negative) 0.673 (0.340–1.335) 0.257 0.676 (0.337–1.356) 0.271 0.471 (0.180–1.229) 0.124 
PR (positive vs. negative) 0.875 (0.402–1.907) 0.737 0.924 (0.465–1.836) 0.822 1.800 (0.557–5.822) 0.326 
CD66b+ TINs in parenchyma (high vs. low) 3.331 (1.827–6.073) <0.001*** 2.055 (1.132–3.731) 0.018* 4.546 (1.869–11.056) 0.001** 
DFS
Training cohort (n = 170)Validation cohort (n = 171)Independent cohort (n = 105)
Factor HR (95% CI) P HR (95% CI) P HR (95% CI) P 
Tumor grade (III vs. I–II) 1.282 (0.718–2.289) 0.401 1.123 (0.644–1.959) 0.682 2.228 (0.935–5.307) 0.070 
Tumor size (>2 cm) 2.168 (1.152–4.079) 0.016* 1.976 (1.054–3.704) 0.034* 4.424 (1.274–15.368) 0.019* 
LN (positive vs. negative) 3.098 (1.535–6.251) 0.002** 2.970 (1.525–5.782) 0.001** 4.947 (1.619–15.116) 0.005** 
ER (positive vs. negative) 0.673 (0.340–1.335) 0.257 0.676 (0.337–1.356) 0.271 0.471 (0.180–1.229) 0.124 
PR (positive vs. negative) 0.875 (0.402–1.907) 0.737 0.924 (0.465–1.836) 0.822 1.800 (0.557–5.822) 0.326 
CD66b+ TINs in parenchyma (high vs. low) 3.331 (1.827–6.073) <0.001*** 2.055 (1.132–3.731) 0.018* 4.546 (1.869–11.056) 0.001** 

NOTE: P value was calculated by two-sided log-rank test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Abbreviations: ER, estrogen receptor; LN, lymph node; PR, progesterone receptor; TNM, tumor–node–metastasis.

To assess the sensitivity and specificity of parenchymal TINs as a prognostic biomarker, we calculated the area under the receiver operating characteristic curve (AUC). TINs in tumor parenchyma showed a good prognostic value of DFS in the training [AUC = 0.72; 95% confidence interval (CI): 0.63–0.83], validation (AUC = 0.77; 95% CI: 0.69–0.85), and independent set (AUC = 0.84; 95% CI: 0.76–0.92), respectively (Fig. 1D). Similarly, parenchymal TINs also showed a good prognostic value of OS in the training (AUC = 0.79; 95% CI: 0.67–0.91), validation (AUC = 0.78; 95% CI: 0.69–0.88), and independent set (AUC = 0.84; 95% CI: 0.75–0.93), respectively (Supplementary Fig. S1B).

Similarly, we also used X-tile plots to generate the optimum cut-off value (5.0 cells/HPF) for the TINs in stroma in the training set (Supplementary Fig. S1C). We found that TINs at stroma was not significantly associated with DFS or OS in all three cohorts (Supplementary Fig. S1D). High TINs in stroma was significantly associated with tumor size in the training set and lymphovascular invasion in internal validation set, which were not validated by other cohorts (Supplementary Table S6). Collectively, our data indicated that TINs in parenchyma, but not ones in stroma, are associated with poor prognosis of patients with breast cancer.

Primary TINs promote metastasis of breast cancer cells

Given TINs in tumor parenchyma were a prognostic factor predicting poor prognosis, we further investigated its underlying mechanism. We isolated primary TINs from freshly resected human breast cancer tissues as reported previously (17, 22). The purity of isolated TINs determined by CD11b+CD66b+ cells were more than 90% (Fig. 2A). The isolated CD11b+CD66b+ cells from tumors exhibited deeply lobed nucleus (Fig. 2B and C). The expression of CD66b remained unchanged after magnetic sorting and up to 48 hours of culture (Supplementary Fig. S2A). The apoptosis of neutrophils increased after 48 hours of culture alone. However, coculture with tumor cells significantly reduced the apoptosis of neutrophils (Supplementary Fig. S2B). We observed that primary TINs, rather than untreated autologous neutrophils isolated from peripheral blood, significantly promoted migration and invasion of multiple breast cancer cell lines (Fig. 2D and E). In BALB/c mice, depletion of neutrophils by anti-Ly6G inhibited the metastasis of synergic 4T1 cells to both lung and liver (Fig. 2F and G).

Figure 2.

Primary human TINs promote EMT and metastasis. Fresh carcinoma tissues were digested by collagenase. The TINs and autologous neutrophils from peripheral blood were isolated by CD66b magnetic beads. A–C, The neutrophil purity was determined by flow cytometry (A) and immunofluorescent staining (B) for CD66b and CD11b. Arrows indicate the area of magnification. Scale bars, 10 μm. C, The Giemsa staining of isolated neutrophils. Scale bars, 200 μm. D and E, Migration and invasion assays of cancer cells with indicated treatments were performed in the transwell system. Cancer cells were added in the top chamber, whereas primary TINs or autologous peripheral blood neutrophils were added to the bottom chambers. Three independent experiments were performed for the neutrophils isolated from each of the 5 patients. D, Representative images. Scale bars, 100 μm. E, Statistical analysis, mean ± SEM. ***, P < 0.001 by Student t test. F, Neutrophils in 4T1 tumor-bearing mice were depleted by anti-Ly6G intraperitoneal injection. Representative images of lung and liver metastasis evaluated by PET-CT. SUV, standardized uptake value. G, Statistical analysis of metastasis calculated by the standardized uptake value (SUV). n = 8 for each group, mean ± SEM. **, P < 0.01 compared with untreated group by Student t test. H and I, MCF-7 cells were cultured alone (−), or treated with conditioned media (CM) of primary TINs or autologous blood neutrophils. Three independent experiments were performed for the neutrophils isolated from each of the three patients. H, Representative images for morphology, E-cadherin, and vimentin expression. Scale bars, 20 μm. I, Representative Western blotting images for EMT markers (n = 3).

