Keratinocyte-derived small extracellular vesicles supply antigens for CD1a-resticted T cells and promote their type 2 bias in the context of filaggrin insufficiency

Abstract

Introduction: Exosome-enriched small extracellular vesicles (sEVs) are nanosized organelles known to participate in long distance communication between cells, including in the skin. Atopic dermatitis (AD) is a chronic inflammatory skin disease for which filaggrin (FLG) gene mutations are the strongest genetic risk factor. Filaggrin insufficiency affects multiple cellular function, but it is unclear if sEV-mediated cellular communication originating from the affected keratinocytes is also altered, and if this influences peptide and lipid antigen presentation to T cells in the skin.

Methods: Available mRNA and protein expression datasets from filaggrin-insufficient keratinocytes (shFLG), organotypic models and AD skin were used for gene ontology analysis with FunRich tool. sEVs secreted by shFLG and control shC cells were isolated from conditioned media by differential centrifugation. Mass spectrometry was carried out for lipidomic and proteomic profiling of the cells and sEVs. T cell responses to protein, peptide, CD1a lipid antigens, as well as phospholipase A2-digested or intact sEVs were measured by ELISpot and ELISA.

Results: Data analysis revealed extensive remodeling of the sEV compartment in filaggrin insufficient keratinocytes, 3D models and the AD skin. Lipidomic profiles of shFLGsEV showed a reduction in the long chain (LCFAs) and polyunsaturated fatty acids (PUFAs; permissive CD1a ligands) and increased content of the bulky headgroup sphingolipids (non-permissive ligands). This resulted in a reduction of CD1a-mediated interferon-γ T cell responses to the lipids liberated from shFLG-generated sEVs in comparison to those induced by sEVs from control cells, and an increase in interleukin 13 secretion. The altered sEV lipidome reflected a generalized alteration in the cellular lipidome in filaggrin-insufficient cells and the skin of AD patients, resulting from a downregulation of key enzymes implicated in fatty acid elongation and desaturation, i.e., enzymes of the ACSL, ELOVL and FADS family.

Discussion: We determined that sEVs constitute a source of antigens suitable for CD1a-mediated presentation to T cells. Lipids enclosed within the sEVs secreted on the background of filaggrin insufficiency contribute to allergic inflammation by reducing type 1 responses and inducing a type 2 bias from CD1a-restricted T cells, thus likely perpetuating allergic inflammation in the skin.

Keywords: CD1a; T cell; allergic inflammation; atopic dermatitis; exosome; filaggrin; sEV.

Figure 1

mRNA and protein expression signatures signify alterations in GO terms for proteins enriched in the exosomal/EV compartment in filaggrin-insufficient keratinocytes. (A) Volcano plot depicting mRNA expression changes in shFLG keratinocytes; n = three biological replicates; moderated t-test; FC values were log₂-transformed and p-values were log₁₀-transformed; differentially expressed genes with p <0.05 in red; symbols of selected epidermal barrier- and immune response-related genes are shown; (B–D) FunRich analysis showing differential expression of genes encoding proteins enriched within cellular compartments; (B) total, (C) upregulated and (D) downregulated in shFLG; (E–G) Gene Ontology and Reactome terms related to genes encoding proteins identified in sEVs by FunRich, differentially expressed in shFLG keratinocytes; analysis by Panther tool; enrichment in GO terms related to: (E) biological process, (F) molecular function and (G) Reactome terms; (H) Volcano plot depicting protein expression changes in shFLG keratinocyte cultures; n = four biological replicates; Benjamini-Hochberg FDR; FC values were log₂-transformed and p-values were log₁₀-transformed; differentially expressed proteins with p <0.05 in red; symbols of selected epidermal barrier-, lipid metabolism-, and immune response-related proteins are shown; (I–K) FunRich analysis showing differential expression of proteins enriched within cellular compartments; (I) total, (J) upregulated, and (K) downregulated in shFLG; (L–N) Gene Ontology and Reactome terms related to proteins identified by FunRich in sEVs, differentially expressed in shFLG keratinocytes; analysis by Panther tool; enrichment in GO terms related to (L) biological process, (M) molecular function and (N) Reactome terms; FC, fold change.

