NF-κB-dependent secretome of senescent cells can trigger neuroendocrine transdifferentiation of breast cancer cells

Abstract

Cellular senescence is characterized by a stable proliferation arrest in response to stresses and the acquisition of a senescence-associated secretory phenotype, called SASP, composed of numerous factors including pro-inflammatory molecules, proteases, and growth factors. The SASP affects the environment of senescent cells, especially during aging, by inducing and modulating various phenotypes such as paracrine senescence, immune cell activity, and extracellular matrix deposition and organization, which critically impact various pathophysiological situations, including fibrosis and cancer. Here, we uncover a novel paracrine effect of the SASP: the neuroendocrine transdifferentiation (NED) of some epithelial cancer cells, evidenced both in the breast and prostate. Mechanistically, this effect is mediated by NF-κB-dependent SASP factors, and leads to an increase in intracellular Ca²⁺ levels. Consistently, buffering Ca²⁺ by overexpressing the CALB1 buffering protein partly reverts SASP-induced NED, suggesting that the SASP promotes NED through a SASP-induced Ca²⁺ signaling. Human breast cancer dataset analyses support that NED occurs mainly in p53 WT tumors and in older patients, in line with a role of senescent cells and its secretome, as they are increasing during aging. In conclusion, our work, uncovering SASP-induced NED in some cancer cells, paves the way for future studies aiming at better understanding the functional link between senescent cell accumulation during aging, NED and clinical patient outcome.

Keywords: aging; breast cancer; cellular senescence; neuroendocrine transdifferentiation; senescence-associated secretory phenotype.

FIGURE 1

SASP promotes neuroendocrine transdifferentiation in MCF‐7 breast cancer cells. MCF‐7 cells were treated every 3 days with control (CTL) or SASP conditioned medium for 6 days. (a) Bright‐field micrographs of MCF‐7 cell morphology (representative of more than 5 independent experiments). (b) immunostaining against α‐tubulin and staining of nuclei with Hoechst were performed. High content imaging Operetta system was used to automatically acquire images and Columbus software was used to perform analysis of cell shape (colorful segments). (c) total neurite length per cell and (d) number of segments per cell were quantified (n = 3, unpaired non‐parametric Mann–Whitney t‐test, mean ± SEM). (e–f) qPCR was performed to quantify mRNA levels of the neuroendocrine markers (NEM) (e) SCG2 and (f) CHGB (n = 4, one sample t‐test, mean ± SEM)

FIGURE 2

NF‐κB‐dependent pro‐inflammatory SASP molecules drive NED in MCF‐7. Six days after treatment with conditioned medium from MRC‐5/RAF:ER cells (CTL or SASP) transfected with control siRNA (sictl) or siRNA directed against RELA (siRELA), images of MCF‐7 cells (a) were taken. Neurite‐like structures were quantified, either (b) total neurite length or (c) number of segments per cell, using an Operetta system and Columbus software after immunostaining against α‐tubulin (n = 1408 cells, 3 independent experiments, non‐parametric Kruskal–Wallis, mean ± SEM). (d–e) RT‐qPCR on NEM (d) SCG2 and (e) CHGB were performed (n = 5 or 4, unpaired parametric Mann–Whitney t‐test, mean ± SEM)

FIGURE 3

Ca²⁺ signaling mediates SASP‐induced NED. Three or 6 days after treatment with conditioned medium from MRC‐5/RAF:ER (CTL or SASP), (a) total intracellular Ca²⁺ content was assessed with the ratiometric probe Fura2‐AM after ionomycin stimulation (n = 194 for CTL and n = 205 for SASP at D3; n = 181 for CTL and n = 91 for SASP at D6, 3 independent experiments, non‐parametric Mann–Whitney t‐test, mean [in red] ± 95% of confidence). (b–g) MCF‐7 cells were infected with a lentiviral vector encoding CALB1 (pLV‐CALB1) or GFP (pLV‐GFP) as control vector. (b) CALB1 protein levels were validated by Western blot. After 6 days of treatment with MRC‐5/RAF:ER (CTL or SASP), (c) images of MCF‐7 were taken and (d,e) neurite‐like structures (total neurite length and number of segments per cell) quantified using the Operetta system and Columbus software after α‐tubulin immunostaining (n = 881 cells, 3 independent experiments, non‐parametric Kruskal–Wallis test, mean ± SEM). (f–g) RTqPCR on NEM (f) SCG2, and (g) CHGB were performed (n = 7 or 5, unpaired parametric Mann–Whitney t‐test, mean ± SEM)

FIGURE 4

Breast tumors presenting neuroendocrine features are ER+, p53 wild‐type, and their proportion in breast cancer patients increases with aging. METABRIC data on breast tumor patients (clinical data and gene expression for 1904 tumors) were extracted and analyzed. (a) Based on the low (negative, neg) or high (positive, pos) expression of the four neuroendocrine markers (NEM) SCG2, CHGB, CHGA, and SYP, Venn diagrams were used to determine two groups of tumors: The “non‐Neuroendocrine Breast Carcinomas” (non‐NBC) (n = 1635) and the “Neuroendocrine Breast Carcinomas” (NBC) (n = 45). (b‐c) Gene set enrichment analysis (GSEA) between the two groups of patients. The y‐axis represents enrichment score (ES) and on the x‐axis are genes (vertical black lines) represented in gene sets from C5 MSigDB collection related to neuroendocrine features in (b) (“GO‐NEUROSTRANSMITTER TRANSPORT” and “GO_SIGNAL_RELEASE”) or in (c) CALCIUM signaling (CALCIUM_ION_REGULATED_EXOCYTOSIS”). FDR q‐Value is indicated in red. (d) ER status measured by immunohistochemistry was examined (n = 1635 for non‐NBC and n = 45 for NBC, Fisher's exact test). Further analyses were performed only on ER‐positive breast tumors. (e) Status of p53 (wild‐type, WT, or mutated) was analyzed in non‐NBC and NBC (n = 1210 for non‐NBC and n = 45 for NBC, Fisher's exact test). (f) Tumor grade was compared between non‐NBC and NBC patients (n = 1210 for non‐NBC and n = 45 for NBC, Fisher's exact test). (g) Mean mRNA levels of proliferation marker Ki67 in the two groups of patients was determined (n = 1210 for non‐NBC and n = 45 for NBC, non‐parametric unpaired Mann–Whitney t‐test). (h) Percentage of patients with the different types of tumors (NBC, non‐NBC or partial‐NBC presenting 1, 2, or 3 positive NEM) was examined in different age groups (n = 1445)