Mammary-specific expression of Trim24 establishes a mouse model of human metaplastic breast cancer

Abstract

Conditional overexpression of histone reader Tripartite motif containing protein 24 (TRIM24) in mouse mammary epithelia (Trim24^COE) drives spontaneous development of mammary carcinosarcoma tumors, lacking ER, PR and HER2. Human carcinosarcomas or metaplastic breast cancers (MpBC) are a rare, chemorefractory subclass of triple-negative breast cancers (TNBC). Comparison of Trim24^COE metaplastic carcinosarcoma morphology, TRIM24 protein levels and a derived Trim24^COE gene signature reveals strong correlation with human MpBC tumors and MpBC patient-derived xenograft (PDX) models. Global and single-cell tumor profiling reveal Met as a direct oncogenic target of TRIM24, leading to aberrant PI3K/mTOR activation. Here, we find that pharmacological inhibition of these pathways in primary Trim24^COE tumor cells and TRIM24-PROTAC treatment of MpBC TNBC PDX tumorspheres decreased cellular viability, suggesting potential in therapeutically targeting TRIM24 and its regulated pathways in TRIM24-expressing TNBC.

Fig. 1. TRIM24 over-expression promotes hyperplasia, increases ductal branching, and tumorigenesis.

a Top panels show H&E staining and bottom panels carmine alum staining of mammary glands of 2-month-old Trim24^COE and control MMTV-Cre^Tg/0. b Quantified comparison of mammary gland branch points per ductal main branch of 2-month old (n = 4 mice each) MMTV-Cre^Tg/0 (blue) and Trim24^COE (red) mice. Data represented as mean with SD and p-value (*<0.05) is calculated using two-tailed paired t test. c Tumor-free percent survival curve of MMTV-Cre^Tg/0, Trim24^LSL, and, Trim24^COE mice. The p-value is calculated based on Log-rank method. d Comparison of TRIM24 IHC and morphology of age-matched MMTV-Cre^Tg/o normal mammary gland and Trim24^COE tumor sections. Images are representative of >50 experiments. Scale bar 100 μM. e IHC of Trim24^COE tumor sections (left) and mammary glands of MMTV-Cre^Tg/0 mice (right) with antibodies recognizing FLAG, TRIM24, and p53 at two magnifications: ×1 and ×10. Images are representative of 15 experiments. Scale bar = 3 mM (1×) and 300 μM (10×).

Fig. 2. TRIM24 over-expression promotes development of mammary metaplastic carcinosarcomas with similarity to tumors borne by MpBC patients with poor probability of survival.

a Pie chart showing quantified classifications (% of total) of Trim24^COE mammary tumors along with representative H&E stained sections of Trim24^COE mammary tumors: metaplastic carcinosarcomas (blue); carcinoma (green); and adenoma (red). Images are representative of 15 experiments. Scale bar = 300 μM. b Representative IHC of metaplastic carcinosarcomas: vimentin and E-Cadherin at ×1 and ×10 magnification (left panel and right panel, respectively). Images are representative of >25 experiments. Scale bar = 3 mM (1×) and 300 μM (10×). c Comparison of H&E stained sections of (i) murine Trim24^COE tumors and (ii) human MpBC tumor. Each tumor has a cohesive epithelial component (open arrow) and a discohesive spindle cell component (solid arrow). Images are representative of >10 experiments. Scale bar = 30 μM d Representative IHC staining of TRIM24 in human MpBC biopsies, illustrating (i) negative nuclear and cytoplasmic staining; (ii) positive nuclear staining and negative cytoplasmic staining; (iii) negative nuclear staining and positive cytoplasmic staining; and (iv) positive nuclear and cytoplasmic staining (immunoperoxidase with 3,3′-diaminobenzidine chromogen). Images are representative of 47 experiments. Scale bar = 15 μM e Kaplan–Meier plot of probability of overall survival versus time (days) for MpBC patient (n = 47) shows nuclear TRIM24 is linked to poor MpBC patient survival regardless of cytoplasmic TRIM24 protein levels. Each line in the graph represents survival data for MpBC patients segregated by pathologist-scored TRIM24 nuclear (Nuc) and cytoplasmic (Cyt) levels.

Fig. 3. TRIM24 over-expression drives expression of EMT-related genes.

a Volcano plot of all genes. Green dots represent all genes which exceeded FDR cutoff. Y-axis represents −log₁₀ of FDR and X-axis represents –log₂ fold change; genes up-regulated on the right and genes down-regulated on the left. Dashed line represents FDR < 0.05 threshold b Unsupervised hierarchical clustering identifies three distinct groups from RNA deep sequencing. The heatmap is prepared using diverging scale where up-regulated genes are in red and down-regulated genes are in blue. The first group represents MMTV-Cre^Tg/0 mammary glands (green), second cluster is Trim24^COE driven mammary tumors (both carcinomas and metaplastic carcinosarcomas), and the third group is mixed non-TRIM24 driven mammary tumors and Trim24^COE carcinomas. These clusters are further classified into normal MMTV-Cre^Tg/0 mammary glands (white), carcinomas (teal), and metaplastic carcinosarcomas (gray). c Gene set enrichment analysis of up-regulated (red) and down-regulated (blue) differentially expressed pathways between Trim24^COE metaplastic carcinosarcoma tumors compared to MMTV-Cre^Tg/0 mammary glands. The dotted line represents FDR < 0.05 threshold. d Heatmap of top differentially expressed pathways calculated by ssGSEA scores between N - MMTV-Cre^Tg/0 mammary glands, C - Trim24^COE carcinomas and M - Trim24^COE metaplastic carcinosarcoma tumors.

