Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2025 Aug 13;267(2):213–224. doi: 10.1002/path.6463

Activation of PI3K and deletion of p53 in keratin 15‐expressing mouse mammary cells induces tumor heterogeneity and plasticity modeling metaplastic breast cancer

Khoa A Nguyen 1, Lauren E McLemore 1, Yao Ke 1,2, Lisa N DePledge 1, Lynelle P Smith 1, Li Bian 1, John D Aleman 1, David Debretsion 1, Erica Wong 1, Nicole Manning 1, Xiao‐Jing Wang 2,3, Christian D Young 1,✉
PMCID: PMC12438019  PMID: 40804766

Abstract

Although relatively rare, metaplastic breast cancer responds poorly to traditional therapies compared to other subtypes. Accordingly, there is a need for novel animal models to understand its pathogenesis and plasticity. Since alterations in PIK3CA and TP53 genes are common in metaplastic breast cancer, we generated a mouse model of metaplastic breast cancer by driving Pik3ca activation and Trp53 loss in keratin 15‐expressing mammary cells. In this model, male and female mice developed spontaneous mammary lesions, with malignancy reliant on loss of both Trp53 alleles. Importantly, tumors of this model are heterogeneous and resemble the mixed histology of metaplastic breast cancer by exhibiting squamous cell carcinoma, carcinosarcoma, and sarcoma features. We developed mammary cell lines from mouse tumors representing these different histological subtypes. These Pik3ca‐activated tumor cells were more sensitive to alpelisib, a p110α‐selective inhibitor approved by the FDA for the treatment of some PIK3CA mutant cancers, compared to Pik3ca WT cells. Additionally, some of these cell lines expressed the androgen receptor, a hormone receptor targeted in prostate cancer and currently under investigation as a therapeutic target in breast cancer. Transplantation of these cell lines into recipient mice maintained histological heterogeneity. Additionally, transplantation of either Epcam+ or Epcam− sorted cells, representing epithelial cell‐like and nonepithelial cell‐like, respectively, from a carcinosarcoma cell line, initiated tumor formation. Both sorted populations formed tumors with mixed histologic features, demonstrating plasticity arising from different tumor‐initiating components. These new models of metaplastic breast cancer from relevant genetic drivers serve as a platform for identifying mechanisms driving plasticity that could inform therapeutic strategies based on histology and reveal how plasticity alters treatment efficacy. © 2025 The Author(s). The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Keywords: adenofibroma, benign, carcinosarcoma, Cre, Krt15, malignant, PIK3CA, progesterone receptor, SCC, spindle, stem cell, TP53

Introduction

Amplification or mutation of the PIK3CA oncogene (encoding p110α) and inactivation of TP53 tumor suppressor gene (encoding p53) are two frequent genomic events in a variety of cancers, including breast cancer, colorectal cancer, lung cancer, endometrial cancer, and head and neck squamous cell carcinoma [1, 2, 3, 4, 5]. Approximately 80% of PIK3CA mutations occur in three hotspot sites (E542, E545, and H1047), which hyperactivate phosphoinositide‐3‐kinase (PI3K) activity and its downstream signaling, thereby promoting cell proliferation, apoptosis resistance, and anabolic activities [6]. Similarly, PIK3CA amplification is also associated with increased PI3K signaling and occurs more frequently in squamous cell carcinomas (head and neck, lung and cervix) [2, 7, 8]. The TP53 gene encodes p53 and is the most frequently mutated gene in cancer [4]. As a key mediator of response to cellular stress and damage, p53 regulates cell cycle arrest, apoptosis, metabolic adaptation, and DNA damage repair, often acting as a counterbalance to oncogenic events [9].

Breast cancer is a heterogeneous disease characterized by distinct molecular, clinical, and histological subtypes that guide prognosis and treatment. Key markers used in the clinical diagnosis of breast cancer include expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) [1]. The expression of these receptors guides clinical subtyping and treatment strategy. Invasive ductal carcinoma and invasive lobular carcinoma are the two most frequent histologic subtypes of breast cancer, accounting for ~75% and ~10% of breast cancer diagnoses, respectively [10]. Several less common histological subtypes include cribriform carcinoma, mucinous carcinoma, medullary carcinoma, papillary carcinoma, apocrine carcinoma, and metaplastic carcinoma [10]. Across all subtypes of breast cancer, PIK3CA and TP53 are the two most often mutated genes, with mutation rates of 36% and 37%, respectively [1]. PIK3CA mutations are particularly common in ER/PR+ and HER2+ subtypes, while TP53 mutations are often associated with HER2+ and triple‐negative (lacking ER, PR, and HER2 expression) subtypes. Although PIK3CA and TP53 mutations are not strictly confined to specific breast cancer subtypes, co‐mutation of both genes is rare [11].

One specific, challenging subtype of breast cancer with high rates of both PIK3CA and TP53 mutations is metaplastic breast cancer. This rare and aggressive breast cancer is characterized by metaplasia, where the invasive epithelium presents as squamous, spindle, or sarcomatous morphologies, often alongside other epithelial features [12]. The mutation rates for PIK3CA and TP53 are especially high in these tumors, with reported PIK3CA mutation rates up to 47% and TP53 mutation rates up to 75% [13, 14, 15, 16, 17]. Thus, while PIK3CA and TP53 are highly mutated in all subtypes of breast cancers [1], the high prevalence of these mutations in metaplastic breast cancer suggests a potential link to the aggressive nature or metaplastic phenotype of this subtype.

Most metaplastic breast cancers are triple‐negative and are thus treated with standard of care matching other triple‐negative breast cancers: chemotherapy. However, outcomes for metaplastic breast cancer patients are often poorer due to a higher incidence of chemotherapy resistance [18, 19]. This resistance is likely attributed to the significant heterogeneity and lineage plasticity of these tumors [14]. Notably, the various subtypes of metaplastic breast cancer (spindle, squamous, sarcomatous, and mixed) exhibit distinct molecular characteristics that tend to correlate with treatment response [13, 15, 20, 21]. Establishing metaplasia models that mirror the plasticity seen in patient tumors is necessary for testing mechanistic preclinical hypotheses, identifying features of progression, efficacious drug combinations, and clinical correlations. Considering this need, our study focused on the generation of a genetically engineered mouse model (GEMM) of metaplastic breast cancer, based on Pik3ca activation and Trp53 loss.

