Suppression of p16 alleviates the senescence-associated secretory phenotype

Abstract

Oncogene-induced senescence (OIS) is characterized by increased expression of the cell cycle inhibitor p16, leading to a hallmark cell cycle arrest. Suppression of p16 in this context drives proliferation, senescence bypass, and contributes to tumorigenesis. OIS cells are also characterized by the expression and secretion of a widely variable group of factors collectively termed the senescence-associated secretory phenotype (SASP). The SASP can be both beneficial and detrimental and affects the microenvironment in a highly context-dependent manner. The relationship between p16 suppression and the SASP remains unclear. Here, we show that knockdown of p16 decreases expression of the SASP factors and pro-inflammatory cytokines IL6 and CXCL8 in multiple models, including OIS and DNA damage-induced senescence. Notably, this is uncoupled from the senescence-associated cell cycle arrest. Moreover, low p16 expression in both cancer cell lines and patient samples correspond to decreased SASP gene expression, suggesting this is a universal effect of loss of p16 expression. Together, our data suggest that p16 regulates SASP gene expression, which has implications for understanding how p16 modulates both the senescent and tumor microenvironment.

Keywords: LMNB1; inflammation; interleukin-6; interleukin-8; melanoma.

Figure 1

Knockdown of p16 decreases IL6 and CXCL8 expression in oncogene-induced senescent cells. IMR90s expressing either BRAF^V600E or HRAS^G12V alone or in combination with a shRNA targeting p16 (shp16 hairpin #1). An empty pBabe retroviral vector and a shRNA targeting GFP lentiviral vector were used as controls. See Supplementary Figure 1A for an experimental timeline. (A) Immunoblot of BRAF and p16. Vinculin was used a loading control. (B) Representative images of senescence-associated β-galactosidase (β-GAL) staining and colony formation (CF). (C) Quantification of β-GAL in (B). (D) Quantification of CF in (B). (E) Immunoblot of RAS and p16. β-actin was used as loading control. (F) Representative images of β-GAL staining and colony formation (CF). (G) Quantification of β-GAL in (F). (H) Quantification of CF in (F). (I, J) IL6 and CXCL8 mRNA expression (fold change relative to control mean). Expression of target genes was normalized against multiple reference genes. Data normalized against MRPL9 are shown. n=3/group and mean±SD. 1 out of 3 experiments is shown. *p<0.05.

Figure 2

Knockdown of p16 at later timepoints decreases IL6 and CXCL8 expression without bypassing oncogene-induced senescence. IMR90s expressing BRAF^V600E alone or in combination with shRNAs targeting p16 (shp16 hairpin #1 and #2). An empty pBabe retroviral vector and a shRNA targeting GFP lentiviral vector were used as controls. See Supplementary Figure 2A for an experimental timeline. (A) Immunoblot of BRAF and p16. β-actin was used as loading control. (B) Representative images of senescence-associated β-galactosidase (β-GAL) staining and colony formation (CF). (C) Quantification of β-GAL in (B). (D) Quantification of CF in (B). (E) IL6 and CXCL8 mRNA expression (fold change relative to control mean). Expression of target genes was normalized against multiple reference genes. Data normalized against PMSC4 are shown. n=3/group and mean±SD. 1 out of 3 experiments is shown. *p<0.05. ns = not significant.

Figure 3

Knockdown of p16 in melanoma cells decreases IL6 and CXCL8 expression. The melanoma cell lines SKMel28, RPMI-7951, and Hs 600T expressing wildtype p16 were infected with lentivirus expressing a shRNA targeting p16 (shp16 hairpin #1). An shRNA targeting GFP lentiviral vector was used as control. (A) Immunoblot of p16. Vinculin was used as loading control. (B–D) mRNA expression of IL6 and CXCL8 (fold change relative to control mean) in SKMel28 (B), RPMI-7951 (C), and Hs 600T (D) melanoma cells. Expression of target genes was normalized against multiple reference genes. Data normalized against MRPL9 are shown. n=3/group and mean±SD. 1 out of 3 experiments is shown. (E–H) p16 was stably knocked down in SKMel28 melanoma cells with a shRNA (shp16 hairpin #1). An shRNA targeting GFP lentiviral vector was used as control. Cells were treated with 1μM etoposide for 6 days. (E) Representative images of senescence-associated β-galactosidase (β-GAL) staining and colony formation (CF). (F) Quantification of β-GAL in (E). (G) Quantification of CF in (E). (H) IL6 and CXCL8 mRNA expression (fold change relative to control mean). Expression of target genes was normalized against multiple reference genes. Data normalized against PMSC4 are shown. n=3/group and mean±SD. 1 out of 2 experiments is shown. *p<0.05.

Figure 4

Tumors with low CDKN2A expression have decreased expression of SASP. (A) Percentage of SASP genes significantly upregulated and downregulated in CDKN2A-low (i.e., p16-low) expressing tumors when compared to CDKN2A-high (i.e., p16-high) expressing tumors. (B) Negatively enriched SASP term among the six studied tumor types in Gene Set Enrichment Analysis (GSEA) between CDKN2A-low and CDKN2A-high expressing tumors. SKCM (skin cutaneous melanoma), PAAD (pancreatic adenocarcinoma), COADREAD (colorectal adenocarcinoma), MESO (mesothelioma), BLCA (bladder urothelial carcinoma), GBM (glioblastoma multiforme), NES (negative enrichment score). (C) Correlation between CDKN2A and LMNB1 expression for each tumor type. Data are shown as Log₂ of RSEM. Coefficient of correlation (r) and p-value were calculated using Pearson’s correlation.