p53R172H and p53R245W Hotspot Mutations Drive Distinct Transcriptomes in Mouse Mammary Tumors Through a Convergent Transcriptional Mediator

Abstract Aggressive breast cancers harbor TP53 missense mutations. Tumor cells with TP53 missense mutations exhibit enhanced growth and survival through transcriptional rewiring. To delineate how TP53 mutations in breast cancer contribute to tumorigenesis and progression in vivo, we created a somatic mouse model driven by mammary epithelial cell-specific expression of Trp53 mutations. Mice developed primary mammary tumors reflecting the human molecular subtypes of luminal A, luminal B, HER2-enriched, and triple-negative breast cancer with metastases. Transcriptomic analyses comparing MaPR172H/− or MaPR245W/− mammary tumors to MaP−/− tumors revealed (1) differences in cancer-associated pathways activated in both p53 mutants and (2) Nr5a2 as a novel transcriptional mediator of distinct pathways in p53 mutants. Meta-analyses of human breast tumors corroborated these results. In vitro assays demonstrate mutant p53 upregulates specific target genes that are enriched for Nr5a2 response elements in their promoters. Co-immunoprecipitation studies revealed p53R172H and p53R245W interact with Nr5a2. These findings implicate NR5A2 as a novel mediator of mutant p53 transcriptional activity in breast cancer. Significance: Our findings implicate NR5A2 as a novel mediator of mutant p53 transcriptional activity in breast cancer. NR5A2 may be an important therapeutic target in hard-to-treat breast cancers such as endocrine-resistant tumors and metastatic triple-negative breast cancers harboring TP53 missense mutations.


Introduction
Breast cancer comprises four subtypes: luminal A, luminal B, HER2enriched, and basal-like breast cancer (1,2).Luminal breast cancers are driven by estrogen receptor (ER) and/or progesterone receptor (PR).HER2enriched breast cancers are driven by amplification and constitutive signaling of HER2.Basal-like breast cancers are often termed triple-negative breast cancer (TNBC) owing to the lack of expression of ER, PR, and HER2.
A molecular driver associated with aggressive breast cancer is the mutation of TP53.Recent studies show that TP53 mutations are the most frequent alteration in metastatic breast cancers (9,10).TP53 mutations are associated with endocrine resistance in luminal breast cancers, targeted therapy resistance in HER2-enriched breast tumors; and chemoresistance in TNBCs (11)(12)(13)(14).Overall, TP53 mutations are present in 34% of all breast cancers: luminal A (12%); luminal B (29%), HER2-enriched (72%); and TNBC (88%; ref. 15).Therefore, mutations in TP53 are early events in mammary tumorigenesis that likely play a role in both initiation and progression of aggressive breast tumors.
The TP53 tumor suppressor gene encodes a transcription factor that functions as a central sensor of cell signals and is a master regulator of cell response to DNA damage (16,17).The most common TP53 alterations are missense mutations in the DNA-binding domain (16,18).Modification of any five arginine residues that are major hotspots for p53 mutations in human cancers disables p53 sequence-specific DNA binding (19).In addition to causing loss of wild-type (WT) tumor suppressor activity, some p53 missense mutations also confer growth and survival advantages including enhanced transformation, greater invasion, increased metastatic potential, and chemoresistance that surpasses what is observed in cells lacking wildtype p53 (20)(21)(22)(23)(24)(25).Mutant p53 activities are often mediated through interactions with other DNA-binding proteins to activate the transcription of genes (26,27), many of which play roles in tumor development.In breast cancer, mutant p53 interacts with SREBP2 to activate the cholesterol biosynthesis pathway driving the disorganized cell morphology of spheroids (21).A functional p53 N-terminal activation domain is required for oncogenic activities of mutant p53 (21,28,29).
We developed a physiologically relevant somatic breast cancer mouse model, in which Trp53 R172H and Trp53 R245W missense mutations (orthologs to the human breast cancer hotspot missense mutations TP53 R175H and TP53 R248W ), are focally induced (25).Trp53 R172H/� , Trp53 R245W/� , and Trp53 �/� mice were generated via mammary gland injection of adenovirus containing Cre-recombinase (Ad-Cre).Recombination in the mammary gland results in excision of WT Trp53 complementary DNA (cDNA) sequences from the endogenous locus, resulting in mammary ductal epithelial cell-specific mutant p53 expression of Trp53 R172H or Trp53 R245W and biallelic loss of Trp53 (referred to as MaP R172H , MaP R245W , or MaP � from hereon).
Mice with these mutant Trp53 alleles develop breast tumors that spontaneously disseminate and form metastases (25).
Transcriptomic analyses comparing tumors from MaP R172H/� and MaP R245W/� mice to those of MaP �/� mice revealed (i) dysregulation of unique cancer-associated pathways in tumors and (ii) Nr5a2 (nuclear receptor subfamily 5 group A member 2) as a predicted mediator of mutant-p53 transcriptional programs.
NR5A2 is an orphan nuclear hormone receptor that belongs to the NR5A (Ftz-F1) subfamily of nuclear receptors and is predominantly expressed in the enterohepatic axis and ovary (30,31).Studies support a role in embryonic stem cell differentiation and development of several cancers (32)(33)(34)(35)(36)(37)(38).Although NR5A2 has been associated with poor prognosis in breast cancer (39,40), its role in breast cancer is not fully understood.Our findings suggest NR5A2 is a novel mediator of mutant p53 transcriptional rewiring in breast cancer.