Materials and methods

Ethics approval

All animal work was performed in accordance with protocols approved by the University of Colorado Anschutz Medical Campus Institutional Animal Care and Use Committee (IACUC; protocol 1223).

Mouse models

The following mouse strains were interbred to establish trigenic animals: Krt15‐CrePR1 (The Jackson Laboratory, Bar Harbor, ME, USA; strain #005249; RRID:IMSR_JAX:005249) [22]; Pik3ca*, encoding a fusion of the iSH2 domain of p85 to the p110α coding region, resulting in the expression of constitutively active p110α (The Jackson Laboratory; strain #012343; RRID:IMSR_JAX:012343) [23]; and p53 flox (The Jackson Laboratory; strain #008462; RRID:IMSR_JAX:008462) [24]. Genotyping details are provided in the Supplementary materials and methods. Mice were monitored for tumor development and euthanized when overall tumor volume exceeded 1,500 mm3, or when other humane endpoints were met. As tumors were typically apparent and palpable by 2–4 months of age, all mice were monitored for a minimum of 5 months to determine if tumors were present. Tissue was harvested from euthanized mice for histologic analyses, cell line establishment, transplantation, and drug treatments, as detailed in the Supplementary materials and methods.

Analysis of patient samples

Data used to evaluate associations between human metaplastic breast cancer histology subtypes with TP53 and PIK3CA genetic alterations were retrieved from the supplemental materials in three studies [15, 16, 17]. We included only samples with both pathology diagnoses and next‐generation sequencing mutation analyses (n = 73 patients in total).

Histology and pathology

Tissues were collected in 10% neutral buffered formalin, embedded in paraffin, and sectioned following standard techniques. Tissue processing, sectioning, and hematoxylin and eosin (H&E) staining was performed by the University of Colorado Pathology Shared Resource. Pathology diagnoses were performed by anatomic pathologists blinded to sex, genotype, and other variables. Details of immunohistochemistry detection of p53, PR, keratin 8, keratin 5, and keratin 15 are provided in the Supplementary materials and methods.

Flow cytometry and immunoblotting

Epcam+ and Epcam– G1330L or G1319 cells were stained and isolated by flow cytometry sorting at the Flow Cytometry shared resource at The University of Colorado Anschutz Medical Center. Details of flow cytometry sorting and immunoblotting analyses are detailed in the Supplementary materials and methods.

Statistical analyses

Statistical analysis was performed using GraphPad Prism software (Boston, MA, USA; version 10.3). Differences between histology subtypes were assessed by Fisher's exact test. Differences between tumor latency was assessed by the log‐rank test. Drug sensitivity IC50 analyses were determined by nonlinear regression and differences between two groups was assessed by Student's t‐test. Analysis of more than two groups was determined by one‐way ANOVA with Tukey's multiple comparison test. Symbols representing p values were applied as follows: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001.

Results

Activation of PI3K and deletion of p53 in Krt15‐expressing mammary cells induces malignant, heterogenous tumors

Using three published reports that describe the histology and mutation rates of PIK3CA and TP53 in 73 metaplastic breast cancer samples [15, 16, 17], we found that all three histologic subtypes had co‐occurring mutations in both TP53 and PIK3CA. However, the 32% rate of dual mutations observed in samples with squamous metaplasia was 2–3× higher than the other metaplastic subtypes (Figure 1A; supplementary material, Tables S1–S3). We previously described the generation of a GEMM with three genes [25]: (1) An RU486‐inducible Cre recombinase, CrePR1, under control of the keratin 15 (Krt15) promoter, ‘Krt15CrePR1’ [22]; (2) an oncogenic Pik3ca allele preceded by a lox‐stop‐lox cassette, ‘Pik3ca*’ [23]; and (3) the endogenous Trp53 allele flanked by LoxP sites [24] (Figure 1B). In this model, CrePR1 is expressed in Krt15‐expressing cells, which recombines the lox‐stop‐lox cassette leading to expression of oncogenic Pik3ca* and deletion of the Trp53 coding region, eliminating p53 expression. As Krt15 is expressed in cutaneous stem cells, this model induces cutaneous squamous cell carcinoma (SCC) with topical application of RU486 [25]. We bred a cohort of mice expressing Krt15CrePR1, Pik3ca*, and one or two floxed Trp53 alleles, Trp53 F/W and Trp53 F/F , respectively. While wildtype (WT) control C57BL/6J mice have normal mammary tissue (Figure 1C), we noted that both male and female Krt15CrePR1, Pik3ca*, Trp53 F/F mice had spontaneous development of mammary tumors without treatment of the inducing agent, RU486 (Figure 1D). Mammary tumor development was dependent upon both Krt15CrePR1 and Pik3ca* expression, as mice lacking either of these two genes did not develop mammary lesions. Importantly, while Cre activity is enhanced by RU486 induction, ‘leaky’ Cre activity in the absence of RU486 induction is observed in Krt15CrePR mice [26], explaining Krt15CrePR‐dependent mammary tumor development in the absence of RU486 induction. Hereafter, we refer to Krt15‐CrePR1, Pik3ca*, Trp53 F/W mice as “KPP F/W ” and Krt15‐CrePR1, Pik3ca*, Trp53 F/F mice as “KPP F/F ”. In female mice, loss of both Trp53 alleles accelerated mammary tumor formation by 34 days compared to the loss of one allele, and loss of both Trp53 alleles was necessary to observe mammary tumors in male mice (Figure 1E).

Figure 1.

