ICAM-1 and CD11b mediate tumor cell and neutrophil interaction
We previously have reported that CTC clusters are present in several patient-derived-xenograft (PDX) breast cancer mouse models by immunohistochemistry (IHC) staining of lung sections, and breast cancer patients by FDA-approved Cellsearch platform [10]. Interestingly, we frequently observed CTC-neutrophil clusters (identified by the unique fragmented shape of the neutrophil nucleus) in these PDX models and breast cancer patients (Fig. 1A&1B). In support of our findings, the CTC-neutrophil clusters have also been observed by other groups, and were negatively correlated with breast cancer patient outcomes [14]. Using CRISPR–Cas9-based loss-of function screening, the vascular cell adhesion molecule 1 (VCAM-1) was identified as a molecule mediating CTC– neutrophil cluster formation[14]. Recently, we found that another adhesion molecule ICAM-1 mediates homotypic CTC cluster formation[29]. Interestingly, some ICAM-1+ CTCs were also associated with neutrophils, suggesting ICAM-1 may be involved in CTC-neutrophils cluster formation.
It is well known that Mac-1 (heterodimer of CD11b/CD18) on activated neutrophils can bind to ICAM-1 on endothelial cells to mediate their firm adhesion to endothelial cells during acute and chronic inflammatory diseases[36–39]. This raises a possibility that ICAM-1 on tumor cells may mediate their binding with neutrophils. Indeed, knockdown of ICAM-1 (shICAM-1) in E0771 breast cancer cells significantly reduced the binding of E0771 cells to WT neutrophils (Fig. 1D&E). Furthermore, the CD11b−/− neutrophils isolated from CD11b−/− mice displayed much lower binding ability with E0771 cells compared with WT neutrophils (Fig. 1D&E). Together, these data suggest that ICAM-1 on tumor cells binds to the CD11b on neutrophils, thereby promoting firmly association of neutrophils with ICAM-1+ tumor cells.
CD11b promotes lung metastasis of TNBC, and mediates H 2 O 2 production of neutrophil in response to tumor stimulation.
When tumor cells intravasate into the bloodstream, these CTCs are confronted with a multiple environmental stress, including shear forces, oxidative stress and immune cell attack [40]. Ultimately, only a very small fraction of CTCs survives and forms metastasis. Since CTC-associated neutrophils enhance survival and metastatic ability of CTCs[14, 41], and CD11b mediates association of neutrophils with tumor cell (Fig. 1D), it raises a possibility that CD11b may play a role in metastasis. To test this hypothesis, the L2T-lalebed (expressing both luciferase and tdTomato red fluorescent protein) E0771 cell were injected into WT or CD11b-/- C56BL/6 mice via tail vein. After 24 hours, the E0771 cells colonized in the lungs were dramatically reduced in CD11b-/- mice compared with WT mice, which were determined by bioluminescence imaging (BLI) (Fig. 2A&B). Consequently, the lung metastases were significantly decreased in CD11b-/- mice at 25 days (D25) after intravenous injection of tumor cells (Fig. 2C). Previous studies have demonstrated that neutrophils primarily impact CTCs in the post-intravasation processes of metastasis [42]. Consistently, when the tumor cells were injected into the mammary fat pads of the mice, there was no difference in primary tumor growth between WT or CD11b-/- C57BL/6 mice, but spontaneous lung metastasis was significantly reduced in CD11b-/- mice (Fig. 2D&E). These data suggest that CD11b deficiency most likely inhibits metastasis after tumor cell intravasation. Since currently there is no neutrophil-specific CD11b knockout mice available, the other CD11b-expressing myeloid cells (such as macrophages) may also contribute to the lung metastasis, which warrants further investigation by generating CD11b conditional knockout mice.
NK cells are the main immune cells to kill CTCs[42, 43], but CTC clusters exhibit higher resistance to NK cell killing[44]. Recent studies have demonstrated that neutrophils can suppress the tumoricidal activity of NK cells by producing reactive oxygen species (ROS) to promote breast cancer lung metastasis[30]. Given CD11b is involved in ROS production of macrophage [45], we tested whether CD11b could also regulate ROS production in neutrophils. The neutrophils were stimulated by tumor-cells secretions (conditioned mediums collected from cultured TNBC cells), and then the extracellular H2O2 production from WT or CD11b-/- neutrophils was measured. Surprisingly, the production of H2O2 was completely deficient in CD11b-/- neutrophils, but significantly increased in WT neutrophils (Fig. 2F). Similarly, the H2O2 production in response to phorbol myristate acetate (PMA, a common used potent stimulus for neutrophils[46]) was also compromised in CD11b-/- neutrophils (Supplemental Fig. 1). These data suggest that CD11b is involved in the production of H2O2 from neutrophils in response to tumor stimulation.
ICAM-1 promotes uPAR protein expression and secretion in TNBC cells.
Neutrophils can sense extracellular chemical gradients and exhibit directional migration toward higher concentrations in a process termed chemotaxis[47, 48]. This directed neutrophil recruitment is orchestrated by chemoattractant, such as chemokines (cytokines with chemotactic activities) and lipids [49]. Therefore, we expect that ICAM-1+ tumor cells should secret specific chemoattractant to recruit neutrophils. To identify which chemoattractant may be involved in tumor cell and neutrophil interaction, we performed cytokine array analysis using conditioned mediums from control and ICAM-1 knockdown MDA-MB-231 cells. Among 61 cytokines, uPAR in the conditioned medium collected from ICAM-1 knockdown MDA-MB-231 cells were dramatically reduced compared with control cells (Fig. 3A), which was further confirmed by uPAR ELISA assay (Fig. 3B). Interestingly, ICAM-1 knockdown also decreased uPAR protein levels (Fig. 3C&3D, and Supplemental Fig. 2A). But the uPAR mRNA levels were not significantly changed by ICAM-1 (Supplemental Fig. 2B). These data suggest that ICAM-1 regulates uPAR protein expression and secretion.