Figure 2.

Primary human TINs promote EMT and metastasis. Fresh carcinoma tissues were digested by collagenase. The TINs and autologous neutrophils from peripheral blood were isolated by CD66b magnetic beads. A–C, The neutrophil purity was determined by flow cytometry (A) and immunofluorescent staining (B) for CD66b and CD11b. Arrows indicate the area of magnification. Scale bars, 10 μm. C, The Giemsa staining of isolated neutrophils. Scale bars, 200 μm. D and E, Migration and invasion assays of cancer cells with indicated treatments were performed in the transwell system. Cancer cells were added in the top chamber, whereas primary TINs or autologous peripheral blood neutrophils were added to the bottom chambers. Three independent experiments were performed for the neutrophils isolated from each of the 5 patients. D, Representative images. Scale bars, 100 μm. E, Statistical analysis, mean ± SEM. ***, P < 0.001 by Student t test. F, Neutrophils in 4T1 tumor-bearing mice were depleted by anti-Ly6G intraperitoneal injection. Representative images of lung and liver metastasis evaluated by PET-CT. SUV, standardized uptake value. G, Statistical analysis of metastasis calculated by the standardized uptake value (SUV). n = 8 for each group, mean ± SEM. **, P < 0.01 compared with untreated group by Student t test. H and I, MCF-7 cells were cultured alone (−), or treated with conditioned media (CM) of primary TINs or autologous blood neutrophils. Three independent experiments were performed for the neutrophils isolated from each of the three patients. H, Representative images for morphology, E-cadherin, and vimentin expression. Scale bars, 20 μm. I, Representative Western blotting images for EMT markers (n = 3).

Close modal

EMT plays a crucial role in breast cancer metastasis. Therefore, we investigated whether TINs induce EMT of breast cancer cells. We cultured epithelial-like breast cancer cell MCF-7 alone, or with 30% conditioned media from untreated neutrophils or autologous primary TINs. We found that MCF-7 cells treated with the conditioned media of TINs, but not the ones cultured alone or treated with the conditioned media of control neutrophils became stretched and elongated (Fig. 2H), downregulated in epithelial biomarker E-cadherin expression, while increased in mesenchymal biomarker vimentin expression (Fig. 2H and I). In addition, stimulation with the conditioned media of TINs increased levels of EMT transcriptional factor Twist, but not Snail, Slug, or ZEB1 in tumor cells (Fig. 2I). Together, these results indicated that TINs induce EMT and metastasis of breast cancer cells.

Primary TINs promote metastasis of breast cancer cells by TIMP-1

To identify the cytokines secreted by TINs that induce EMT, we used antibody array to compare the cytokine production profile of primary TINs to the one of autologous peripheral neutrophils. Three cytokines, TIMP-1, GM-CSF, and angiogenin, increased in TINs (Fig. 3A; Supplementary Table S7), which was confirmed by ELISA (Fig. 3B).

Figure 3.

TINs promote metastasis by TIMP-1. A, Representative cytokine arrays for the conditioned media of TINs and autologous neutrophils from peripheral blood. Boxes indicate the cytokines with significant changes and table on the right summarizes the relative signal intensity of indicated cytokines. Mean ± SEM, n = 5 patients. B, Indicated cytokines levels were validated by ELISA. Mean ± SEM, *, P < 0.05, ***, P < 0.001 compared with peripheral neutrophils by Student t test. Three independent experiments were performed for each of the 6 patients. C, Neutralizing antibodies against GM-CSF, angiogenin, or TIMP-1 were added in the transwell cocultured system of MCF-7 cells and primary TINs. Representative images and quantification of migration and invasion assays of MCF-7 cells with indicated treatment are shown. Scale bars, 100 μm. Mean ± SEM. ***, P < 0.001 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 3 patients. D and E, Conditioned media of TINs were preincubated with neutralizing antibodies against GM-CSF, angiogenin, or TIMP-1. D, Representative immunofluorescent images for E-cadherin and vimentin (E-cad/Vim) expression in MCF-7 cells with indicated treatments. Scale bars, 20 μm (n = 3). E, The representative images of Western blotting analysis for the expression of EMT markers in MCF-7 cells with indicated treatments (n = 3). F, MCF-7 cells and primary TINs or autologous peripheral neutrophils were cocultured in a Matrigel three-dimensional system. The representative images of immunofluorescent staining for E-cadherin, vimentin, CD66b, and TIMP-1 are shown. Scale bars, 20 μm (n = 3). G, Representative images of Western blotting analysis for EMT markers in MCF-7 cells with indicated treatments (n = 3). H, 4T1 cells were injected into the mammary fat pads of BALB/c mice and TIMP-1–neutralizing antibodies were administrated via intraperioneal route once per week at 0.4 mg/kg after the syngeneic grafts were palpable. The metastasis was evaluated by IVIS. Representative bioluminescence images (left) and the quantification of bioluminescence signal (right). n = 8 for each group, mean ± SEM (**, P < 0.01 by Student t test).

