Potent, p53-independent induction of NOXA sensitizes MLL-rearranged B-cell acute lymphoblastic leukemia cells to venetoclax

The prognosis for B-cell precursor acute lymphoblastic leukemia patients with Mixed-Lineage Leukemia (MLL) gene rearrangements (MLLr BCP-ALL) is still extremely poor. Inhibition of anti-apoptotic protein BCL-2 with venetoclax emerged as a promising strategy for this subtype of BCP-ALL, however, lack of sufficient responses in preclinical models and the possibility of developing resistance exclude using venetoclax as monotherapy. Herein, we aimed to uncover potential mechanisms responsible for limited venetoclax activity in MLLr BCP-ALL and to identify drugs that could be used in combination therapy. Using RNA-seq, we observed that long-term exposure to venetoclax in vivo in a patient-derived xenograft model leads to downregulation of several tumor protein 53 (TP53)-related genes. Interestingly, auranofin, a thioredoxin reductase inhibitor, sensitized MLLr BCP-ALL to venetoclax in various in vitro and in vivo models, independently of the p53 pathway functionality. Synergistic activity of these drugs resulted from auranofin-mediated upregulation of NOXA pro-apoptotic protein and potent induction of apoptotic cell death. More specifically, we observed that auranofin orchestrates upregulation of the NOXA-encoding gene Phorbol-12-Myristate-13-Acetate-Induced Protein 1 (PMAIP1) associated with chromatin remodeling and increased transcriptional accessibility. Altogether, these results present an efficacious drug combination that could be considered for the treatment of MLLr BCP-ALL patients, including those with TP53 mutations.

