NF-κB inhibition in keratinocytes causes RIPK1-mediated necroptosis and skin inflammation

Abstract

Tumor necrosis factor receptor 1 (TNFR1) activates NF-κB-dependent pro-inflammatory gene expression, but also induces cell death by triggering apoptosis and necroptosis. Inhibition of inhibitor of NF-κB kinase (IKK)/NF-κB signaling in keratinocytes paradoxically unleashed spontaneous TNFR1-mediated skin inflammation in mice, but the underlying mechanisms remain poorly understood. Here, we show that TNFR1 causes skin inflammation in mice with epidermis-specific knockout of IKK2 by inducing receptor interacting protein kinase 1 (RIPK1)-dependent necroptosis, and to a lesser extent also apoptosis, of keratinocytes. Combined epidermis-specific ablation of the NF-κB subunits RelA and c-Rel also caused skin inflammation by inducing TNFR1-mediated keratinocyte necroptosis. Contrary to the currently established model that inhibition of NF-κB-dependent gene transcription causes RIPK1-independent cell death, keratinocyte necroptosis, and skin inflammation in mice with epidermis-specific RelA and c-Rel deficiency also depended on RIPK1 kinase activity. These results advance our understanding of the mechanisms regulating TNFR1-induced cell death and identify RIPK1-mediated necroptosis as a potent driver of skin inflammation.

Figure 1.. IKK2 deficiency induces TNF-mediated death of primary keratinocytes.

(A) Skin sections from P1-P8 pups stained with TUNEL and CC3. Representative images are shown (control and IKK2^E-KO n = 3). Scale bars, 50 μm. (B) Primary keratinocytes from Control or IKK2^E-KO (n = 6) mice were treated with TNF (20 ng/ml) for 18 h. Cell viability was determined by WST-1 assay. Graphs show mean ± SEM from pooled data from three independent experiments. Multiple comparisons of groups were evaluated by Kruskal–Wallis one-way ANOVAs with post-Dunn corrections. (C) Skin sections from P7-P8 pups stained with TUNEL and CC3. Representative images are shown (control, IKK2^E-KO and IKK2^E-KO TNFR1^E-KO n = 3). Scale bars, 50 μm. (D, E) qRT-PCR analysis of the mRNA expression of the indicated genes in RNA isolated from the P4 epidermis of the mice with the indicated genotypes. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005. Source data are available for this figure.

Figure S1.. Microbiota colonization does not affect the skin inflammation in IKK2^E-KO mice.

(A) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies or TUNEL. Mice were raised in either specific pathogen-free or germ-free conditions. Representative images are shown (specific pathogen-free; IKK2^E-KO n = 6 [P7-P8] for H&E and n ≥ 5 for immunostainings & germ-free; IKK2^E-KO n = 11 [P7-P9] for H&E and n ≥ 3 for immunostainings). Scale bars, 50 μm. (B) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) on the skin sections from 7- to 11-d-old mice with the indicated genotypes. Each dot represents individual mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005.

Figure S2.. IKK2 deficiency sensitizes keratinocytes to TNF-induced necroptosis and apoptosis, and FADD ablation alone is not sufficient to protect skin inflammation in IKK2^E-KO mice.

(A) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (IKK2^E-KO FADD^E-KO n = 3 or H&E, and n = 2 for immunostainings). Scale bars, 50 μm. (B) Primary keratinocytes isolated from control or IKK2 ^E-KO pups (n = 4) were treated with TNF (20 ng/ml) for 18 h with or without pre-treatment with Z-VAD (20 μM) for 30 min. Graphs show mean ± SEM of pooled data from three independent experiments. Multiple comparisons of groups were evaluated by ordinary one-way ANOVA with post-Dunn corrections. (C) Primary keratinocytes isolated from pups with the indicated genotypes were treated with TNF (20 ng/ml) for 18 h. Graphs show mean ± SEM of pooled data from five independent experiments (IKK2 ^E-KO n = 7; IKK2^E-KO Ripk1^D13N/D138N n = 5; IKK2^E-KO Ripk3^−/− n = 7; and IKK2^E-KO FADD^E-KO Ripk3^−/− n = 4). Cell viability was determined by WST-1 assay. Graphs show mean ± SEM of pooled data from five independent experiments. Multiple comparisons of groups were evaluated by Kruskal–Wallis’s one-way ANOVA with post-Dunn corrections. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005.

Figure 2.. Inhibition of RIPK3 and epidermal FADD prevents skin inflammation in IKK2^E-KO mice.

(A) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (IKK2^E-KO n = 6 for H&E, and n ≥ 5 for immunostainings; IKK2^E-KO FADD^E-KO Ripk3^−/− n = 9 [P7-P8] & n = 22 [P134-P385] for H&E, and n ≥ 3 for immunostainings). Scale bars, 50 μm. (B) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) as well as quantification of TUNEL & CC3 positive cells on the skin sections from 7- to 8-d-old mice with the indicated genotypes. (C) qRT-PCR analysis of the mRNA expression of the indicated cytokines and chemokines in RNA isolated from the epidermis of 4-d-old mice with the indicated genotypes. Each dot represents individual mice. Source data are available for this figure.