Figure 3.

TINs promote metastasis by TIMP-1. A, Representative cytokine arrays for the conditioned media of TINs and autologous neutrophils from peripheral blood. Boxes indicate the cytokines with significant changes and table on the right summarizes the relative signal intensity of indicated cytokines. Mean ± SEM, n = 5 patients. B, Indicated cytokines levels were validated by ELISA. Mean ± SEM, *, P < 0.05, ***, P < 0.001 compared with peripheral neutrophils by Student t test. Three independent experiments were performed for each of the 6 patients. C, Neutralizing antibodies against GM-CSF, angiogenin, or TIMP-1 were added in the transwell cocultured system of MCF-7 cells and primary TINs. Representative images and quantification of migration and invasion assays of MCF-7 cells with indicated treatment are shown. Scale bars, 100 μm. Mean ± SEM. ***, P < 0.001 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 3 patients. D and E, Conditioned media of TINs were preincubated with neutralizing antibodies against GM-CSF, angiogenin, or TIMP-1. D, Representative immunofluorescent images for E-cadherin and vimentin (E-cad/Vim) expression in MCF-7 cells with indicated treatments. Scale bars, 20 μm (n = 3). E, The representative images of Western blotting analysis for the expression of EMT markers in MCF-7 cells with indicated treatments (n = 3). F, MCF-7 cells and primary TINs or autologous peripheral neutrophils were cocultured in a Matrigel three-dimensional system. The representative images of immunofluorescent staining for E-cadherin, vimentin, CD66b, and TIMP-1 are shown. Scale bars, 20 μm (n = 3). G, Representative images of Western blotting analysis for EMT markers in MCF-7 cells with indicated treatments (n = 3). H, 4T1 cells were injected into the mammary fat pads of BALB/c mice and TIMP-1–neutralizing antibodies were administrated via intraperioneal route once per week at 0.4 mg/kg after the syngeneic grafts were palpable. The metastasis was evaluated by IVIS. Representative bioluminescence images (left) and the quantification of bioluminescence signal (right). n = 8 for each group, mean ± SEM (**, P < 0.01 by Student t test).

Close modal

To further explore which of these cytokines is responsible for EMT induced by TINs, we added neutralizing antibodies against these cytokines and observed that neutralization of TIMP-1, but not GM-CSF or angiogenin, significantly reduced the migration and invasion of tumor cells induced by TINs (Fig. 3C). Moreover, blockade of TIMP-1, rather than GM-CSF or angiogenin, markedly inhibited EMT of tumor cells treated with the conditioned media of TINs (Fig. 3D and E).

It has been reported that TGFβ within the tumor microenvironment induces a population of TINs with N2 phenotype (27). To investigate the difference of N1 and N2 cytokines between neutrophils and TINs, we evaluated the levels of TGFβ1, N1 cytokine (TNFα), and N2 cytokines (VEGF, CCL5, and CCL2) in TINs and paired peripheral neutrophils by ELISA. Our data showed that CCL2, rather than TNFα, significantly increased in TINs compared with peripheral neutrophils (Supplementary Fig. S3A). VEGF and CCL5 exhibited a trend of elevation in TINs. However, the difference was not statistically significant (Supplementary Fig. S3A). Neutralization of CCL2 did not significantly reduce the migration and invasion of tumor cells induced by TINs (Supplementary Fig. S3B). The TGFβ1 secreted by TINs was very low (Supplementary Fig. S3A), suggesting TGFβ1 may not be mainly derived from neutrophils in the tumor microenvironment.

Consistently, three-dimensional culture demonstrated that the TINs with increased TIMP-1 production compared with control neutrophils induced EMT of cocultured tumor cells (Fig. 3F). In addition, treatment of rhTIMP-1 upregulated vimentin and downregulated E-cadherin of tumor cells. rhTIMP-1 stimulation increased the level of Twist, rather than other EMT factors such as Snail and ZEB1 (Fig. 3G). Moreover, rhTIMP-1 enhanced migration and invasion of tumor cells (Supplementary Fig. S3C). Silencing Twist abrogated the EMT (Fig. 3G) and migration and invasion (Supplementary Fig. S3C) induced by rhTIMP-1. In vivo, administration of TIMP-1 neutralization antibody intraperitoneally markedly inhibited tumor metastasis (Fig. 3H). Collectively, these data demonstrated that TINs promote metastasis of breast cancer cells via TIMP-1.

TINs promote EMT by a CD90-TIMP-1 juxtacrine–paracrine loop

Given that TINs induced EMT by TIMP-1, we then asked whether cancer cells undergoing EMT induce TIMP-1 overexpression in neutrophils in turn. We induced EMT of MCF7 cells (MCF7EMT) by TIN conditioned media for 5 days. Then, we cocultured MCF7EMT cells with unstimulated neutrophils isolated from peripheral blood in the transwell system. We observed that although TIMP-1 in cocultured neutrophils slightly increased, its level was much less than the one in TINs (Fig. 4A). Therefore, we mixed neutrophils with MCF7EMT cells for 24 hours and retrieved the neutrophils by CD66b magnetic beads. We found that the secretion of TIMP1 in neutrophils mixed with MCF7EMT cells significantly increased, which was comparable with the level in TINs (Fig. 4A). Thus, these data suggest that cancer cells undergoing EMT induce TIMP-1 overexpression in neutrophils in a cell-contact manner.

