Cell lines and molecules

Seven established ABC DLBCL cell lines (HBL1, OCI-LY-10, OCI-LY-3, RI-1, SU-DHL-2, TMD8, and U2932) were all validated for their cell identity by short tandem repeat (STR) DNA fingerprinting (Promega GenePrint 10 System kit) [43]. All media were supplemented with fetal bovine serum (FBS; 10%), penicillin-streptomycin-neomycin (~5,000 units penicillin, 5 mg streptomycin and 10 mg neomycin/mL, Sigma), and L-glutamine (1%). MZ1 and cisMZ1 were synthesized following a previously established synthetic route [42], while birabresib (OTX015) was purchased (Selleck Chemicals, Houston, TX, USA). All the molecules were dissolved in dimethyl sulfoxide (DMSO) at 10 mmol/L stock solution.

Cell proliferation assay

The antiproliferative activity of MZ1, cisMZ1, and birabresib as a single agent was assessed as previously described [9]. The compounds were serially diluted in the appropriate tissue culture medium, at a range of 0.95–1,000 nmol/L, and added to cells (two biological replicates) at a maximum of 0.1% DMSO (v/v). Cells were incubated for 72 h at 37°C, 5% CO₂. DMSO alone was added to negative control (untreated) cells. Wells containing medium only were included on each plate and served as blanks for absorbance readings. 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazoliumbromide (MTT, Sigma-Aldrich Chemie GmbH, Buchs, Switzerland) was prepared as a 5 mg/mL stock in phosphate buffered saline (PBS) and filter-sterilized. MTT solution (20 μL) was added to each well and tissue culture plates were incubated at 37°C for 4 h. Cells were then lysed with 25% sodium dodecyl sulfate (SDS) lysis buffer (50 μL) and absorbance was read at 570 nm using Cytation3 plate reader (Biotek Instruments, VT, USA). The doses corresponding to the half maximal inhibitory concentration (IC₅₀) and area under the curve (AUC) were estimated by fitting a sigmoidal model through the dose-response curve using the R statistical package (www.r-project.org).

Immunoblotting analysis

Immunoblots of cell line protein extracts were performed as previously described [44]. Proteins were extracted in an appropriate volume of mammalian protein extraction reagent (M-PER) lysis buffer added with protease and phosphatase inhibitors (Thermo Fisher Scientific, MA, USA). The protein content was determined using the bicinchoninic acid (BCA) protein assay (Pierce Chemical Co). Lysates were fractionated by 8% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Membranes were incubated with the primary antibodies overnight, followed by the appropriate horseradish peroxidase-conjugated anti-mouse or anti-rabbit secondary antibodies (Amersham Life Science) for 1 h. Enhanced chemiluminescence detection was then done following the manufacturer’s instructions (Amersham Life Science). Equal loading of samples was confirmed by probing for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or actin. Used antibodies are pSTAT3-Tyr705 (CST 9131), STAT3 (CST 9139), MYC (CST D3N8F), (Cell Signaling Technology CST, Danvers, MA, USA); BRD4 (SC-48772), BRD3 (SC-81202), BRD2 (SC-81825), (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and GAPDH (FF26A, Centro Nacional De Investigaciones Oncologicas CNIO, Madrid, Spain).

Cell cycle and apoptosis assay

Cell cycle was performed as previously described [9]. Briefly, 1 × 10⁶ cells/mL were cultured for 72 h, harvested and washed twice in PBS with 1% FBS, permeabilized with 70% ethanol, and resuspended in PBS with propidium iodide (PI; 1 mg/mL, Sigma-Aldrich) and ribonuclease (RNase). After 20 min of incubation, cells were analyzed using flow cytometer [BD Fluorescent-Activated Cell Sorter (FACS) Canto I]. The analysis of the percentage of cell death the percentages of cells in G1, S, and G2/M phases of the cell cycle was determined using the Watson Pragmatic model and the FlowJo software (TreeStar Inc., Ashland, OR, USA). Cells (3 × 10⁵ cells/mL) were cultured for 72 h, and the percentage of viable and apoptotic cells [Annexin V positive/7-Aminoactinomycin D (7AAD) negative and Annexin V positive/7AAD positive] was determined by double staining the cells with Annexin V-FITC/7AAD.