Figure 1

Open in a new tab

Krt15‐CrePR1 induces mammary tumors in Pik3ca*, Trp53 flox/flox mice. (A) The rates of dual PIK3CA and TP53 mutation in different histology subtypes of metaplastic breast cancer across three published studies and 73 samples was determined. (B) Diagram of the trigenic mouse model: the keratin 15 (Krt15) promoter drives mRNA expression coding for the CrePR1 fusion protein. The activated Pik3ca* oncogene is preceded by a lox‐stop‐lox cassette and the Trp53 tumor suppressor gene is flanked by LoxP sites. Thus, activated CrePR1 induces Pik3ca* expression and eliminates floxed Trp53 alleles in Krt15‐expressing cells. (C,D) Gross photographs of mouse mammary glands: (C) WT mouse; (D) Krt15‐CrePR1, Pik3ca*, Trp53 flox/flox mouse. (E) Time to palpation of mammary tumors for each genotype and sex of mice is presented and assessed by the log‐rank test to compare KPPF/F to KPPF/W; *p < 0.05. (F) Immunofluorescent staining of mouse mammary gland or skin sections to detect Krt15, Krt8, and Krt5.

There are mixed reports of Krt15 expression in the mammary gland [27, 28], so we performed Krt15 staining of WT mouse mammary glands using WT skin as a positive control. Krt15 was detected in the hair follicle bulge of skin, as expected [22] and the basal layer of mammary gland ducts, similar to Krt5, while Krt8 labeled the luminal mammary epithelial cells (Figure 1F). Thus, CrePR1 expression under control of the Krt15 promoter would be expected in the mammary gland.

Tumor sections were next diagnosed by pathologists blinded to the genotype and sex of the source tissue. While we had limited numbers of male mice in our cohort, we observed no mammary lesions in KPPF/W male mice and malignant lesions in >75% of KPPF/F male mice (Figure 2A). We observed mammary lesions in >90% of KPPF/W and KPPF/F female mice (Figure 2B). All but one lesion in KPPF/W female mice were benign and all lesions in KPPF/F female mice were malignant (Figure 2B). Tumors in KPPF/F male mice were mostly sarcomas, but we did observe one carcinosarcoma and two SCCs (Figure 2C). All benign lesions in female KPPF/W mice were adenofibromas and the one malignant lesion was SCC (Figure 2D). To determine if this SCC lost p53 expression, we performed immunohistochemistry (IHC) using an anti‐p53 antibody. This KPPF/W SCC maintained nuclear p53 expression similar to the KPPF/W adenofibromas (supplementary material, Figure S1). The malignant lesions in female KPPF/F mice were most often carcinosarcomas and sarcomas, with fewer numbers of SCCs and carcinomas (Figure 2D). Benign adenofibromas were characterized by a nodular expansion of both a benign glandular (ductal and/or acinar) epithelial component admixed with a stromal component, which may be comprised of fibrous and adipose tissue (Figure 2E). The liposarcoma consisted of mature‐appearing adipose tissue with fibrous septae containing stromal cells with enlarged, irregularly shaped, hyperchromatic nuclei; with entrapped normal, benign, ductal epithelium noted in the top right (Figure 2E). SCC consisted of infiltrating nests, cords, and sheets of squamous type epithelium with areas of abrupt keratinization and keratin pearl formation, with brisk stromal inflammation (Figure 2E). Adenocarcinoma was characterized by proliferation of irregularly shaped angulated glands lined by malignant cuboidal to columnar epithelial cells infiltrating through a desmoplastic stroma (Figure 2E). Carcinosarcoma included an admixed malignant population of cells comprised of a carcinomatous (malignant epithelial) component arranged in nests and glandular structures (top), and a sarcomatous (malignant stromal) component (bottom) arranged in sheets of atypical stellate and rhabdoid stromal cells with a myxoid to chondroid‐like stromal matrix (Figure 2E). Other examples of carcinosarcomas included SCC differentiated epithelium and osteosarcoma‐appearing sarcomatous components (Figure 2F). We lysed four different KPPF/F tumors and evaluated expression of p110α* from the Pik3ca* transgene, p53 and epithelial cell marker, E‐cadherin, by immunoblot using a carcinogen‐induced oral SCC as a control. This confirmed the expression of p110α* and loss of p53 in all KPPF/F tumors and higher levels of E‐cadherin in the carcinosarcomas (Figure 2G). In summary, in both male and female mice, loss of both alleles of p53 dramatically increased malignancy in Krt15CrePR1, Pik3ca* mice and these tumors were histologically heterogenous. The presence of squamous differentiation and sarcomatous components in most KPPF/F tumors resembles metaplastic breast cancer [10, 12].

Figure 2.

Figure 2

Open in a new tab

Loss of both Trp53 alleles and activation of Pik3ca* in Krt15+ mammary cells induces malignant tumors with diverse histology. (A, C) KPPF/W and KPPF/F male and (B, D) KPPF/W and KPPF/F female mice were monitored for mammary tumor formation. H&E‐stained tumor sections were assessed by genotype‐ and sex‐blinded pathologists to determine (A,B) malignancy and (C,D) histologic subtypes. Data are presented as pie charts reflecting the percentage of each subgroup per genotype, and assessed by Fisher's exact test to compare Trp53 F/W to Trp53 F/F ; *p < 0.05; ****p < 0.0001. (E) Representative photomicrographs of each indicated histological subtype. (F) Carcinosarcomas contain components of carcinoma and sarcoma. Osteosarcoma (‘Osteosarc.’). (G) Lysates generated from a sarcoma, two carcinosarcomas, and a squamous cell carcinoma (SCC) from the KPPF/F model were analyzed by immunoblotting for the indicated markers. Lysate of an oral SCC generated by carcinogenesis in WT C57BL/6 was used as a control. ‘*’ indicates the protein encoded by the Pik3ca* transgene and ‘E’ indicates endogenous p110α.

