Discussion
Breast cancer metastases in brain (BCM/B) show significant morphological and genomic heterogeneity (Hubalek, Hubalek et al., 2017; Yam, Mani et al., 2017; Yates, Knappskog et al., 2017; Omabe, Ezeani et al., 2014; Da Silva, Cardoso Nunes et al., 2020; Jézéquel, Kerdraon et al., 2019; Hu, Kang et al., 2009; Cacho-Díaz, García-Botello et al., 2020; Roma-Rodrigues, Mendes et al., 2019; Belli, Trapani et al., 2018; Nakamura & Smyth, 2017). Patients with BCM/B have high mortality resulted from the brain lesions and are resistant to chemotherapy treatments (Hu, Kang et al., 2009; Cacho-Díaz, García-Botello et al., 2020). Metastatic breast tumor cells transmigrate across the blood-brain barrier (BBB) and form colonies of tumor cells in brain (Boire, Brastianos et al., 2020; Hu, Kang et al., 2009; Cruceriu, Baldasici et al., 2020). Under normal conditions, the BBB is a highly selective barrier due to existence of tight junctions (TJs) between adjacent brain microvascular endothelial cells (BMECs). However, in the process of invasion of tumor cells to the brain, inflammatory cytokines and chemokines secreted by these infiltrating tumor cells disrupt the BBB integrity by causing damage to the continuous TJ structures formed by endothelial cells of the brain parenchyma and astrocytes. Although the BBB prevents the delivery of most therapeutic drugs into the brain, the circulating metastatic tumor cells invade the damaged BBB to form colonies in the brain (Hu, Kang et al., 2009; Cacho-Díaz, García-Botello et al., 2020).
During tumor progression, cells undergo reprogramming of metabolic pathways that regulate glycolysis and the production of lipids (Omabe, Ezeani et al., 2014; Omabe, Ezeani et al., 2015; Zhang, Guo et al., 2020; Li, Gao et al., 2019; Nomura, Long et al., 2010; Tyukhtenko, Ma et al., 2020; Wang, Nag et al., 2015; Beloribi-Djefaflia, Vasseur et al., 2016). Since MAGL is important in assigning the lipid stores in the direction of pro-tumorigenic signaling lipids in cancer cells, we studied the mechanisms by which MAGL contributes to promoting TNBC metastases. The energy supply of lipids comes from de novosynthesis rather than circulating lipids during tumor development, which are modulated and regulated by MAGL (Wang, Nag et al., 2015; Beloribi-Djefaflia, Vasseur et al., 2016; Taïb, Aboussalah et al., 2019; Yecies & Manning, 2010).
The assessment of MAGL inhibitors potency and selectivity was performed to select compounds for the in vitro and in vivo studies using the application of protein-based real-time 1H NMR spectroscopy (Senga, Kobayashi et al., 2018; Carbonetti, Wilpshaar et al., 2019; Zhu, Zhao et al., 2016), which allowed us to determine the residence time on target for potent MAGL inhibitor AM9928 (46 h) (Figure 1). The longer inhibitory effect of AM9928 strongly suggest prolonged pharmacodynamic effect of AM9928 which is required for our in vivo studies.
Acute and chronic inflammation are regulated by interferons, the interleukins, the chemokine family, mesenchymal growth factors, the tumor necrosis factor family and adipokines (Tuo, Leleu-Chavain et al., 2017; Karki, Man et al., 2017). Key pro-inflammatory cytokines involved in inflammation include interleukin-1 (IL-1), IL-6, tumor necrosis factor (TNFα), IL-8, IL-12, IFN-γ and IL-18. Here, we analyzed the anti-inflammatory profile by targeting MAGL. Based on our results, AM9928 (Figure 4) target tumor-inflammatory molecules IL-6 and IL-8 and the angiogenic factor VEGF-A. These effects were like the effects observed by siRNA-MAGL, but not with control siRNA (data not shown).
Several genes were shown to be involved in the development of brain metastases which include cyclooxygenase COX-2, EGFR ligand HBEGF and α-2,6-sialyltransferase ST6GALNAC5 (Hu, Kang et al., 2009; Bos, Zhang et al., 2009), all involved in facilitating cancer cell passage through the BBB. We have previously reported the roles of the proinflammatory peptide P in impairing the BBB integrity and the angiogenic factor Angiopoietin-2 in mediating activation of brain microvascular endothelial cells and BBB impairment, resulting in infiltration of TNBCs into the brain (Rodriguez, Jiang et al., 2014; Avraham, Jiang et al., 2014). The TME cells are the source of proinflammatory cytokines (Cacho-Díaz, García-Botello et al., 2020; Omabe, Ezeani et al., 2015), which are secreted by both autocrine and paracrine manner though activation of STAT3 (Karki, Man et al., 2017; Bahiraee, Ebrahimi et al., 2019). Indeed, the strong inhibition of IL-6 by AM9928 indicates the potential of AM9928 to target inflammation associated with TNBC. Therefore, MAGL may regulate lipid quality and/or quantity to promote aggressiveness such as migration and inflammation in breast cancer cells.
VEGF-A/VEGF-R signaling functions as an important survival pathway in breast cancer cells 48-50 (Huang, Ouyang et al., 2018; Mery, Rowinski et al., 2019; Banys-Paluchowski et al., 2018). VEGF-A promoted tumor cell self-renewal through VEGF-A/VEGF-R/STAT3 signaling resulting in activation of Myc, Sox2 and STAT3 (Huang, Ouyang et al., 2018; Mery, Rowinski et al., 2019; Banys-Paluchowski, Witzel et al., 2018). High VEGF-A levels strongly correlated with both STAT3 and Myc expression as well as with tumor metastatic potential. We found significant inhibition of VEGF-A expression by AM9928 (Figure 4c), suggesting the effectiveness of targeting MAGL in inhibition of angiogenesis.
Breast cancer-derived exosomes accumulate in the lung, spleen, and bone marrow of naïve mice. At these sites, the exosomal content rich in IL-6 causes pro-metastatic alterations associated with reduction of T cell proliferation, NK cell cytotoxicity and modulation of immune responses. However, the role of TNBC-derived exosomes on HBMECs is not known. Here we show that HBMECs were activated by TNBC-derived exosomes which resulted in secretion of IL-8 and VEGF-A, thereby facilitating tumor transmigration across HBMEC, and promoting tumor colonization. Further, AM9929 treated TNBC-derived exosomes inhibited HBMECs activation as indicated by the secretion levels of IL-8 and VEGF-A. Thus, TNBC-derived exosomes may act as important mediators of intercellular communication between TNBCs and brain endothelium facilitating the formation of pre-metastatic niches and contributing to tumor dissemination in brain.
The BCM/B gene signature includes ANGLT4, MMP1, PTGS2 and TNC16 . This gene signature facilitates the metastatic process and changes in endothelial tight junction, cell migration and invasion. In vivo studies demonstrated that AM9928 significantly inhibited tumor growth in the mammary fat pads (Figure 5b), as compared to untreated mice or mice treated with vehicle control. Further, AM9928 inhibited tumor cell colonization in the brain as detected by GFP positive tumor cells ( Figure 5c + 5d). Indeed, quantitation of GFP–4T1‐BrM5 tumor cell colonization in brain showed that the expression of GFP-expressing Pan-CK positive tumor cells was significantly lower in the mice group treated with AM9928, while the presence of GFP–4T1‐BrM5 tumor cells in brain was significantly higher in the untreated GFP-4T1BrM5 cells and the vehicle control treated mice (Figure 5d). Since decrease in tumor colonization in the brain by AM9928 was observed, we then decided to determine BBB permeability. The permeability of the BBB plays a crucial role in brain metastasis (Boire, Brastianos et al., 2020; Bos, Zhang et al., 2009; Rodriguez, Jiang et al., 2014; Avraham, Jiang et al., 2014). The changes in permeability of the BBB‐BMEC‐TJs are a dynamic process dependent on the interaction between TNBCs and BMECs. Our results showed that TNBC transmigration across the BBB resulted in increased BBB permeability and these changes in BBB permeability were reduced by AM9928 (Figure 6a) . Further, tumor cells were shown to be in closed proximity to BMECs and the expression of the tight junctions ZO-1 and Claudin 5 were impaired in the BBB from mice untreated or treated with vehicle control (figure 6c), while treatment of mice with AM9928, the expression of both ZO-1 and claudin-5 was observed as less damaged structures. Quantitation of TJs showed there was lower expression levels of both ZO-1 and claudin-5 in mice untreated or treated with vehicle control as compared to AM9928 treated mice (Figure 6e-6f). However, these differences in TJ numbers were not statistically significant between the untreated mice and the treated AM9928 mice. Further, importantly, the number of alive mice at day 28 was higher in the AM9928 treated mice as compared to the vehicle control group (Table 1).
Since AM9928 inhibited tumor growth in the mammary fat pads and prevented changes in BBB integrity in vivo, resulting in reduced tumor cell colonization in brain, we conclude that MAGL plays a role in TNBC tumor growth and TNBC transmigration across the BBB. Taken together, we suggest that targeting MAGL may provide a new approach to reducing expression of proinflammatory factors as well as inhibiting tumor growth and tumor cell colonization in brain.

