Exosomal MMP-1 transfers metastasis potential in triple-negative breast cancer through PAR1-mediated EMT

Triple-negative breast cancer (TNBC) is a subtype of breast cancer with high risk of distant metastasis, in which the intercellular communication between tumor cells also plays a role. Exosomes can be released by tumor cells and promote distant metastasis through intercellular communication or changes in tumor microenvironment, it is an optimized transportation facility for biologically active payloads. This was a hypothesis-generating research on role of exosomal payload in TNBC distant metastasis.

initiate a series of downstream cascade reactions to enhance their metastasis ability. Our experiments in vitro showed that after co-cultivation of exosomes derived from classic TNBC breast cancer cells (MDA-MB-231) and exosomes derived from high lung metastasis breast cancer cells built in our laboratory (MDA-MB-231-HM) [14][15][16] , the invasion capability of MDA-MB-231 was signi cantly improved. Therefore, we speculate that the key protein carried in exosomes derived from highly metastatic cell lines is one of the relevant factors that promote invasion and distant metastasis. This study used mass spectrometry to screen the differential proteins in exosomes derived from the high lung-metastatic cell line MDA-MB-231-HM and the classic TNBC cell line MDA-MB-231, found MMP-1 to be the most different protein. MMPs are a family of zinc-dependent endopeptidases which are crucial to ECM degradation. Several MMPs have been detected in exosomes from tumor cells 17 . MT1-MMP existing in exosomes from tumor cells is involved in activation of proMMP-2 as well as degradation of ECM proteins 18, 19 . It was also reported that combined expression of miR-21 and MMP-1 in urinary exosomes detects 95% of breast cancer 20 . We suppose that MMP-1may be a key protein to mediate the transfer of metastatic ability between tumor cells through exosomes.

Tissue samples
Tissue specimens(n=134) were collected from female patients with TNBC, de ned as ER negative, PR negative, HER2 negative or 1+ or 2+ but uorescence in situ hybridization (FISH) negative, who underwent operation of tumor in Fudan University Shanghai Cancer Center (Shanghai, China) between August of sections, 5 tissue microarrays (TMA) blocks with 50 cores per TMA were constructed. Sections were cut at 4µM thick for immunohistochemistry. Clinical characteristics were obtained from the Electronic Medical Record System and telephone follow-up. The research protocol was approved by the Ethics Committee of the Fudan University Shanghai Cancer Center, and written informed consent was obtained from all patients prior to enrollment.
The TMAs were digitized via whole-slide scanning with the Aperio T2 scanner (Aperio Technologies) and IOD values were assessed independently by two pathologists in from the Department of Pathology (Fudan University Shanghai Cancer Center); discrepancies were resolved by discussion until a consensus was reached. Lung tumor sections of nude mice were observed and imaged by microscope (OLYMPUS, BX43). Tumor cell migration and invasion assays 24-well plates inserted with 8-mm pore size lters (Corning Life Sciences, Corning, NY, USA), for tumor cell invasion ability, the lter membranes were pre-coated with 20µl Matrigel (dilution, 1:5; BD Biosciences, Franklin Lakes, NJ, USA). Cells were seeded into the upper chambers, with L-15 containing 20% FBS in the lower chambers. Cells were incubated for 24-36h at 37˚C, and cells that had invaded or migrated to the reverse side of the membrane were detected by staining with crystal violet, viewed and counted in at least ve random elds under a light microscope (OLYMPUS, IX51).

Cell proliferation
A Cell Counting Kit-8 (PN812; Dojindo, Japan) was used for cell proliferation assays. Cells were seeded into a 96-well plate at a density of 2×10 3 cells per well in quintuplicate wells. At day 1-6 after culture, 10μl of CCK-8 solution was added to each well and incubated for 2h at 37°C, the absorbance of cells was measured at a wavelength of 450nm for calculation of the optical density (OD) values (Synergy H1, Biotek).

