Background
Inflammatory activation, an interaction between host and cancer cell, is often triggered in cancer development and metastasis [
1,
2]. Meanwhile, anti-inflammation drugs have displayed certain anti-metastasis effects through less clarified mechanism [
3]. Neutrophils are the most abundant immune cells and have a fundamental role in inflammatory responses, but their contribution to metastasis is still controversial [
4‐
7]. It is reported that tumor-entrained neutrophils may induce an inhibitory process at the metastatic site [
8]. However, more studies indicate that neutrophil recruitment to the pre-metastatic site play critical roles in the metastatic initiation [
9].
Neutrophil extracellular traps (NETs) are extensive released extracellular web-like structures that are composed of cytosolic protein assembled on a scaffold of released chromatin from the activated neutrophils [
10]. They are originally discovered as the innate immune defensive process to trap and kill invading pathogens and then found wildly associated with various pathological conditions including autoimmune response, clot-related disorders, wound healing, organ impairment, and sterile inflammation [
11,
12]. The recognition of NETs in cancer is just emerging. Links have been made between NETs and metastasis in some mice models, and evidences of NETs’ presence have been reported in some certain tumors [
13‐
16]. Despite these, how NETs affect metastasis remains to be explored.
Hepatocellular carcinoma (HCC) accounts for the second leading cause of cancer-related death. The most frequent metastasis destinations of HCC are the liver and lung [
17]. High infiltration of tumor-associated neutrophils and elevated neutrophil-lymphocyte ratio (NLR) has been observed in HCC, which were correlated with worse outcome [
18‐
20]. But little is known about the specific role of neutrophils, especially NETs, on HCC metastasis. Thus, we designed this study to reveal participation of NETs in HCC metastasis and uncover the mechanism of NETs’ role on metastasis cascade. Moreover, we studied the inflammatory response triggered by NETs and combined breaking NETs plus anti-inflammation therapies against HCC metastasis.
Methods
Human specimens, animal models, and cell lines
Surgical samples or peripheral blood were obtained from HCC patients or healthy donors (HD) in our institute. Six to 8 weeks old of C57BL/6 male mice or null mice were used in animal studies. The human cell line HepG2 and L02 were obtained from Chinese Academy of Sciences. The human cell line MHCC97H and mice cell line Hepa1-6 were obtained from the Liver Cancer Institute, Fudan University. Detail information is described in Additional file
1.
To evaluate NET formation capacity, freshly isolated human or mice neutrophils were adjusted to a concentration of 5 × 10
5 cells/ml and stimulated with
Phorbol 12-myristate 13-acetate (PMA, 20 nM,
Sigma-Aldrich) for indicated hours, with or without DNase 1(100 U/ml,
Sigma-Aldrich) to allow NET formation. In CM-induced and plasma-induced NET formation assay, neutrophils were incubated with corresponding CM (1:2) for 30 min before PMA stimulation or incubated in plasma of HCC patients or HD. For co-culture, 1 × 10
5 indicated HCC cells were seeded on upper chamber, 5 × 10
5 species-matched normal neutrophils were seeded on lower chamber of 8-μM Transwell system for 16–20 h incubation. In LPS-induced NET model, isolated neutrophils were directly incubated for 4 h to form NETs. In some assays, the neutrophils were pretreated with hydroxychloroquine (HCQ, 50 μM, R&D) and Aspirin (5 mM,
Sigma-Aldrich) for 30 min before CM administration to inhibit CM-induced NET formation.
For visualization, neutrophils were seeded on 96-well plates for corresponding incubation, and cell-impermeable DNA dye SytoxGreen (Thermo Fisher Scientific, 1:10000) and cell-permeable DNA dye Hoechst33342 (Thermo Fisher Scientific, 1:1000) were added to the incubation system. At the end of incubation, the plates were directly moved to fluorescence microscope (Leica) for NET formation visualization. In some cases, neutrophils were seeded on coverslips in 24-well plates to generate NETs as described above, and then, the formed NETs were fixed for further immunofluorescence detection.
