Human neutrophils are the most abundant leukocytes and serve as key effectors in the first-line host defence against infectious micro-organisms 1, 2. They circulate in the bloodstream after leaving the bone marrow, whereupon they undergo spontaneous apoptosis 3, 4. Under the influence of local conditions, they acquire specialized phenotypic characteristics with diverse functional programmes. N1 or conventional neutrophils are capable of efficiently killing micro-organisms and tumour cells. In contrast, N2 neutrophils are distinctly polarized by anti-inflammatory molecules, such as TGF-β, and have low cytotoxicity, produce significant pro-angiogenic factors and matrix-degrading enzymes, and also suppress the anti-tumour immune response 5, 6.
Neutrophils are the common inflammatory infiltrate in tumours, and tumour-associated neutrophils (TANs) are derived almost entirely from circulating blood neutrophils 7, 8. In our latest study, we demonstrated that pro-inflammatory IL-17-producing cells recruited blood neutrophils into the peritumoural stroma of hepatocellular carcinoma (HCC) by epithelium-derived CXC chemokines, and those neutrophils in turn promoted angiogenesis at the adjacent tumour-invading edge via MMP-9 signalling. Notably, exposure of neutrophils to tumour-derived factors resulted in a prolonged release of MMP-9 in vitro9. These results strongly indicate that, besides attracting neutrophils, tumour microenvironments may also administrate the lifespan of those cells to perform sustained pro-tumourigenic effects, although the relative mechanism remains largely unknown.
Tumours can mimic some of the signalling pathways of the immune system to propagate conditions for disease progression 10. PI3K/Akt signalling has been implicated in regulating the activation of antigen-presenting cells (APCs), and activation of PI3K/Akt signalling has been shown to induce the production of pro-inflammatory cytokines and chemokines, potentially linking innate and adaptive immunity 11. Furthermore, although not directly related to TANs, selective pathways that are triggered by some TLRs, in particular, TLR2, TLR4, and TLR9, can activate the PI3K pathway in neutrophils, and such activation of cell signalling cascades elicits responses leading to angiogenesis and tissue remodelling 12, 13. Thus, immune signalling data of neutrophils in the cancer environment are essential for understanding their roles and potential mechanisms in tumour immunopathogenesis.
Tumour microenvironments comprise both cellular and non-cellular (matrix) components 14, 15. Hyaluronan (HA), one of the major glycosaminoglycans of the extracellular matrix that relate to cell motility, has been shown to undergo rapid degradation at sites of inflammation 16–18. We have recently found that intermediate-sized HA fragments released by human solid tumours can alter the normal development of APCs that is intended to trigger the transient early activation of monocytes in the peritumoural region 19. The present study shows that neutrophils accumulated in the peritumoural stroma and the cancer nest of several types of human solid tumours. Tumour-derived HA fragments triggered the activation of neutrophils by TLR4/PI3K interactions. These activated neutrophils up-regulated the metastatic potential of malignant cells via a contact-dependent mechanism. Moreover, we found that blocking PI3K activity in neutrophils effectively attenuated their pro-tumourigenic effects.
Materials and methods
Patients and specimens
Tumour samples were obtained from 120 patients with pathologically confirmed hepatocellular, cervical, colorectal, and gastric carcinomas (n = 30 for each) at the Cancer Center of Sun Yat-sen University. None of the patients received anti-cancer therapy before sampling. Clinical stages were classified according to the International Union against Cancer. The clinical characteristics of all patients are summarized in the Supporting information, Supplementary Table 1. All samples were anonymously coded in accordance with local ethical guidelines (as stipulated by the Declaration of Helsinki). Written informed consent was obtained from the patients and the protocol was approved by the Review Board of Sun Yat-sen University.