Figure 4.

TINs promote EMT by a CD90-TIMP-1 juxtacrine–paracrine loop. A, Neutrophils from peripheral blood were cocultured with untreated MCF-7 cells or MCF-7 cells underwent EMT (MCF7EMT) by pretreatment of TIN-conditioned media in transwell system or directly. After 24 hours, neutrophils were isolated by CD66b magnetic bead isolation and their level of TIMP-1 was assessed by ELISA. Mean ± SEM. n.s., not statistically significant; **, P < 0.01; ***, P < 0.001 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 4 patients. B and C, Peripheral neutrophils were mixed with MCF7EMT in the presence of indicated neutralizing antibody. After 24 hours, neutrophils were isolated by CD66b magnetic bead isolation and their level of TIMP-1 was assessed by ELISA. Mean ± SEM. **, P < 0.01 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 3 patients. D and E, 4T1 cells were injected into the fat pads of BALB/c mice and neutralizing antibodies against TIMP-1 (D) or CD90 (E) were administrated after the syngeneic grafts were palpable. Representative images of IHC staining for CD90 (D) and TIMP-1 (E) in xenograft (left) and the quantification (right) are shown. Scale bars, 100 μm. n = 8 for each group, mean ± SEM. ***, P < 0.001 by Student t test. F, 4T1 cells were injected into the mammary fat pads of BALB/c mice. The indicated antibodies were administrated after the syngeneic grafts were palpable. The lung and liver metastases were evaluated by IVIS. Representative bioluminescence images (left) and the quantification of bioluminescence signal (right) demonstrated the lung and liver metastasis. n = 6 for each group, mean ± SEM (n.s., not statistically significant; *, P < 0.05; **, P < 0.01 by Student t test).

Figure 4.

TINs promote EMT by a CD90-TIMP-1 juxtacrine–paracrine loop. A, Neutrophils from peripheral blood were cocultured with untreated MCF-7 cells or MCF-7 cells underwent EMT (MCF7EMT) by pretreatment of TIN-conditioned media in transwell system or directly. After 24 hours, neutrophils were isolated by CD66b magnetic bead isolation and their level of TIMP-1 was assessed by ELISA. Mean ± SEM. n.s., not statistically significant; **, P < 0.01; ***, P < 0.001 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 4 patients. B and C, Peripheral neutrophils were mixed with MCF7EMT in the presence of indicated neutralizing antibody. After 24 hours, neutrophils were isolated by CD66b magnetic bead isolation and their level of TIMP-1 was assessed by ELISA. Mean ± SEM. **, P < 0.01 by Student t test. Three independent experiments were performed for the neutrophils isolated from each of the 3 patients. D and E, 4T1 cells were injected into the fat pads of BALB/c mice and neutralizing antibodies against TIMP-1 (D) or CD90 (E) were administrated after the syngeneic grafts were palpable. Representative images of IHC staining for CD90 (D) and TIMP-1 (E) in xenograft (left) and the quantification (right) are shown. Scale bars, 100 μm. n = 8 for each group, mean ± SEM. ***, P < 0.001 by Student t test. F, 4T1 cells were injected into the mammary fat pads of BALB/c mice. The indicated antibodies were administrated after the syngeneic grafts were palpable. The lung and liver metastases were evaluated by IVIS. Representative bioluminescence images (left) and the quantification of bioluminescence signal (right) demonstrated the lung and liver metastasis. n = 6 for each group, mean ± SEM (n.s., not statistically significant; *, P < 0.05; **, P < 0.01 by Student t test).

Close modal

A panel of membrane proteins was elevated in cancer cells after EMT, including Frizzled2, TMPRSS4, CD90, and EphA4 (19, 28, 29). Among them, neutralization of CD90 abrogated TIMP-1 overexpression in neutrophils mixed with MCF7EMT cells (Fig. 4B). Consistently, neutralization of Mac-1, the neutrophil receptor of CD90, in both a cis- and trans- manner (19, 30), also exerted similarly effects (Fig. 4C).

To investigate whether TIMP-1 induced upregulation of CD90 in tumor cells, we induced EMT of MCF7 cells by treatment of rhTIMP-1 and evaluated CD90 by flow cytometry, Western blotting, and immunostaining. We observed that CD90 level was markedly elevated in rhTIMP-1–treated tumor cells compared with untreated ones (Supplementary Fig. S4A–S4C). Of note, treatment of TGFβ1, another EMT inducer, also upregulated CD90 expression (Supplementary Fig. S4A–S4C). Moreover, we found that addition of TIN conditioned media, rather than peripheral neutrophil one, significantly increased CD90 in MCF7 cells (Supplementary Fig. S4D). More importantly, TIMP-1 blockade abolished the CD90 induction in MCF7 cells by the conditioned media of TINs (Supplementary Fig. S4E).

