Activity of BET-proteolysis targeting chimeric (PROTAC) compounds in triple negative breast cancer

TNBC and ovarian cell lines, MDA-MB-231 and BT549 and SKOV3, respectively, were cultured in DMEM, and ovarian cells OVCAR3 were cultured in RPMI supplemented with inactivated fetal bovine serum (10%), antibiotics (100 U/mL penicillin and 100 /mL streptomycin) and L-glutamine (2 mM) (Gibco (Thermofisher), Sigma-Aldrich) (37 °C, 5% CO₂). All cell lines used were provided by Drs. J. Losada and A. Balmain, who purchased them from the ATCC, in 2015. Cells authenticity was confirmed by STR analysis at the molecular biology unit at the Salamanca University Hospital. MDA-MB-231-derived resistant cell line (MDA-MB-231R) was obtained by pulsed exposure to increasing doses of JQ1 (72 h pulses every 2 weeks for 6 months).

BET inhibitors (JQ1 (HPLC: 99.6% purity) and OTX-015 (HPLC: 99.82% purity) and PROTACs-BRD4 (MZ1 (HPLC: 99.5% purity) and ARV-825 (LCMS: 99.37% purity)), together with the inactive form of MZ1, cis-MZ1 (HPLC: 98.6% purity), were purchase from Selleckchem (Houston, TX) and Tocris Bioscience (Bio-Techne R&D Systems, S.LU).

For MTT assay colorimetric assay, after treatments, cell medium was replaced with MTT solution (red phenol-free DMEM with MTT 0.5 μg/μL) (45 min, 37 °C). DMSO was then used to solubilize the samples. Absorbance values were recorded in a multiwell plate reader (555 nm with a reference wavelength of 690 nm). For synergy studies, we used the Chou-Talalay algorithm, which allows to obtain the combination index (CI) to determined which combinations were synergistic (CI < 1), additive (CI = 1), or antagonic (CI > 1) using Calcusyn 2.0 software.

For clonogenic assays, 24 h-treated cells were counted and seeded in triplicates for each condition. After 10 days, cells were fixed with glutaraldehyde (0.5%, 15 min) and, then, stained with crystal violet (0.05%, 15 min). Colonies were quantified using Image J software. For 3D invasion assays, cells were seeded on 48-wells plates containing a 1 mm layer of Matrigel (Sigma-Aldrich) and treated for 72 h. Matrigel generates a net that mimics the extracellular matrix. Invading 3D structures were evaluated using an inverted microscope and their diameter was quantified using Image J software.

For cell cycle analysis, after 12 h of treatment, cells were fixed in 70% ethanol in PBS (15 min). Cell pellets were washed in PBS + 2% BSA and incubated with Propidium iodide/RNAse staining solution (1 h, 4 °C, in dark; Immunostep).

For cell death studies, after 48 or 96 h of treatment, adherent and floating cells were collected and, after a wash with PBS, stained with Annexin Binding Buffer containing Annexin V-DT-634 and Propidium iodide (2 mg/mL) (1 h, RT, in dark; Immunostep). For caspase assays, cells were pre-treated with the pan-caspase inhibitor QVD (10 μM, 45 min; Sigma Aldrich) prior drug exposure.

All analyses were performed on a FACSCanto™ II flow cytometer using the FACS Diva software.

For the evaluation of protein levels, MDA-MB-231 and MDA-MB-231R cells were seeded (500.000 cell/100 mm dish) and, the following day, treated sequentially: first, the 48 h points; the following day, the 24 h points; and finally, the 12 h points. All treatments were collected in parallel 72 h post-seeding together with their common non-treated control.

For the evaluation of cell cycle and apoptosis-related proteins, cells were treated for 12 h and 96 h, respectively. Then cells were lysed, and protein extracts (25–60 μg) were used for Western blot analyses with the indicated antibodies (Additional file 2).

Caspase reaction buffer was added to the protein extract (50 μg, 1 h, 37 °C, in the dark). Then, fluorescence was measured (400/505 nm).

BALB/c nu/nu mice (4–5 weeks old, n = 13) mammary fat pads were injected with MDA-MB-231R (2.5 × 10⁶). Daily treatment with JQ1 (25 mg/kg, i.p.) was initiated when tumors reached a volume of 80–150 mm³. After 1 week of treatment with JQ1, a group of animals (n = 6) continued under this compound regime, while another group (n = 7) received a treatment of MZ1 (10 mg/kg, i.p.). Tumor growth was monitored for two more weeks. Then, tumors were collected, weighted, and stored at − 80 °C. For Western blot analysis, tumor samples (JQ1-treated n = 5; MZ1-treated n = 7) were homogenized with a sonicator Dispomix in ice-cold lysis buffer (1.5 mL/100 mg of sample). For protein levels evaluation, 60 μg of protein were used.

We used t-test for independent samples non-parametric assay, together with the Levenne test to consider, or not, equal variances or ANOVA assay with Tukey subtype. The level of significance was considered 95% (* p ≤ 0.05; ** p ≤ 0.01 and *** p ≤ 0.001). Software GraphPad Prism and SPSS were used.