Immunohistochemistry and immunofluorescence
Paraffin-embedded samples were cut into 5 µm sections and processed for immunohistochemistry as previously described 20. Following incubation with anti-CD15 Ab (Lab Vision Corporation; Fremont, CA, USA), the sections were stained using the Envision System with diaminobenzidine (DakoCytomation, Glostrup, Denmark). Thereafter, to evaluate the density of CD15+ cells, five respective areas were measured at × 400 magnification and the nucleated CD15+ cells in each area were counted and expressed as the number of cells per field. For immunofluorescence analysis, frozen sections of human HCC tissues were stained with anti-CD15 Ab, followed by Alexa Fluor 488-conjugated goat anti-mouse IgG and propidium iodide. Cytocentrifuged neutrophils were stained with mouse anti-PI3K p85α Ab (Santa Cruz Biotechnology, Santa Cruz, CA, USA) followed by Alexa Fluor 488-conjugated goat anti-mouse IgG.
Cell lines and preparation of tumour culture supernatants (TSNs)
Human hepatocellular (SK-Hep-1 and HepG2), cervical (HeLa), colorectal (Caco2), and gastric (SGC-7901) carcinoma cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA); normal liver (L02) cells were obtained from the Institute of Biochemistry and Cell Biology at the Chinese Academy of Sciences; and the stable HA synthase 2-knockdown SK-Hep-1 clones (SiHAS2) and mock transfectants were established in our previous study 19. All cells were tested for mycoplasma contamination using a single-step PCR method 21. The TSN or culture supernatant from normal liver cells was prepared as previously described 19.
Preparation of the HA fragments
Intermediate-sized HA fragments (50–200 kD) were prepared by partial digestion of high-molecular mass HA with limited amounts of testicular hyaluronidase as previously described 22. In short, high-molecular mass HA from rooster comb (H1751) was dissolved in digestion buffer [10 mM CH3COONa (pH 4.0), 500 U/ml hyaluronidase from bovine testes, and 5 mg of HA/ml] and incubated at 37 °C for 30 min. The reaction was terminated by putting the digestion mixture into a boiling water bath. The sizes of the HA fragments were determined by 0.5% agarose gel electrophoresis and visualized with the cationic dye Stain-All.
Neutrophil isolation and culture
Human neutrophils were isolated from peripheral blood of healthy donors as previously reported 9. Neutrophils of ∼98% purity in DMEM containing 10% FBS were plated at 5 × 106 per well in the presence or absence of HA or 30% cell culture supernatants from different cell lines. In some experiments, the cells were pretreated with polymyxin B (an LPS sequestering agent), a PI3K inhibitor (LY 294 002), an IκBα inhibitor (Bay 11–7082), an Erk1/2 inhibitor (PD 98 059), an anti-CD44 antibody (Ab) (Lab Vision Corporation; Fremont, CA, USA), an anti-TLR4 Ab (R&D Systems, Abingdon, UK), or a control Ab, or with HA-specific blocking peptide Pep-1 19 or control peptide at indicated concentrations before exposure to HA or cell culture supernatants.
Assessment of neutrophil apoptosis
Neutrophil apoptosis was quantified using an annexin V apoptosis detection kit according to the manufacturer's instructions (R&D Systems). The binding of annexin V-FITC and PI to the cells was measured by flow cytometry (FACS Vantage-SE, BD Immunocytometry Systems, San Diego, CA, USA) using Cell Quest software.
Concentrations of TNF-α and IL-1β in the culture supernatants were determined using ELISA kits according to the manufacturer's instructions (eBioscience, San Diego, CA, USA).
The proteins from neutrophils were extracted as previously described 19, 23. Equal amounts of cellular proteins were separated by 10% SDS-PAGE; immunoblotted with Abs agaist Bax, Mcl-1, and β-actin (Santa Cruz Biotechnology), or total Akt, STAT3, IκBα, p38, Erk1/2, JNK, and their phosphorylated forms (Cell Signaling, Beverly, MA, USA); and visualized with an ECL kit.