For quantification, NET DNA generated by neutrophils was digested with 500 mU/ml micrococcal nuclease (MNase). The nuclease activity was stopped with Ethylenediaminetetraacetic acid (EDTA, 5 mM), and the culture supernatants were collected and stored at – 80 °C until further use. NET DNA in the supernatants was quantified by PicoGreen® dsDNA Quantitation Reagent (Thermo Fisher Scientific) with fluorescence spectrometry under filter setting of 480 nm/520 nm excitation/emission and semi-quantitatively standardized to control group.
Preparation of NETs
Neutrophils were isolated and seeded on 6-well plates (1 × 107/well). Human neutrophils were stimulated with PMA (20 nM) for 4 h, and neutrophils from LPS-treated C57BL/6 mice were incubated in medium for 4 h to form NETs. Then, the supernatants were discharged carefully by slow suction and washed twice to eliminate residual PMA or NET-unassociated substances without disturbing NETs. RPMI (1 mL) containing MNase (1 U/mL) was then added to digest NETs at 37 °C for 20 min followed by 5 mM EDTA to stop nuclease activity. The supernatant containing NETs was collected and centrifuged to eliminate cell debris. Isolated NETs were stored at − 80 °C for further use.
Measurement of serum MPO-DNA level
We measured MPO-DNA complexes in human and mice serum using a well-adopted capture ELISA assay with some modification [
21]. Briefly, as the capturing antibody, 5 μg/mL anti-MPO monoclonal antibody was coated to 96-well plates overnight at 4 °C. After blocking in 1% BSA, 100 μl of diluted serum was added per well and incubated at room temperature on a shaking device for 2 h. After washing five times with PBST, PicoGreen® dsDNA Quantitation Reagent was added according to manufacturer’s directions. The values were then read with a fluorometer with a filter setting of 480 nm/520 nm excitation/emission and semi-quantitatively standardized to healthy donor or control group.
Mice model: LPS-induced NET model
We adopted the well-used lipopolysaccharide (LPS)-induced NET model [
14,
22]. Briefly, LPS (Sigma, 10 ug/mouse) was intraperitoneally injected to induce systemic inflammation in C57BL/6 mice. DNase 1 (100 U/mouse) was given intraperitoneally daily as abrogation 24 h prior to LPS. A neutrophil-depleting antibody (rat anti-Ly6G; clone 1A8 from BioXcell; 12.5 μg/mouse, intravenously) was given 24 h prior to LPS to deplete neutrophils. To verify NET formation in the inflammation model, mice were sacrificed in 6 h after LPS injection, serum was then collected for MPO-DNA level detection, and peripheral neutrophils were isolated for NET formation assay or generate NETs. The liver and lung were removed to prepare single-cell suspension for neutrophil infiltration detection using flow cytometry. The lung was also fixed by tracheal perfusion with 4% paraformaldehyde (PFA) for 15 min and embedded in O.C.T. compound for frozen sections and subsequent in situ immunofluorescence staining of NETs.
Mice model: establishment of experimental metastasis in LPS-induced NET model
In 6 h after establishment of the LPS-induced inflammation model in C57BL/6 mice, 2 × 106 Hepa1-6 cells were injected through the tail vein or spleen. DNase 1 (100 U/mouse) abrogation was then given daily. The mice were sacrificed, and intrahepatic/lung metastasis burden was assessed in 20 days. Experimental intrahepatic metastasis burden was assessed by calculating the percentage of hepatic tissue replaced by tumor (the hepatic replacement area, HRA). The lung metastatic lesions were directly counted on tissue sections using H&E staining.
In vitro assays on invasion, death rate, adhesion, and proliferation of HCC cells
Detail of invasion, death rate, adhesion, and proliferation assay of HCC cells under NET stimulation were described in Additional file
1.