Wound healing assay
Cells were seeded in 6-well plates and grew to almost complete con uence, then a 10μl pipette tip was used to scratch a gap on the cell layer. Cells were incubated in serum-free medium. The gap widths were measured and imaged under an inverted microscope (OLYMPUS, IX51) at 0h, 24h and 36h after scratching.
Puri cation of exosomes from cell and human plasma When cells grew at about 40-50% con uence, they were moved to medium containing exosome-depleted FBS, which was obtained by ultracentrifugation of standard FBS at 100,000g for 10h at 4°C followed by ltration through a 0.22µm vacuum ltration bottle. Supernatants were collected from 24-36h cell cultures. Exosomes were puri ed by differential centrifugation at 3,000g for 20 min to remove cell debris and dead cells. Vesicles were pelleted after centrifugation at 16,500g for 45 min and resuspended in PBS, then centrifuged at 100,000g for 2h at 4°C (Beckman Coulter, Optima XPN-100). After resuspending in an appropriate amount of PBS, the protein concentration was measured, recorded and stored for subsequent experiments.
For puri cation of circulating exosomes by differential centrifugation, blood from patients was centrifuged at 1,500g for 30min to obtain cell-free plasma. Plasma was centrifuged at 15,000g for 30min. The pelleted vesicles were suspended in PBS and then centrifuged at 100,000g for 2h at 4 °C.
Transmission electron microscopy (TEM) and Nanosight analysis (NTA) of exosomes Puri ed exosomes were prepared on copper TEM grids (3.05mm; 200 mesh) by negative staining. 10μl sample was dropped on copper grids and incubated for 3mins. 10μl 2% uranyl acetate was pipetted on the grid and incubated for 10min. Excess solution was removed by with lter paper. Grid was stored in the dark grid box at RT until imaging. Imaging was conducted using a 120kV Biology Transmission Electron Microscope (Tecnai G2 SpiritBiotwin).
Nanoparticle tracking analysis was conducted by NanoSight NS300. Exosomes in 20μl PBS were diluted to 1000μl with DPBS, then loaded with a 1ml syringe to be tested. Three recordings of 30s at 37°C in camera were obtained and processed using NTA software.

Exosome labeling
Exosomes were resuspended in PBS and stained with DIO dye for 20 min at room temperature. The

Lung colonization study
Six-to eight-week-old nude mice were injected in tail vein with 1×10 6 MDA-MB-231 cells. Cell-derived exosomes (15μg in 100µl PBS) were intravenous injected every other day for 2 weeks. The mice were euthanized six weeks after the cancer cell injection and their lungs were xed, sectioned and analyzed for H&E quantify the metastatic tumor burden. Paraformaldehyde-xed lungs were sectioned into 8-μm-thick sections, respectively, at 100μm intervals. Three large lung sections were stained with H&E and tumor nodules were counted and their area measured using the OlyVIA (Olympus) and ImageJ software. The number of the metastatic nodules were calculated by averaging data from individual sections.

Statistical analysis
All experiments in the study were performed in triplicate. Statistical analyses were performed with the SPSS v20.0 software (SPSS Inc., Chicago, IL, USA). Data are presented as the mean±standard deviation (SD). Quantitative data were compared with a two-tailed Student's t-test between groups and a one-way analysis of variance among multiple groups followed by Lease Signi cant Difference post hoc test. Kaplan-Meier curves of disease-free survival were plotted and survival in the groups was compared by log-rank test. The signi cance levels of the data are denoted by * symbols as follows: *p<0.05; **p<0.01; ***p<0.001; and ****p<0.0001. p<0.05 was considered to indicate a statistically signi cant difference.

Bioinformatics analysis
The association between MMP1 expression and prognosis in triple-negative breast cancer patients was analyzed by using the Kaplan Meier plotter, which is an online database that provides assessment of the effect of 54,675 genes on survival using 10,293 cancer samples, including 22,277 genes in 5,143 breast cancer samples (http://kmplot.com/analysis/).

