The transcription regulators ZNF750 and LSD1/KDM1A dampen inflammation on the skin's surface by silencing pattern recognition receptors

Abstract

The surface of the skin is continually exposed to pro-inflammatory stimuli; however, it is unclear why it is not constantly inflamed due to this exposure. Here, we showed undifferentiated keratinocytes residing in the deep epidermis could trigger a strong inflammatory response due to their high expression of pattern recognition receptors (PRRs) that detect damage or pathogens. As keratinocytes differentiated, they migrated outward toward the surface of the skin and decreased their PRR expression, which led to dampened immune responses. ZNF750, a transcription factor expressed only in differentiated keratinocytes, recruited the histone demethylase KDM1A/LSD1 to silence genes coding for PRRs (TLR3, IFIH1/MDA5, and DDX58/RIG1). Loss of ZNF750 or KDM1A in human keratinocytes or mice resulted in sustained and excessive inflammation resembling psoriatic skin, which could be restored to homeostatic conditions upon silencing of TLR3. Our findings explain how the skin's surface prevents excessive inflammation through ZNF750- and KDM1A-mediated suppression of PRRs.

Keywords: KDM1A; LSD1; MDA5; RIG1; TLR3; UV damage; ZNF750; Zfp750; cytokines; differentiation; histone demethylase; innate immune response; psoriasis; skin inflammation.

Figure 1.. ZNF750 suppresses inflammatory gene expression in differentiated keratinocytes.

(A) Differentiated primary human keratinocytes were treated with control (CTLi) or ZNF750 (ZNF750i) siRNAs and the remaining ZNF750 mRNA expression were measured by RT-QPCR. Cells were stimulated with Poly IC for four hours before harvesting. Two distinct siRNAs (ZNF750i-A and ZNF750i-B) targeting ZNF750 were used. QPCR results were normalized to L32. (B-C) RT-QPCR of the inflammatory gene expression of CTLi and ZNF750i differentiated keratinocytes after treatment with Poly IC. (D) Western blot of ZNF750, PRRs (MDA5, RIG1) and interferon response proteins after ZNF750 loss (Poly IC treated differentiated samples). (E) Measurement of IFN-β secreted into the culture medium in CTLi and ZNF750i differentiated cells. (F) RNA-Seq analysis of differentiated CTLi and ZNF750i cells stimulated with Poly IC. Volcano plot shows 2,602 genes increased (red) and 1,633 genes decreased (blue) on a log2 scale upon ZNF750 depletion. (G) Gene ontology (GO) terms of the 1,633 decreased genes using Enrichr. Dotted red line is p<0.05. (H) GO terms of the 2,602 increased genes upon ZNF750 loss. (I) Heatmap of the 66 genes found in the top 4 Gene Ontology terms found in Figure 1H. N=3 independent experiments performed in Figure 1. Mean values are shown with error bars=SEM. *p <0.05, ** p <0.01, ***p <0.001 (1 way ANOVA followed by Tukey’s multiple comparison). Please also see Figures S1 and S2.

Figure 2.. Epidermal specific deletion of Zfp750 in mice results in an increased and prolonged inflammatory response to imiquimod treatment and UVB mediated tissue damage.