In vivo experiment

Mice maintenance and animal experiments were performed under the institutional guidelines established for the Animal Facility and with study protocols approved by the local Cantonal Veterinary Authority (license TI-23/2015). NOD.CB17-Prkdcscid/NCrHsd (NOD-SCID) mice were obtained from The Harlan Laboratory (S. Pietro al Natisone, Udine, IT). Xenografts were established by injecting TMD8 lymphoma cells (15 × 10⁶ cells/mouse, 200 μL of PBS) into the left flanks of female NOD-SCID mice (6 weeks of age, approximately 20 g of body weight). Compounds were dissolved in 25% hydroxypropyl-beta-cyclodextrin (HP-β-CD), and adjusted pH 6.0 in an application volume of 4 ml/kg, as previously done [45]. Tumor size was measured on regular basis and until tumors reached around 5 mm in diameter (100 mm³). The body condition scoring was used to assess mice’s health status [46]. Differences in tumor volumes [(length × width × width)/2] were calculated using the Wilcoxon rank-sum test (Stata/SE 12.1 for Mac, Stata Corporation, College Station, TX). The P-value (P) for significance was < 0.05.

Transcriptome profiling

RNA was extracted and processed for RNA sequencing (RNA-Seq; stranded, single-ended 75 bp-long sequencing reads) using the NEBNext Ultra Directional RNA Library Prep Kit for Illumina (New England BioLabs Inc., Ipswich, MA, USA) on a NextSeq 500 (Illumina, San Diego, CA, USA) as previously described [47]. Data mining was performed as previously described [48]. Differentially expressed transcripts were defined as those with an average normalized log expression [counts per million reads mapped (cpm)] of at least 2, presenting an absolute |logFC| > 1 and Benjamini-Hochberg (BH) multiple tests corrected P value [false discovery rate (FDR)] < 0.05. Functional annotation was done using Gene Set Enrichment Analysis (GSEA) on fold-change pre-ranked lists of genes considered differentially expressed. Genesets from the MSigDB collection (hallmark, c2cpg, c2cp, c6) [49], SignatureDB [50] and from different experimental conditions [9, 10, 44], applying as thresholds P < 0.05 and FDR values < 0.1. Omics playground tool [51] was used to compare the expression profiles against the L1000 drug expression database [52]. Expression data are available at the National Center for Biotechnology Information Gene Expression Omnibus (GEO number, GSE152497; http://www.ncbi.nlm.nih.gov/geo) database. The transcriptome study was performed in the context of a larger work assessing also other agents; hence, the DMSO and the birabresib raw data have been previously reported [53].

The BET degrader MZ1 has in vitro cytotoxic activity in ABC DLBCL

Seven cell lines derived from ABC DLBCL were exposed for 72 h to the BET degrader MZ1, its negative control epimer cisMZ1 and, as a comparison, the pan-BET inhibitor birabresib which has shown early clinical activity [12, 27]. MZ1 was very active with a median IC₅₀ of 49 nmol/L [95% confidence interval (CI), 10.16–126.76 nmol/L] and a median AUC of 14,789 (95% CI, 1,302–32,997) (Figure 1A, Table 1, Figure S1). The BET degrader was more active than the BET inhibitor, which presented a median IC₅₀ of 126 nmol/L (95% CI, 77.83–446.64 nmol/L; P = 0.024) and a median AUC of 33,067 (95% CI, 17,690–53,025; birabresib vs. MZ1, P = 0.012). No activity was seen with cisMZ1, which showed a median AUC of 89,072 (95% CI, 86,715–99,547; cisMZ1 vs. birabresib, P < 0.001; cisMZ1 vs. MZ1, P < 0.001).

Display full size

Figure 1. 

