TGFβ1-TNFα-regulated secretion of neutrophil chemokines is independent of epithelial-mesenchymal transition in breast tumor cells

Abstract

Neutrophils exert tumor-promoting roles in breast cancer and are particularly prominent in aggressive breast tumors. The proinflammatory signals TGF-β1 and TNF-α are upregulated in breast tumors and induce epithelial-to-mesenchymal transitions (EMT), a process linked to cancer cell aggressiveness. Here, we investigated the roles of TGF-β1 and TNF-α in the recruitment of neutrophils by breast cancer cells. Dual-treatment with TGF-β1 and TNF-α induces EMT signatures in premalignant M2 cells, which are part of the MCF10A breast cancer progression model. Conditioned media (CM) harvested from M2 cells treated with TGF-β1/TNF-α gives rise to amplified neutrophil chemotaxis compared with CM from vehicle-treated M2 cells. This response correlates with higher levels of the neutrophil chemokines CXCL1 and CXCL8, in a p38MAPK-dependent manner, and is attenuated by CXCL8-neutralizing antibodies. We combined gene editing, immunological, and biochemical assays to show that neutrophil recruitment and EMT are uncoupled in treated M2 cells. Finally, analysis of transcriptomic databases of cancer cell lines revealed a significant correlation between CXCL8 and TGF-β1/TNF-α-regulated or effector genes in breast cancer. These findings establish a novel role for the TGF-β1/TNF-α/p38 MAPK signaling axis in regulating neutrophil recruitment in breast cancer, independent of their profound impact on EMT.

FIGURE 1:

TGF-β1/TNF-α treatment amplifies the neutrophil-recruiting activity of M2 cells by inducing chemokine secretion. (A) Representative bright-field (i and ii) and IF images (iii–viii) (n = 3) of control (CTRL) M2 cells (i, iii, v, and vii) or M2 cells treated with a combination of 20 ng/ml T-β1 (TGF-β1) and 100 ng/ml T-α (TNF-α) [T-β1+ T-α] for 72 h (ii, iv, vi, and viii). Airyscan confocal microscopy images showing MIP (iii and iv) or a single z image (v–viii) of fixed M2 cells stained for F-actin with phalloidin-TRITC (red) (iii and iv), E-Cad (red)/Vim (green) (v and vi), N-Cad (green)/Fn (red) (vii and viii), and nuclei with DAPI (blue). Individual channel images are provided in Supplemental Figure S4, C–E. Scale bar 50 µm for bright-filed images and 20 µm (iii–viii) for IF images. (B) Graphs depicting cytosolic (i) and nuclear (ii) area or cortex/cytosolic intensity ratio of E-Cad (iii) or integrated intensity measures of Vim, N-Cad, Fn (iv, v, and vi) in control versus T-β1/T-α–treated M2 cells. Each dot represents an average of all cells in each image (≥3 images/condition/experiment). Total number of cells (n) analyzed is reported under each condition. (C) Top: Representative Western blots of the respective markers from CTRL or T-β1/T-α–treated cells. Bottom: Graphs showing band intensities of the markers normalized to the loading controls (mean values ± SEM from n = 3). (D) Graph depicting the percentage of neutrophils that migrated into the bottom chamber of Transwells containing equal volume of CM from CTRL or T-β1/T-α–treated M2 cells or positive control fMLF (mean values ± SEM from n = 9). Each dot represents the response of neutrophils from an independent donor. (E) Graphs showing the amount (pg/ml) of CXCL1 (i), CXCL2 (ii), and CXCL8 (iii) secreted by CTRL or T-β1/T-α–treated M2 cells (mean values ± SEM from n = 8–10). Each dot represents the value from one experiment. (F) Graphs showing the percentage of neutrophils that migrated in response to CXCL8 or CM derived from T-β1/T-α–treated M2 and preincubated with anti-CXCL8 antibody or isotype control or to control fMLF (mean ± SEM from n = 4). P values were determined using unpaired t test (B and C) or paired t test (D–F).