Acknowledgments

We thank Shuxian Jiang for her help in the in vivo studies, Yigong Fu for his technical support and help, Lili Wang for the quantitative analysis of tight junctions in brain microvascular endothelial cells, Lauren Graham and Elizabeth C. Ethier for text editing, compilation, and figures preparations. We thank the following members at the CDD for providing the MAGL inhibitors: Drs. Shakiru O. Alapafuja, Vidyanand Shukla, Alexander Zvonok, and Kiran Vemuri.
The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. This study was supported by departmental grants to Hava Karsenty Avraham and Shalom Avraham.

Author contributions

HKA and SA designed experiments, analyzed data, and wrote the manuscript; OB performed in vitro experiments and analyzed data; SJ helped with the in vivo experiments. ST helped with the NMR analysis of MAGL inhibitors, AM for the synthesis of MAGL inhibitors, MKW, CY and AM helped with the concept of MAGL as modulator of the BBB breaching.
* All in vivo studies were performed in the Division of Experimental Medicine, BIDMC, Harvard Medical School, Boston and allin vitro studies were performed in CDD at Northeastern University.
All in vivo studies were approved by BIDMC institutional Animal care and use committee (IACUC) and the mice were handled in accordance with the animal care policy of BIDMC, Harvard Medical School and in accordance with the Declaration of Helsinki.