Mice
Female MaP R172H/� , MaP R245W/� , and MaP �/� mice in an F1 hybrid 50% BALB/c and 50% C57BL6/J background were generated via mammary gland injection of adenovirus containing Cre-recombinase (Ad-Cre) at the age of 10 to 12 weeks, as previously described (25).Genotyping analysis of all mice was performed by PCR as previously described (25).Mouse cohorts were monitored daily for tumor development.

Molecular subtyping
Molecular subtyping of mammary tumors to define the expression of Esr1 (ER), Pgr (PR), and Erbb2 (HER2) was performed via qRT-PCR analysis as previously described (25).

Histology
Mammary tissues harvested from mice were fixed in 10% neutral buffered formalin, followed by paraffin embedding; 4-μm tissue sections were stained with hematoxylin and eosin by the MD Anderson Cancer Center Department of Veterinary Medicine Surgery and Histology Laboratory.Tissue sections were analyzed by a pathologist.Hematoxylin and eosin bright field images were taken at 40X using the Nikon Eclipse Ni microscope.

RNA extraction
A portion of the mouse mammary tumor was homogenized using TRIzol and the Qiagen RNeasy Mini Kit (Qiagen Cat # 74104, RRID:SCR_008539) to isolate total RNA using a modified extraction protocol.TRIzol was added to tissue homogenates and incubated at room temperature for 5 minutes.A 1:5 volume of chloroform was added to the tissue/TRIzol mixture and vortexed briefly.Samples were incubated at room temperature for 3 minutes, followed by centrifugation at 12,000 g at 4 °C for 30 minutes.The aqueous phase was transferred to a new collection tube before mixing with 1.5 volume of 100% ethanol and loading onto an RNeasy spin column (Qiagen, CA).Columns were centrifuged at >8,000 g for 15 seconds.Flow-through for each sample was discarded.Each column was washed with buffer RW1, treated with DNase I, and then washed with buffers RW1 and RPE, respectively.Residual ethanol was dried with a final spin and RNA was eluted in 60 μL of nuclease-free water.
Total RNA was isolated from cultured cell lines either overexpressing p53R172H or p53R245W or harboring Nr5a2 knockdown using TRIzol reagent and the Zymo Research Direct-zol RNA Microprep kit (Zymo Research Cat #R2060, RRID:SCR_008968), following the manufacturer's instructions.
Barcoded Illumina-compatible stranded total RNA libraries were prepared using the TruSeq Stranded Total RNA Sample Preparation Kit (Illumina).
Briefly, 197 to 250 ng of DNase I treated total RNA was depleted of cytoplasmic and mitochondrial ribosomal RNA using Ribo-Zero Gold (Illumina).After purification, the RNA was fragmented using divalent cations, and double-stranded cDNA was synthesized using random primers.The ends of the resulting double-stranded cDNA fragments were repaired, 5 0phosphorylated, 3 0 -A tailed, and Illumina-specific indexed adapters were ligated.The products were purified and enriched with 12 cycles of PCR to create the final cDNA library.The libraries were quantified using the Qubit dsDNA HS Assay Kit and assessed for size distribution using the Agilent TapeStation System (RRID:SCR_018435; Agilent Technologies).The libraries for MaP R127H/� tumors were then pooled, 7 to 8 libraries per pool for a total of three pools.The libraries corresponding to MaP R245W/� and MaP �/� tumors were then pooled, with 18 samples per pool.All library pools were quantified by qPCR using the KAPA Library Quantification Kit (Kapa Biosystems).The library pools for MaP R172H/� tumors were sequenced, one pool per lane on the Illumina HiSeq 4000 sequencer using the 76-nt paired-end format.The library pool for MaP R245W/� and MaP �/� tumors was sequenced on the NovaSeq 6000 SP-200 flow cell using the 100-nt paired-end format.