Isolation and veri cation of exosomes
In our previously study, MDA-MB-231-HM (MDA-231-HM), a highly pulmonary metastatic variant of parental MDA-MB-231 (MDA-231) cells, was derived within the presented model system by six cycles of the pulmonary metastasis implantation to the mammary fat pad (MFP) 14 , which was a valuable model system for the study of the molecular events underlying breast cancer metastasis 15,16 . The differences in pulmonary metastatic potential of the two cell lines make them valuable systems for understanding the molecular mechanisms underlying breast cancer metastasis. Transwell assay showed that migration and invasion capabilities of MDA-231-HM are stronger than MDA-231 cells ( Figure 1A).
We puri ed exosomes from supernatants of MDA-231 and MDA-231-HM cells by differential centrifugation 21,22 , and veri ed them by Transmission electron microscopy (TEM) ( Figure 1B), Western blot ( Figure 1C) and nanoparticle tracking analysis (NTA) ( Figure 1D). After co-cultivating MDA-231 cells with 231-HM-exo for 24h, the exosomes were observed by confocal microscopy close to the nucleus, indicating that the exosomes have been taken up by MDA-231 cells ( Figure 1E). In order to nd the key protein transferred by exosomes that changes the ability of invasion and migration capability, we screened the differential proteins between 231-exo and 231-HM-exo by mass spectrometry. From the heat map we found that MMP-1 is the protein with the most signi cant difference in expression ( Figure 2B).

MMP
MMP-1 belongs to matrix metalloproteinase (MMP) family, which includes a series of zinc-and calciumdependent endopeptidases. MMPs are of crucial importance for invasive cancer cells to break extracellular matrix (ECM) barriers and start the metastatic cascade 23,24 . MMP1 has also been shown to be related to invasive phenotype, metastasis and response to chemotherapy in human breast cancer, and is related to poor prognosis [25][26][27] . We veri ed the results of mass spectrometry by western blot of MMP-1 in MDA-231, MDA-231-HM, 231-exo and 231-HM-exo and saw signi cant difference in both cells and exosomes ( Figure 2C).
After observing the differences in migration and invasion ability of cells, we focused on cell-to-cell transmission of MMP-1 through exosomes. Transwell assay showed that co-cultivation of MDA-231 with 231-HM-shMMP1-exo for 24h signi cantly reduced the migration and invasion of 231 cells co-cultured with 231-HM-exo. Considering the heterogeneity of TNBC, we added MDA-MB-468 (MDA-468) and BT549 to repeat the above experiment and the results were similar, proving that the role of MMP-1 in exosomes is not limited to MDA-231 cell line ( Figure 3F, G).
MMP-1 binding to PAR1 promotes the metastasis of TNBC possibly through EMT The function of MMP-1 on ECM degradation has been reported 23,24 , which cannot fully explain the enhanced invasion and migration activity both. MMP-1/PAR1 as a key signal to activate downstream signaling pathways to facilitate vascular intravasation and metastatic dissemination has also been proposed in several tumors 28-30 . Protease-activated receptor 1 (PAR1) is a G protein-coupled receptor that is classically activated by the serine protease thrombin cleavage of the N-terminal outer domain, can also be directly activated by MMP-1 and MMP-9 31 . PAR1 is also considered as an independent factor for the poor prognosis of tumors and a potential therapeutic target 32 .