Activity of MZ1, birabresib and cisMZ1 in ABC DLBCL cell lines. A. Seven ABC DLBCL lymphoma cell lines were exposed to increasing concentrations of MZ1, birabresib or cisMZ1 for 72 h. IC₅₀ and AUC were calculated and showed in the in the box plots; B. cell cycle effects observed after 24 h and 72 h of treatment of OCI-LY-10 cells with birabresib, MZ1 and cisMZ1 (500 nmol/L); C. representative plots obtained after 72 h of treatment at 500 nmol/L; D. apoptotic effect of the BET degrader MZ1 and birabresib in four ABC DLBCL cell lines tested at their IC₅₀s, as in Table 1. Early and late apoptotic effect for 72 h was evaluated. DMSO vs. birabresib, P < 0.05. DMSO treatment was used as a control. MZ1, cisMZ1 and birabresib were dissolved in DMSO and maximum 0.1% DMSO was used for treatments. The Mann-Whitney U test was used to compare groups. ^* P < 0.05; ^*** P < 0.001

MZ1 (500 nmol/L) induced an accumulation of OCI-LY-10 cells in the subG0 phase already after 24 h of treatment, while birabresib (500 nmol/L) induced G0/G1 accumulation in agreement with what previously reported (Figure 1B–C) [9]. Both effects were more pronounced at 72 h. No changes in the cell cycle occurred with cisMZ1. MZ1 tested at IC₅₀ concentration in four ABC DLBCL cell lines for 72 h showed that MZ1 induces cell death in only 1/4 cell lines more than birabresib (TMD8), while the latter is inducing more cell death in 2/4 cells (HBL1, SUDHL2) (Figure 1D).

The BET degrader MZ1 strongly suppresses BRD4 and MYC protein levels

Since MZ1 presented a stronger anti-tumor activity than birabresib, we assessed its effect on proteins known to be affected by BET inhibitors. We exposed four ABC DLBCL cell lines to the two compounds for 4 h (Figure 2). MZ1 determined a strong down-regulation of its direct target BRD4 and of the other BRD proteins (BRD2 and BRD3) and abrogated MYC protein levels. Birabresib-induced MYC down-regulation was less than after MZ1 and induced an up-regulation of BRD2, BRD3, and BRD4. The MZ1 negative control epimer cisMZ1 did not cause any protein change. Based on the reported mechanism of action of BET inhibitors in ABC DLBCL [7, 9], we analyzed the effect of MZ1 on phosphorylated STAT3 (pSTAT3). We detected reduced phosphorylation in STAT3-Tyr705 after birabresib treatment, stronger with MZ1, while the protein expression level of total STAT3 was not affected by any treatment.

Display full size

Figure 2. 

BRD proteins, MYC and pSTAT3 immunoblotting after birabresib or MZ1 treatment. Four ABC-DLBCL cell lines were treated with 500 nmol/L of DMSO, birabresib, MZ1 or cisMZ1. Protein extraction was performed after 4 h of exposure to the compound. GAPDH was used as a loading control. A. Representative immunoblottings; B. relative quantification of protein levels normalized to GAPDH

The BET degrader MZ1 has in vivo antitumor activity in ABC DLBCL

To confirm the anti-tumor activity of MZ1 in an in vivo model, we engrafted the ABC DLBCL cell line TMD8 in NOD-SCID mice and we treated them with MZ1 (100 mg/kg, i.p.; 3 days ON/4 days OFF), its negative control epimer cisMZ1 (100 mg/kg, i.p.; 3 days ON/4 days OFF) or with vehicle only (Figure 3). MZ1 presented ant-tumor activity at day 10 (P = 0.023) and 13 (P = 0.014) when compared to vehicle. The cisMZ1 did not decrease the growth of the tumor cells. Tumor weight was significantly different between MZ1 (median = 1,098.4 mg, 95% CI 2,059.4–358) and vehicle (median = 2,388.7 mg, 95% CI 2,645–1,716.2; P = 0.012), and between MZ1 and cisMZ1 (median = 2,059.4 mg, 95% CI 2,498.3–1,945; P = 0.024). Treatments were well tolerated in mice, without significant signs of toxicity. Throughout treatment, mice were well-conditioned with a body condition score of body conditioning score 3 (BC3) for all groups.

Display full size

Figure 3. 