Fig. 4. Protein profiling shows TRIM24 over-expression correlates with increased PI3K-AKT-mTOR signaling.

a Reverse Phase Protein Array shows distinct protein profiles between metaplastic carcinosarcoma tumors driven by TRIM24 compared to MMTV-Cre^Tg/0 mammary glands shown by heatmap. b The bar plot represents up-regulated hallmark pathways on proteins identified by RPPA. X-axis represents −log₁₀ FDR values for each hallmark pathways. c Validation of protein expressions by CyTOF where allograft tumors derived of TRIM24 overexpressing metaplastic carcinosarcoma primary cell lines where compared to MMTV-Cre^Tg/0 mammary glands for TRIM24, phospho-PI3K, phospho-AKT, and mTOR. d Quantification of overall median protein expression in Trim24^COE allograft tumors compared to MMTV-Cre^Tg/0 mammary glands (n = 2 biologically independent samples). Data represented as mean with SD.

Fig. 5. TRIM24 directly activates c-Met to support metaplastic carcinosarcoma cell viability.

a qRT-PCR validation of cMET RNA expression in TRIM24 overexpressing tumors compared to MMTV-Cre^Tg/0 mammary gland (03) (n = 2 technical replicates for each tumor samples). b Snapshot of Integrated Genome Viewer (IGV) showing tracks and peaks of ATAC-Seq-determined chromatin accessibility of MMTV-Cre spontaneous tumor cell line 823 (blue) and TRIM24 overexpressing primary metaplastic carcinosarcoma cell line 897 (red). The scales of the tracks are adjusted to 168. The color bar below the tracks shows positions of primers used to assess TRIM24 interactions with cMET chromatin. c Chromatin immunoprecipitation (ChIP) PCR showing enrichment of TRIM24 on Met promoter to validate ATAC-Seq on 823 and 897 cell lines. The color of each bars indicates primers corresponding to arrows shown in B (n = 3 biological replicates). (*p = 0.0209, **p = 0.0014, ***p = 0.0004). d Cell viability assay on 4T1 (mouse TNBC cell line- non-metaplastic) and TRIM24 overexpressing primary cell lines (567, 64, and 897) upon treatment with Crizotinib, an FDA-approved ATP-competitive selective small molecule inhibitor of c-Met and ALK receptor. Percent cell viability is measured using a luminescence reader; values are normalized to DMSO, a vehicle control (n = 3 biological replicates). Data represented as mean with SD in a, and mean ± SEM in c, d. p-values are calculated in c based on two-way ANOVA for multiple comparisons using Tukey method and in d using two-way ANOVA for multiple comparisons using Holm–Sidak method. (***p < 0.001, ****p < 0.0001).

Fig. 6. A murine TRIM24 metaplastic carcinosarcoma gene signature correlates with human TNBC MpBC patient signature, EMT, and chemorefractory mesenchymal subtype.

a Heatmap of expression of genes from the TRIM24 signature in Trim24^COE tumors and human metaplastic TNBCs. C - Trim24^COE carcinomas and M - Trim24^COE metaplastic carcinosarcomas. b Statistical significance of Hallmark pathways most strongly enriched in the TRIM24 signature. Red: up in TRIM24, Blue: down in TRIM24. The dotted line represents FDR < 0.05 threshold. c TRIM24 metaplastic score matched to MpBC patient gene expression (red) compared to non-MpBC TNBC patient gene expression (black). d Correlation of percentage of Vimentin staining in patient biopsies from the ARTEMIS TNBC cohort. MpBC tumors are in red. e Association of TRIM24 scores with response to primary chemotherapy in TNBC and MpBC patients from the ARTEMIS trial. pCR: pathological complete response, RCB-I-III – residual cancer burden categories 1–3. MpBC patients are shown in red. f Distribution and classification of the ARTEMIS patient cohort and their correlation with TRIM24 metaplastic score to various subtypes of TNBC. BL basal like, IM immunomodulatory, LAR luminal androgen receptor, M* mesenchymal, MSL mesenchymal stem-like, UNS unstable tumor type. Data represented as mean in c–e and p-values are calculated based on two-tailed unpaired t-test.

Fig. 7. Human MpBC patient-derived xenografts (PDX) overexpress TRIM24 and show reduced cell viability when treated with a TRIM24 protein degrader.

a Representative images of TRIM24 IHC of human MpBC PDXs (PIM056, PIM010, PIM077, and PIM228) compared to control samples, non-MpBC-TNBC PDXs, (PIM091, PIM172, and PIM262). Scale bar = 200 μM for PIM056, PIM010, and PIM077. 300 μM for PIM228 and all TNBC PDXs. b Representative graphs show luminescence-based assays of cell viability with TRIM24 degrader (dTRIM24) or negative control eTRIM24 treatment of human MpBC PDX-derived suspension cells compared to TNBC non-metaplastic PDX cell suspensions. The percentage cell viability is calculated by normalizing luminescence to DMSO treatment. Significance is calculated using ANOVA between eTRIM24 and dTRIM24 of each sample (n = 3 technical replicates). Data represented as mean ± SEM and p-values are calculated by two-way ANOVA for multiple comparison using Holm-Sidak method (*p = 0.0351, ****p < 0.0001, ns non-significant).