Mutant Ras and inflammation-driven skin tumorigenesis is suppressed via a JNK-iASPP-AP1 axis

Abstract

Concurrent mutation of a RAS oncogene and the tumor suppressor p53 is common in tumorigenesis, and inflammation can promote RAS-driven tumorigenesis without the need to mutate p53. Here, we show, using a well-established mutant RAS and an inflammation-driven mouse skin tumor model, that loss of the p53 inhibitor iASPP facilitates tumorigenesis. Specifically, iASPP regulates expression of a subset of p63 and AP1 targets, including genes involved in skin differentiation and inflammation, suggesting that loss of iASPP in keratinocytes supports a tumor-promoting inflammatory microenvironment. Mechanistically, JNK-mediated phosphorylation regulates iASPP function and inhibits iASPP binding with AP1 components, such as JUND, via PXXP/SH3 domain-mediated interaction. Our results uncover a JNK-iASPP-AP1 regulatory axis that is crucial for tissue homeostasis. We show that iASPP is a tumor suppressor and an AP1 coregulator.

Keywords: AP1/JNK; CP: Cancer; RAS; iASPP; inflammation-driven tumorigenesis; p63; skin cancer; target selective transcription.

Graphical abstract

Figure 1

iASPP suppresses skin tumorigenesis driven by chemically induced mutant Ras and inflammation (A) Diagram of the DMBA/TPA two-stage tumor induction protocol. Krt14-Cre;Ppp1r13l^+/+, Krt14-Cre;Ppp1r13l^Δ8/+, and Krt14-Cre;Ppp1r13l^Δ8/Δ8 mice are abbreviated as WT, HET, and KO, respectively. Arrows indicate treatment time points. (B) Representative images showing a dorsal view of papillomas on WT and KO mice. (C) Papilloma incidence represented as a percentage of mice bearing papillomas in iASPP WT (7%, n = 14), HET (11%, n = 9), and KO (91%, n = 11) mice. (D) Mean number of papillomas per mouse in WT, HET, and KO mice during DMBA/TPA treatment. Line plot shows mean values, shading represents SEM. (E) H&E staining of papillomas in iASPP WT and KO mice. Absence of invasion into dermis is shown on enlarged images. Scale bar, 200 μm. For p calculations for (C) and (D), see STAR Methods.

Figure 2

Loss of iASPP selectively enhances p63 binding to regions containing AP1 motifs (A) Left, study workflow diagram. Right, IF staining of iASPP in WT and KO primary mouse keratinocytes. Scale bar, 10 μm. (B) Left, in the volcano plot, genes are assigned a color depending on whether they are determined to be significant (adjusted p < 0.05) and in which condition they are significantly enriched. Right, heatmap showing scaled VST-normalized counts for the 50 genes most significantly enriched in iASPP KO. In orange: five genes from the GO-BP keratinocyte differentiation pathway. (C) MA plot showing the average signal (A values) and log₂-transformed fold changes (M values) for each p63 peak. Using MAnorm-derived p with a significance threshold of 5 × 10⁻³, we defined each peak as not enriched (gray) or significantly enriched in iASPP KO (orange) or WT (blue). (D) Dot plot showing the correlation between TF motif scores (FIMO) and enhanced p63 binding in iASPP KO. Among all 579 JASPAR 2018 CORE vertebrate motifs, AP1 (MA0490.1/JUNB) and E2F1 motifs have the highest and lowest association scores, respectively. p53 and NF-κB show intermediate ranking.

Figure 3

iASPP depletion affects expression of epidermal differentiation genes (A) Peaks annotated to expressed genes were categorized in regions according to their distance from TSS: promoter region (within 2 kb either side of TSS), enhancer region (outside of promoter but within 20 kb upstream or downstream), or other. Pie chart shows the percentage of peaks in each region. (B) Left, diagram of the mouse EDC. Right, volcano plot with significantly differentially expressed EDC genes after iASPP KO labeled. (C) p63 binding (by ChIP-seq) and TF motifs (FIMO) at example peak regions as indicated. ChIP-seq signal values are presented as signal per million reads. VST-transformed RNA-seq expression values for the highlighted genes are shown below the genome schematic. Adjusted p values for expression were calculated by DESeq2 (see STAR Methods for calculation).

Figure 4

iASPP deficient keratinocytes induce pro-inflammatory gene expression in vitro and attract macrophages in vivo (A) qPCR analysis of mRNA expression levels of inflammatory genes (and iASPP gene Ppp1r13l) in iASPP WT and KO primary keratinocytes at 0, 1, and 6 h after treatment with TNF-α. Values are mean + SD. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001; n = 3 biological replicates. (B) Left, IHC staining of S100a8 and S100a9 in untreated skin sections from iASPP WT and KO mice (adjacent sections). Scale bar, 50 μm. Right, IB of iASPP and S100a9 expression levels in iASPP WT (−tamoxifen) and KO (+tamoxifen) primary keratinocytes. (C) H&E analysis of acetone- or TPA-treated skin sections from iASPP WT and KO mice. Scale bar, 50 μm. Histograms below show epidermal thickness in the same samples. Values are mean ± SD. n = 3 (WT) and n = 4 (KO) mice in acetone cohort; n = 4 (WT), n = 3 (KO) mice in TPA cohort. (D) IHC staining of F4/80-positive macrophages in acetone- or TPA-treated skin sections from iASPP WT and KO mice. Scale bar, 50 μm. Histograms below show quantification of the same samples. Values are mean + SD. Same cohort as in (C). See STAR Methods for p calculations for (A), (C), and (D).