Statistical analysis
The results are expressed as the means ± SEM. The statistical significance of differences between groups was determined by Student’s t tests. Pearson correlation test was used for correlation analysis. Kaplan-Meier method and log-rank test were used for follow-up data. GraphPad statistical software (version 5.0) was used for all statistical analyses. All data were analyzed using two-tailed tests unless otherwise specified, and P < 0.05 was considered statistically significant.
Further details of materials and methods are described in Additional files
1 and
3.
Discussion
Metastasis is a complex multistep cascade, which is related to both biological features of tumor cells and non-malignant tumor stroma. Accumulating evidences suggest that inflammation status of tumor microenvironment bridges host and cancer cells to affect metastasis cascade [
1]. Neutrophils, the most abundant host inflammatory cells, may influence multiple steps of metastasis. Neutrophils are often found in high numbers in human tumors and mice models, but there is a controversy with their roles in metastasis, depending on the different neutrophil subtypes or the different tumor types and the microenvironment studied [
7]. Neutrophils are found to be accumulated in HCC and associated with a worse outcome [
18]. Some non-specific inflammatory mediators from neutrophils have been found to influence HCC progression [
19,
27]. However, the specific functions of neutrophils in HCC metastasis remains to be illustrated. In this study, we demonstrated that neutrophils promoted HCC metastasis through forming enhanced NETs, which trapped HCC cells and further provoked their metastasis potential. Mechanically, NETs triggered a tumorous inflammatory response through activation of TLR4/9-COX2 axis to fuel metastasis.
NET formation is a unique functional process of neutrophils first described in host defense to trap and kill invading pathogen, with emerging recognition of which in non-infectious diseases and sterile inflammation [
10,
11]. The important roles of NETs have been described in some kinds of solid malignancies [
13,
15,
28]. One study has linked sterile inflammation-driven NETs with HCC tumorigenesis in mice with steatohepatitis [
29]. But the roles of NETs in HCC metastasis remain to be illustrated. In the present study, through various detection means including isolated neutrophils, sera and pathological samples in both mice models and human patients, we provide solid evidence to support that NET formation of neutrophils is enhanced in patients with HCC, especially those with metastatic HCCs. However, the really involved mechanism is not understood. We proposed that several secreted factors from HCC cells activated neutrophils towards NET formation or prime neutrophils for enhanced NETs with “second-hit” such as infection or stress. Some studies have indicated that NET-promoting effect is attributed to certain tumor-released cytokines or vesicles [
30,
31]. However, the pattern of secreted factors is highly distinct among various cancers. When expanded beyond cancer, the range of NET-promoting factors may even cover chromatin and lipid products [
32,
33]
.
The link between NETs and metastasis is getting appreciated. Biologically, metastasis is a low-efficient process where most of the disseminated cancer cells fail to seed and cease following cascade. Accumulating studies have suggested a seeding-supporting role of NETs to optimize the early adhesion of tumor cells to favor metastasis in different mice models [
14,
34–36]
. In consistent with these reports, here we also proved the dominant role of NETs to trap more disseminated HCC cells from circulation was necessary for the establishment of experimental metastasis of HCC and further found this effect was diminished in the case with normal neutrophils or NETs disturbed. But how NETs facilitate metastasis after trapping tumor cells is largely less known. NETs are equipped with toxic protease that cause cell damage [
24], which raises a possibility that NETs may restrict metastasis by killing trapped cancer cells with cytotoxicity in a similar pattern to eliminate pathogen [
24,
28]. This possibility is excluded, since the present study has demonstrated that the trapped HCC cells are not affected by the potential cytotoxicity of NETs. And more, the invasiveness of HCC cells is enhanced after surviving from NETs. These suggest that certain key defense/survival event is triggered by NET challenge in trapped HCC cells which acquired a higher metastasis potential. The acquired invasiveness enhancement as a survival mechanism has been described in cancer cells upon potential deadly extracellular stress [
37]. NETs may act as a beneficial stress on HCC cells in a similar pattern.