FIGURE 2:

Neutrophil recruitment by TGF-β1/TNF-α–treated M2 cells is independent of snail and twist. (A) Top: Representative Western blots showing the expression of (i) E-Cad, (ii) N-Cad, and (iii) Fn in CTRL versus T-β1/T-α–treated cells. Bottom: Graphs showing band intensities of the markers normalized to the respective loading controls (mean values ± SEM from n = 3). (B) Graphs depicting normalized fluorescence intensity measurements of migrated CTRL or T-β1/T-α–treated cells (mean values ± SEM from n = 3–4). (C) Representative (n = 3) bright-field images showing morphological changes in M2 twist KO cells with T-β1/T-α treatment. Bar, 50 µm. (D) Top: Representative Western blots of N-Cad and Fn expression in M2 twist KO cells. Bottom: Graphs showing band intensities of the markers normalized to the respective loading controls (mean values ± SEM from n = 3). (E and F) Graphs depicting the percentage of neutrophils that migrated into the bottom chamber of Transwells containing equal volume of CM from CTRL or T-β1/T-α–treated cells or positive control fMLF (mean values ± SEM from n = 3–5). Each dot represents response of neutrophils from independent donors. (G–I) Graphs depicting the percentage of neutrophils that migrated into the bottom chamber of Transwells containing equal volume of CM from WT or twist KO cell lines for M4 (G) and BT549 (H), and from the SCR or snail sh cell line for BT549 (I) or positive control fMLF (mean values ± SEM from n = 3–4). Each dot represents response of neutrophils from independent donors. P values were determined using two-way ANOVA with Sidak (A) or one-way ANOVA with Dunnett's (B) or Turkey (E) multiple comparisons test or unpaired (D) or paired (F–I) t test.

FIGURE 3:

TNF-α treatment amplifies the neutrophil-recruiting activity of M2 cells without inducing EMT-associated changes. (A) Representative (n = 3) bright-field images showing the morphology of CRTL, T-β1, T-α, or T-β1/T-α–treated M2 cells; scale bar, 50 µm. (B) Top: Representative Western blots showing the expression of (i) N-Cad and (ii) Fn in CTRL and T-β1, T-α, or T-β1/T-α–treated cells. Bottom: Graphs showing band intensities of the markers normalized to the respective loading controls (mean ± SEM from n = 3). (C) Graph depicting the percentage of neutrophils that migrated into the bottom chamber of Transwells containing equal volume of CM from CTRL, T-β1, T-α, or T-β1/T-α–treated M2 cells (mean ± SEM from n = 4). Each dot represents response of neutrophils from independent donors. (D) Graphs showing the amount of CXCL1 (i), CXCL2 (ii), and CXCL8 (iii) secreted from CTRL, T-β1, T-α, or T-β1/T-α–treated M2 cells (mean ± SEM from n = 3). P values were determined using one-way ANOVA with Dunnett's multiple comparisons test (B–D).

FIGURE 4:

Signaling pathways activated in M2 cells treated with TGF-β1/TNF-α. (A) Top: Representative Western blots showing P-Smad3–specific bands over the course of 1, 3, and 6 h of treatments. Bottom: Graphs depicting band intensities of P-Smad3 normalized to the loading control (mean ± SEM from n = 3). Western blot of total Smad3 is provided in Supplemental Figure S3. (B) Representative IF images showing MIPs of fixed CRTL M2 cells or M2 cells treated for 30 min with T-β1, T-α, or T-β1/T-α and stained for p65 (green), F-actin with phalloidin-TRITC (red), and nucleus (blue) (i–iv); scale bar, 10 µm. Graph (v) depicting the nuclear/cytosolic intensity ration for p65. Each dot represents average of all cells in each image (three images/condition/experiment). Total number of cells (n) analyzed is depicted under each condition. Individual channel images are provided in Supplemental Figure S4F. (C) Representative Western blots showing P-p38MAPK (i), P-ERK (ii), P-JNK (iii), or P-AKT (iv) specific bands over the course of 1, 3, and 6 h of treatments. Graphs show band intensities of the respective phospho-proteins normalized to the loading control (mean values ± SEM from n = 3). Western blots of total p38MAKP, ERK, JNK, and AKT are provided in Supplemental Figure S3. P values were obtained using one-way ANOVA with Dunnett's multiple comparisons test (Bv).