Transcriptome analysis of mouse mammary tumors
FASTQ files were analyzed for read and base quality using FastQC (RRID: SCR_014583, v 0.11.9).Files that passed QC were analyzed for gene expression.

Microarray data analysis
The METABRIC study was queried for human breast tumor samples harboring a p53 R175H or p53 R248Q/W mutation in cBioPortal (47,48).This study was also queried for human breast tumors with biallelic deletion of TP53.Sample IDs for each tissue were used to acquire corresponding microarray data.Gene expression microarray results matching the identified samples were processed using the Bioconductor package limma for differential expression (RRID:SCR_006442; ref. 49).GSEA was performed using Hallmark pathways from the Molecular Signature Database and DEGs between samples with TP53 missense mutations and samples with homozygous TP53 deletions.

Transcription factor motif discovery analysis
FASTA files containing sequences from the 10,000 base pairs upstream of the transcriptional start site of significantly upregulated genes were downloaded using the Ensembl Biomart (RRID:SCR_002344; ref. 50) and used as input for MEME or MEME-SEA (MEME-Suite, Motif-based sequence analysis tools, RRID:SCR_001783) to identify significantly enriched motifs (bioRxiv 2021.08.23.457422; refs. 51,52).In a reciprocal analysis, FIMO (MEME-Suite, Motif-based sequence analysis tools, RRID:SCR_001783) was used to identify genes with promotors that harbor at least one motif corresponding to a defined transcription factor's known DNA-binding sequence (53).

Overexpression of murine p53 mutants
Lentivirus expressing murine p53R172H-or p53R245W-mutant proteins (and enhanced green fluorescent protein) was synthesized by Vector-Builder.A cell line was derived from a Trp53-null mammary tumor using previously described methods (54).The cell line was transduced with lentivirus, selected with 4-μg puromycin for 2 days, and assessed for transduction by enhanced green fluorescent protein expression on a fluorescence microscope.Transduced cells were washed with PBS and collected for protein or RNA extraction.The in-house generated cell line used for transduction experiments was authenticated for the Trp53-null genotype and for murine origin.Cells were cultured for 10 passages between thawing, transduction, and RNA extraction.

Protein isolation, immunoprecipitation, and immunoblotting
Protein was isolated from cell pellets using NP-40 lysis buffer containing a protein inhibitor cocktail (Sigma-Aldrich).Supernatant was collected and protein was quantified utilizing the Pierce BCA Protein Assay Kit (Thermo Fisher).The supernatant was mixed with 6� SDS-PAGE gel loading buffer.

Real-time qRT-PCR of p53 mutant-specific target genes
Total RNA isolated from cell lines overexpressing p53R172H and p53R245W or transfected with control and Nr5a2 siRNAs was used to prepare cDNA via reverse transcription using the iScript cDNA synthesis kit (Bio-Rad).
qRT-PCR was performed using SYBR green (Bimake) on the Bio-Rad CFX384 Real-Time Detection System (RRID:SCR_018057; Bio-Rad).Relative expression levels for murine p53 mutant-specific target genes of interest or Nr5a2 were normalized to Actab.A description of the primers used is in Supplementary Table S1.