In vivo, CD90 staining was predominantly observed in tumor cells, whereas TIMP-1 blockade significantly reduced CD90 expression (Fig. 4D). Reciprocally, administration of CD90 neutralization antibody intraperitoneally markedly reduced TIMP-1 levels in tumors (Fig. 4E) and distant metastasis (Fig. 4F). In contrast, no difference in metastasis was observed in the mice treated with IgG and the ones treated with anti-CD90 antibodies when anti-TIMP-1 neutralizing antibody was coinjected (Fig. 4F). Collectively, these data suggested that tumor-contacted neutrophils promote EMT by a CD90-TIMP-1 juxtacrine–paracrine loop.

Both TIMP-1 and CD90 are associated with poor prognosis of patients with breast cancer

To investigate the clinical significance of the CD90–TIMP-1 loop in human breast cancer, we investigate the expression of CD90, TIMP-1, EMT marker E-cadherin, and neutrophil marker CD66b in human breast cancer. We observed that CD90 is overexpressed in tumor cells with downregulation of E-cadherin compared with the ones with abundant E-cadherin expression (Fig. 5A). In addition, TIMP-1 is predominantly expressed in TINs in tumor parenchyma, rather than the ones in stroma (Fig. 5A). Consistent with the ex vivo data, TIMP-1 levels were strongly correlated with CD90 levels (Fig. 5B). More importantly, both CD90 and TIMP-1 were associated with long-term poor prognosis (Fig. 5C).

Figure 5.

Both TIMP-1 and CD90 are associated with poor prognosis. A, Representative images of H&E staining and immunofluorescence staining of CD66b, TIMP-1, E-cadherin, CD90, and cytokeratin (CK) in serial sections of breast cancer samples. Scale bars, 50 μm. B, The correlation between the number of TIMP-1+ TINs (TIMP-1+ CD66b+) and the number of CD90+ tumor cells (CD90+CK+) in human breast cancer (n = 341). CK, cytokeratin. C, Kaplan–Meier curves of patients with low and high numbers of TIMP-1+ TINs and CD90+ tumor cells (n = 341). D, Schematics highlighting the major findings of this study. Neutrophils induced EMT via TIMP-1. Reciprocally, tumor cells undergoing EMT enhanced neutrophils' TIMP-1 by CD90 in a cell-contact manner. Blocking this juxtacrine–paracrine loop reduces metastasis in vivo.

Figure 5.

Both TIMP-1 and CD90 are associated with poor prognosis. A, Representative images of H&E staining and immunofluorescence staining of CD66b, TIMP-1, E-cadherin, CD90, and cytokeratin (CK) in serial sections of breast cancer samples. Scale bars, 50 μm. B, The correlation between the number of TIMP-1+ TINs (TIMP-1+ CD66b+) and the number of CD90+ tumor cells (CD90+CK+) in human breast cancer (n = 341). CK, cytokeratin. C, Kaplan–Meier curves of patients with low and high numbers of TIMP-1+ TINs and CD90+ tumor cells (n = 341). D, Schematics highlighting the major findings of this study. Neutrophils induced EMT via TIMP-1. Reciprocally, tumor cells undergoing EMT enhanced neutrophils' TIMP-1 by CD90 in a cell-contact manner. Blocking this juxtacrine–paracrine loop reduces metastasis in vivo.

Close modal

Neutrophils are the first responders to tissue damage. It has been well documented that they also play a critical role in multiple types of cancer, the “unhealed wound” (31, 32). Evidences for both pro- and antitumor roles have been reported (4, 5, 33, 34), highlighting our lack of detailed knowledge about their diversity in tumors. Therefore, careful dissections of interaction between neutrophils and cancer cells by more clinically relevant approaches are of paramount importance for developing neutrophil-targeted oncology treatment (3, 8). By analyzing the clinical samples from three cohorts of patients with breast cancer, we found that neutrophils in tumor parenchyma, rather than the ones in stroma, were an independent risk factor, especially in the triple-negative subtype, which is notoriously poor prognostic due to the lack of targeted therapy. By isolating primary TINs from clinical samples and cytokine microarray screening, we uncovered that tumor-contacted neutrophils induced EMT by a CD90-TIMP-1 juxtacrine-paracrine cycle (Fig. 5D). More importantly, we highlighted the therapeutic opportunities by showing blockade of either TIMP-1 or CD90 inhibited breast cancer metastasis in vivo.

TINs can be found in almost every type of malignancies. Although their presence is undeniable, their prognostic value is controversial (2, 3). Positive (35), negative (36), or no (37) association between TINs and patient prognosis have been reported, highlighting their heterogeneity and diversity. Intriguingly, we observed that neutrophils in tumor parenchyma, rather than the ones in stroma, were significantly correlated with worse long-term survival of patients with breast cancer. To our knowledge, this is the first study to identify the different prognostic values of TINs in different tissue compartments. It has been reported that TINs can be classified as N1 (anti-) and N2 (pro-) subtypes, highlighting the heterogeneity of TINs (27). Our report provided the clinical evidence to support the hypothesis that the plasticity of neutrophils is constantly mediated by different signals within the tumor microenvironment (38, 39).