. Effects of p53 downregulation on VEN+AUR efficacy and p21 induction upon AUR and VEN+AUR co-treatment. A, RS4;11 cells, either wild type (WT, n=1), siNTC (n=2) or with downregulation of p53 (siTP53, n=2) were treated with indicated concentrations of VEN, AUR, and a combination of both for 24 h. The number of dead cells was evaluated by assessment of PI-positive cells in flow cytometry. The matrices present CI which was calculated in CompuSyn software. CI < 1 indicates synergistic interaction between tested drugs. Colour intensity determines the potency of the synergistic effects, according to legends. B, Immunoblotting showing p53 and p21 protein levels in RS4;11 cells, either with siRNA mediated knockdown of TP53 (siTP53) or control (siNTC), after 6 h exposure to 10 μM nutlin-3, 1.5 μM AUR, 50 nM VEN and combination of VEN and AUR. All blots are representative of 2 independent knockdown experiments. Supplementary Fig. 7. Effects of TXNRD1 and TXNRD2 downregulation on VEN cytotoxicity and cytoplasmic ROS levels in cells incubated with VEN, AUR, and VEN+AUR. A, B, Control SEM cells (siNTC) and cells with siRNA-mediated knockdown of TXNRD1 (siTXNRD1) or TXNRD2 (siTXNRD2) were treated with increasing concentrations of VEN for 48 h, and then the viability was determined by MTT assay. The data are presented as means from 2 independent knockdown experiments + SEM. *P < 0.05, nsnot significant by 2-way ANOVA test (left panels) for the interaction presented, n=2. The efficiency of siRNA-mediated knockdown was determined by RT-qPCR and immunoblotting. The graphs in the middle panel show mean mRNA expression of either TXNRD1 or TXNRD2 relative to control genes (GUSB, RPL29) + error. Representative immunoblots show TXNRD1 and TXNRD2 protein levels from 2 independent knockdown experiments. -actin serves as a loading control. C, SEM cells were treated with indicated doses of VEN, AUR, or both drugs for 1, 2, and 4 h. After the treatment cytoplasmic ROS levels were detected by CM-H2-DCFDA dye using flow cytometry. The graph shows mean fold change in mean fluorescence intensity (MFI) over untreated control (DMSO) + SEM, n=4. *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA with Dunnett's post-hoc test. Supplementary Fig. 8. AUR enhances VEN-induced apoptosis. A, SEM cells were treated with VEN, AUR, and both drugs for 16h and subjected to MultiCaspase assay. Activation of caspases and cell death was determined by staining of the cells with a fluorescently-labeled inhibitor of caspases (FLICA) and 7AAD dye and analyzed using Muse Cell Analyzer. Plots present percentages of viable and dead cells with activated caspases. B, SEM cells were treated with 0.8 μM AUR, 400 nM VEN, or a combination of both drugs for 8 h and 24 h. The level of cleaved PARP protein in response to the drugs was determined by immunoblotting. An equal sample loading was assessed by detection of α-tubulin, n=3. C, SEM cells were exposed to 0.8 μM AUR, 400 nM VEN, and a combination of both drugs for 4, 8, and 24 h. The levels of members of selected pro-apoptotic BCL-2 family proteins were determined by immunoblotting. Representative blots of at least 2 independent experiments are presented. Supplementary Fig. 10. AUR strongly induces NOXA at mRNA and protein levels in various BCP-ALL cell lines. A, The levels of PMAIP1 mRNA in BCP-ALL cell lines (BV-173, REH, 697) exposed to EC50 concentration of AUR for 1 and 3 h were assessed by RT-qPCR. Data are presented as a mean expression of the target gene relative to RPL29 and GUSB control genes. Bars show mean values + SEM, n=2. *P < 0.05, **P < 0.01 by 1-way ANOVA with Dunnett's post-hoc test. B, Same BCP-ALL cell lines as in (A) plus NALM-16 were treated with EC50 or EC80 of AUR and collected after 8 h. NOXA protein levels were assessed by immunoblotting. Representative blots of 2 independent experiments are shown. C, Human primary PBMC, PDX#4, and PDX#5 cells were incubated with 1.5 µM AUR and collected after 4 h. NOXA protein levels were assessed by immunoblotting. Supplementary Fig. 11. ATF4 only partially contributes to AUR-mediated NOXA upregulation. A, SEM cells were exposed to either 1.5 μM AUR or 5 µg/ml tunicamycin (TUNI) for the indicated period of time. ATF4 and NOXA proteins levels were determined by immunoblotting. GAPDH served as a loading control. B, SEM cells were treated with 1.5 μM AUR, 5 µg/ml TUNI or DMSO (control) for 4 h, and ATF4 transcription factor binding within PMAIP1 loci was determined by ChIP-qPCR. SCL7A1 is a known ATF4 target and served as a positive control. The bars indicate mean + SEM. *P < 0.05 by paired t-test, n=4. C, D, 24 h following siRNA-mediated ATF4 downregulation, SEM cells were exposed to 1.5 μM AUR for 4 h. ATF4 and PMAIP1 mRNA expression (C), as well as their protein levels (D), were determined in control (siNTC) and in ATF4 siRNA-treated SEM cells (siATF4). The bars indicate the mean relative expression of the target to reference genes (RPL29, GUSB) + SEM. **P < 0.01 by paired t-test, n=3. E, SEM cells were exposed to 1.5 μM AUR or 5 µg/ml tunicamycin (TUNI) for 1, 2, and 4 h. The graph shows mean PMAIP1 gene expression relative to GAPDH over control + SEM, n=2. Mean expression values were calculated using cycle threshold (Ct) values and presented as 2 -ΔΔCt . ***P < 0.001 by 1-way ANOVA with Dunnett's post-hoc test; nsnot significant.

Cell culture
Human BCP-ALL cell lines representing various genetic subtypes of the disease were purchased from

MTT assay
The cytostatic/cytotoxic effects of single drugs and drugs combinations in BCP-ALL cell lines were tested using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. The assay was performed as described in (2) and the cells were exposed to particular drugs for 48 h. Based on the results from single drug treatment (at least 2 independent experiments), EC20 and EC50 values were determined by nonlinear regression dose-response analysis using GraphPad Prism 7. EC20 and EC50 values of each drug were further used for combinatorial drug testing.

Combination index (CI) calculation
In order to determine the potency of drug combinations, combination index (CI) was calculated using CompuSyn software. The software employs Chou-Talalay method as described in (3). The algorithm defines synergistic (CI < 1), additive (CI = 1) and antagonistic (CI > 1) effects of drug combinations.

Primografts generation
For propagation of primografts, also referred as patient-derived xenografts (PDXs), primary material isolated from BCP-ALL patients was used. The blasts were isolated as described in (2) Table 1 and Sup Table 2 for detailed information about generated PDX samples and for selected antibodies.

Drug testing in BCP-ALL PDXs in monoculture
BCP-ALL PDXs were thawed and their viability was determined using trypan blue exclusion method.
When the percentage of dead cells exceeded 20%, the viable cell population was enriched using The number of CD19 + 7AADcells was calculated based on the equation provided by the beads' manufacturer. Finally, for each PDX EC50 value of VEN was calculated as described above for BCP-ALL cell lines.