Statistical analysis
Kaplan-Meier survival analyses for each mouse cohort were performed using Prism 9 software (GraphPad Software, v 9, RRID:SCR_002798, CA).Fisher's exact tests comparing primary tumor incidence were performed using Prism

Somatic breast cancer mouse models driven by p53 mutations recapitulate human tumor diversity
To examine how p53R172H and p53R245W drive mammary tumor development and progression, we generated and examined mice with the following mutations: , and MaP �/� (n ¼ 6).Adenovirus containing Cre-recombinase (Ad-Cre) was delivered to the mammary epithelial duct via intraductal injection.Ad-Cre allows recombination of the WT Trp53 allele to express mutant p53.Thus, mutant p53 is only expressed in the epithelial cells, while the stroma and immune components retain WT Trp53 (25).MaP R172H/� mice resulted in 100% mammary tumor formation, with a median latency of 13.9 months postinjection (Table 1; Fig. 1A and B).Moreover, 100% of MaP R245W/� mice developed primary mammary tumors with a median latency of 14 months postinjection (Table 1; Fig. 1A and B).Only 67% (4/6) of Ad-Cre injected MaP -/ -mice developed mammary tumors, with a median latency of 14.7 months postinjection (Table 1; Fig. 1A and B).No statistically significant differences in tumor-free survival were observed (Fig. 1B).
Distant metastases in our cohorts were determined by gross dissection and histopathology and, in some cases, visualized using TdTomato as a fluorescent reporter (Table 1; Fig. 2).Among the MaP R172H/� mice, only 28% of the mice (2 out of 7) bearing mammary tumors developed metastasis to the lungs (Table 1; Fig. 2A and B).In the MaP R245W/� cohort, 100% of the mammary tumors (n ¼ 5) were metastatic to the lungs (Table 1; Fig. 2A and C).Immunofluorescence staining for TdTomato revealed recombination in epithelial cells of primary mammary tumors and their lung metastases in both MaP R172H/� and MaP R245W/� mice (Fig. 2B and C).In the MaP �/� cohort, four mice developed primary breast tumors and three of these mice developed metastasis, two with metastasis to the lungs, and one mouse with metastasis to both the lung and liver (Table 1; Fig. 2A).

Transcriptomes of mutant p53-driven tumors demonstrate mutant-specific pathway enrichment
Mutant p53 oncogenic activities have been attributed to its interactions with other transcription factors and disruption of transcriptional networks (27).The transcriptional output associated with these activities becomes and MaP R245W/� (n ¼ 5) mammary tumors and compared with MaP �/� (n ¼ 6) mammary tumors (Table 1).Comparison of MaP R172H/� mammary tumors to MaP �/� mammary tumors revealed 324 DEGs, of which 106 genes are upregulated and 218 genes are downregulated using an FDR of 5% and a log 2 fold change of 2 (Fig. 3A and B).To assess pathways dysregulated by p53R172H, GSEA was performed using the Hallmark database of pathway signatures (Fig. 3C; Supplementary Table S2

Promoter analysis nominates candidate mediators of p53 mutant-specific transcriptional programs in breast cancer
The transcriptomes of p53R172H-and p53R245W-driven breast tumors demonstrate little overlap.Supervised hierarchical clustering of DEGs from MaP R172H/� , MaP R245W/� , and MaP �/� mice suggests activation of distinct target genes across the mutants (Fig. 4A).InteractiVenn identified commonly shared DEGs between MaP R172H/� and MaP R245W/� mammary tumors (68).Only one upregulated gene and 18 downregulated genes were shared between the DEGs for p53R172H and p53R245W, respectively (Fig. 4B   and C).
Motif enrichment analysis of upregulated DEGs for each mutant was performed using MEME-SEA to determine which transcription factor binding motifs were enriched in the 10 kb region upstream of the transcription start sites of all DEGs (bioRxiv 2021.08.23.457422).Genes from analysis of MaP R172H/� breast tumors show enrichment for 100 transcription factor motifs (FDR < 5.0%; Table 2).Upregulated genes in MaP R245W/� tumors demonstrate enrichment for 40 transcription factor motifs (FDR < 5.0%; Table 2).To focus on which candidate transcriptional regulators were potentially mediating the transcriptomic changes associated with p53R172H and p53R245W, we identified factors known to activate Wnt signaling and oxidative phosphorylation.
Our analysis revealed NR5A2 as a candidate upstream regulator of both Wnt signaling and oxidative phosphorylation (30,69).
We then used FIMO to determine how many DEGs contained the Nr5a2 motif (53).We found 96% (104/106) of upregulated DEGs in MaP R172H/� tumors and 100% (28/28) of upregulated DEGs in MaP R245W/� harbor at least one Nr5a2 binding motif in the promoters (Fig. 4D and E; Supplementary Tables S6 and S7).Our transcriptomic analyses unexpectedly nominate Nr5a2 as a common mediator of the otherwise distinct transcriptional output from p53R172H and p53R245W mutants.

