Fig 2: MRTFA and MRTFB have specific expression patterns based on ER status.(A) Box plots (Tukey presentation) for mRNA expression of MRTFA and MRTFB among estrogen receptor-positive (n = 1431) and estrogen receptor-negative (n = 435) invasive breast cancer cases. Data accessed via Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) and Mann-Whitney test used to calculate p-values. (B) Box plots (Tukey presentation) for mRNA expression of MRTFA and MRTFB among Luminal A (n = 500), Luminal B (n = 196), Her2 (n = 78) and Basal (n = 171) molecular subtypes of breast cancer. Data accessed via The Cancer Genome Atlas (TCGA) and Kruskal-Wallis test used to calculate p-values. (C) Box plots (Tukey presentation) for nuclear staining intensity and total staining intensity (D) of MRTFA and MRTFB in panCK-positive tumor cells of primary tumor cores. Immunohistochemistry was used to identify tumors’ estrogen-receptor status as weak or moderate positive (1–10% and 11–50% staining, n = 35), strong positive (51–100% staining, n = 168), or negative (0% staining, n = 88). Kruskal-Wallis test used to calculate p-values. (E) Dot plot of all TCGA invasive breast cancer samples’ tumor purity as calculated by the TIMER algorithm as a function of MRTFA (MKL1) and MRTFB (MKL2) gene expression. (F) Box plots (Tukey presentation) for nuclear staining intensity and total staining intensity (G) of MRTFA and MRTFB in cancer and stromal compartments of primary tumor cores. Mann-Whitney test used to calculate p-values.

Fig 3: Clinical and demographic correlates of MRTFA and MRTFB expression in lymph node metastases.(A) Pearson correlation plots of primary tumor nuclear staining intensity and matching lymph node metastasis nuclear staining intensity of MRTFA and MRTFB for 53 cases. Dotted lines signify 99% confidence interval. Cases were categorized into “Low MRTFA/B” and “High MRTFA/B” groups based on falling above or below the median nuclear lymph node staining intensity. (B) Chart with results from a series of Fisher’s exact tests run for patient demographic and tumor characteristics variables. Cases that did not have data available for every variable tested for were excluded from the analysis (n = 8). (C) Pearson correlation plots of primary tumor nuclear staining intensity and matching lymph node nuclear staining intensity of MRTFA and MRTFB for self-identified Black/African American patients (n = 31) and White patients (n = 18). Dotted lines signify 99% confidence interval. (D) Multiple linear regression model for nuclear MRTFA lymph node intensity with patient demographic and tumor characteristics predictor variables. (E) Multiple linear regression results with top five predictor variables for nuclear MRTFA lymph node intensity based on p-value. (F) Multiple linear regression model for nuclear MRTFB lymph node intensity with patient demographic and tumor characteristics predictor variables. (G) Multiple linear regression results with top five predictor variables for nuclear MRTFB lymph node intensity based on p-value. Significance level for p-values: * (p<0.05), ** (p<0.01), *** (p<0.001), **** (p<0.0001).

Fig 4: MRTFA shows a distinct expression pattern in antigen presenting cells in the tumor microenvironment.(A) Single-cell transcriptomic data showing scaled mean expression of and percent of cells expressing MRTFA and MRTFB in immune, structural, and antigen-presenting cell types of the tumor microenvironment (n = 26). Data accessed via Broad Institute Single Cell Portal. (B) Dendritic Cell (DC) and cancer associated fibroblast (CAF) enrichment scores based on TIMER algorithm as a function of MRTFA and MRTFB expression without tumor purity adjustments. (C–E) Multiplex image of (C) an invasive breast cancer sample from one patient included in the BRWG UIC-001-TMA (Scale bar, 50μm) and (D) in a ductal carcinoma in situ sample (Scale bar, 100μm for large view and scale bar, 50μm for insets) and (E) in a lymph node metastasis sample from CHTN_BrCaProg3 TMA showing cancer cells (panCK, red), endothelial cells (CD31, magenta), antigen presenting cells (HLA-DRA, yellow) and MRTFA (green) and MRTFB (cyan). (E) * shows the nearby DCIS lesion, rectangle shows the area of inset for the bottom panels.

Fig 5: Subcellular segmentation reveals a correlation between MRTFA and MRTFB staining in tumor cells and abundant nuclear localization.(A) Workflow for nuclear and cytoplasmic segmentation of 4-color multiplex immunofluorescence in preparation for statistical and bioinformatic analyses, including training of HALO AI to recognize panCK-positive tumor regions versus panCK-negative benign regions of cores. (B) Violin plot of calculated staining intensity ratios (quotient of nuclear staining intensity and cytoplasmic staining intensity) for MRTFA and MRTFB in panCK-positive tumor cells of primary tumor cores (n = 291). Median staining intensity ratios greater than 1 suggest activation and abundant nuclear expression. (C) Violin plot showing distribution of Pearson correlation coefficients of average nuclear MRTFA and MRTFB staining intensities in panCK-positive tumor cells of primary tumor cores (n = 289). Mann-Whitney test used to calculate p-values. (D) Pearson correlation plot and simple linear regression of average nuclear MRTFA and MRTFB staining intensities in panCK-positive tumor cells of primary tumor cores (n = 289).