Primary BM-MSC isolation and culture conditions
Before sample collection appropriate individual written consent was obtained and all the procedures were approved by Bioethical Committee of Medical University of Warsaw. The cells were isolated and maintained in culture for the maximum of 5 passages as described in (5)

Drug testing in BCP-ALL and BM-MSC stromal cells co-culture model
To assess cytotoxic/cytostatic effects of the drugs on MLLr BCP-ALL PDXs in the presence of stromal cells, we employed the co-culture model described in (2,5), with some modifications. Specifically, bone marrow-derived mesenchymal stem cells (BM-MSC) were seeded onto 96-well plate at 0.2 × 10 4 cells per well. The next day, after removal of the medium, PDX cells processed as described above for monoculture conditions were seeded onto BM-MSC (1.0 × 10 5 cells per well in 100 µl of their medium).
Next, single drugs or VEN+AUR combination were added (total volume of 200 µl), in 2 independent wells. After additional 4-5 days, the total amount of viable cells was assessed by flow cytometry using CountBright™ Absolute Counting Beads (ThermoFisher Scientific, Waltham, MA, USA). The results are presented as the percentage of untreated control (DMSO). To assess drug interaction, combination index (CI) was calculated using CompuSyn software.

Drug testing in BCP-ALL-OP9 cells co-culture model
The activity of single drugs and VEN+AUR combination on Philadelphia (Ph) + and Ph-like BCP-ALL cells was tested in the presence of OP9 stromal cells. For this purpose, GFP-expressing OP9 cells were seeded at the density of 0.3 × 10 4 cell per well onto 96-well plate in 100 µl of their medium.
After 6 h leukemic cells were thawed and seeded as depicted for PDXs grown in monoculture. The same day cells were exposed to single and VEN+AUR drugs combination and following 3 days of the treatment they were stained with 7AAD dye. The viable PDX cells were determined by flow cytometry using CANTO II (BD Biosciences) and presented as the percentage of GFP -/7ADDcells. The type and potency of drug interactions was presented as CI, which was calculated as described above.

CD19 + cells isolation and drug testing
Peripheral blood of healthy donors was obtained from Regional Blood and Hemotherapy Center in Warsaw. Normal peripheral blood mononuclear cells (PBMC) were isolated from the blood using density gradient medium -Lymphoprep™ (1.077 g/ml, STEMCELL Technologies). Next, normal CD19 + cells were isolated from PBMC using EasySep™ Human CD19 Positive Selection Kit II (STEMCELL Technologies). For the viability assay, the enriched cells (more than 75% of CD19 + detected by flow cytometry) were seeded onto 96-well plate, 2.0 × 10 5 cells per well in 100 µl of SFEM II medium containing 20% FBS and 1% penicillin/streptomycin. The same day, the cells were

Primary samples:
1 μg RNA was used for cDNA synthesis using the High-Capacity cDNA Reverse Transcription kit (ThermoFisher Scientific) and 60 ng cDNA was used for each RT-qPCR reaction. The expression of TP53, MDM2 and CDKN2A target genes and ACTB and B2M as internal control genes was measured in duplicate by a fluorescence-based kinetic RT-qPCR using the following TaqMan™Gene Expression Assays: Hs01034249_m1, Hs00923894_m1, Hs01066930_m1, Hs01060665_g1, Hs00187842_m1 Transfection of human HEK293T cells and lentivirus production was performed as described in (7), however in this study plasmid delivery was performed using polyethylenimine (31.5 µg PEI per 10.5 µg DNA). Supernatant containing lentiviral particles was added to 0.5 × 10 6 of SEM cells and then cells underwent spinoculation for 1 h (750 × g at 32°C) in the presence of 5 µg/ml polybrene. Following transduction, clonal lines were selected using 2 µg/ml puromycin.

Immunoblotting
Whole cell extracts from BCP-ALL cell lines were prepared in Triton X-100-based buffer and subjected to immunoblotting as previously described in (2). Extraction and detection of histone proteins was performed as in (8). Antibodies used for detection of particular proteins are listed in Sup Table 2.