Human breast tumors harboring p53 mutations recapitulate Nr5a2 as a transcriptional mediator
To assess if Nr5a2 is a mediator of mutant p53 transcriptional rewiring in human breast cancers, we performed gene expression analyses of the METABRIC consortium, a highly annotated and comprehensive omics analysis of breast cancers (47).We compared human breast tumors harboring either a TP53 R175H or TP53 R248Q/W mutation, the human counterparts to mouse Trp53 R172H and Trp53 R245W , respectively, to TP53-null tumors.This analysis identified 106 DEGs (59 upregulated and 47 downregulated) in breast tumors harboring a TP53 R175H mutation (Fig. 5A and B).Motif enrichment analysis of upregulated DEGs identified the NR5A2 motif (FDR ¼ 2.64%) as the most significantly enriched (Fig. 5E; Supplementary Table S8;
To further assess the relevance of NR5A2 to human breast cancer, we generated an Oncoprint across breast tumors from the METABRIC study (47).This was done to determine whether NR5A2 is recurrently mutated or amplified in breast cancer and whether NR5A2 alterations co-occur with TP53 alterations (48).The METABRIC study is the only publicly available human breast cancer dataset that has mutation and copy number analyses, with 2068 samples representing the four molecular subtypes of breast cancer.Twenty-three percent of these breast cancers harbor an amplification of NR5A2; no tumors have NR5A2 deletions (Fig. 5F).TP53 is altered in 36% of these tumors, 21% of which harbor a p53 missense mutation.Fifteen percent (65/437) of breast tumors with a TP53 missense mutation also harbor a concurrent NR5A2 amplification (Fig. 5F).Each breast tumor with co-occurrence of TP53 missense mutations and NR5A2 amplification was annotated for its ER, PR, and HER2 status to examine prevalence across breast cancer subtypes.Our annotations revealed cooccurrence in 38% luminal A, 26%, luminal B, 15% HER2-enriched, and 21% TNBC (Fig. 5G).Thus, this co-occurrence is most prevalent in luminal A breast tumors.

Discussion
Mutant p53 proteins exert oncogenic activities in many cases through interactions with other transcription factors altering the cellular milieu (27).
This study describes a novel association of mutant p53R172H and p53R245W proteins with the transcription factor Nr5a2 in a somatic model of breast cancer.In this model, expression of p53R172H leads to the generation of hormonally driven metastatic breast tumors, whereas MaP R245W/- (28)       Our transcription factor motif analysis yielded many transcription factor motifs enriched upstream of p53R172H and p53R245W specific target genes.
Many of these factors were shared across the p53 mutants.We examined each transcription factor on our list to determine its association with Wnt signaling and/or oxidative phosphorylation.Nr5a2 was the only transcriptional factor candidate that has been shown to activate genes involved in both Wnt signaling and oxidative phosphorylation in other contexts (30,69).
Our meta-analyses of human breast cancer tumors harboring a p53R172H or p53R245W/Q missense mutations corroborate our findings that NR5A2 is a mediator of mutant p53 transcriptional reprogramming.NR5A2 was the top motif identified for both mutants.NR5A2 is an orphan nuclear receptor that regulates embryonic stem cell differentiation and a broad range of functions such as steroidogenesis, cholesterol homeostasis, and tumorigenesis in adult tissues (31).Both NR5A2 and SREBP2 have similar roles in cholesterol homeostasis (72).SREBP2 also binds mutant p53 and disrupts cholesterol biogenesis, leading to disorganized acinar morphology in breast spheroids (21).
Our data indicate that p53R172H and p53R245W lead to transcriptional activation of a nonoverlapping transcriptional program through a single transcription factor Nr5a2.It is possible that p53 mutants engage with Nr5a2 differently to facilitate chromatin accessibility through distinct transcriptional regulators, possibly via one or more of the factors in which motifs were also significantly enriched upstream of p53 mutantspecific target genes.Alternatively, NR5A2 has been shown to bind to enhancers and participate in epigenetic modification of chromatin.Thus, the methylation status of the genome may contribute to the differential gene signatures.Further genomic studies such as methylation profiling or ATAC-seq are necessary to determine how the distinct transcriptomes are activated.Lastly, NR5A2 is regulated by ER in luminal breast tumors, demonstrating that hormonal regulation may influence NR5A2 activity (73).The majority of Trp53 R172H/� breast tumors are of luminal origin (Table 1).Thus, it is possible that hormone receptors direct NR5A2 to mediate distinct transcriptional landscapes in breast cancer.
In conclusion, our transcriptome profiling of primary tumors from our somatic model of breast cancer suggests Nr5a2 is a key mediator involved in mutant p53-driven transcriptional rewiring in breast cancers.
Moribund mice were euthanized according to Institutional Animal Care and Use Committee guidelines, and tissues were collected in 10% v/v formalin and paraffin-embedded.Additionally, a portion of mammary tumors and matching metastases were flashfrozen on dry ice and stored for downstream analyses.All mouse experiments were performed in compliance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care and the US Public Health Service Policy on Human Care and Use of Laboratory Animals.All animal studies and procedures were approved by the Institutional Animal Care and Use Committee at MD Anderson Cancer Center.