First, we evaluated the BET-PROTAC inhibitor MZ1, based on the chemical structure of JQ1 [14], on parental MDA-MB-231 cells and on an exclusive model of JQ1-resistant TNBC cells derived from the above-mentioned. A profound downregulation of BRD4 and BRD2 was observed in both cell models upon MZ1 treatment, despite the fact that BRD4 basal expression levels were found to be much higher in the resistant model (Fig. 1a). Similarly, a reduction of the BET proteins levels was observed after ARV-825 treatment, a PROTAC that uses OTX-015, another BETi, as a backbone [15]. However, this effect was milder, particularly on BRD4 in the resistant model, what could be due to this elevated levels of this protein displayed by these cells (Fig. 1a).

Next, we explored the antiproliferative effect of the BET-PROTAC inhibitors MZ1 and ARV-825 when compared with their counterparts JQ1 and OTX015, in MDA-MB-231 and MDA-MB-231R. Both BET-PROTACs induced a clear antiproliferative activity at different time points for both cell models, being this effect even greater in the resistant one (Fig. 1b). The effect of MZ1 and ARV-825 was further confirmed in matrigel invasion (Fig. 1c) and clonogenic assays (Fig. 1d), demonstrating a similar efficacy in both cell lines.

Given the activity of MZ1 and ARV-825 in naïve MDA-MB-231, we decided to explore their effect on other representative TNBC cell lines and extent this evaluation to ovarian cancer, given the molecular characteristics shared between both tumor types. A relevant antiproliferative activity of MZ1 and ARV-825 when compared with JQ1 and OTX-015 was observed in the TNBC cell line BT549 and in two ovarian cell lines, SKOV3 and OVCAR3 (Additional file 1: Figure. S1 A, B).

Bearing in mind that BET-PROTACs compounds showed a relevant antiproliferative effect in TNBC and ovarian cancer cell lines, and that these agents were able to overcome resistance to BETi, we decided to explore the molecular mechanism behind their activity in both naïve and resistant cell lines. To assess their impact on cell cycle, cells were first exposed to JQ1. This BETi was able to induce cell arrest in G1 in the sensitive cell line, while failed to activate this checkpoint in the resistant one (Fig. 2a). In contrast, MZ1 and ARV-825 increased G2/M in the sensitive model when compared with its resistant counterpart (Fig. 2a). Biochemical evaluation of cell cycle components showed that while PROTACs clearly augmented the expression of p21 in naïve cells, this increase was very slight in the resistant cells (Fig. 2b). Also, while JQ1 increased cdc25c levels in naïve MDA-MB-231, MZ1 and ARV-825 do not induce significant changes in this cyclin. Contrary, PROTACs decreased its levels in resistant cells, which actually exhibited higher basal levels of cdc25c, what correlates with a higher presence of resistant cells in G1. As expected, JQ1 do not strongly impact this cyclin in the resistant model (Fig. 2b). Globally, this data suggests that PROTACs mainly act on early mitosis arresting cells at the G2 phase.

We then explored the effect of PROTACs on cell death in both sensitive and resistant cells. Both MZ1 and ARV-825 were able to induce a marked increase in apoptosis at different time points, which was slightly higher in the sensitive model (Fig. 2c). Administration of a pan-caspase inhibitor reverted apoptosis in both models, suggesting that the mechanism was mainly caspase-dependent (Fig. 2d). In fact, the two PROTACs compounds activated caspase 3 in both cell lines, being this effect more clear in the naïve (Fig. 2e and f), as was also observed in the flow cytometry studies. This data support the effect observed after caspase inhibitor administration in this model. Biochemical evaluation of the mechanism of action showed that the levels of the antiapoptotic protein MCL1 were also reduced in both cells lines. Moreover, PROTACs were able to induce DNA damage by H2AX activation and PARP cleavage in both sensitive and resistant cell lines (Fig. 2f).

Most therapies approved for solid tumors are based on combinations of anti-cancer agents [17]. In this context, we compared the anti-proliferative activity of MZ1 with other agents used in the clinical setting, including cisplatin, docetaxel, and the recently approved PARP inhibitor olaparib. As can be observed in Fig. 3a, MZ1 showed a significant antitumoral activity, being docetaxel the only agent that displayed a higher effect. This effect was similar in MDA-MB-231 and MDA-MB-231R. MZ1 combination with the mentioned therapies did not lead to a clear synergistic interaction for any combination in the either of the two cell models (Fig. 3b).

To explore the efficacy of MZ1 on JQ1-resistant tumors in vivo, MDA-MB-231R xenograft models were used. All animals were initially treated with JQ1 to ensure tumor resistance and, then, they were randomized into two groups. While tumors from animals that pursued JQ1 treatment continued growing, MZ1 prevented tumor progression in the other group (Fig. 3c). Evaluation of BRD4 levels in tumors lysates showed that average expression of this BET protein was lower in MZ1-treated mice when compared with mice treated with JQ1, confirming the effect of this BET-PROTAC in vivo. No such effect was observed for BRD2 (Fig. 3d).