In vitro assays of migration
The migration assay was performed in a 24-well Boyden chamber with an 8-µm pore size polycarbonate membrane (Corning Life Sciences - Axygen Inc, Union City, CA, USA) as previously described 24. Tumour cells (3 × 104 in 100 µl of serum-free medium) were added to the upper compartment of the chamber, whereas the lower compartment was filled with 600 µl of DMEM containing 10% FBS. After incubation for 10 h, the tumour cells remaining on the upper surface of the membrane were removed. The migrated tumour cells on the lower surface of the membrane were rinsed with PBS containing 5 mM EDTA for 5 min to remove residual neutrophils and subsequently stained with crystal violet after fixation and then counted under a light microscope. For co-culture studies, the neutrophils (1.5 × 105) were pre-seeded in the lower or upper compartment.
The data on cytokine concentrations and cell apoptotic percentages are given as means ± SEM and SD, respectively. Statistical significance was determined by Student's t-test. A value of p < 0.05 was considered statistically significant.
Infiltration patterns of neutrophils in human solid tumours
In individuals with untreated malignant hepatocellular, cervical, colorectal, and gastric carcinomas (n = 30 for each), we identified a substantial population of CD15+ cells that were considered neutrophils by immunohistochemical staining in tumour samples (Figure 1A and Supporting information, Supplementary Figure 1). In all the tumours analysed, the CD15+ cells were more abundant in the peritumoural stroma than in the cancer nest (Figure 1B). In contrast, only a small fraction of CD15+ cells was observed in paired non-tumour tissues (2–5 cells per field; Figures 1A and 1B). Moreover, we found that peritumoural stroma CD15+ cell density was also associated with intrahepatic metastasis in liver cancer or lymph node metastasis in gastric cancer (Figure 1C). These data suggest that neutrophils represent an important myeloid cell component in human tumour stroma that may contribute to the disease progression.
Induction of long-lived neutrophils by supernatants from cells of solid tumours
The lifespan of neutrophils is quite short. They circulate in the bloodstream after leaving the bone marrow, whereupon they undergo spontaneous apoptosis 3. The accumulation of neutrophils in human cancer specimens suggests that the tumour microenvironment may affect the survival of neutrophils. To test this assumption, neutrophils of ∼98% purity were left untreated or cultured with supernatants from four tumour and one normal liver cell lines. The neutrophils cultured alone for 12 h underwent apoptosis, and 25.7 ± 4.5% of the cells were positive for annexin V. Exposure of neutrophils to 30% TSNs originating from HepG2, HeLa, Caco2, and SGC-7901 cells, respectively, significantly attenuated the spontaneous apoptosis (Figure 2A). In contrast, the culture supernatant from normal liver (L02) cells did not affect the survival of neutrophils (Figure 2A), even when used at a high concentration (50%).
Neutrophils exhibited a delayed onset of apoptosis after they were activated 25 and hence we examined the activation status of neutrophils after exposure to the TSN. The results showed that TSN originating from tumour cells effectively activated neutrophils to produce significant levels of TNF-α and IL-1β (Figure 2B). Polymyxin B did not affect the TSN-induced activation of neutrophils, whereas it effectively blocked the activity of LPS in parallel experiments (Figure 2B), which suggests that the TSN that we used triggered neutrophil activation via a mechanism different from that associated with LPS. Supporting the above result that TSN inhibited apoptosis of neutrophils, TSN-exposed neutrophils expressed significant Mcl-1, an anti-apoptotic protein structurally related to Bcl-2, whereas they exhibited attenuated expression of Bax, a pro-apoptotic protein of the Bcl-2 family (Figure 2C). Collectively, these results show that soluble factors released by cancer cells induce activation of neutrophils and in that way lead to the accumulation of those cells in tumours.