Antitumor in vivo activity of MZ1 in an ABC DLBCL model. Mice were treated with vehicle (i.p., 3 days ON/4 days OFF), MZ1 (100 mg/kg i.p., 3 days ON/4 days OFF), cisMZ1 (100 mg/kg, i.p., 3 days ON/4 days OFF). (A) Lines show median values per timepoint with the corresponding upper interquartile range. Y-axis, tumor volume in mm³; X-axis, days of treatment. (B) Tumor weight represented at autopsy illustrated as Tukey box plot showing the median and the 25–75 percentile of the treatments. ^* P < 0.05

The BET degrader MZ1 causes distinct transcriptome changes compared to birabresib

To assess the differences in transcription regulation, we exposed two ABC DLBCL cell lines (OCI-LY-10 and TMD8) to MZ1 (100 nmol/L), birabresib (100 nmol/L), or DMSO for 6 h (Figure 4, Figure S2, Table S1A-B). Considering coding genes, the number of differentially expressed transcripts was greater after MZ1 than after birabresib (MZ1 n = 885, birabresib n = 213; P = 0.0001, one tail Fisher’s exact test), indicating that the BET degrader MZ1 had a bigger impact on the transcriptome of ABC DLBCL than the BET inhibitor birabresib.

Display full size

Figure 4. 

Volcano plots showing significantly upregulated (right quadrant in each plot) or down-regulated (left quadrant in each plot) transcripts after exposure to MZ1 (right panel) or birabresib (left panel) in two ABC DLBCL cell lines. Both MZ1 and birabresib were compared with DMSO. Y axis, log10-adjusted P-value; X axis, log2-fold changes after birabresib or MZ1 treatment

MYC signatures and MYC itself were downregulated by both molecules (Figure S3A, Table S2). Both MZ1 and birabresib also downregulated signatures related to interferon, JAK/STAT pathway, MYD88, NF-κB and B-cell receptor (BCR) signaling (Figure S3B, Table S2). Different to birabresib, MZ1 upregulated transcripts related to translation, ribosomal proteins, and metabolism of RNA (Table S2).

The changes induced by both compounds overlapped with signatures for BET and histone deacetylase (HDAC) inhibitors and mammalian/mechanistic target of rapamycin (mTOR) inhibitors, but only the effect given by MZ1 was similar to what was reported for BET degraders such as ARV-771 and multiple cyclin-dependent kinase (CDK) inhibitors (Figure 5, Figure S4, Tables S2–S3).

Display full size

Figure 5. 

Drug connectivity enrichment plots showing the top pharmacological targets having similar or opposite enrichment in birabresib (left panel) or MZ1 (right panel) signatures obtained in two ABC DLBCL cell lines exposed to the two compounds or to DMSO, as control. RPL3: ribosomal protein L3; PDE3B: phosphodiesterase 3B; GSK3A: glycogen synthase kinase 3 alpha; NQO2: N-ribosyldihydronicotinamide:quinone reductase 2; PTGS1: prostaglandin-endoperoxide synthase 1; ESRRB: estrogen related receptor beta; ESRRG: estrogen related receptor gamma; MET: MET proto-oncogene, receptor tyrosine kinase; TNF: tumor necrosis factor; PDE5A: phosphodiesterase 5A; ESR2: estrogen receptor 2; MAP2K1: mitogen-activated protein kinase kinase 1; EGFR: epidermal growth factor receptor; CHRM2: cholinergic receptor muscarinic 2; HTR1E: 5-hydroxytryptamine receptor 1E; FFAR1: free fatty acid receptor 1; PRKACA: protein kinase CAMP-activated catalytic subunit alpha; ADORA3: adenosine A3 receptor; PSMB1: proteasome 20S subunit beta 1; HTR2A: 5-hydroxytryptamine receptor 2A; AURKA: aurora kinase A

While genes coding for histones were upregulated by both BET degrader and inhibitor, HEXIM P-TEFb complex subunit 1 (HEXIM1) and sestrin 3 (SESN3), two other transcripts previously reported as commonly upregulated by BET inhibitors in multiple models [15, 36, 54–59], were upregulated only by birabresib but not by MZ1 (Table S1). As previously reported [60], the transcription factor 4 (TCF4) was downregulated by the BET inhibitor and especially by the degrader (Table S1).