(A) Diagram representing the strategy used for generating the conditional Zfp750 genetically ablated mouse. (B) Schematic diagram of tamoxifen inducible deletion of Zfp750 followed by 2 consecutive days of UVB treatment. (C) Hematoxylin and eosin staining of control (Zfp750^{fl/wt or wt/wt} K14-Cre-ER^tam) and Zfp750^−/− (Zfp750^fl/fl K14-Cre-ER^tam) dorsal skin harvested from mice 24 hours after UV treatment. Brackets denote areas of neutrophil accumulation in the stratum corneum (Munro’s microabscess). (D) Staining of dorsal skin with antibodies against CD11b (myeloid cell marker: red) and Keratin 5 (K5; green), a marker of the basal layer of the epidermis and hair. Merged image includes Hoechst 33342 staining of nuclei (blue). Tissue was harvested 24 hours after the last UVB treatment. (E) Flow cytometry analysis of the percent of cells in the dorsal skin that are neutrophils (Ly6G+, CD11b+), 24 hours after the last UVB treatment. (F) Percent of cells in the dorsal skin that are inflammatory monocytes (CD45+, CD11b+, Ly6C+, Ly6G−) 24 hours after the last UVB treatment. (G) Weight of the skin draining (axillary) lymph nodes 24 hours after the last UVB treatment. (H) Percent of Th17 cells (CD4+, Il17+) in the lymph nodes 24 hours after the last UVB treatment. (I) Percent of neutrophils in the skin 6 days after the last UVB treatment. (J) Percent of inflammatory monocytes 6 days after the last UVB treatment. (K) Weight of the skin draining lymph nodes 6 days after the last UVB treatment. (L) Percent of Th17 cells 6 days after the last UVB treatment. (M) RT-QPCR of the inflammatory gene expression of control and Zfp750^−/− mice. QPCR results were normalized to Hprt. Epidermis was harvested 6 days after the last UV treatment. (N) Hematoxylin and eosin staining of skin harvested from mice 6 days after UV treatment. Black arrows denote areas of retention of nuclei in the stratum corneum (parakeratosis). (O-P) Staining of dorsal skin with antibodies against CD11b and Keratin 5 after UV (O) or imiquimod (P) treatment. (Q) Hematoxylin and eosin staining of skin. (R) Weight of the lymph nodes 24 hours after the last imiquimod treatment. (S) Percent of Th17 cells in the lymph nodes, 24 hours after the last imiquimod treatment. (T) RT-QPCR of the inflammatory gene expression of control and Zfp750^−/− epidermis harvested 24 hours after the last imiquimod treatment. Mean values are shown with error bars=SEM. NS=not significant, *p <0.05, ** p <0.01, ***p <0.001 (t-test). N is a minimum of at least 3. All individual dots in bar graphs represent data from an individual mouse. Scale bar=150um for all skin staining except for left/middle panels of Figure 2O where Scale bar=50um. Please also see Figures S2 and S3.

Figure 3.. ZNF750 dampens immune sensing by directly inhibiting the expression of PRRs such as TLR3.

(A) Venn diagram showing overlap of genes directly bound by ZNF750 (ChIP-Seq: red circle) and the genes upregulated upon siRNA silencing of ZNF750 when treated with Poly IC (RNA-Seq: green circle). (B) Gene ontology terms of the overlapped genes from (A). Dotted red line is p<0.05. (C) Top de novo motif of the 1,156 overlapped genes containing 2,753 ZNF750 binding peaks. (D) Chromatin immunoprecipitations (ChIP) was performed on CTLi (blue) and ZNF750i (red) differentiated keratinocytes (treated with Poly IC) using a ZNF750 antibody. ChIP was also performed using IGG in CTLi (black) and ZNF750i (green) cells as a specificity control. ZNF750 binding was calculated as a percentage of input. Genomic regions for TLR3, IFIH1, and DDX58 are represented as distance away from the transcriptional start site (TSS). (E) Differentiated primary human keratinocytes were silenced with scrambled control (CTLi), ZNF750 (ZNF750i), or ZNF750 and TLR3 (ZNF750i+TLR3i) siRNAs and the remaining ZNF750 mRNA expression were measured by RT-QPCR. Cells were stimulated with Poly IC for four hours. QPCR results were normalized to L32. (F) RT-QPCR of the inflammatory gene expression of CTLi, ZNF750i, ZNF750i +TLR3i differentiated keratinocytes after treatment with Poly IC. (G) Western blot of ZNF750, and interferon response proteins from CTLi, ZNF750i and ZNF750+TLR3i depleted cells after Poly IC treatment. (H) Measurement of IFN-β secreted into the culture medium in CTLi, ZNF750i, and ZNF750i+TLR3i differentiated cells after Poly IC treatment. (I) Flow cytometry analysis of the percent of cells in the dorsal skin that are neutrophils (Ly6G+, CD11b+), 24 hours after the last UVB treatment from control (Zfp750^{fl/wt or wt/wt} K14-Cre-ERtam Tlr3^+/+ or ^+/−), Zfp750^−/− (Zfp750fl/fl K14-Cre-ERtam Tlr3^+/+ or^+/−), and Zfp750^−/− Tlr3^−/− (Zfp750^fl/fl K14−Cre-ER^tam Tlr3^−/−) mice. (J) Staining of dorsal skin with antibodies against CD11b and Keratin 5. Tissue was harvested 24 hours after the last UVB treatment. Brackets denote areas of neutrophil accumulation in the stratum corneum (Munro’s microabscess). Scale bar=150um. (K) Hematoxylin and eosin staining of dorsal skin. Brackets denote Munro’s microabscess. Scale bar=150um. (L) RT-QPCR of the inflammatory gene expression of control, Zfp750^−/−, and Zfp750^−/− Tlr3^−/− mice. QPCR results were normalized to Hprt. Epidermis was harvested from mice 24 hours after the last UV treatment. For Figure 3, mean values are shown with error bars=SEM and N is a minimum of at least 3. Each dot from graphs represents a single replicate of data or data from an individual mouse. *p <0.05, ** p <0.01, ***p <0.001, ****p <0.0001 (1 way ANOVA followed by Tukey’s multiple comparison for 3 groups and T-test was performed for comparison between two groups). Please also see Figure S3.