Despite the progress in the treatment of early-stage cancer, metastatic diseases are still incurable. EMT has been identified as a critical regulator of metastasis (12). We and others have previously revealed a crosstalk between cancer cells with EMT traits and inflammatory cells, including monocytes, macrophages, and fibroblasts (16, 17, 19). In this study, we discovered that neutrophils, another major cellular component of the tumor microenvironment, induced EMT of breast cancer cells by TIMP-1. In turn, breast cancer cells undergoing EMT maintained and reinforced TIMP-1 production from neutrophils by CD90 in a cell-contact fashion. The cell–cell contact–dependent interactions identified here provided the mechanistic insight to the clinical observation that neutrophils in tumor parenchyma, but not the ones in stroma, were strongly associated with worse clinical outcomes in patients with breast cancer. Indeed, TIMP-1 and CD90 were positively correlated in human breast cancer and both associated with long-term poor prognosis. Our data here provide new evidences to support the emerging notion that various signaling molecules secreted extracellularly or located on the membrane of multiple stromal cell subsets in the milieu facilitate, perhaps in a coordinated manner, EMT induction and metastasis of cancer cells.

Neutrophils are a major source of matrix metalloproteinase 9 (MMP9), which facilitates tumor cell invasion in the extracellular matrix (9). Unexpectedly, we found that neutrophils produced TIMP-1, the inhibitor of MMP9, which directly enhanced tumor cell motility by inducing EMT. More broadly, the crosstalk between neutrophils and cancer cells with EMT traits observed here may represent a general mechanism that also operates in other carcinoma types. Elevated TIMP-1 has been correlated with poor prognosis of multiple cancers, including breast (40), lung (41), and lymphoma (42). In addition, inhibition of TIMP-1 reduces liver metastasis of colorectal cancer (43, 44). Therefore, our and others' data suggested that tumor cells utilize the bilateral properties of TIMP-1 for cancer metastasis.

CD90 (Thy-1) is a membrane-anchoring protein that faces the extracellular matrix. It can bind to Mac1 (CD11b/CD18) expressed in neutrophils and macrophages, and initiate signaling cascades in both cis- and trans-manner (30). CD90 is mainly expressed in endothelial cells and mediates extravasation of myeloid cells in inflammation (45). Despite the fact that its function in tumor metastasis remains unclear, accumulating evidences indicate that CD90 is overexpressed in cancer stem cells of various tumor types (19, 46, 47), which are closely linked to EMT. We extend the knowledge of this emerging field by showing (i) CD90 in cancer cells with EMT traits induces TIMP-1 production in neutrophils, which reciprocally reinforced EMT and (ii) blockade of CD90 inhibited breast cancer metastasis in vivo. How CD90 induces TIMP-1 expression in neutrophils and the therapeutic value of CD90 blockade in other cancer types warrant further studies.

Collectively, our study demonstrated a juxtacrine–paracrine loop between neutrophils and tumor cells with EMT traits, which is important for breast cancer metastasis and a potential therapeutic target for cancer treatment.

No potential conflicts of interest were disclosed.

Conception and design: S. Su

Development of methodology: Y. Wang, J. Chen, J. Li, W. Wu, M. Huang, S. Su

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Wang, J. Chen, L. Yang, J. Li, W. Wu, M. Huang, L. Lin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Wang, J. Chen, J. Li, M. Huang

Writing, review, and/or revision of the manuscript: J. Chen, M. Huang, S. Su

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Wang, J. Li, W. Wu, L. Lin, S. Su

