Regulation of MYC by CARD14 in human epithelium is a determinant of epidermal homeostasis and disease

Abstract

Caspase recruitment domain family member 14 (CARD14) and its variants are associated with both atopic dermatitis (AD) and psoriasis, but their mechanistic impact on skin barrier homeostasis is largely unknown. CARD14 is known to signal via NF-κB; however, CARD14-NF-κB signaling does not fully explain the heterogeneity of CARD14-driven disease. Here, we describe a direct interaction between CARD14 and MYC and show that CARD14 signals through MYC in keratinocytes to coordinate skin barrier homeostasis. CARD14 directly binds MYC and influences barrier formation in an MYC-dependent fashion, and this mechanism is undermined by disease-associated CARD14 variants. These studies establish a paradigm that CARD14 activation regulates skin barrier function by two distinct mechanisms, including activating NF-κB to bolster the antimicrobial (chemical) barrier and stimulating MYC to bolster the physical barrier. Finally, we show that CARD14-dependent MYC signaling occurs in other epithelia, expanding the impact of our findings beyond the skin.

Keywords: CARD14; CP: Developmental biology; MYC; allergic disease; atopic dermatitis; atopy; barrier homeostasis; epidermis; epithelium; keratinocyte; skin barrier.

Figure 1.. The AD-associated R820W variant affects NF-κB signaling and suppresses MYC transcriptional signatures

(A and B) Maximum relative expression values of NL FLG (A) and S100A8 mRNA in the skin of children with AD in the MPAACH cohort by rs11652075 genotype. CC: n = 208; CT: n = 227; TT: n = 69. Median and interquartile range are shown. The simple linear regression b coefficients and p values of the indicated groupings on log-transformed expression values are shown. (C) Relative luciferase activity in HEK293T cells transfected with the indicated plasmid and stimulated with PMA and ionomycin for 24 h. Values normalized to cells transfected with WT plasmid. Means and SDs are shown. Each symbol represents matched values from an independent replicate (n = 5). Each replicate was normalized to WT; one-sample t tests were used to test whether the mean of each experimental group was different from 1 (represented by the dotted line). (D) Schematic of the RNA-seq experiment between WT and isogenic R820W HaCaT keratinocytes treated with 24 h mepazine or DMSO vehicle (n = 3). (E) GSEA plots of the Hallmark MYC Target V1 and V2 pathways in R820W + V vs. WT + V cells. (F) IPA Upstream Regulators analysis. (G) IPA Molecular and Cellular Functions analysis. (H) Graphical summary of the IPA Network Analysis highlighting the central location of the MYC node. See also Figure S2.

Figure 2.. MYC levels are reduced, but MYC autophagy is elevated in R820W and mepazine-treated keratinocytes

(A and B) Quantification (A) and representative blot (B) of relative expression of total cellular MYC, MAD, and their ratio under each treatment condition. Each symbol represents matched values from an independent replicate (n = 8). (C and D) Quantification (C) and representative blot (D) of relative expression of nuclear MYC, MAD, and their ratio under each treatment condition. Each symbol represents matched values from an independent replicate (n = 4). (E and F) Quantification of relative autophagic flux of cellular MYC, MAD, and their ratio (E, n = 4) and LC3B (F, n = 5) under each treatment condition ±3 h bafilomycin. Each symbol represents matched values from an independent replicate. (G) Representative blot quantified in (E) and (F). Means and SDs are shown in all graphs. Means separated by paired two-way ANOVAs and Tukey post hoc tests. See also Figures S3 and S4.

Figure 3.. Direct CARD14 and MYC binding in keratinocytes is dependent on CARD14 activation status and the presence of disease-associated CARD14 variants

(A) Representative blot of IPs between tagged CARD14 and MYC expressed in KO HaCaT keratinocytes (n = 3). (B and C) Representative images (B) and quantification (C) of PLAs between tagged MYC and CARD14 expressed in KO keratinocytes. Each point represents puncta counts from one cell. Matching symbols represent data points from cells within the same replicate (≥5 co-transfected cells per condition per replicate, n = 3 replicates). Means separated by a paired two-tailed t test of average puncta counts per replicate. Means and SDs are shown. Scale bars, 50 μM. (D) In silico model of the predicted interaction between the membrane-associated guanylate kinase domain of CARD14 and the basic-helix-loop-helix domain of MYC. The 10058-F4 binding site on MYC is indicated. (E) Representative blot of IPs of KO HaCaT keratinocytes transfected with CARD14 and MYC ±10058-F4 treatment 24 before harvest (n = 3). Relative intensities of MYC bands are shown. (F and G) Representative images (F) and quantification (G) of PLAs between endogenous MYC and tagged CARD14 constructs transfected into KO keratinocytes with 3 h bafilomycin (to prevent MYC degradation) and 1 h PMA and ionomycin (to activate CARD14) treatment prior to fixation. Each symbol represents matched values from an independent replicate (≥6 cells per condition per replicate; n = 3 replicates). Each replicate was normalized to WT; one-sample t tests were used to test whether the mean of each experimental group was different from 1 (represented by the dotted line). Means and SDs are shown. Scale bars, 20 μM. See also Figure S5.

