Results
Assessment of The Residence Time of AM4301 and AM9928 : In vitro binding and unbinding of potent AM9928, as compared to another MAGL inhibitor AM4301, carbamate-based inhibitors to human MAGL (hMAGL) was monitored by 1H NMR spectroscopy using downfield1H NMR spectral pattern as a probe of conformational changes (Zhu, Zhao et al., 2016; Tyukhtenko, Ma et al., 2020; Tyukhtenko, Karageorgos et al., 2016) associated with binding. The chemical shift perturbations (CSPs) induced by binding in the characteristic downfield NMR pattern are shown in Figure 1. Addition of equimolar amounts of inhibitor in both cases caused significant upfield chemical shift (from 14.9 to ~13.9 ppm) for the resonance peak corresponding to the catalytic His269 and smaller shifts for other downfield resonances. Following the fast formation of hMAGL covalent adducts with AM4301 (Figure 1a) and AM9928 (Figure 1b), slow reactivation of the enzyme due to hydrolytic decomposition of adducts was monitored by real-time NMR spectroscopy. The residence time (lifetime) of ligand on a target is equal to the lifetime of covalent adduct which is a period required for adduct to decrease to 1/e (0.367) of its original concentration. The estimated residence time based on NMR time course data for AM4301 is 1.4 hours whereas AM9928 residence time is 46 hours. Significantly longer in vitro residence time of AM9928 in comparison with AM4301 could be translated to its prolonged pharmacodynamic effect. Based on these observations we selected AM9928 for our studies.
The toxicity effects and time kinetics of treatment with AM9928:The toxicity of AM9928 on TNBC cells were analyzed. The following doses were examined: 10, 100, 250, or 500 nM, for durations of 24, 48 or 72 hours. No significant toxicity of AM9928 (Figure 2) was observed on the tested cells.
MAGL mediates adhesion and transmigration of breast cancer cells through human brain microvascular endothelial cells (HBMECs): To determine the effects of MAGL on TNBCs co-cultured with HBMECs, in vitro adhesion assay was performed using co-cultures of the brain-seeking variant MDA-MB-BrM2 cells and HBMECs in the presence of vehicle control or AM9928. MDA-MB-BrM2 cells adhered to the untreated monolayer by as early as 10 minutes (Figure 3a). Pre-treatment of HBMECs monolayers with AM9928 (100 ng/ml) blocked the adhesion of the tumor cells at time points of 10 and 30 minutes (Figure 3a). Next, we examined the role of MAGL in transmigration of TNBC cells across HBMECs using anin vitro model where HBMECs were co-cultured with breast tumor MDA-MB-231 cells or with MDA-MB-BrM2 cells (Avraham, Jiang et al., 2014). Tumor cell transmigration across HBMECs was significantly inhibited (about 50%) by AM9928 (Figure 3b).
Targeting MAGL in human breast cancer cells: Cytokines are produced by both immune cells and cancer cells (Parker, Limebeer et al., 2016; Nagarsheth, Wicha et al., 2017). To characterize the potential activity of MAGL inhibitor AM9928 on the inflammatory responses in breast cancer, we have used TNBC cells, the highly invasive MDA-MB-231 and the brain- seeking variant MDA-MB-BrM2 cells, as compared to the immortalized normal breast epithelial MCF-10A cells.
First, we examined the effects of AM9928 as compared to the vehicle control on secretion of cytokines/chemokines in cells treated with either the vehicle control or with AM9928 (250nM) for 72 hours. The supernatants were collected and analyzed by specific ELISA assays for IL-6, IL-4, INF-alpha2, TGF-alpha, TNF-alpha, and IL-8. We observed significant inhibitory effects of AM9928 on IL-6 and IL-8 (Figure 4a-4b). Both inflammatory cytokines are important in tumor-associated inflammation (Omabe, Ezeani et al., 2015; King, Mir et al., 2017; Masjedi, Hashemi et al., 2018; Banerjee & Resat, 2016; Atretkhany, Drutskaya et al., 2016).
Second, we analyzed the effects of AM9928 on the angiogenic factor VEGF-A secretion and observed significant inhibitory effects of AM9928 on VEGF-A secretion (Figure 4c).
