PIK3CA Cooperates with KRAS to Promote MYC Activity and Tumorigenesis via the Bromodomain Protein BRD9

Tumor formation is generally linked to the acquisition of two or more driver genes that cause normal cells to progress from proliferation to abnormal expansion and malignancy. In order to understand genetic alterations involved in this process, we compared the transcriptomes of an isogenic set of breast epithelial cell lines that are non-transformed or contain a single or double knock-in (DKI) of PIK3CA (H1047R) or KRAS (G12V). Gene set enrichment analysis revealed that DKI cells were enriched over single mutant cells for genes that characterize a MYC target gene signature. This gene signature was mediated in part by the bromodomain-containing protein 9 (BRD9) that was found in the SWI-SNF chromatin-remodeling complex, bound to the MYC super-enhancer locus. Small molecule inhibition of BRD9 reduced MYC transcript levels. Critically, only DKI cells had the capacity for anchorage-independent growth in semi-solid medium, and CRISPR-Cas9 manipulations showed that PIK3CA and BRD9 expression were essential for this phenotype. In contrast, KRAS was necessary for DKI cell migration, and BRD9 overexpression induced the growth of KRAS single mutant cells in semi-solid medium. These results provide new insight into the earliest transforming events driven by oncoprotein cooperation and suggest BRD9 is an important mediator of mutant PIK3CA/KRAS-driven oncogenic transformation.


Whole Genome Sequencing
High molecular weight genomic DNA was prepared from MCF-10A cells using Ultra-Pure phenol:chloroform:isoamyl alcohol (ThermoFisher). Purity, quantity and size of DNA were assessed by NanoDrop and agarose gel electrophoresis. DNA was sequenced by GeneWiz on an Illumina Novaseq. Raw sequencing data were processed into fastq files using CASAVA 1.8 (Illumina). The raw reads used for variant calling are available from the National Center for Biotechnology Information Sequence Read Archive (NCBI SRA) as Project PRJNA526932. The resulting reads were aligned to HG38 using Burrows-Wheeler Aligner version 0.7.10 [55].

Copy Number Variation Analysis
Whole genome sequencing data were used to identify regions of the genome with significant copy number variations (CNVs). The aligned sequences were analyzed using ExomeDepth version 1.0.5 [56] in R. The regions with a Bayes factor of >20 were selected as significant and annotated with the Conrad's frequent CNV regions and overlapping genes [57].
For large regions, variant calls were used to determine CNV by changes in SNP allele frequency from the whole genome sequencing data. The allele frequencies present in wild type and mutant MCF-10A cell lines were plotted against genomic coordinates in a way analogous to the methods used for CNV determination by SNP microarray [58]. Regions of copy number gain are visible as changes in heterozygous allele frequency from 1/2-1/3, 2/3, 1/4, 3/4, etc.

Growth on a non-binding surface
To assess anchorage-independent growth, 96-well plates were coated with 20 mg/ml poly2hydroxymethacrylate (polyHEMA) (Sigma) in 95% ethanol. Plates were dried overnight in a nonhumidified 37°C incubator prior to UV sterilization for 30 min. Exponentially growing cells were resuspended at 5x104 cells/ml complete EGF-free growth media, and 200 µl were plated per well to yield 10,000 cells per well. Cell viability was quantified after 72 h by addition of Resazurin dye (Sigma) and absorbance at 595 nm on a plate reader.

CRISPR/Cas9
Initial experiments to knock out PIK3CA or KRAS were performed using pSpCas9(BB)-2A-GFP (PX458) (provided by Feng Zhang, Addgene plasmid #48138). Six 20-nucleotide guide sequences, including a 5' PAM, were designed to target open reading frames of each gene, using the CRISPR Design Tool (http://tools.genome-engineering.org). Oligos with 40 nt homology arms (Integrated DNA Technologies) were phosphorylated, annealed and ligated into PX458 for co-expression with Cas9. Plasmids were transfected into MCF-10A cells using Lipofectamine 3000 (ThermoFisher) and FACSsorted for GFP expression. Guide RNA validation was evaluated by High-Resolution Melting on a Roche LightCycler 96 Instrument, comparing PCR-amplified DNA from transfected and non-transfected cells using PIK3CA and KRAS exon-specific primers.

PCR amplification of KRAB
DNA was isolated from DKI cells expressing dCas9/KRAB and sgRNA targeting KRAS or PIK3CA using a DNeasy Blood and Tissue kit (Qiagen). KRAB was amplified using gene-specific primers and the Expand HiFi PCR system (Roche) prior to resolution on a 1% agarose gel.

Cell viability assay and cell cycle analysis
For detection of cell death, cells were stained with propidium iodide (PI) and annexin V-conjugated to Alexa Fluor™ 488 according to the manufacturer's instruction (dead cell apoptosis kit, Invitrogen). Immediately after staining, the cells were analyzed on a flow cytometer using 488-nm excitation for Alexa Fluor™ 488 and a 640 nm filter for PI detection. 10,000 cells were acquired in an LSR II flow cytometer using FACS Diva 6 software (BD Biosciences). Percentages of dead cells were determined by dual-color analysis.
For cell cycle analysis, cells were removed from plates using 0.25% Trypsin and resuspended into EBSS. Cells were fixed by dropping into 70% ethanol solution. The fixed cells were centrifuged and resuspended in 1 mL of PBS containing 0.1% Triton X-100. RNA was removed by treatment with 50 µg RNAse A. Cells were stained with 20 µg/mL Propidium Iodide. Stained cells were analyzed on an LSR II flow cytometer using FACS Diva 6 software (BD Biosciences). Cell cycle analysis was performed using FlowJo v 10.1 using the Watson model [59].

IC50 calculations
In order to understand the potency of the inhibitors A66 (Tocris), AZD6244 (Selleckchem), I-BRD9 (Tocris) and KJ-Pyr-9 in MCF-10A cells, dose-response curves were generated in GraphPad Prism v. 6.0. Exponentially growing cells were seeded into 96-well plates in complete media at a density of 1500 cells/well. The media was exchanged 24 h later for EGF-free media containing inhibitor in log-fold increases in concentration. After 72 h, cells were stained with 10 µg/ml Resazurin (Sigma) and viability assessed by quantifying the fluorescence at 595 nm on a plate reader. This data was fit to a curve using nonlinear regression, log (inhibitor) vs. response, variable slope (four parameters). Here, the relative IC50 is defined as the concentration that defines the point halfway between the top (no inhibitor) and bottom plateaus (maximum cell death). S-4 MCF-10A cells were grown in EGF-free media to subconfluence prior to immunoblotting using antibodies that characterize PI3K and KRAS activation. (B) MCF-10A cells were grown in EGF-free media for 72h on tissue culture-coated or polyHEMA-coated 96-well plates and stained using Alamar blue. $ and ****, p = significant by 2-way ANOVA using Tukey's multiple comparison test (α = 0.05. Mean ± SEM, n = 3). (C) DKI cells migrate faster across a scratched surface. Near-confluent MCF-10A cell monolayers were scratched and allowed to grow for 48h in EGF-free media prior to imaging by light microscope at 50x magnification.