To further confirm that tumor cells can secret uPAR, the plasma from tumor-free and tumor-bearing mice was collected for uPAR ELISA assay. Indeed, the plasma levels of uPAR were significantly elevated in both MDA-MB-231 tumor-bearing NSG mice, and E0771 tumor-bearing C57BL/6 mice (Fig. 3E&F). Of note, plasma levels of both human uPAR (secreted by human MDA-MB-231 TNBC cells) and mouse (secreted by host) uPAR were elevated in the MDA-MB-231-bearing NSG mice measured by human uPAR and mouse uPAR ELISA kits, respectively (Fig. 3E), suggesting in addition to tumor cells, the host cells also secret uPAR, contributing to the elevated plasma levels of uPAR in tumor-bearing mice.
suPAR functions as a chemoattractant for neutrophils.
We have known that ICAM-1 promotes uPAR secretion (Fig. 3A&B). Next, to further test the possibility that uPAR is involved in the binding of ICAM-1+ tumor cells with neutrophils, we investigated whether suPAR could act as a chemoattractant for neutrophils via chemotaxis assay. We found that suPAR (at 10ng/ml) in the lower chamber significantly increased mouse neutrophil migration, compared to the control group (Fig. 4A&B). When E0771control or uPAR knockdown (KD) cells (generated by lentiviral vector-mediated transduction of uPAR short hairpin RNA) were injected into the mice, the neutrophils recruited into the lungs were also dramatically reduced in the uPAR KD group compared to the control group (Fig. 4C), further suggesting that uPAR promotes neutrophil recruitment. In line with our finding, the uPAR expression is also positively correlated with neutrophil infiltration level in the TCGA breast cancer-basal subtype (n = 191; p < 0.05) (Fig. 4D). Together, these data reveal a novel role of suPAR as a chemoattractant for neutrophils.
uPAR plays a direct role in promoting lung metastasis of TNBC.
Since we recently demonstrated that ICAM-1 promotes TNBC metastasis [29], and we have shown that ICAM-1 promotes uPAR expression and secretion (Fig. 3), we wonder whether uPAR could also plays a role in TNBC metastasis. To that end, the L2T-labelled MDA-MB-231 and E0771 uPAR knockdown (KD) cells were generated. Knockdown of uPAR not only reduced uPAR expression in the cells, but also decreased secreted-uPAR levels which were measured by ELISA (Supplemental Fig. 3A&B). Next, to determine the role of uPAR in metastasis, we subcutaneously injected either L2T-labelled MDA-MB-231 and E0771 control or uPAR KD cells close to the fourth mammary fat pad areas of NSG and C57/BL6 mice, respectively. Although there is no significant difference in tumor growth between control and uPAR KO groups (Supplemental Fig. 3C&D), the spontaneous lung metastases were dramatically reduced in both MDA-MB-231 and E0771 uPAR KO groups (Fig. 5A-D). Furthermore, the experimental lung metastasis of TNBC via tail vein injection was also inhibited by uPAR knockdown (Supplemental Fig. 3E&F). Consistent with uPAR KO in tumor cells, the tumor-secreted human uPAR and mouse uPAR plasma levels were also significantly reduced in mice implanted with uPAR KO MDA-MB-231 and uPAR KO E0771 cells, respectively (Fig. 5E&F). However, the mouse uPAR plasma levels were not significantly changed between control and uPAR KO MDA-MB-231 groups, suggesting tumor cell-secreted uPAR does not change the uPAR secretion from the host cells.
uPAR positively correlates with ICAM-1 expression in TNBC patients, and high uPAR expression correlates with worse breast cancer patient’s outcome.
Finally, we evaluated the potential clinical association of uPAR in breast cancer patients. We first compared ICAM-1 and uPAR protein expression between different breast cancer subtypes using Clinical Proteomic Tumor Analysis Consortium (CPTAC) data sets. We found that both ICAM-1 and uPAR are highly expressed in TNBC patients compared with other subtypes (Fig. 6A&B). Consistent with our mouse studies, the ICAM-1 gene expression also positively correlates with PLAUR gene expression in TCGA breast cancer data sets (Fig. 6C). Furthermore, PLAUR is highly expressed in breast cancer tissues compared with matched TCGA normal mammary tissues ([Log2FC] > 1, p < 0.01) (Supplemental Fig. 3). We also performed ELISA assay to measure the plasma levels of suPAR in breast cancer. Interestingly, we found that the average plasma levels of suPAR are higher in patients with primary tumors, compared with those whose primary tumors have been resected (Fig. 6D), suggesting tumor-secreted uPAR may be a major source for elevated plasma levels of suPAR in breast cancer patients. At last, we evaluated the clinical relevance of uPAR with breast cancer patient’s survival. Using PrognoScan analyses [35], we observed that high levels of PLAUR mRNA expression in breast tumors are associated with poor distant metastasis–free survival (DMFS) and overall survival (OS) (Fig. 6E). Thus, high uPAR is clinically associated with worse breast cancer patient’s outcome.