To further characterize these heterogenous tumors, we performed staining to detect the expression of PR and expression/localization of keratins 5, 8, and 15 in a panel of tumor sections representing distinct pathologies. The levels of ER and PR are routinely determined in breast cancer cases, but we were unable to optimize a reliable staining protocol to detect mouse ER. PR staining was most prominent in epithelial cells of adenofibromas, with lower levels also observed in the carcinoma component of carcinosarcomas. However, sarcomas and SCCs were generally negative for PR staining (Figure 3A,B). Since Krt15 is expressed in stem cells of the skin and mammary gland [22, 28], it is expected that normal tissue and malignant tissue originating from Krt15+ cells may express different keratins as they proliferate and differentiate [22, 29]. Indeed, we observed expression of Krt15 in carcinoma components of tumors (carcinosarcomas, SCCs, and adenosquamous carcinoma), but not sarcomatous components of tumors (Figure 3C, supplementary material, Figure S2). Interestingly, Krt15 staining was found to overlap with a subset of Krt5+ cells in carcinosarcoma, but Krt15 staining overlapped with a subset of Krt8+ cells in SCC (supplementary material, Figure S2). Tumors expressed basal Krt5 at the tumor−stroma and basal boundaries, while Krt8 was expressed more inwardly (Figure 3C, supplementary material, Figure S2). We observed punctate Krt8 staining in some tumors, pathologically identified as sarcoma (Figure 3C), consistent with spindle cell pathology.

Figure 3.

Figure 3

Open in a new tab

Diversity of PR expression and keratin expression and distribution in mammary tumors from Krt15‐CrePR1, Pik3ca*, Trp53 F/W /Trp53 F/F mice. (A) Primary tumors representing the indicated histologic subtypes were stained with an antibody against the progesterone receptor (PR) and counterstained with hematoxylin. Representative images are presented. (B) The number of PR+ nuclei per 10× field of view was quantified in 6–10 samples per histologic subtype, with each point representing one sample. **p < 0.01; ****p < 0.0001 determined by one‐way ANOVA with Tukey's multiple comparison test. (C) Primary tumors representing the indicated histologic subtypes were stained with antibodies against keratin 15, keratin 5, and keratin 8 with a DAPI nuclear counterstain and imaged. (D) A carcinosarcoma cell line from mouse G1330L was established, transplanted into a WT C57BL/6J mouse and the resulting tumor was harvested and stained as described for panel (C).

PI3Kα inhibitor sensitivity and tumor plasticity is maintained in KPPF /F mammary carcinosarcoma tumor cells

While harvesting mammary tumors from KPPF/F mice, we generated cell lines from primary tumors prior to knowing the pathology of the tumor. We were unable to establish cell lines from KPPF/W mice, likely due to their benign nature. A cell line established from G1330L carcinosarcoma was transplanted into recipient female C57BL/6J mice and formed tumors with similar morphology and Krt5/Krt8/Krt15 staining as the parent tumor (Figure 3D). Additionally, we established cell lines from G1330R SCC, G1319 carcinosarcoma, and M935C and C1362 sarcomas. We used carcinogen‐induced epithelial cell‐appearing A1206 oral SCC cells [30] as a cell line control (Figure 4).

Figure 4.

Figure 4

Open in a new tab

Cell line models of Pik3ca*, Trp53 F/F mammary tumors are sensitive to p110α inhibition. Five tumor cell lines were established from five independent lesions arising in the mammary glands of four different Krt15‐CrePR1, Pik3ca*, Trp53f/f mice. A1206 SCC cells, derived from 4NQO‐induced tongue carcinogenesis in C57BL/6J mice, were used as a control. Two subclones of G1319 were established as early trypsinizing, fibroblast/mesenchymal‐appearing (‘F’) cells and more adherent, epithelial‐appearing (‘A’) cells. (A) Photomicrographs capturing the morphology of each cell line. The histology descriptor represents the histology of the tumor from which each cell line was derived. (B) Cell lysates were harvested and analyzed by immunoblotting for the indicated markers. ‘*’ indicates the protein encoded by the Pik3ca* transgene and ‘E’ indicates endogenous p110α expressed by all cell lines. (C) Sensitivity of the indicated cell lines to Alpelisib was determined by 3‐day proliferation assay assessed by SRB staining and IC50 determination evaluated by Student's t‐test; *p < 0.05. (D) Tumorigenicity of each cell line in immune‐compromised nude mice or syngeneic C57BL/6J mice.

The SCC cell lines maintained a cuboidal morphology, while the carcinosarcoma and sarcoma cell lines were more spindle‐shaped. The G1319 carcinosarcoma formed two different‐appearing populations, which we isolated by differential trypsinization times to establish a more mesenchymal/fibroblast‐appearing ‘F’ subclone, and a more tightly adherent, cuboidal ‘A’ subclone (Figure 4A). Analysis of cell lysates using immunoblotting analysis confirmed expression of the p110α* protein (from the Pik3ca* transgene) at the expected higher molecular weight compared to endogenously expressed p110α. All mammary cell lines established from KPPF/F models lacked p53 expression (Figure 4B). The two sarcoma cell lines lack E‐cadherin expression, a marker of epithelial cells. However, M935C sarcoma cells do express moderate levels of Krt5 and Krt14, markers of epithelial cells. All SCC and carcinosarcoma cells express E‐cadherin and Krt5, Krt8, and Krt14. Enrichment of more epithelial‐appearing G1319 ‘A’ cells had higher levels of E‐cadherin and Krt8 compared to their more spindly, mesenchymal‐appearing ‘F’ siblings, which still maintained these epithelial markers, but at reduced levels (Figure 4B). All cell lines express vimentin, a marker of mesenchymal cells or mesenchymal differentiation, demonstrating some level of differentiation (or epithelial‐mesenchymal transition, EMT) away from the epithelial origin (Figure 4B).

As hormone receptor status guides breast cancer diagnosis and treatment, we performed immunoblotting analysis to detect hormone receptors in these KPPF/F cell lines. Using mouse uterus lysate as a positive control, no cell line expressed ER or PR, but G1319 and G1330R both expressed androgen receptor (AR) (supplementary material, Figure S3A). Antiandrogen therapies, including enzalutamide, are being tested to treat AR+ breast cancers [31, 32]. Even high doses of enzalutamide failed to inhibit >50% of the growth of these cell lines, although AR+ G1330R cells were the most sensitive of these relatively resistant cell lines (supplementary material, Figure S3B). Alpelisib, a PI3Kα‐selective inhibitor, is approved by the FDA for the treatment of breast cancer patients with PIK3CA mutant cancers meeting other criteria [33, 34]. We determined the sensitivity of our Pik3ca* KPPF/F cell lines to alpelisib using Pikc3a WT 4T1 mouse mammary cells, 4NQO‐induced oral SCC cell lines (A1206 and A1419), and KrasG12D, Smad4−/−‐induced SCCs (A223 and P029) as controls. We found that the KPPF/F cell lines were more sensitive to alpelisib (with lower IC50s) than the control cell lines (Figure 4C).