FIGURE 5:

P-p38MAPK regulates the secretion of neutrophil-recruiting chemokines in TGF-β1/TNF-α–treated M2 cells. (A) Top: Representative Western blot showing P-p38MAPK–specific bands in CTRL M2 cells or M2 cells treated with T-β1, T-α or T-β1+T-α over the course of 1, 3, and 6 h. Bottom: Graph depicting band intensities of P-p38MAPK normalized to the loading control (mean ± SEM from n = 3). Representative Western blot of total p38MAKP is provided in Supplemental Figure S3. (B) Left: Representative Western blot showing P-p38MAPK–specific bands in M2 cells pretreated with 10 or 50 nM of DMPM or vehicle control and stimulated with T-β1+T-α for 72 h. Right: Graph depicting band intensities of P-p38MAPK normalized to loading control and represented as the percentage of vehicle control (mean ± SEM from n = 3). Representative Western blot of total p38MAKP is provided in Supplemental Figure S3. (C) Representative bright-filed (i–iii) or IF (iv–ix) images (n = 3) showing MIPs of M2 cells pretreated with DMPM or vehicle control and stimulated with T-β1+T-α for 72 h. Cells were stained for N-Cad (green)/Fn (red) (iv–vi) or Vim (green) (vii–ix) and nucleus (blue); scale bar, 100 µm (bright-field) or 20 µm (IF). Individual channel images are provided in Supplemental Figure S4, G and H. (D) Graphs depicting integrated intensity measures of Vim (i), N-Cad (ii), and Fn (iii). Each dot represents average of all cells in each image (≥3 images/condition/experiment). Total number of cells (n) analyzed is depicted under each condition. (E) Graphs showing amount of CXCL1 (i) and CXCL8 (ii) secreted by T-β1+T-α–treated M2 cells in the presence of DMPM or vehicle control (mean values ± SEM from n = 3). P values were determined using two-way (A) and one-way (B, D, and E) ANOVA with Dunnett's multiple comparisons test.

FIGURE 6:

Computational analysis of gene expression datasets identifies the association between CXCL8 and members of TGF-β1/TNF-α pathways. (A) Workflow for the analysis of RNAseq dataset from the Harvard Medical School Breast Cancer Profiling Project (ID: 20352). (B and C) Expression level of CXCL8 (B) and CXCL1 (C) in HR⁺, HER2⁺, or TNBC cell lines available in the LINCs dataset. (D–H) Distribution of linear correlations (Pearson r) of all genes surveyed in the LINCs dataset against CXCL8 (D and F). Individual gene correlation between CXCL8 and CXCL1 (E), CXCL8 and TGFBI (TGFBI) (G) and CXCL8 and BIRC3 (H) across the 35 cell lines. (I) Workflow for the analysis of Broad institute DepMap gene expression profiles. (J and K) Distribution of linear correlations (Pearson r) between CXCL8 and all genes surveyed in the DepMap dataset, across 1450 available cell lines (J). Correlation between CXCL1 and CXCL8 across available cell lines (K). (L) Analysis of the top 100 CXCL8 correlation genes using panther overrepresentation assay. T-β–related pathways are marked in color and their Reactome pathway ID provided. (M) DepMap Analysis of CXCL8 correlations against T-α genes in breast cancer lines versus all lineages. The heatmap presents correlation coefficients of (Pearson r). TNF gene is highlighted in red.