Periodontal inflammation recruits distant metastatic breast cancer cells by increasing myeloid-derived suppressor cells

Abstract

Periodontal diseases can lead to chronic inflammation affecting the integrity of the tooth supporting tissues. Recently, a striking association has been made between periodontal diseases and primary cancers in the absence of a mechanistic understanding. Here we address the effect of periodontal inflammation (PI) on tumor progression, metastasis, and possible underlining mechanisms. We show that an experimental model of PI in mice can promote lymph node (LN) micrometastasis, as well as head and neck metastasis of 4T1 breast cancer cells, both in early and late stages of cancer progression. The cervical LNs had a greater tumor burden and infiltration of MDSC and M2 macrophages compared with LNs at other sites. Pyroptosis and the resultant IL-1β production were detected in patients with PI, mirrored in mouse models. Anakinra, IL-1 receptor antagonist, limited metastasis, and MDSC recruitment at early stages of tumor progression, but failed to reverse established metastatic tumors. PI and the resulting production of IL-1β was found to promote CCL5, CXCL12, CCL2, and CXCL5 expression. These chemokines recruit MDSC and macrophages, finally enabling the generation of a premetastatic niche in the inflammatory site. These findings support the idea that periodontal inflammation promotes metastasis of breast cancer by recruiting MDSC in part by pyroptosis-induced IL-1β generation and downstream CCL2, CCL5, and CXCL5 signaling in the early steps of metastasis. These studies define the role for IL-1β in the metastatic progression of breast cancer and highlight the need to control PI, a pervasive inflammatory condition in older patients.

Fig. 1

Periodontal inflammation promotes micrometastasis and shows enhanced MDSCs in cervical lymph nodes. a Timeline of the PI model, LPS was injected directly in the gingiva of the mice for 3 weeks. b Gingiva from PBS and LPS treated mice were stained by F4/80 and FACS quantification of M2 macrophages (F4/8+/CD206+) was assessed (n = 3; Bar represents 100 μm). c Timeline of the early metastasis model 1 established by intracardiac injection of 4T1 cells. d In early metastasis model 1, the axillary and cervical lymph nodes (LNs) were collected for luciferase staining by IHC (Bar represents 200 μm). e The numbers of visible LN count (left), as well as the ratio of luciferase-positive LN to total LN were shown (right; n = 4). f The percentage of metastasis was determined by measuring the ratio of metastasis area to total lymph node area per histologic field. (n ≥ 4; LNs lymph nodes; student’s t test). g Timeline of the early metastasis model 2, mice were injected in the mammary fat pad with 4T1 cells and animals were euthanized at 5 weeks. h MDSC and macrophages were measured by FACS in the LN and in spleen. The level of MDSC and macrophages in tumor group were set as 1, then other groups were counted as relative ratio to the tumor group. MDSC and M2 macrophages were significantly elevated in the cervical LN of PI + tumor animals (n ≥ 7; LNs lymph nodes; student’s t test). (*P < 0.05; ****P < 0.0001)

Fig. 2

LPS leads to pyroptosis and IL-1β expression of gingival fibroblasts in vitro. a Gingival fibroblasts were treated with LPS (1 μg/ml) for 24 h. The mRNA level of TLR2 and TLR4 were measured by RT-PCR. b Reactive oxygen species (ROS) were determined by DCFH-DA staining (bar = 100 µm). c Gingival fibroblasts were treated with LPS for 6 and 24 h. Protein expressions of IKKβ (phosphorylated and total), p65 NF-κB (phosphorylated and total), NLRP3, caspase 1 (pro- and cleaved p10), caspase 3 (pro- and cleaved p20), and IL-1β were measured by western blot. Cleaved p10 caspase-1 is depicted as C-cas1. Band intensities were assessed using Quantity One. Data (n = 3) are shown as means ± SEM. d Gingival fibroblasts were treated with LPS (1 μg/ml) for 24 h and 14 d. IL-1β mRNA level was measured by real-time PCR. (n = 3) (E) Gingival fibroblasts were treated with LPS for 24 h, exogenously secreted IL-1β was measured by ELISA. (n ≥ 4; student’s t test, *P < 0.05; **P < 0.01; ****P < 0.0001)