We transplanted cell lines into the syngeneic immune‐competent C57BL/6J mice and found that the carcinosarcoma cell lines (G1319 and G1330L) were tumorigenic, but the SCC and sarcoma cell lines (G1330R, M935C, and C1362) were not. C1362 sarcoma cells and the G1330R SCC cells were tumorigenic in immune‐compromised nude mice (Figure 4D), suggesting either moderate tumorigenic potential or an inability to escape immune clearance.

Since carcinosarcomas and sarcomas developed from a model that originally depended upon Krt15‐Cre activity (obligately expressed in epithelial cells by definition), we performed flow cytometry sorting of Epcam+ G1330L cells, and G1319A cells to establish populations of epithelial cells, discarding any potential nonepithelial cells in the populations (supplementary material, Figure S4A). The Epcam+ populations were expanded for 4 weeks in vitro, and then transplanted to syngeneic female C57BL/6J mice. The tumors arising from G1319A Epcam+ cell transplants were uniformly sarcomas, while tumors arising from G1330L Epcam+ cell transplants were either carcinosarcomas or sarcomas (with osteosarcoma features) (supplementary material, Figure S4B). These data suggest that even Epcam+ cells can result in the generation of tumors that appear as sarcomas, despite originating from epithelial cells as defined by Epcam‐positivity. To account for the possibility that sorted Epcam+ cells became Epcam− and then generate sarcomas at transplant, we sorted G1330L and G1319 Epcam+ cells and evaluated Epcam levels after routine passaging for 8 weeks. While both lines maintained >75% Epcam+ after 8 weeks, up to 25% of the cells were Epcam− (supplementary material, Figure S4C), suggesting that pure populations of Epcam+ cancer cells can lose Epcam expression over time.

We next performed flow cytometry sorting of G1330L cells to isolate the Epcam+ and Epcam− populations without in vitro expansion to test their tumorigenic potential and plasticity to induce carcinoma‐like and sarcoma‐like tumor features. We confirmed that the flow sorted Epcam+ population uniquely expressed epithelial cell marker E‐cadherin, and both the Epcam− and Epcam+ populations expressed the mesenchymal marker vimentin, albeit higher in the Epcam− population. Importantly, both the Epcam− and Epcam+ populations expressed the p110α* transgene, supporting the hypothesis that both populations are derived from Krt15‐CrePR activity to activate the p110α* transgene (Figure 5A). All transplants grew out as tumors, with the Epcam− population forming tumors with diverse histology (SCC, sarcoma, carcinoma, and carcinosarcoma), while the Epcam+ population formed mostly adenosquamous carcinomas, although one SCC and two carcinosarcomas were also observed (Figure 5B,C). In conclusion, histological plasticity was maintained within both the Epcam+ and Epcam− cells of the G1330L KPPF/F carcinosarcoma model.

Figure 5.

Figure 5

Open in a new tab

Pik3ca*, Trp53 F/F carcinosarcoma cell line maintains plasticity. G1330L cells were stained with an Epcam antibody and sorted by flow cytometry to isolate the Epcam− and Epcam+ populations. (A) Sorted cells were lysed immediately and analyzed by immunoblotting for the indicated markers. ‘*’ indicates the protein encoded by the Pik3ca* transgene and ‘E’ indicates endogenous p110α. (B,C) Epcam− or Epcam+ sorted G1330L cells were transplanted to the flanks of female C57BL/6J mice. Tumors were allowed to grow for 96 days and tumor histology was determined by H&E analysis. (B) Histology subtypes of tumors formed from Epcam− and Epcam+ transplants. Fisher's exact test was performed to compare Trp53 F/W with Trp53 F/F ; *p < 0.05. (C) Representative photomicrographs of each indicated histological subtype: squamous cell carcinoma (SCC), sarcoma, adenosquamous carcinoma, and carcinosarcoma.

Discussion

Direct activation of PI3K and loss of p53 to drive multipotent mammary tumorigenesis, modeling metaplastic breast cancer, has been observed in other mouse models [35, 36, 37, 38]. Adams et al [35] used MMTV‐Cre to induce activation of PIK3CA H1047R and found that mammary tumors were more metaplastic and aggressive with the deletion of one Trp53 allele. However, deletion of both alleles of Trp53 in this model resulted in high rates of lymphoma or thymoma, precluding the study of later‐occurring mammary pathology [35]. Our previous study using MMTV‐controlled, doxycycline inducible PIK3CA H1047R demonstrated metaplastic mammary tumor formation without direct manipulation of p53 levels [36]. Two studies published side‐by‐side demonstrated that activation of PIK3CA H1047R from a luminal keratin 8 (Krt8) promoter induced malignant mammary tumors and loss of p53 exacerbated malignancy, increasing metaplasia characteristics [37, 38]. These results align with our findings in Krt15‐Cre driven mammary tumors, where the activation of PI3K (without p53 loss) drove adenofibroma formation, but activation of PI3K coupled with a loss of p53 drove malignancy and metaplastic tumor formation (Figure 2). In fact, this is the first study we are aware of using Krt15‐Cre to model breast cancer. Human breast adenofibroma formation is most typical in younger, premenopausal women, regulated by hormones, and often express hormone receptors [39, 40]. While benign, there may be an increased risk of breast cancer in patients with breast adenofibromas [41]. As adenofibromas were only found in female KPPF/W mice (with intact p53) and were often PR+, this suggests that female hormone regulation and hyperplasia induced by activated PI3K are critical to their development, while maintenance of p53 prevented malignancy.