Then comes an interesting question that how these trapped HCC cells withstand and utilize NETs to enhance their metastatic potential. Here, we have found NETs induce an aggressive inflammatory response in the trapped HCC cells featured as COX2 upregulation through activating TLR4/9 to enhance metastatic potential of the trapped HCC cells. Through RNA-seq, we identified COX2 as the key event of NET-triggered metastasis potential. NETs are of strong immunostimulatory capacity and known to license macrophages and other host cells for cytokine production in vitro [
38]. This pro-inflammatory effect of NETs is also reported in mice [
39]. For the first time, we found the effects of NETs on inflammatory response in boosting metastasis behavior of cancer cells. Elevation of COX2 is associated with a higher metastasis behavior, including protection from cell death, induction of invasion, stimulation of angiogenesis, and inhibition of immunosurveillance [
40‐
42]. COX2 is also an appealing therapeutic target, and targeting COX2 has certain anti-metastasis effects even as a single intervention alone [
3]. We have found when COX2 is blocked, trapped HCC cells lose the counteraction to NET potential cytotoxicity and acquire no enhanced invasiveness from NETs. These strongly suggested the induction of COX2 as the key molecular event responsible for NET-enhanced metastatic capacity in trapped HCC cells. Moreover, we revealed activation of TLR4/9 as the intermediate link between NETs and induction of the inflammatory response in trapped HCC cells. TLR4/9 are important sensors of several damage-associated molecular patterns (DAMPs) and mediate cellular communication among host cells, and their activation also represents a highly metastatic phenotype [
25,
43]. NETs contain several DAMPs that may be recognized by TLR4/9 [
44]. TLR signals mediate the pro-inflammatory effect of NETs on several host cells [
45,
46]. Many reports have demonstrated that NETs could upregulate TLR9 expression in colon cancer cells and that TLR is an upstream regulator of COX2 expression [
36,
47,
48]. By blocking TLR4/9, we have found NETs failed to induce COX2 or trigger metastasis capacity in trapped HCC cells. These findings suggest TLR4/9 activation and subsequent COX2 induction as the key signaling in NET-triggered metastasis potential.
DNase 1 is well acknowledged to digest extracellular chromatin and NETs. Endogenous DNase 1 is a vital physiological regulation of NETs in host and in adequate clearance of NETs due to low level or bio-activity of endogenous DNase 1 which may lead to dysregulation of NETs, thus causing autoimmune disease and other inflammatory disorders [
49–51]. Many studies have revealed an association of DNase 1 polymorphism with the susceptibility of autoimmune disease such as systemic lupus erythematosus (SLE), but the correlation between endogenous DNase 1 and cancer remains to be studied [
52]. As a therapeutic mean, DNase 1 has demonstrated a satisfied effect in digesting NETs in several preclinical models and confirmed safety in cystic fibrosis and SLE [
53]. However, DNase 1 alone against NETs has certain limitation. The blood concentration of given DNase 1 is found less stable [
54]. Besides, the fact that DNase 1 dismantles NET structure but does not totally degrades protein components of NETs indicate its less effectiveness in abrogating NET-triggered inflammatory response [
53]. Combination of DNase 1 and other available means provides a solution. Targeting COX2 has been well acknowledged to have both anti-inflammatory capacity and anti-tumor effect through multiple mechanisms [
3]. A capacity of anti-inflammatory drugs to decrease NETs is also demonstrated [
40]. Based on our finding of NETs fueling HCC metastasis through activating tumorous inflammatory response, we adopted anti-inflammatory drugs aspirin and HCQ to block COX2 and upstream TLR4/9 activation complementary to DNase 1 and proved well efficiency in inhibition of HCC metastasis through multiple perspectives. These combination therapies could block or digest NETs and abrogate the triggered metastasis potential of trapped HCC cells by undissolved NETs, featuring a new use of old anti-inflammatory drugs. More combination strategies with DNase 1 against metastasis are to be developed.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.