Figure 5

Cross-regulation between iASPP and JNK/AP1 (A) Upper, luciferase assay in HaCaT cells using involucrin enhancer reporter and 3xAP1 reporter after knockdown of iASPP or p63. Luciferase luminescence was normalized over renilla luminescence. Values are mean fold change over scramble + SD. For p calculation, see STAR Methods. Lower, IB of iASPP and p63 expression in the same samples. (B) Upper, diagram of iASPP structure and epitopes of anti-iASPP antibodies LX49.3 and LX128.5 as indicated. Lower, IF staining of iASPP in HaCaT cells ± 5 mJ/cm² UV irradiation, stained with either LX49.3 or LX128.5 as indicated. Arrow indicates nuclear iASPP. Scale bar, 10 μm. (C) IF staining of iASPP (LX49.3) and cytokeratin 10 (CK10) in organotypic epidermis tissue from immortalized Ker-CT cells. Tissues were collected 24 h post 150 mJ/cm² UV irradiation. Scale bar, 25 μm. (D) IHC analysis of iASPP expression in a sample of human skin from healthy donors or patients with psoriasis or eczema. Scale bar, 50 μm. (E) Barplot showing ratio of intensity of 75/100 kDa iASPP bands determined by IB using LX49.3 antibody in HaCaT cells upon UV irradiation and subjected to subcellular fractionation. Values are mean ratio + SD, n = 5 biological replicates. (F) IB of iASPP, p53, and p63 expression levels in p53 wild-type (HEKn, EPC2, SKML23), p53 mutant (HaCaT, SKML37, Hap1), or p53 null (H1299, Saos-2) cell lines upon UV irradiation. FL-iASPP and CL-iASPP refer to full-length and cleaved iASPP, respectively. (G) IB of iASPP and caspase-3 expression levels in whole lysates or nuclear fractions of HaCaT and MCF7 cells 24 h ± 5 mJ/cm² UV irradiation. (H) IF staining of iASPP localization in HaCaT cells pre-treated with either JNK inhibitor or p38 inhibitor for 1 h and collected 24 h ± 5 mJ/cm² UV irradiation as indicated. Arrow heads indicate nuclear iASPP. Scale bar, 10 μm. (I) IB of iASPP expression in cytoplasmic and nuclear fractions of HaCaT cells treated with JNK activator anisomycin for 1 or 24 h.

Figure 6

iASPP SH3 domain interacts with JunD N-terminal PxxP motif to inhibit AP1 transcriptional activity (A) Luciferase activity assay in H1299 cells using 3xAP1 reporter after overexpression of FL-tagged (1–295) or V5-tagged (295–828) iASPP combined with p53, HA-tagged TAp63α, or HA-tagged ΔNp63α. Luciferase activity was normalized over renilla activity. Values are mean fold change over empty vector ± SD. (B) Left, IB of pull down assay using immobilized Halo-tagged ASPP CTDs to pull down in vitro translated (IVT) Myc-tagged AP1 proteins. PD, pull down. Right, barplot showing the mean PD efficiency relative to input determined by densitometric quantification of the IB from left panel. Values are mean ± SD. n = 3 biological replicates. (C) Isothermal titration calorimetry (ITC) results of JunD peptide binding to WT iASPP CTD (left) or iASPP CTD N813A/Y814A mutant (right). Top diagrams show raw titration profiles and bottom diagrams show integrated heat. Best fit of single-site binding model is shown as a solid black line with the resulting equilibrium binding constant (KD). Fit parameters are found in Table S2. (D) IB of pull down assay using Halo-tagged iASPP CTD to pull down IVT Myc-tagged JunD WT and mutants. (E) IB of pull down assay using Halo-tagged iASPP CTD to pull down IVT Myc-tagged cJun, JunB, JunD, or p63 ± of either JunD peptide or PP1 peptide. (F) Left, co-IP using anti-iASPP antibodies in lysates of control or UV-irradiated HaCaT cells. IB for JunD. Right, co-IP using anti-JunD antibody in lysates of control or UV-irradiated HaCaT cells. IB for iASPP (LX49.3). Short and Long refer to short and long forms of JunD, respectively. (G) Co-IP using LX49.3 in lysates of control or UV-irradiated HaCaT cells treated with DMSO, JNK inhibitor, or p38 inhibitor. IB for JunD. Short and Long refer to short and long forms of JunD respectively. (H) Luciferase activity of WT JunD and indicated JunD mutants ± iASPP in H1299 cells using 3xAP1 reporter as a readout. Luciferase activity was normalized over renilla activity. Values are mean fold change over empty vector ± SD. n = 4 biological replicates. For p calculation for (A), (B), and (H), see STAR Methods.

Figure 7

Proposed models of iASPP-mediated regulation of tumorigenesis and the JNK-iASPP-AP1 axis Diagrams to show (A) how loss of iASPP in keratinocytes may enhance inflammation and create a pro-tumorigenic microenvironment; (B) how iASPP may regulate selective transcription of p63 and AP1 targets either alone or in combination; (C) iASPP SH3 domain and JunD PxxP motif mediate iASPP-JunD interaction; (D) regulatory loop of JNK/iASPP/AP1 axis.