Our mouse KPPF/F model resembles metaplastic breast cancer. The pathological diagnoses of H&E‐stained sections from our mouse KPPF/F model used broader, descriptive terms of pathological features, not constrained to subtypes of metaplastic breast cancer. However, we did observe SCC tumors and carcinosarcomas with SCC components in our KPPF/F model, similar to human metaplastic breast cancer with squamous metaplasia—a subtype with the highest rate of co‐occurring TP53 and PIK3CA mutations (Figure 1A). This rate of PIK3CA activation may be under‐represented, as our analysis only accounted for mutation and not amplification of PIK3CA, which is especially frequent in SCCs [42, 43, 44]. The formation of sarcomas and tumors in male mice in the KPPF/F model is striking, but less applicable to modeling metaplastic breast cancer. However, some KPPF/F sarcomas appear to be spindle cell, a subtype of metaplastic breast cancer, as defined by cellular features and expression of keratin. The different histology patterns of tumors in male versus female KPPF/F mice is notable and warrants further investigation into whether hormone differences drive these changes. We noted nearly no PR expression in sarcomas and SCCs, more prominent in male mice, while female mice had higher amounts of carcinomas and carcinosarcomas, which had moderate levels of PR confined to epithelial cells.

The cell of origin for metaplastic breast cancer is controversial; however, mouse models may provide some clues. The three most frequently discussed ‘theories’ regarding the cell or origin in metaplastic breast cancer are: (1) the collision theory, stating that carcinoma and sarcoma arise separately and then mix; (2) the combination theory, stating that a multipotent progenitor cell gives rise to both carcinomatous and sarcomatous cells; and (3) the metaplasia theory, stating that the sarcomatous components are derived from carcinomatous components [12]. While our model is limited to the genetics we chose and is of mouse origin, the collision theory seems unlikely, as all lesions were induced in Krt15‐CrePR+ cells, which are obligately defined as epithelial. Our findings instead support characteristics of both the combination theory and the metaplasia theory: all tumors originated from Krt15‐expressing epithelial cells, yet both Epcam+ (epithelial‐like) and Epcam− (mesenchymal‐like) populations could initiate histological diversity. This plasticity suggests that either a multipotent progenitor cell or epithelial−mesenchymal transition and mesenchymal−epithelial transition contribute to the observed diversity. However, as we observed only endstage tumors, the progression and cellular transitions over time were not characterized, and the two theories are not necessarily mutually exclusive. Additionally, as our Epcam− and Epcam+ modeling used cell line transplantation into the flank, it is possible that the tumor microenvironment and the site of tumor initiation, the niche, may play a role, which we did not assess.

The molecular and cellular features of breast cancer dictate which therapeutic approach is best suited for each patient, and modeling these features permits preclinical testing to guide later clinical trials. Therapies targeting ER activity have long been used to treat ER+ breast cancers. Clinical trials are underway testing antiandrogen therapies to treat AR+ breast cancers [31, 32]. Two of our cell lines were positive for AR expression, but their growth was only modestly reduced by enzalutamide, an antiandrogen therapy approved by the FDA for the treatment of AR+ prostate cancer. It remains to be determined whether AR activity in these cells is important for features other than in vitro growth, such as signaling, metabolism, or in vivo plasticity. Recent clinical and preclinical research into metaplastic breast cancer has shown that inhibition of nitric oxide synthase (NOS) may benefit metaplastic breast cancer patients refractory to chemotherapy, and combining inhibition of NOS with alpelisib was synergistic in PDX models with mutations driving PI3K pathway activation [45, 46]. Similarly, the mouse cell lines developed in this study were sensitive to alpelisib, although we did not test NOS inhibition. As basket trials for PIK3CA‐mutant cancer patients explore PI3K‐directed therapies, our novel cell lines may be valuable tools for preclinical testing to identify potential mechanisms of resistance or sensitivity, rational combination therapies, or correlates of responsiveness [47, 48].

Mouse models of human disease can lead to better understanding of human disease to improve diagnostic, prognostic, and therapeutic approaches clinically. In metaplastic breast cancer patients, it appears that squamous tumors have a better prognosis, while mixed tumors have a worse prognosis [49], although the data remains incomplete. Metaplastic breast cancer is also generally more resistant to chemotherapy than triple‐negative breast cancer [12, 19], but our results with alpelisib, and those of others, suggest that PIK3CA mutant metaplastic breast cancer may be sensitive to p110α‐targeted therapies, warranting further investigation [19, 46]. In summary, we developed a novel model capturing the histological plasticity of metaplastic breast cancer. The transplantable cell lines derived from this model provide a powerful tool for hypothesis‐driven research, offering new opportunities to deepen our understanding and improve the treatment of this aggressive and challenging disease.

Author contributions statement

KAN, YK, LND, JDA, DD, EW, NM and CDY carried out experiments, interpreted data, and compiled reports. LEM, LPS and LB performed pathological diagnoses. X‐JW and CDY conceived, supported and directed the study. CDY and KAN wrote the article with input, edits, and approval from all authors.

Supporting information

Supplementary materials and methods

Figure S1. Pik3ca*, Trp53 F/W squamous cell carcinoma retains p53 expression

Figure S2. Diversity of keratin expression and distribution in mammary tumors from Krt15‐CrePR1, Pik3ca*, Trp53 F/F mice

Figure S3. Some KPPF/F tumors express androgen receptor, but none are sensitive to enzalutamide

Figure S4. Histology of carcinosarcoma cell lines after sorting for Epcam+ cells

PATH-267-213-s001.docx (4MB, docx)

Table S1. Summary of histology diagnosis and mutation status of PIK3CA and TP53 across 73 samples in three studies.

Table S2. Number of mutant cases in each histology subtype

Table S3. Percentage of mutant cases in each histology subtype

PATH-267-213-s002.xlsx (18.2KB, xlsx)

Acknowledgments

This work used shared resources at the University of Colorado Anschutz Medical Campus funded in part by Cancer Center Support Grant P30CA046934, including the Pathology Shared Resource, Flow Cytometry Shared Resource, and the Office of Laboratory Animal Resources. CDY and XJW were supported by NIH SPORE P50CA261605. CDY was supported by start‐up funds provided by the Department of Pathology and awards from the University of Colorado Head and Neck Cancer Research Program and Head and Neck Cancer SPORE developmental research program. XJW was supported by DE028420 (NIH) and VA merit award I01 BX003232 and a Research Career Scientist award IK6BX006039 from the Department of Veterans Affairs. JDA was supported by NRSA 5F31DE032592. We thank Nicole Spoelstra and Dane Sessions of the laboratory of Jennifer Richer at the University of Colorado Anschutz Medical Campus for their advice and discussions regarding hormone receptor staining and enzalutamide testing.