Fig. 3

Pyroptosis/IL-1β pathway is enhanced in vivo and can be modulated by anakinra. a In the PI model, phosphorylated-p65 NF-κB, NLRP3, 8-OHdG, IL-1β, and TUNEL were detected by IHC (Bar, 100 μm). b IL-1β was immuno-localized in gum tissues of mice with PI alone and with orthotopic mammary tumor introduction (Bar, 100 μm). c The mRNA expression of IL-1β was measured in the gingiva of the indicated groups. Serum IL-1β was detected by ELISA (n ≥ 5; Student’s t test). d The size of primary tumors was measured (n ≥ 7). e Early metastasis in lymph nodes were measured by luciferase assay and normalized by protein concentration, where anakinra reduced the metastasis to the axillary and cervical lymph nodes. MDSC content was also decreased by anakinra in cervical lymph node. (n ≥ 7; Ank anakinra; ANOVA, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001)

Fig. 4

LPS increases expressions of inflammatory genes and/or chemoattractants. Anakinra partly diminishes the effects of LPS. a Gingival fibroblasts were treated with LPS for 24 h, with or without anakinra in vitro. Cytokine assay showed anakinra inhibited the expressions of CXCL12 and CCL2, which were increased by LPS. b, c mRNA level of CXCL5, CCL2, CCL5, and CXCL12 were measured in vitro. LPS enhanced all the chemokines and anakinra inhibited them except for the mRNA level of CXCL12 (n = 3; ANOVA or student’s t test). d, e mRNA level of CXCL5, CCL2, CCL5, and CXCL12 in the gingival tissue were measured in vivo in the 3-animal group. (n = 5; Ank anakinra, ANOVA or student’s t test, *P < 0.05; **P < 0.01; ****P < 0.0001)

Fig. 5

Human gingiva from periodontitis patients show activation of MDSC and express inflammatory chemokines. a NLRP3, 8-OHdG, cleaved-caspase 1, and IL-1β were detected in periodontitis specimens by IHC. (Bar represents 50 μm) b human CD33+ and CD11b+ MDSC subsets were detected by IHC (three healthy gingiva and 5–8 periodontitis specimens). The number of CD33+ or CD11b+ cells were increased in periodontitis (student’s t test; Bar represents 50 μm) c human gingival fibroblasts were cultured in vitro and were stimulated by E.coli LPS or P. gingivalis LPS for 24 h. CCL2, CCL5, CXCL5, and CXCL12 in cell supernatant were measured by ELISA. E.coli LPS and P. gingivalis LPS highly increased the secretion of CCL2, CCL5 compared with the control. The secretion of CXCL5 was merely increased by E.coli LPS. (n = 5; student’s t test, *P < 0.05; **P < 0.01; ****P < 0.0001)

Fig. 6

Periodontal inflammation can promote head and neck metastasis. a Timeline of the late metastasis model, PI was induced with LPS for 3 weeks then mice were injected in the mammary fat pad with 4T1 cells, the primary tumor was let to grow until it reached 100 mm³ in size, surgery was accomplished to extract this tumor and animals were let to grow metastasis and euthanized at 10.5 weeks. b Metastasis were observed and counted in lungs and lymph node in the individual animal groups in the presence or absence of anakinra. (n ≥ 3; ANOVA). c IHC staining of Pan-cytokeratin and Gr1 from the lymph node and the cervical tissue of late metastasis model. (n ≥ 3; Ank anakinra, ANOVA, *P < 0.05; **P < 0.01; ****P < 0.0001)

Fig. 7

Multiplex protein detection of head and neck and lymph node metastasis. CyTOF based detection of phenotypic and functional markers on paraffin embedded tissue sections of the late stage tumor-bearing mouse control, periodontitis, and periodontitis model administered anakinra. a Evaluation of CD11b, Ly6C, Ly6G, TGF-β, CD163, IL-10, and IL-2 expression were detected in lymph nodes from the late breast cancer metastasis model. b Detection of these same markers were performed on the neck tissues of the same model. Arrowhead indicates tumor metastasis. The overlaid pseudo-colored images for MDSC and macrophage here are individually illustrated in Supplementary Fig. S3

Fig. 8

Possible mechanism underling periodontal inflammation-induced metastasis. Bacterial wall component, LPS, induces caspase 1 activation and proptosis pathway signaling in gingival fibroblasts in instances of PI. The proinflammatory cytokine, IL-1β, as well as CCL2, CCL5, and CXCL5 are produced and initiate M2 macrophage. In the early stages of primary breast tumor progression, where there is no expected metastasis, IL-1β and the produced cytokines recruit M2 macrophages, as well as MDSC to the lymph node and head and neck area. In the later stages when metastasis has established, MDSC are the main cells implicated and have a role in increasing the tumor burden in the lymph node and head and neck area, where the role of IL-1β is dispensable