Figure 3.. RIPK3-mediated keratinocyte death drives skin inflammation in IKK2^E-KO mice.

(A, B) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (IKK2^E-KO n = 6 for H&E, and n ≥ 5 for immunostainings; IKK2^E-KO Ripk3^−/− n = 8 [P7-P8] & n = 16 [P42-P98] for H&E, n ≥ 3 for immunostainings; IKK2^E-KO RIPK3^E-KO n = 7 [P7-P8] & n = 12 [P58-P191] for H&E, and n ≥ 3 for immunostainings; IKK2^E-KO Mlkl^−/− n = 9 [P7-P8] & n = 7 [P66-P103] for H&E, and n ≥ 3 for immunostainings). Scale bars, 50 μm. (C) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) as well as quantification of TUNEL and CC3-positive cells on the skin sections from 7- to 8-d-old mice with the indicated genotypes. (D) qRT-PCR analysis of the mRNA expression of the indicated cytokines and chemokines in RNA isolated from the epidermis of 4-d-old mice with the indicated genotypes (data for Ikk2^fl/fl and IKK2^E-KO mice are same as in Fig 1 and are shown for comparison). Each dot represents individual mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005. Arrows indicate skin lesions. Source data are available for this figure.

Figure 4.. Mixed lineage kinase like-dependent necroptosis drives skin inflammation in IKK2^E-KO mice.

(A) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (IKK2^E-KO n = 6 for H&E, and n ≥ 5 for immunostainings; IKK2^E-KO Mlkl^−/− n = 9 [P7-P8] & n = 7 [P66-P103] for H&E, and n ≥ 3 for immunostainings). Scale bars, 50 μm. (B) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) as well as quantification of TUNEL & CC3 positive cells on the skin sections from 7- to 8-d-old mice with the indicated genotypes. (C) qRT-PCR analysis of the mRNA expression of the indicated cytokines and chemokines in RNA isolated from the epidermis of 4-d-old mice with the indicated genotypes (data for Ikk2^fl/fl and IKK2^E-KO mice are same as in Fig 1 and are shown for comparison). Each dot represents individual mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005. Arrows indicate skin lesions. Source data are available for this figure.

Figure 5.. Kinase activity of RIPK1 triggers skin inflammation in IKK2^E-KO mice.

(A) Representative macroscopic mouse pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (IKK2^E-KO n = 6 for H&E and n ≥ 5 for immunostainings; IKK2^E-KO Ripk1^D138N/D138N n = 12 [P7-P8] & n = 20 [P136-P410] for H&E and n ≥ 3 for immunostainings). Scale bars, 50 μm. (B) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) as well as quantification of TUNEL & CC3 positive cells on the skin sections from 7- to 8-d-old mice with the indicated genotypes. (C) qRT-PCR analysis of the mRNA expression of the indicated cytokines and chemokines in RNA isolated from the epidermis of 4-d-old mice with the indicated genotypes (data for Ikk2^fl/fl and IKK2^E-KO mice are same as in Fig 1 and are shown for comparison). Each dot represents individual mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005. Arrows indicate skin lesions. Source data are available for this figure.

Figure 6.. NF-κB inhibition triggers RIPK1 kinase activity and mixed lineage kinase like-dependent necroptosis and skin inflammation.

(A, B) Representative macroscopic pictures and skin sections from mice with the indicated age and genotype stained with H&E or immunostained with the indicated antibodies. Representative images are shown (RelA^E-KO c-Rel^E-KO n = 12 [P7-P10] for H&E and n ≥ 3 for immunostainings; RelA^E-KO c-Rel^E-KO Mlkl^−/− n = 3 [P7-P11] & n = 6 [P130-P300] for H&E and n ≥ 3 for immunostainings; RelA^E-KO c-Rel^E-KO Ripk1^D138N/D138N n = 11 [P7-P8] & n = 26 [P100-P364] for H&E and n ≥ 3 for immunostainings; RelA^E-KO c-Rel^E-KO TNFR1^E-KO n = 3 [P45] & n = 4 [P161-P210] for H&E and n ≥ 3 for immunostainings). Scale bars, 50 μm. (C) Microscopic quantification of epidermal thickness (Epi. th.) and inflamed skin area (Infl. area) on the skin sections from 7- to 11-d-old mice with the indicated genotypes. Each dot represents individual mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005. (D) Primary keratinocytes from Control (n = 6) or RelA^E-KO c-Rel^E-KO (n = 5) pups were treated with TNF (20 ng/ml) for 18 h. Cell viability was determined by WST-1 assay. Graphs show mean ± SEM from pooled data from four independent experiments. (E) Representative macroscopic pictures of mice with the indicated age and genotype. Multiple comparisons of groups were evaluated by Kruskal–Wallis one-way ANOVAs with post-Dunn corrections. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005.