No conflicts of interest were declared.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  • 1. Cancer Genome Atlas Network . Comprehensive molecular portraits of human breast tumours. Nature 2012; 490: 61–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Cancer Genome Atlas Network . Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature 2015; 517: 576–582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Mendiratta G, Ke E, Aziz M, et al. Cancer gene mutation frequencies for the U.S. population. Nat Commun 2021; 12: 5961. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. ICGC/TCGA Pan‐Cancer Analysis of Whole Genomes Consortium . Pan‐cancer analysis of whole genomes. Nature 2020; 578: 82–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Kinnersley B, Sud A, Everall A, et al. Analysis of 10,478 cancer genomes identifies candidate driver genes and opportunities for precision oncology. Nat Genet 2024; 56: 1868–1877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Forbes SA, Beare D, Gunasekaran P, et al. COSMIC: exploring the world's knowledge of somatic mutations in human cancer. Nucleic Acids Res 2015; 43: D805–D811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Cancer Genome Atlas Research Network . Comprehensive genomic characterization of squamous cell lung cancers. Nature 2012; 489: 519–525. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Cancer Genome Atlas Research Network , Albert Einstein College of Medicine , Analytical Biological Services , et al. Integrated genomic and molecular characterization of cervical cancer. Nature 2017; 543: 378–384. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Levine AJ. p53: 800 million years of evolution and 40 years of discovery. Nat Rev Cancer 2020; 20: 471–480. [DOI] [PubMed] [Google Scholar]
  • 10. Makki J. Diversity of breast carcinoma: histological subtypes and clinical relevance. Clin Med Insights Pathol 2015; 8: 23–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Vellichirammal NN, Tan YD, Xiao P, et al. The mutational landscape of a US midwestern breast cancer cohort reveals subtype‐specific cancer drivers and prognostic markers. Hum Genomics 2023; 17: 64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Reddy TP, Rosato RR, Li X, et al. A comprehensive overview of metaplastic breast cancer: clinical features and molecular aberrations. Breast Cancer Res 2020; 22: 121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Afkhami M, Schmolze D, Yost SE, et al. Mutation and immune profiling of metaplastic breast cancer: correlation with survival. PLoS One 2019; 14: e0224726. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Hennessy BT, Gonzalez‐Angulo AM, Stemke‐Hale K, et al. Characterization of a naturally occurring breast cancer subset enriched in epithelial‐to‐mesenchymal transition and stem cell characteristics. Cancer Res 2009; 69: 4116–4124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Ng CKY, Piscuoglio S, Geyer FC, et al. The landscape of somatic genetic alterations in metaplastic breast carcinomas. Clin Cancer Res 2017; 23: 3859–3870. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Ross JS, Badve S, Wang K, et al. Genomic profiling of advanced‐stage, metaplastic breast carcinoma by next‐generation sequencing reveals frequent, targetable genomic abnormalities and potential new treatment options. Arch Pathol Lab Med 2015; 139: 642–649. [DOI] [PubMed] [Google Scholar]
  • 17. Zhai J, Giannini G, Ewalt MD, et al. Molecular characterization of metaplastic breast carcinoma via next‐generation sequencing. Hum Pathol 2019; 86: 85–92. [DOI] [PubMed] [Google Scholar]
  • 18. Chen IC, Lin CH, Huang CS, et al. Lack of efficacy to systemic chemotherapy for treatment of metaplastic carcinoma of the breast in the modern era. Breast Cancer Res Treat 2011; 130: 345–351. [DOI] [PubMed] [Google Scholar]
  • 19. Aydiner A, Sen F, Tambas M, et al. Metaplastic breast carcinoma versus triple‐negative breast cancer: survival and response to treatment. Medicine (Baltimore) 2015; 94: e2341. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Djomehri SI, Gonzalez ME, da Veiga LF, et al. Quantitative proteomic landscape of metaplastic breast carcinoma pathological subtypes and their relationship to triple‐negative tumors. Nat Commun 2020; 11: 1723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Han M, Salamat A, Zhu L, et al. Metaplastic breast carcinoma: a clinical‐pathologic study of 97 cases with subset analysis of response to neoadjuvant chemotherapy. Mod Pathol 2019; 32: 807–816. [DOI] [PubMed] [Google Scholar]
  • 22. Morris RJ, Liu Y, Marles L, et al. Capturing and profiling adult hair follicle stem cells. Nat Biotechnol 2004; 22: 411–417. [DOI] [PubMed] [Google Scholar]
  • 23. Srinivasan L, Sasaki Y, Calado DP, et al. PI3 kinase signals BCR‐dependent mature B cell survival. Cell 2009; 139: 573–586. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Marino S, Vooijs M, van Der Gulden H, et al. Induction of medulloblastomas in p53‐null mutant mice by somatic inactivation of Rb in the external granular layer cells of the cerebellum. Genes Dev 2000; 14: 994–1004. [PMC free article] [PubMed] [Google Scholar]
  • 25. Chen SMY, Li B, Nicklawsky AG, et al. Deletion of p53 and hyper‐activation of PIK3CA in Keratin‐15(+) stem cells Lead to the development of spontaneous squamous cell carcinoma. Int J Mol Sci 2020; 21: 6585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Li S, Park H, Trempus CS, et al. A keratin 15 containing stem cell population from the hair follicle contributes to squamous papilloma development in the mouse. Mol Carcinog 2013; 52: 751–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Mikaelian I, Hovick M, Silva KA, et al. Expression of terminal differentiation proteins defines stages of mouse mammary gland development. Vet Pathol 2006; 43: 36–49. [DOI] [PubMed] [Google Scholar]