Study supervision: S. Su

This work was supported by grants from the National Key Research and Development Program of China (2017YFA0106300), the Natural Science Foundation of China (81622036, 81472468, 81672614, 81802656, 81802645, and 81672620), Science Foundation of Guangdong Province (2016A030306023, 2017A030313878, and 201710010083), the Tip-top Scientific and Technical Innovative Youth Talents of Guangdong special support program (2016TQ03R553), and the Guangzhou Science Technology and Innovation Commission (201508020008 and 201508020249).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Gregory
AD
,
Houghton
AM
. 
Tumor-associated neutrophils: new targets for cancer therapy
.
Cancer Res
2011
;
71
:
2411
6
.
2.
Jenne
CN
,
Kubes
P
. 
Gastrointestinal cancer: neutrophils and cancer: guilt by association
.
Nat Rev Gastroenterol Hepatol
2016
;
13
:
381
2
.
3.
Coffelt
SB
,
Wellenstein
MD
,
de Visser
KE
. 
Neutrophils in cancer: neutral no more
.
Nat Rev Cancer
2016
;
16
:
431
46
.
4.
Coffelt
SB
,
Kersten
K
,
Doornebal
CW
,
Weiden
J
,
Vrijland
K
,
Hau
CS
, et al
IL-17-producing gammadelta T cells and neutrophils conspire to promote breast cancer metastasis
.
Nature
2015
;
522
:
345
8
.
5.
Wculek
SK
,
Malanchi
I
. 
Neutrophils support lung colonization of metastasis-initiating breast cancer cells
.
Nature
2015
;
528
:
413
7
.
6.
Granot
Z
,
Henke
E
,
Comen
EA
,
King
TA
,
Norton
L
,
Benezra
R
. 
Tumor entrained neutrophils inhibit seeding in the premetastatic lung
.
Cancer Cell
2011
;
20
:
300
14
.
7.
Albanesi
M
,
Mancardi
DA
,
Jonsson
F
,
Iannascoli
B
,
Fiette
L
,
Di Santo
JP
, et al
Neutrophils mediate antibody-induced antitumor effects in mice
.
Blood
2013
;
122
:
3160
4
.
8.
Nicolas-Avila
JA
,
Adrover
JM
,
Hidalgo
A
. 
Neutrophils in homeostasis, immunity, and cancer
.
Immunity
2017
;
46
:
15
28
.
9.
Treffers
LW
,
Hiemstra
IH
,
Kuijpers
TW
,
van den Berg
TK
,
Matlung
HL
. 
Neutrophils in cancer
.
Immunol Rev
2016
;
273
:
312
28
.
10.
Ethier
JL
,
Desautels
D
,
Templeton
A
,
Shah
PS
,
Amir
E
. 
Prognostic role of neutrophil-to-lymphocyte ratio in breast cancer: a systematic review and meta-analysis
.
Breast Cancer Res
2017
;
19
:
2
.
11.
Mouchemore
KA
,
Anderson
RL
,
Hamilton
JA
. 
Neutrophils, G-CSF and their contribution to breast cancer metastasis
.
FEBS J
2017
;
285
:
665
79
.
12.
Nieto
MA
,
Huang
RY
,
Jackson
RA
,
Thiery
JP
. 
EMT: 2016
.
Cell
2016
;
166
:
21
45
.
13.
Kalluri
R
,
Weinberg
RA
. 
The basics of epithelial-mesenchymal transition
.
J Clin Invest
2009
;
119
:
1420
8
.
14.
Roussos
ET
,
Condeelis
JS
,
Patsialou
A
. 
Chemotaxis in cancer
.
Nat Rev Cancer
2011
;
11
:
573
87
.
15.
Kudo-Saito
C
,
Shirako
H
,
Takeuchi
T
,
Kawakami
Y
. 
Cancer metastasis is accelerated through immunosuppression during Snail-induced EMT of cancer cells
.
Cancer Cell
2009
;
15
:
195
206
.
16.
Huang
D
,
Song
SJ
,
Wu
ZZ
,
Wu
W
,
Cui
XY
,
Chen
JN
, et al
Epstein-Barr virus-induced VEGF and GM-CSF drive nasopharyngeal carcinoma metastasis via recruitment and activation of macrophages
.
Cancer Res
2017
;
77
:
3591
604
.
17.
Su
S
,
Liu
Q
,
Chen
J
,
Chen
J
,
Chen
F
,
He
C
, et al
A positive feedback loop between mesenchymal-like cancer cells and macrophages is essential to breast cancer metastasis
.
Cancer Cell
2014
;
25
:
605
20
.
18.
Su
S
,
Wu
W
,
He
C
,
Liu
Q
,
Song
E
. 
Breaking the vicious cycle between breast cancer cells and tumor-associated macrophages
.
Oncoimmunology
2014
;
3
:
e953418
.
19.
Lu
H
,
Clauser
KR
,
Tam
WL
,
Frose
J
,
Ye
X
,
Eaton
EN
, et al
A breast cancer stem cell niche supported by juxtacrine signalling from monocytes and macrophages
.
Nat Cell Biol
2014
;
16
:
1105
17
.
20.
Shibue
T
,
Weinberg
RA
. 
EMT, CSCs, and drug resistance: the mechanistic link and clinical implications
.
Nat Rev Clin Oncol
2017
;
14
:
611
29
.
21.
Jung
HY
,
Fattet
L
,
Yang
J
. 
Molecular pathways: linking tumor microenvironment to epithelial-mesenchymal transition in metastasis
.
Clin Cancer Res
2015
;
21
:
962
8
.
22.
Sionov
RV
,
Assi
S
,
Gershkovitz
M
,
Sagiv
JY
,
Polyansky
L
,
Mishalian
I
, et al
Isolation and characterization of neutrophils with anti-tumor properties
.
J Vis Exp
2015
;
100
:
e52933
.
23.
Singhal
S
,
Bhojnagarwala
PS
,
O'Brien
S
,
Moon
EK
,
Garfall
AL
,
Rao
AS
, et al
Origin and role of a subset of tumor-associated neutrophils with antigen-presenting cell features in early-stage human lung cancer
.
Cancer Cell
2016
;
30
:
120
35
.
24.
Su
S
,
Chen
J
,
Yao
H
,
Liu
J
,
Yu
S
,
Lao
L
, et al
CD10(+)GPR77(+) cancer-associated fibroblasts promote cancer formation and chemoresistance by sustaining cancer stemness