We found that after ingestion of exosomes with different levels of MMP-1, the level of free MMP-1 secreted by MDA-231 cells into the supernatant was positively correlated with the level of MMP-1 in exosomes ( Figure 4A). Then, we demonstrated through co-immunoprecipitation that MMP-1 can directly interact with PAR1 to stimulate downstream metastasis related pathways ( Figure 4B). The PAR1 irreversible antagonist, Vorapaxar, was added at a concentration gradient to the medium of 231 cells cocultured with 231-HM-exo and could effectively reverse the effect of 231-HM-exo. Vorapaxar inhibited 58% and 79% of the migration, 45% and 57% of the invasion at 75μM and 100μM, respectively ( Figure  4C). This strongly explains that MMP-1 can not only degrade collagen to promote invasion, but mainly rely on PAR1 to exert its invasion effect. Inhibiting PAR1 can greatly inhibit this acquired capability of invasion. In order to eliminate that the change in cell proliferation ability brought by PAR1 may interfere the results of migration and invasion assays, we rst used CCK-8 assay to detect the proliferation of MDA-231 cells co-cultured with 231-HM-exo and Vorapaxar. Results showed that cell proliferation was independent of 231-HM-exo and Vorapaxar ( Figure 4D).
According to previous reports, exosome-associated MMPs are involved in the processes of EMT in some  Table 1. The patients were followed up to 2020.8, median follow-up time was 56.8 months. MMP-1 expression was determined by immunohistochemistry and a quarter one fourth of the OD value of MMP-1 (OD=4.009) was chosen as a cut-off value to de ne 'high' and 'low' expression ( Figure 6A). In the univariate Kaplan-Meier analysis, TNBC showing high expression of MMP-1 had a signi cantly shorter disease-free survival (DFS) (logrank p=0.028) ( Figure 6B). Survival curve retrieved from Kaplan-Meier plotter showed that high level MMP-1 expression is associated with signi cantly worse DFS (logrank p=6e-13) in TNBC patients ( Figure 6C).
Blood samples from patients at baseline of pre-surgery (n=18) and initial metastasis (n=30), baseline characteristics are shown in Table 2. The MMP-1 expression level in serum and exosomes puri ed from serum was detected by ELISA. We found that MMP-1 expression in initial metastasis group was signi cantly higher than that in pre-surgery group, whether in serum (p<0.0001) or exosomes (p<0.0001).
It was con rmed that the MMP-1 carried in exosomes was related to the recurrence and systemic metastasis of TNBC ( Figure 6D, E).
Furthermore, we distinguished patients with single metastatic site (n=13) and multiple metastatic sites (n=17) in the initial metastasis group to research the correlation between MMP-1 and systemic metastasis. The concentration of MMP-1 in the serum has no statistical difference (p=0.1488) while in exosomes, MMP-1 detected in patients with multiple metastases was signi cantly higher than that in single metastatic site (p<0.01), which suggests that for patients with distant recurrence and metastasis, MMP-1 detected in exosomes would be more sensitive than in serum ( Figure 6F, G).
Interestingly, we collected blood samples from several patients (n=11) before and after rst-line chemotherapy and tumor response was evaluated in accordance with the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) guidelines by computed tomography scanning or magnetic resonance imaging after 2 cycles 38 .We found that MMP-1 level in exosomes was also related to response to chemotherapy. The expression of MMP-1 in exosomes of patients with partial response (PR)(n=4) decreased while increased in patients with progressive disease (PD)(n=3). In patients with stable disease (SD)(n=4), the expression of MMP-1 in exosomes both increased and decreased, which suggests that MMP-1 in exosomes even has the potential to be an indicator of therapeutic e cacy ( Figure 6H).