Essential role of PI3K/Akt signalling in TSN-mediated activation of neutrophils
PI3K/Akt signalling has been implicated in the regulation of the activation of APCs 11. To investigate whether such a mechanism is also responsible for the activation of neutrophils in the cancer environment, we used immunofluorescence and immunoblotting to analyse the PI3K/Akt activity in TSN-treated neutrophils. In control neutrophils, the regulatory subunit p85 of PI3K was equably distributed in the cytoplasm. Upon exposure to the TSN from HepG2 cells, this subunit anchored to the membrane of cells, suggesting an activated form (Figure 3A and Supporting information, Supplementary Figure 2). Consistent with this result, Akt, an important PI3K downstream effector, was vigorously phosphorylated soon after exposure to TSN (Figure 3B). Similar results were obtained when we exposed neutrophils to the TSNs from HeLa, Caco2, and SGC-7901 cells (data not shown).
To assess the effect of PI3K/Akt signalling on tumour neutrophil activation, we pretreated neutrophils with LY 294 002, a selective phosphatidylinositol inhibitor for PI3K, and subsequently exposed them to the TSN from HepG2 cells. As shown in Figure 3C, such treatment, at concentrations up to 10 mM, completely abolished the TSN-induced phosphorylation of Akt in neutrophils, whereas it did not affect that in untreated neutrophils (Figure 3C). As expected, blocking PI3K/Akt activity could effectively inhibit the activation of TSN-treated neutrophils showing decreased production of TNF-α and IL-1β as well as impaired expression of Mcl-1 and attenuated survival (Figures 3D–3F). Of note, the PI3K inhibitor, even at concentrations up to 30 mM, did not up-regulate the spontaneous apoptosis of untreated neutrophils (Figure 3F). These data suggest that PI3K/Akt signalling selectively contributes to tumour-mediated neutrophil activation.
To further elucidate the mechanisms involved in the induction of long-lived neutrophils by the tumour environment, we analysed the activation of the STAT3, NF-κB, and MAPK (Erk1/2, JNK/SAP, and p38) signalling pathways in neutrophils in response to the TSN. As shown in Figure 4A, among these pathways, only Erk1/2 (an important MEK downstream effector) and IκBα (an NF-κB-inducing kinase) were more activated in TSN-exposed neutrophils. Moreover, blocking the activation of NF-κB using Bay 11–7082, an inhibitor of IκBα, effectively inhibited the survival of untreated and TSN-exposed neutrophils, whereas inhibition of Erk1/2 activation had no effect (Figures 4B and 4C). Thus, these findings imply that the NF-κB pathway is also involved in the intrinsic survival of neutrophils.
Tumour-derived HA caused early activation and subsequent longevity of neutrophils
We have recently observed that HA fragments constitute a common factor produced by several types of human tumours, including hepatocellular and cervical carcinomas, to stimulate the pro-inflammatory responses of APCs 19, 23, 26. In the present study, we found that intermediate-sized HA fragments, but not high-molecular-weight HA fragments (data not shown), can mimic the effect of TSN to induce the activation of neutrophils that exhibited marked activation of PI3K (Figure 5A and Supporting information, Supplementary Figure 2), increased production of TNF-α and IL-1β (Figure 5B), and up-regulation of Mcl-1 (Figure 5C). Similarly, these HA-conditioned neutrophils displayed a delayed onset of apoptosis, as mediated by the TSN (Figure 5D), and this phenomenon was not affected by polymyxin B (data not shown).
To further confirm the role of HA in TSN-induced activation of neutrophils, we pretreated neutrophils with agents that antagonize the interactions between HA and its receptors, and then exposed the cells to the TSN. In support, pretreatment with HA-specific blocking peptide Pep-1 effectively attenuated the cytokine production and restored the spontaneous apoptosis of neutrophils (Figures 5E and 5F). Both CD44 and TLR4 are the major cell surface receptors for HA fragments 27, 28; thus, we conducted experiments using the Abs against these receptors, respectively. Interestingly, the activation of TSN-exposed neutrophils was partially inhibited by the anti-TLR4 Ab, whereas it was not affected by treatment with anti-CD44 Ab (Figures 5E and 5F). Moreover, we also observed that such activation and increased survival of cells were significantly impaired by reducing HA levels in the TSN via the silencing of HA synthase 2 in tumour cells (Figures 5G and 5H). These findings indicate that HA fragments released by tumours stimulate the activation of neutrophils via the TLR4 receptor.