Figure 4.. CARD14-MYC signaling regulates keratinocyte proliferation and epidermal barrier function

(A and B) Quantification and representative flow plots of the proliferation of WT vs. R820W HaCaT cells (±mepazine) as determined by dye dilution (A, n = 3) and % Ki67 expression (B, n = 4). Each symbol represents matching values from independent replicates. Means and SDs are shown. (C) The ALI model. (D–F) Longitudinal TEERs of homozygous WT or R820W ALIs at baseline (D), and with mepazine and/or 10058-F4 treatment (E and F). Maximum TEERs and CIs are shown. (G) Ratio of day 8:day 6 TEERs. (H) Representative ALI H&Es. (I) Quantification of the combined thicknesses of basal, spinous, and granular layers as a percentage of full thickness; inset normalizes to untreated to highlight the effect of mepazine. (J and K) Quantification (J) and representative blot (K) of MYC protein expression in ALIs under the indicated condition. (L and M) Quantification (L) and representative blot (M) of FLG protein expression in ALIs under the indicated condition. For all ALI experiments, n = 3 keratinocyte strains per genotype. Each strain and genotype is represented by a different symbol and color (purple = WT, red = R820W). Means and SDs are shown. For inset in (I) and WT vs. R820W FLG in (L), means were separated by a two-tailed paired t test. In (L), the ALIs of each genotype that were grown concurrently were used to pair the samples for statistical purposes. For all other quantifications, means were separated by paired two-way ANOVAs and Tukey post hoc tests. F4, 10058-F4; SB, stratum basale; SG, stratum granulosum; SS, stratum spinosum. See also Figures S6 and S7.

Figure 5.. CARD14-MYC signaling promotes epidermal differentiation in vivo, and MYC signaling is altered downstream of pathogenic CARD14 variants

(A and B) Uniform manifold approximation and projections of scRNA-seq data showing CARD14-positive vs. CARD14-negative cells (A) and epidermal layer cell populations (B) in foreskin epidermal keratinocytes from five donors (GEO: GSE147482; 16,981 total cells). DC, dendritic cell. (C) GSEA results of the Hallmark MYC Target V1 and V2 pathways in CARD14⁺/MYC⁺ vs. CARD14⁻/MYC⁺ cells within each epidermal layer cell population. (D) GSEA normalized enrichment score (NES) and p_adj values of GO terms related to epidermal differentiation in CARD14⁺/MYC⁺ vs. CARD14⁻/MYC⁺ cells within each epidermal layer cell population. (E) Schematic of NF-κB and MYC activity in pathogenic CARD14 variants. (F–H) GSEA results of the Hallmark MYC pathways in the skin of individuals with PsV⁺CARD14^G117S vs. healthy (F), the skin of an individual with GPP⁺CARD14^E138A vs. healthy (G), and the skin of mice 5 days after tamoxifen induction of Card14^E138A vs. WT (H). (I) Schematic of the RNA-seq experiment between fluorescence-activated cell sorting-sorted KO HaCaT keratinocytes transfected with the variants indicated in (E). (J–L) GSEA results of the Hallmark MYC pathways in the KO HaCaT cells transfected with CARD14^R820W (J), CARD14^I593T (K), or CARD14^N737H (L) vs. CARD14^WT (n = 2). See also Figure S8.

Figure 6.. MYC signaling is preserved in the absence of CARD14 protein, and AD is absent in a rare case of human CARD14 deficiency

(A and B) GSEA results of the Hallmark MYC Target V1 and V2 pathways in KO vs. WT (A) or R820W (B) HaCaT keratinocytes. (C and D) Quantification (C) and representative blot (D) of baseline nuclear MYC in each cell type. (E and F) Quantification (E) and representative blot (F) of baseline MYC autophagy in each cell type. Each symbol in (C) and (E) represents matched values from an independent replicate (n = 4). Means separated by paired one-way ANOVAs and Tukey post hoc tests. Means and SDs are shown. (G) GSEA results of the Hallmark MYC pathways in Card14-deficient vs. WT mouse skin. (H and I) Representative images of the gross (H) and histologic (I) phenotypes of the skin of Card14-deficient vs. WT mice. (J) Quantification of epidermal thickness in Card14-deficient (n = 5) vs. WT mice (n = 7). (K) Quantification of TEWL in Card14-deficient (n = 7) vs. WT mice (n = 7). For (J) and (K), means separated with two-tailed t tests; means and SDs are shown. (L) Representative images of MYC expression in the skin of Card14-deficient vs. WT mice. (M) Schematic of the human CARD14^R298X truncation mutation. Baf., bafilomycin. Scale bars, 50 μM.

Figure 7.. Altered CARD14-MYC signaling may contribute to barrier homeostasis and allergic disease in other epithelial tissues

(A) Schematic of calling rs11652075 genotypes using RNA-seq reads. (B–G) GSEA results of the Hallmark MYC pathways in between rs11652075 genotypes within healthy tissues (B and D) or between diseased vs. healthy tissues (C and E–G). (H) Working model representing how representative CARD14 variants impact MYC signaling to alter barrier function and predispose to inflammatory skin disease. See also Figure S9.