.
Cell
2018
;
172
:
841
56
.
25.
Su
S
,
Liao
J
,
Liu
J
,
Huang
D
,
He
C
,
Chen
F
, et al
Blocking the recruitment of naive CD4(+) T cells reverses immunosuppression in breast cancer
.
Cell Res
2017
;
27
:
461
82
.
26.
Wang
J
,
Jia
Y
,
Wang
N
,
Zhang
X
,
Tan
B
,
Zhang
G
, et al
The clinical significance of tumor-infiltrating neutrophils and neutrophil-to-CD8+ lymphocyte ratio in patients with resectable esophageal squamous cell carcinoma
.
J Transl Med
2014
;
12
:
7
.
27.
Fridlender
ZG
,
Sun
J
,
Kim
S
,
Kapoor
V
,
Cheng
G
,
Ling
L
, et al
Polarization of tumor-associated neutrophil phenotype by TGF-beta: "N1" versus "N2" TAN
.
Cancer Cell
2009
;
16
:
183
94
.
28.
Kim
S
,
Lee
JW
. 
Membrane proteins involved in epithelial-mesenchymal transition and tumor invasion: studies on TMPRSS4 and TM4SF5
.
Genomics Inform
2014
;
12
:
12
20
.
29.
Gujral
TS
,
Chan
M
,
Peshkin
L
,
Sorger
PK
,
Kirschner
MW
,
MacBeath
G
. 
A noncanonical Frizzled2 pathway regulates epithelial-mesenchymal transition and metastasis
.
Cell
2014
;
159
:
844
56
.
30.
Herrera-Molina
R
,
Valdivia
A
,
Kong
M
,
Alvarez
A
,
Cardenas
A
,
Quest
AF
, et al
Thy-1-interacting molecules and cellular signaling in cis and trans
.
Int Rev Cell Mol Biol
2013
;
305
:
163
216
.
31.
Singel
KL
,
Segal
BH
. 
Neutrophils in the tumor microenvironment: trying to heal the wound that cannot heal
.
Immunol Rev
2016
;
273
:
329
43
.
32.
Rosowski
EE
,
Huttenlocher
A
. 
Neutrophils, wounds, and cancer progression
.
Dev Cell
2015
;
34
:
134
6
.
33.
Finisguerra
V
,
Di Conza
G
,
Di Matteo
M
,
Serneels
J
,
Costa
S
,
Thompson
AA
, et al
MET is required for the recruitment of anti-tumoural neutrophils
.
Nature
2015
;
522
:
349
53
.
34.
Blaisdell
A
,
Crequer
A
,
Columbus
D
,
Daikoku
T
,
Mittal
K
,
Dey
SK
, et al
Neutrophils oppose uterine epithelial carcinogenesis via debridement of hypoxic tumor cells
.
Cancer Cell
2015
;
28
:
785
99
.
35.
Droeser
RA
,
Hirt
C
,
Eppenberger-Castori
S
,
Zlobec
I
,
Viehl
CT
,
Frey
DM
, et al
High myeloperoxidase positive cell infiltration in colorectal cancer is an independent favorable prognostic factor
.
PLoS One
2013
;
8
:
e64814
.
36.
Rao
HL
,
Chen
JW
,
Li
M
,
Xiao
YB
,
Fu
J
,
Zeng
YX
, et al
Increased intratumoral neutrophil in colorectal carcinomas correlates closely with malignant phenotype and predicts patients' adverse prognosis
.
PLoS One
2012
;
7
:
e30806
.
37.
Carus
A
,
Ladekarl
M
,
Hager
H
,
Pilegaard
H
,
Nielsen
PS
,
Donskov
F
. 
Tumor-associated neutrophils and macrophages in non-small cell lung cancer: no immediate impact on patient outcome
.
Lung Cancer
2013
;
81
:
130
7
.
38.
Wculek
SK
,
Malanchi
I
. 
Neutrophils fan cancer's flames
.
EMBO J
2015
;
34
:
2211
2
.
39.
Granot
Z
,
Fridlender
ZG
. 
Plasticity beyond cancer cells and the "immunosuppressive switch"
.
Cancer Res
2015
;
75
:
4441
5
.
40.
Wurtz
SO
,
Moller
S
,
Mouridsen
H
,
Hertel
PB
,
Friis
E
,
Brunner
N
. 
Plasma and serum levels of tissue inhibitor of metalloproteinases-1 are associated with prognosis in node-negative breast cancer: a prospective study
.
Mol Cell Proteomics
2008
;
7
:
424
30
.
41.
Pesta
M
,
Kulda
V
,
Kucera
R
,
Pesek
M
,
Vrzalova
J
,
Liska
V
, et al
Prognostic significance of TIMP-1 in non-small cell lung cancer
.
Anticancer Res
2011
;
31
:
4031
8
.
42.
Pennanen
H
,
Kuittinen
O
,
Soini
Y
,
Turpeenniemi-Hujanen
T
. 
Clinicopathological correlations of TIMP-1 and TIMP-2 in Hodgkin's lymphoma
.
Eur J Haematol
2004
;
72
:
1
9
.
43.
Seubert
B
,
Grunwald
B
,
Kobuch
J
,
Cui
H
,
Schelter
F
,
Schaten
S
, et al
Tissue inhibitor of metalloproteinases (TIMP)-1 creates a premetastatic niche in the liver through SDF-1/CXCR4-dependent neutrophil recruitment in mice
.
Hepatology
2015
;
61
:
238
48
.
44.
Schelter
F
,
Grandl
M
,
Seubert
B
,
Schaten
S
,
Hauser
S
,
Gerg
M
, et al
Tumor cell-derived Timp-1 is necessary for maintaining metastasis-promoting Met-signaling via inhibition of Adam-10
.
Clin Exp Metastasis
2011
;
28
:
793
802
.
45.
Saalbach
A
,
Haustein
UF
,
Anderegg
U
. 
A ligand of human thy-1 is localized on polymorphonuclear leukocytes and monocytes and mediates the binding to activated thy-1-positive microvascular endothelial cells and fibroblasts
.
J Invest Dermatol
2000
;
115
:
882
8
.
46.
Yang
ZF
,
Ho
DW
,
Ng
MN
,
Lau
CK
,
Yu
WC
,
Ngai
P
, et al
Significance of CD90+ cancer stem cells in human liver cancer
.
Cancer Cell
2008
;
13
:
153
66
.
47.
He
J
,
Liu
Y
,
Zhu
T
,
Zhu
J
,
Dimeco
F
,
Vescovi
AL
, et al
CD90 is identified as a candidate marker for cancer stem cells in primary high-grade gliomas using tissue microarrays
.
Mol Cell Proteomics
2012
;
11
:
M111
.