TSN-activated neutrophils increased the motility of malignant cells via a contact-dependent mechanism
The results described above show that increased peritumoural stromal neutrophil density correlates well with metastasis in patients with HCC or gastric cancer. We therefore investigated the motility of malignant cells in the presence or absence of neutrophils. Direct co-culture of HepG2 cells with neutrophils that were exposed to the TSN from HepG2 cells displayed a five-fold increase in cell motility, and such up-regulated migration was totally lost when the TSN-exposed neutrophils were seeded in the lower chamber (Figures 6A and 6B). In contrast, untreated neutrophils or neutrophils pretreated with the supernatant from normal liver cells only marginally enhanced the cell migration (Figure 6A). These data indicate a cell contact-dependent mechanism during neutrophil-mediated malignant cell migration in the tumour environments. Furthermore, in support of the above-mentioned findings that tumour-derived HA induced neutrophil activation, HA-exposed neutrophils also directly increased the malignant cell migration (Figure 6C). Also, consistent with the results for PI3K/Akt-mediated neutrophil activation, the up-regulated metastatic potential of malignant cells induced by TSN-exposed neutrophils was significantly attenuated by pretreatment with LY 294 002 (Figure 6D). All of these results reveal that activation of neutrophils by tumour-derived HA contributes to malignant cell migration.
After several decades of neglect, myeloid cells are once again an area of active research interest in cancer 29, 30. The best studied myeloid cell population in human cancer is the well-known tumour-associated macrophages, and macrophages in most solid tumours disable antigen-specific T-cell immunity 31–33. We now report another important myeloid cell population in patients with hepatocellular, cervical, colorectal, and gastric carcinomas, namely tumour-associated neutrophils (TANs). The present study shows that HA fragments released by most solid tumour cells activate neutrophils through TLR4/PI3K/Akt signalling interactions to promote tumour motility, suggesting a pro-tumourigenic regulatory loop between different types of cells within a tumour.
Although TANs originate from circulating blood neutrophils, they exhibit distinct functional characteristics with impaired bactericidal activities, but increased potential in regulating angiogenesis and tissue remodelling 34. We recently found that pro-inflammatory IL-17-producing cells recruited blood neutrophils into the peritumoural stroma of HCC by epithelium-derived CXC chemokines 9. In the present study, we observed that the TSN from several different kinds of tumour cells effectively induced the innate activation of neutrophils, which is characterized by high expression of pro-inflammatory cytokines and intracellular Mcl-1 for cell survival, and a delayed onset of apoptosis. However, these activated neutrophils were unable to kill tumour cells and instead they promoted the in vitro migration of malignant cells via a contact-dependent mechanism, which suggests that such neutrophils can actually benefit tumour progression. This notion is supported by our recent finding that the density of neutrophils in the peritumoural stroma was correlated with advanced intrahepatic metastasis in HCC patients 9. Consistent with our results, other investigators reported that breast cancer cells stimulated neutrophils to produce oncostatin M in a cell-contact mechanism, which in turn increased the metastatic potential of malignant cells 35.
PI3K/Akt signalling is tightly associated with advanced progression of human tumours 36, but the nature and regulation of PI3K/Akt signalling in human TANs remain largely unknown. The present study provided evidence that TSN-mediated phosphorylation of PI3K/Akt signalling regulated the activation and sustained pro-tumourigenic effects of neutrophils. Comparison of the kinetics of cytokine production and PI3K/Akt activation in TSN-treated neutrophils revealed that the phosphorylation of PI3K and Akt preceded the accumulation of TNF-α and IL-1β. Furthermore, agents that prevent PI3K/Akt phosphorylation were found to block TNF-α and IL-1β production and Mcl-1 up-regulation. In addition, in vitro migration assays using LY 294 002-treated TSN-exposed neutrophils indicated that activation of PI3K/Akt signalling in neutrophils was essential for the increased motility of malignant cells. Such sequential neutrophil activation and enhanced malignant cell migration in tumours may reflect a novel immune-editing mechanism by which tumours render TANs able to perform a sustained pro-tumourigenic function by stimulating PI3K-mediated neutrophil activation. This hypothesis is compatible with our latest studies showing that tumours educate macrophages to perform a suppressive role by inducing an early pro-inflammatory response of monocytes 19, 26.