Figure 2

Changes in cellular compartments are signified by differential GO term enrichment in the epidermal organotypic models and skin of atopic dermatitis patients. (A–C) Differential expression of proteins enriched in cellular compartments of siFLG organotypic cultures by FunRich tool; (A) total, (B) upregulated, and (C) downregulated in siFLG; (D–F) Gene Ontology and Reactome terms related to the proteins identified by FunRich in sEVs, differentially expressed in siFLG organotypic cultures; analysis by Panther tool; enrichment in GO terms related to (D) biological process, (E) molecular function, and (F) Reactome terms. (G–I) FunRich analysis showing enrichment of differentially expressed genes encoding proteins within cellular compartments in AD skin; (G) total, (H) upregulated, and (I) downregulated in AD skin; (J–L) Gene Ontology and Reactome terms related to the FunRich-identified proteins enriched in sEVs, encoded by genes differentially expressed in AD skin; analysis by Panther tool; enrichment in GO terms related to (J) biological process, (K) molecular function, and (L) Reactome terms.

Figure 3

Filaggrin insufficiency alters the sEV composition of PLA2-digestible lipids. (A) A protocol for isolation of extracellular vesicles by ultracentrifugation; exosome-enriched sEVs are pelleted as 100K fraction and purified by density gradient; (B) Electron microscopy pictures of sEVs preparations; representative of n = 3; (C) Size distribution of purified sEVs by Nanoparticle Tracking Analysis (NTA); representative example shown; (D) Enrichment of exosomal markers in purified sEVs; Western blot; representative blot, n = 2; pooled fractions 1–5 are purified exosome-enriched sEV; pooled fractions 6–10 are smaller microvesicles; (E–G) Lipidomic analysis of PLA2-digestible lipid species in sEVs; (E) heatmap of the detected lipids; (F) lipid species most affected by filaggrin accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (G) boxplots showing lipid species significantly different in abundance; combined data from n = four biological replicates; unpaired t-test, FDR correction; *, p <0.05; (H, I) Fatty acid composition of sEV phospholipids differing in abundance by (H) a chain length or (I) molecular weight; dotted line shows the length and mass benchmarks for highest CD1a-dependent response; (J) Number of the more abundant UFA species in sEVs; (K) Breakdown of UFA species from (J) in shC_sEV by degree of unsaturation; (L) Number of the more abundant SFAs in sEVs; (M) Number of double bonds in the more abundant FA species in sEVs; PLA2, phospholipase A2; VIP, variable importance in projection; FA, fatty acid; UFA, unsaturated fatty acid, SFA, saturated fatty acid; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid; PC, diacylglycerophosphocholine; PCO, ether-linked glycerophosphocholine.

Figure 4

shFLG_sEV demonstrate a reduced capacity to stimulate CD1a-specific T-cell responses. (A) IFNy responses of T cells stimulated with K562-CD1a cells pulsed with 1 µg/ml PLA2 and sEVs from 1 or 2 million keratinocytes overnight measured by ELISpot assay; means ± SEM shown; data normalized to control = 100%; n = seven donors; one-way ANOVA with Šídák’s multiple comparisons test; (B) Extracted Ion Chromatograms (EICs) showing sEV lipid profile before and after digestion with 1 µg/ml PLA2 for 1 h (n = 4; representative data shown); (C–E) Lipidomic analysis of glycerophosphocholine-related products after sEV digestion; (C) heatmap of detected lipids; (D) boxplots showing lipid species significantly different in abundance; data from n = four biological replicates, unpaired t-test, FDR correction; (E) lipid species most affected by filaggrin insufficiency accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (F–H) IFNy responses from (F) ex vivo T cells stimulated with K562-CD1a cells pulsed with 10 µM of lipids overnight; n = six donors; and (G) T cells cultured for 13 days following ELISpot, n = four donors; means from two technical replicates for each individual donor, normalized to the control=100% are shown; (H) comparison of responses between ex vivo and cultured T cells from n = four donors represented both in (F, G); one-way ANOVA with Šídák’s multiple comparisons test. PLA2, phospholipase A2; VIP, variable importance in projection. PC, diacylglycerophosphocholine; Lyso-PC, monoacylglycerophosphocholine; Lyso-PCO, monoalkylglycerophosphocholine; C14:0, tetradecanoic acid; C22:6, docosahexaenoic acid; *, p <0.05; **, p <0.01; ***, p <0.001; ****, p <0.0001.