9
software.Mann-Whitney U tests on real-time qPCR values were performed using Prism 9 software.Statistical analysis of changes in gene expression was performed using the Bioconductor packages limma and DESeq2 in the R Project for Statistical Computing environment (RRID:SCR_001905, v 4.2.0).

FIGURE 2 FFIGURE 3 FIGURE 3 (
FIGURE 2 Metastasis in MaP mice.A, Metastasis incidence in MaP mice.B, Histology and immunofluorescence staining of TdTomato in a primary breast tumor and matched lung metastasis from a MaP R172H/� mouse, YZ2.C, Histology and immunofluorescence staining of TdTomato in a primary breast tumor and matched lung metastasis from a MaP R245W/� mouse, JM1233.H&E, hematoxylin and eosin; IF, immunofluorescence.

FLI1 9 .
92E�04expression of p53R245W drives metastatic TNBC.Although the transcriptional programs initiated by the two different p53 proteins were unique for each mutation, both programs were driven via Nr5a2.Manipulation of cells in culture (increased expression of mutant p53 and downmodulation of Nr5a2) indicates a direct effect on Nr5a2 transcriptional targets.This interaction represents the gain of function as we purposely generated the model without wild-type p53 to avoid the inhibitory effects of mutant p53 on wild-type p53.

FIGURE 5
FIGURE 5 Transcriptomic meta-analysis of human breast tumors with TP53 R175H and TP53 R248W/Q mutations recapitulates NR5A2 as a coregulator.A, Supervised clustering of human breast tumors from METABRIC dataset harboring a TP53 R175H mutation compared with human breast tumors with a homozygous deletion of TP53.B, Pie chart representing significantly DEGs for TP53 R175H breast tumors compared with TP53-null breast tumors (limma), significance criteria were FDR <5% and fold change >1.5.C, Supervised clustering of human breast tumors from METABRIC dataset harboring a TP53 R248Q or TP53 R248W mutation compared with human breast tumors with a homozygous deletion of TP53.D, Pie chart representing DEGs for TP53 R248Q/W breast tumors compared with TP53-null breast tumors (limma), significance criteria were FDR <5% and fold change >1.5.E, NR5A2 binding motif sequences identified by MEME from JASPAR database.F, Oncoprint analysis in cBioportal to assess co-occurrence of TP53 mutations with NR5A2 amplifications.G, Prevalence of co-occurrence of TP53 missense mutations with NR5A2 amplification across human molecular breast cancer subtypes: luminal A (38%), luminal B (26%), HER2-enriched (15%), and TNBC (21%).

TABLE 2
Transcription factor motifs enriched in MaP R172H/� and MaP R245W/� mammary tumors