Discussion
Many reports demonstrated that exosomes can mediate tumor metastasis by horizontal transfer of bioactive molecules to recipient cells 8, 11,[38][39][40] . In this study, we rst reported that TNBC cells with high metastasis potential transform those with low metastasis potential via exosomal MMP-1.
Exosomes have been reported to take part in promoting tumor metastasis and drug resistance 41,42 . Exosomes potentially promote tumor development by regulating biological functions, including angiogenesis, immunity, vascular leakiness, and reprogramming recipient cells to construct premetastatic niche (PMN) and metastasis [43][44][45] . To study the role of exosomes in tumor metastasis, we extracted, veri ed and observed the exosomes taken up by MDA-231 cells. Then mass spectrometry was applied for the most different expressed protein between high and low lung metastasis cell-derived exosomes, which is MMP-1.
Historically, MMPs have been studied for a long time in cancer biology, and accumulated evidence showed that MMPs are related to progression, metastasis, treatment and prognosis of cancer 46-48 . As reported, MMPs mediate the degradation of ECM and basement membrane (BM) at all stages of cancer progression, thereby promoting the development of the surrounding microenvironment and distant metastasis 13,28 . However, the fact that MMP-1 is transmitted between cells through exosomes to enhance metastasis ability of low-metastatic cells has not yet been reported.
Unsurprisingly, our study found that knocking down the MMP-1 of MDA-231-HM with shRNA reduced MMP-1 in MDA-231-HM-secreted exosomes. Hereafter, we observed that exosomes with reduced MMP-1 content actually affected the invasion and migration ability of MDA-231 in vitro and in vivo. Further, we explored the mechanism of the in uence of exosomal MMP-1.The most widely known function of MMP-1 is to degrade collagen I, II and III in the ECM, but it cannot fully explain the enhanced activity of both invasion and migration. In addition to degrading collagen, MMP-1 can also directly PAR1 by cutting the extracellular N-terminus of PAR1 26 . PAR1 is a G protein-coupled receptor that is classically activated by the serine protease thrombin cleavage of the N-terminal outer domain. PAR1 activated by MMP-1 provides a link between the extracellular proteolytic activity important for ECM remodeling and the signal transduction leading to cell migration and invasion 31 . It was reported that MMP-1 and MMP-13 cut the Nterminal outer domain of PAR1 at non-canonical sites, activating the G-protein signaling pathway 39 . In breast cancer cells, thrombin and MMP-1 activated PAR1-dependent phospho-Akt signals 46 . In our study, immunoprecipitation assay showed that MMP-1 directly interacts with PAR1. The PAR1 antagonist, Vorapaxar can also signi cantly reverse the up-regulated metastatic capability of MDA-231 after coculture with 231-HM-exo at a certain concentration. It can also be seen that the EMT process promoted by 231-HM-exo can be reversed by PAR1 antagonist. We took the lung tissues for immunohistochemistry and found that the expression of MMP-1 and PAR1 were higher in the 231-HM-exo-treated group, whereas those were down-regulated with 231-HM-shMMP1-exo. Therefore, it is concluded that free MMP-1 in the supernatant, which sources may be from released MMP-1 from internalized exosomes and production by initial low metastasis capacity MDA-231 cells, binds to the membrane surface receptor PAR1 protein and then promotes the metastasis of the triple-negative breast cancer possibly through EMT.
By searching the Kaplan-Meier database, we observed that the expression level of MMP-1 is related to DFS of TNBC, and this nding was then con rmed by TMA of surgical tissue samples of TNBC patients in our center. There was also signi cantly up-regulated level of MMP-1 in serum and exosomes in patients with metastasis. Moreover, whereas there was no statistical difference in serum MMP-1 expression in patients with solitary lesion and multiple distant metastasis, MMP-1 in exosomes produced statistical differences, indicating that exosomes may be more sensitive and accurate in predicting and evaluating distant metastases. Furthermore, comparing exosomal MMP-1 level before and after rst-line chemotherapy in several patients showed that MMP-1 alteration is associated with the e cacy, indicating implication in predicting the e cacy of chemotherapy.
There are still some limitations in this study. In terms of experimental data, rst of all, due to the limitations of laboratory facilities, we were unable to record the trajectory of cell migration and draw trajectory diagrams to visually display and compare the metastatic ability of cells. Secondly, in the study of the relationship between the changes in the level of MMP-1 in exosomes and the e cacy of rst-line chemotherapy in clinical patients, the number of blood samples was too small to draw a statistical conclusion. In terms of conclusion, the mechanism of EMT caused by the interaction between PAR1 and MMP-1 has not been directly clari ed through experiments. We could only suppose and prove that the cells go through EMT by western blot assay for changes in EMT-related markers before and after the treatment of exosomes. Finally, the research is currently aimed at TNBC, and ER-positive and HER-2positive breast cancer cell lines may be included for further study.

Conclusions
In summary, exosomes secreted by cells of high metastasis ability enriched in MMP-1 can transform cells with low metastasis ability in TNBC into more malignant cells, which is possibly mediated EMT via interaction with PAR1. Clinically, higher MMP-1 level in exosomes in patient's blood indicates more chance of occurrence of distant metastasis, along with a certain potential for the evaluation of therapeutic effects.

Competing interests
The authors declare that they have no competing interests.