In cancer patients, HA concentrations are usually higher in malignant tumours than in corresponding benign or normal tissues 37–39. We and others have recently revealed that not only does HA promote tumour growth and metastasis, but it also modulates the functions of APCs 16, 19, 40. In this context, the results of four sets of experiments in our investigation provide evidence that HA is a common factor that is produced by several types of solid tumour cell lines to induce the activation and pro-tumourigenic effect of neutrophils. First, we found that purified HA fragments were able to mimic the ability of the TSN to induce the activation of neutrophils. Second, pretreatment with an anti-TLR4 mAb or Pep-1 to antagonize the interactions between HA and its receptors partially inhibited the TSN-mediated activation of neutrophils. Third, silencing of HAS2 in tumour cells, which partially reduced the HA levels in the TSN, attenuated the ability of those cells to induce cytokine production in neutrophils. Fourth, exposure of malignant cells to HA-treated neutrophils significantly enhanced their metastatic potential. Therefore, HA fragments generated in tumour environments may constitute a common mediator that regulates TAN functions via TLR4 receptors. This notion is supported by a recent finding that the TLR4 specific agonist is an efficient activator for neutrophil activation and survival in vitro12. Moreover, CD44 receptors have been shown to play an important role in HA-mediated neutrophil recruitment 41, 42. However, blockade of CD44 in our co-culture system could not attenuate the production of TNF-α and IL-1β in neutrophils. These findings suggest distinct roles between CD44 and TLR4 receptors in neutrophils.
In all of our experiments, anti-TLR4 mAb or Pep-1 could only partially inhibit the activation of TSN-exposed neutrophils, which indicates that the effect of the TSN involved additional soluble factors from tumour cells. Besides producing numerous immunosuppressive cytokines (eg GM-CSF and IL-6) and chemokines, tumour cells also secrete other matrix components, such as fibronectin, which have been shown to modulate the migration of neutrophils via TLR4 receptors and ultimately alter the functional characteristics of the cells 8, 43, 44. Therefore, characterization of the underlying mechanisms may provide new avenues for the development of novel immune-based therapies in human cancer.
Our results give important new insights into the formation of TANs in tumours. HA and other soluble factors derived from cancer cells can drive neutrophils to acquire an activated phenotype. After interacting directly with cancer cells, the tumour-activated neutrophils act via a contact-dependent pathway to promote the motility of malignant cells, which suggests a positive regulatory loop between tumours and their stroma. In support of this conclusion, we observed that the number of CD15+ cells in primary HCC was inversely associated with the overall survival of the patients, which can serve as an independent predictor of relapse-free survival and overall survival 9. Therefore, it is possible that studies of the mechanisms that can selectively modulate the activation of neutrophils will provide a novel strategy for anti-cancer therapy 45.
We thank RQ Weaver for linguistic revision of the manuscript. This work was supported by the Fundamental Research Funds for the Central Universities (11lgzd12) and project grants from the National Natural Science Foundation of China (81000915) and the Education Department of Guangdong Province (LYM10008).
Author contribution statement
YW designed and performed most of the research. QZ and CP performed western blot. XFL and LS performed the confocal analysis. DMK designed and supervised the research and wrote the manuscript.
SUPPORTING INFORMATION ON THE INTERNET
The following supporting information may be found in the online version of this article.
Figure S1. Staining of CD15+ cells in peritumoural stroma of HCC tissues.
Figure S2. Colour visions of Figure 3A (A) and Figure 5A (B).
Table S1. Clinical characteristics of patients with hepatocellular, cervical, colorectal, or gastric carcinomas.