Detection of dead cells by propidium iodide staining
Genetically modified SEM and RS4;11 cells were seeded onto 96-well plates (0.2 × 10 6 cells/ml density) in their medium and treated with VEN, AUR or combination of both for 24 or 48 h. The cells were then stained with propidium iodide (PI, final concentration of 1 µg/ml) (Sigma-Aldrich) and analysed in Canto II or Fortessa X-20 flow cytometer (BD Biosciences).

ChIP-qPCR
ChIP assay was performed as described in (8,9). SEM cells were exposed to 1.5 µM AUR for 4 h and 1.0 × 10 7 cells were fixed [1% formaldehyde for 10 min (histone ChIP), or 2 mM disuccinimidyl glutarate for 30 min followed by 1% formaldehyde for 30 min (ATF4 ChIP)] and then DNA was fragmented on Covaris ME220 sonicator according to the manufacturer's protocol. Next, magnetic isolation of antibody-chromatin complexes (for antibodies used see Sup Table 2) was performed using Protein A and G Dynabeads (ThermoFisher Scientific) following a wash step with a solution of 50 mM HEPES-KOH, pH 7.6, 500 mM LiCl, 1 mM EDTA, 1% NP-40 and 0.7% Na-deoxycholate, which was repeated three times. Finally, the beads were washed once with Tris-EDTA, eluted, and incubated with RNase A and Proteinase K and crosslinks were reversed at 65°C overnight. DNA was then isolated using PCR Purification Kit (QIAGEN) and analysed by qPCR (primer sequences are listed in Sup Table   5). The enriched genomic DNA was normalized to inputs as described in (10). SEM H3K27ac ChIP-seq is taken from GSE74812.

ATAC-seq
SEM cells were exposed to AUR or DMSO for 4 h and then 5 × 10 4 cells were harvested. For ATAC-seq the samples were processed as described in (8) using the Nextera Tn5 transposase (Illumina). Paired-end sequencing was performed on DNA libraries using NextSeq 500 (Illumina).
Quality control of FASTQ reads, alignments, PCR duplicate filtering, blacklisted region filtering and UCSC data hub generation were performed using the NGseqBasic pipeline (11).
Leukemia engraftment was detected once a week and determined as percentage of human blasts in murine peripheral blood using specific antibodies and flow cytometry. Once the percentage of human CD45 + /CD19 + cells among murine CD45 + cells exceeded 1% (26 days post engraftment), the mice were randomly selected and treated with either control (60% Phosal 50PG; 30% PEG400; 10% ethanol), or 100mg/kg of VEN (oral gavage). The drugs were given 5 times per week for a total of 13 doses and the mice were sacrificed at the last day of the treatment. The spleens were harvested from all mice and leukemic cells were isolated using EasySep™ Human CD19 Positive Selection Kit II (STEMCELL Technologies). After the magnetic separation, the total amount of viable CD19 + cells was FASTQ files were quality checked using FastQC (v0.11.4). Adapters and poor-quality bases were then trimmed from the reads using trim_galore (v0.4.1). Paired-end reads were mapped to the human genome assembly (hg19) using the STAR aligner (v2.4.2). PCR duplicates were removed using picardtools MarkDuplicates (v1.83). Mapped reads were then quantified over gene exons using subread featureCounts (v1.6.2) to measure gene expression levels.
Statistical analysis was performed in R using the edgeR package (12), and P values were corrected using the Benjamini & Hochberg method (13) to acquire False Discovery Rates (FDR). Genes were considered differentially expressed if they had an FDR of less than 0.05.
Differentially expressed genes were analysed for enriched pathways using PANTHER (v15) (14). p53 pathway genes for visualisation/exploration were extracted from the PANTHER database (14), and the KEGG database (hsa04115). Data visualisations were performed in R. Heatmaps were visualised using log 2 CPM expression values that were Z-score transformed, such that the data are represented by the standard deviations from mean expression.

Data availability
RNA-seq and ATAC-seq data have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE163229.

Statistical Analysis
The n numbers provided in all figure legends state the number of independent biological replicates (cell lines) or the number of individual samples (patients and animal data). Statistical analyses were performed using GraphPad Prism 7 Software (La Jolla, CA, USA), and p values were considered statistically significant when lower than 0.05.
The type of statistical test used is described in Figure legends. All tests were two-sided and for multiple comparisons appropriate corrections were adjusted, as indicated in Figure legends. The exclusion criteria were pre-established. Samples/experiments exclusion from a given analysis was performed only when positive or negative controls did not give appropriate results.

Supplementary Tables
Supplementary