Figure 4.. ZNF750 recruits KDM1A to PRRs to diminish the inflammatory response.

(A) Immunoprecipitations (IPs) were performed using either a ZNF750, KDM1A antibody or IGG and Western blotted for KDM1A protein expression. Keratinocytes were cultured in differentiation medium (DM) and treated with Poly IC for 4 hours before harvest. 5% of the cell lysate was used as input. (B) The 7,210 genes that KDM1A binds (ChIP-Seq) in differentiated keratinocytes were overlapped with the 1,156 genes bound by ZNF750 as well as being increased upon ZNF750 depletion (Poly IC treated). (C) The top 4 WikiPathway terms of the 433 overlapped genes from (B) using Enrichr. (D) Chromatin immunoprecipitations (ChIP) was performed on CTLi (blue) and ZNF750i (red) differentiated keratinocytes (treated with Poly IC for 4 hours) using a KDM1A antibody. ChIP was also performed using IGG in CTLi (black) and ZNF750i (green) cells as a specificity control. Genomic regions for TLR3, IFIH1, and DDX58 are represented as distance away from the transcriptional start site (TSS). (E) Western blot of KDM1A protein expression in CTLi and ZNF750i differentiated keratinocytes treated with Poly IC. Two distinct siRNAs (ZNF750i-A and ZNF750i-B) targeting ZNF750 were used. (F) ChIP was performed on CTLi (blue) and ZNF750i (red) differentiated keratinocytes (treated with Poly IC for 4 hours) using a H3K4me2 antibody. (G) Differentiated keratinocytes were treated with control (CTLi), KDM1A (KDM1Ai), or KDM1A and TLR3 (KDM1Ai+TLR3i) siRNAs and the remaining KDM1A mRNA were measured by RT-QPCR. (H) RT-QPCR of the inflammatory gene expression of differentiated keratinocytes after treatment with Poly IC. (I) Measurement of IFN-β secreted into the culture medium in CTLi, KDM1Ai, and KDM1Ai+TLR3i differentiated cells. Cells were stimulated with Poly IC for 4 hours. (J) Control (Kdm1a^{wt/fl or wt/wt} K14-Cre-ER^tam) and Kdm1a^−/− (Kdm1a^fl/fl K14-Cre-ER^tam) mice were treated with UVB radiation for 2 consecutive days. Mice were sacrificed 6 days after the last UVB exposure. Staining of dorsal skin with antibodies against CD11b and Keratin 5 (K5). White bracket denote areas of neutrophil accumulation in the stratum corneum (Munro’s microabscess). Scale bar=150um. (K) Flow cytometry analysis of the percent of cells in the dorsal skin that are neutrophils (Ly6G+, CD11b+), 6 days after the last UVB treatment. (L) Percent of cells in the dorsal skin that are inflammatory monocytes (CD45+, CD11b+, Ly6C+, Ly6G−). (M) Weight of the skin draining (axillary) lymph nodes 6 days after the last UVB treatment. (N) Percent of Th17 cells (CD4+, Il17+) in the lymph nodes. (O) Scores for erythema and scaling were given by a dermatologist from pictures of skin taken 24 hours after the last imiquimod treatment. (P) Weight of the skin draining lymph nodes after imiquimod treatment. (Q) Percent of Th17 cells in the lymph nodes 24 hours after the last imiquimod treatment. (R) Hematoxylin and eosin staining of dorsal skin harvested from mice 24 hours after the last imiquimod treatment. Brackets denote Munro’s microabscess. Scale bar=150um. (S) RT-QPCR of the inflammatory gene expression of control and Kdm1a^−/− mice. Epidermis was harvested from mice 24 hours after the last imiquimod treatment. For Figure 4, N=3 with the exception of samples in K-Q and S where each dot represents data from an individual mouse. Mean values are shown with error bars=SEM. *p <0.05, ** p <0.01, ***p <0.001, ****p <0.0001 (1 way ANOVA followed by Tukey’s multiple comparison for comparison of 3 samples and t-test for 2 samples). Please also see Figure S4.