  • 28. Soady KJ, Kendrick H, Gao Q, et al. Mouse mammary stem cells express prognostic markers for triple‐negative breast cancer. Breast Cancer Res 2015; 17: 31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Jih DM, Lyle S, Elenitsas R, et al. Cytokeratin 15 expression in trichoepitheliomas and a subset of basal cell carcinomas suggests they originate from hair follicle stem cells. J Cutan Pathol 1999; 26: 113–118. [DOI] [PubMed] [Google Scholar]
  • 30. Nguyen KA, DePledge LN, Bian L, et al. Polymorphonuclear myeloid‐derived suppressor cells and phosphatidylinositol‐3 kinase gamma are critical to tobacco‐mimicking oral carcinogenesis in mice. J Immunother Cancer 2023; 11: e007110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Elias AD, Staley AW, Fornier M, et al. Clinical and immune responses to neoadjuvant fulvestrant with or without enzalutamide in ER+/Her2‐ breast cancer. NPJ Breast Cancer 2024; 10: 88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Traina TA, Miller K, Yardley DA, et al. Enzalutamide for the treatment of androgen receptor‐expressing triple‐negative breast cancer. J Clin Oncol 2018; 36: 884–890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Andre F, Ciruelos E, Rubovszky G, et al. Alpelisib for PIK3CA‐mutated, hormone receptor‐positive advanced breast cancer. N Engl J Med 2019; 380: 1929–1940. [DOI] [PubMed] [Google Scholar]
  • 34. Narayan P, Prowell TM, Gao JJ, et al. FDA approval summary: alpelisib plus fulvestrant for patients with HR‐positive, HER2‐negative, PIK3CA‐mutated, advanced or metastatic breast cancer. Clin Cancer Res 2021; 27: 1842–1849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Adams JR, Xu K, Liu JC, et al. Cooperation between Pik3ca and p53 mutations in mouse mammary tumor formation. Cancer Res 2011; 71: 2706–2717. [DOI] [PubMed] [Google Scholar]
  • 36. Young CD, Pfefferle AD, Owens P, et al. Conditional loss of ErbB3 delays mammary gland hyperplasia induced by mutant PIK3CA without affecting mammary tumor latency, gene expression, or signaling. Cancer Res 2013; 73: 4075–4085. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Koren S, Reavie L, Couto JP, et al. PIK3CA(H1047R) induces multipotency and multi‐lineage mammary tumours. Nature 2015; 525: 114–118. [DOI] [PubMed] [Google Scholar]
  • 38. Van Keymeulen A, Lee MY, Ousset M, et al. Reactivation of multipotency by oncogenic PIK3CA induces breast tumour heterogeneity. Nature 2015; 525: 119–123. [DOI] [PubMed] [Google Scholar]
  • 39. Giani C, D'Amore E, Delarue JC, et al. Estrogen and progesterone receptors in benign breast tumors and lesions: relationship with histological and cytological features. Int J Cancer 1986; 37: 7–10. [DOI] [PubMed] [Google Scholar]
  • 40. Ajmal M, Khan M, Van Fossen K. Breast fibroadenoma. In Treasure Island (FL) Ineligible Companies. Disclosure: Myra Khan Declares no Relevant Financial Relationships with Ineligible Companies. Disclosure: Kelly van Fossen Declares no Relevant Financial Relationships with Ineligible Companies. StatPearls, 2025. [Google Scholar]
  • 41. McDivitt RW, Stevens JA, Lee NC, et al. Histologic types of benign breast disease and the risk for breast cancer. The cancer and steroid hormone study group. Cancer 1992; 69: 1408–1414. [DOI] [PubMed] [Google Scholar]
  • 42. Abu Rous F, Li P, Carskadon S, et al. Brief report: prognostic relevance of 3q amplification in squamous cell carcinoma of the lung. JTO Clin Res Rep 2023; 4: 100486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Kim HS, Lee SE, Bae YS, et al. PIK3CA amplification is associated with poor prognosis among patients with curatively resected esophageal squamous cell carcinoma. Oncotarget 2016; 7: 30691–30701. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Suda T, Hama T, Kondo S, et al. Copy number amplification of the PIK3CA gene is associated with poor prognosis in non‐lymph node metastatic head and neck squamous cell carcinoma. BMC Cancer 2012; 12: 416. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Chung AW, Anand K, Anselme AC, et al. A phase 1/2 clinical trial of the nitric oxide synthase inhibitor L‐NMMA and taxane for treating chemoresistant triple‐negative breast cancer. Sci Transl Med 2021; 13: eabj5070. [DOI] [PubMed] [Google Scholar]
  • 46. Reddy T, Puri A, Guzman‐Rojas L, et al. NOS inhibition sensitizes metaplastic breast cancer to PI3K inhibition and taxane therapy via c‐JUN repression. Nat Commun 2024; 15: 10737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Hasegawa K, Kagabu M, Mizuno M, et al. Phase II basket trial of perifosine monotherapy for recurrent gynecologic cancer with or without PIK3CA mutations. Invest New Drugs 2017; 35: 800–812. [DOI] [PubMed] [Google Scholar]
  • 48. Jhaveri K, Chang MT, Juric D, et al. Phase I basket study of taselisib, an isoform‐selective PI3K inhibitor, in patients with PIK3CA‐mutant cancers. Clin Cancer Res 2021; 27: 447–459. [DOI] [PubMed] [Google Scholar]
  • 49. McCart Reed AE, Kalaw E, Nones K, et al. Phenotypic and molecular dissection of metaplastic breast cancer and the prognostic implications. J Pathol 2019; 247: 214–227. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary materials and methods

Figure S1. Pik3ca*, Trp53 F/W squamous cell carcinoma retains p53 expression

Figure S2. Diversity of keratin expression and distribution in mammary tumors from Krt15‐CrePR1, Pik3ca*, Trp53 F/F mice

Figure S3. Some KPPF/F tumors express androgen receptor, but none are sensitive to enzalutamide

Figure S4. Histology of carcinosarcoma cell lines after sorting for Epcam+ cells

PATH-267-213-s001.docx (4MB, docx)

Table S1. Summary of histology diagnosis and mutation status of PIK3CA and TP53 across 73 samples in three studies.

Table S2. Number of mutant cases in each histology subtype

Table S3. Percentage of mutant cases in each histology subtype

PATH-267-213-s002.xlsx (18.2KB, xlsx)

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Articles from The Journal of Pathology are provided here courtesy of Wiley

RESOURCES