Third, to determine if chemokines/cytokines are directly involved in activation of HBMECs, without the contact of tumor cells, we prepared TNBC-derived exosomes untreated or treated with AM9928 and tested their direct effects on HBMECs. MDA-MB-231 and MDA-MB-BrM2 -derived exosomes caused activation of HBMECs leading to secretion of elevated levels of the pro-inflammatory IL-8 and VEGF-A. AM9928 treated TNBC-derived exosomes significantly inhibited activation of HBMECs, as observed by the secretion levels of IL-8 and VEGF-A (Figure 4d-4e).
In vivo analysis of tumor growth in mammary fat pads and colonization of TNBCs in brain: To examine the in vivo effects of AM9928 on TNBC tumor growth and BBB integrity, GFP-4T1-BrM5 cells were administered (5 x 104 cells) into the mammary fat pads of BALB/c mice and were injected with either the vehicle control or with AM9928 as shown in Figure 5a (n =  10 per group per treatment). The GFP-4T1-BrM5 tumor cell growth in mammary fat pads and tumor cell seeding in the brain was observed by in vivo imaging and immunohistochemistry.
Tumor growth in mammary fat pads was significantly reduced in mice treated with AM9928 as compared to mice treated with vehicle control (Figure 5b). Quantitation of GFP–4T1‐BrM5 tumor cell colonization in brain was determined by ex vivo  imaging of the brain and by immunohistochemistry using GFP antibody. The expression of GFP 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 5c, 5d).
Increased in BBB permeability was found in mice administered with tumor cells or with vehicle treated mice, while mice treated with AM9928 had lower BBB permeability changes as compared to control mice at day 28 (Figure 6a). We then analyzed BMEC-tight junction proteins’ expression of ZO-1 and Claudin-5 in vivo in normal brain, as compared to TJs in tumor brains from syngeneic GFP-4T1-BrM5 mammary tumor cell model. ZO-1 is the main scaffolding protein and claudin-5 is known as the TJ sealer protein of the BBB. As shown in Figure 6b, both expression of ZO-1 and claudin-5 were abundant in the normal brain vasculature. Following GFP–4T1‐BrM5 cells administration to the mammary fat pads, the GFP-tagged tumor cells migrated across the BBB after about 4 weeks. The GFP tagged tumor cells were detected by immunostaining with either Monoclonal Pan-Cytokeratin (Pan-CK) antibodies or GFP antibodies. No staining of normal brain sections with Pan-ck antibodies was observed (data not shown). These tumor cells were usually located around the brain capillaries and in-close proximity to BMECs (BMECS were detected with CD31 antibodies) (figure 6c). Both ZO-1 and claudin-5 structures were damaged in the brain tumor vasculature (Figure 6d) as compared to the control brain (Figure 6b). In some images, both ZO-1 and claudin-5 structures withing BMECs were seen to colocalized with tumor cells, indicating the proximity of BMECs to the tumor cells in vivo. Interestingly, following treatment with AM9928, the expression of both ZO-1 and claudin-5 was less impaired as compared to the mice treated with vehicle control (Figure 6d).
The number of CD31+ vessels expressing TJ proteins, ZO-1, and claudin-5, was performed and images were analyzed using Adobe Photoshop CS2. The expression of ZO-1, as determined by average sum of intensity was 582,096 and following treatment with AM9928 the average sum of intensity was 610,920 (Figure 6e). The average sum of intensity of Claudin-5 expression in untreated group was 518,599 and following treatment with AM9928 the average sum of intensity was 681,457 (Figure 6f). Although these differences in TJ numbers were not statistically significant between the untreated mice and the treated AM9928 mice, both ZO-1 and claudin-5 were observed as less damage’s structures in the AM9928 treated mice, as compared to the vehicle treated mice.
In summary, the number of mice positive with mammary tumors and brain tumors was significantly higher in the vehicle treated group as compared to AM9928 treated mice (Table 1). 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). Thus, AM9928 significantly inhibited changes in the BBB permeability and reduced TNBC colonization in brain.