Figure 5

Non-permissive CD1a lipid antigens are enriched in sEVs secreted by filaggrin-insufficient keratinocytes. (A–C) Lipidomic analysis of PLA2-non-digestible lipid species in sEVs; (A) heatmap of all detected lipids; (B) lipid species most affected by filaggrin insufficiency accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (C) boxplots showing lipid species significantly different in abundance; n = four biological replicates, unpaired t-test, FDR correction; (D–F) Lipidomic analysis of PLA2-non-digestible lipid species in sEVs digested with 1 µg/ml PLA2 for 1 h; (D) heatmap of all detected lipids; (E) lipid species most affected by filaggrin insufficiency accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (F) boxplots showing lipid species significantly different in abundance; n = four biological replicates; unpaired t-test, FDR correction; (G) Relative amounts of permissive and non-permissive species in PLA2-digested sEVs; (H) IFNy responses by T cells stimulated with K562-CD1a cells pulsed overnight with sEVs from 1 or 2 million keratinocytes digested with 1 µg/ml PLA2 for 1 h; n = six donors; data normalized to control = 100%; (I) IL-13 secretion into culture supernatants from (H) measured by ELISA; n = six donors; means ± SEM are shown; one-way ANOVA with Šídák’s multiple comparisons test; *, p <0.05; **, p <0.01; PLA2, phospholipase A2; VIP, variable importance in projection; SMd, sphingomyelin; Cerd, ceramide.

Figure 6

A filaggrin insufficiency background alters the landscape of the PLA2-digestible lipidome in keratinocytes. (A–C) Lipidomic analysis of PLA2-digestible lipid species in shC and shFLG keratinocytes; (A) heatmap of all detected lipid species; (B) lipid species most affected by filaggrin insufficiency accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (C) boxplots showing lipid species significantly different in abundance; n = four biological replicates; unpaired t-test; FDR correction; (D, E) Fatty acid composition of differentially abundant phospholipids in keratinocytes by (D) chain length and (E) molecular weight of fatty acids; dotted line shows the size and mass benchmarks for optimal CD1a-mediated responses; (F) Number of the more abundant lipid species in keratinocytes; (G, H) UFAs represented in (F) found in either shC (G) or shFLG (H) keratinocytes by degree of unsaturation; (I) Number of double bonds in the more abundant UFA species in keratinocytes; PLA2, phospholipase A2; PC, diacylglycerophosphocholine; PCO, ether-linked glycerophosphocholine; PEO, ether-linked glycerophosphoethanoloamine; SFA, saturated fatty acid; UFA, unsaturated fatty acid; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid. *, p<0.05; **, p<0.01.

Figure 7

A filaggrin insufficiency background alters the landscape of non-PLA2-digestible lipidome in keratinocytes. (A–C) Lipidomic analysis of PLA2-nondigestible lipid species in keratinocytes; (A) heatmap of all detected lipid species; (B) lipid species most affected by filaggrin insufficiency accordingly to the PLS-DA analysis; the variance importance for the projection values (VIP) were used to sort lipids accordingly to their contribution to PLS-DA model; (C) boxplots showing lipid species significantly different in abundance; data for n=4 biological replicates, unpaired t-test, FDR correction; (D–F) Differentially abundant sphingomyelin species represented by combined sphingosine and fatty acid chain length; (D), molecular weight (E) and a number of double bonds (F); SMd, sphingomyelin; Cerd, ceramide; LacCerd, lactosylceramide; GlcCerd, glucosylceramide. *, p<0.05; **, p<0.01.

Figure 8

Filaggrin insufficiency in keratinocytes impacts enzymatic pathways of synthesis of lipids acting as substrates for generation of CD1a-dependent lipid neoantigens. (A) ACSL3 protein expression by cultured keratinocytes; n=4 biological replicates, unpaired t-test, Benjamini-Hochberg FDR; (B-I) Analysis of the data from Cole et al. (42) , showing the expression of (B) ACSL1; (C) ACSL3; (D) ACSL5; (E) ACSBG; (F) ELOVL1; (G) ELOVL3; (H) ELOVL4; (I) ELOVL5 mRNA in AD skin; n=26 AD and n=10 healthy subjects; Benjamini-Hochberg FDR; (J) FADS1 mRNA expression in cultured keratinocytes; n=3 biological replicates, t-test; (K–M) Analysis of the data from Cole et al. (42) , showing the expression of (K) FADS1; (L) FADS2; (M) FADS6 mRNA in AD skin; n=26 AD and n=10 healthy subjects; all data are normalized to control (shC or H=1); (N) Summary of the changes in the lipid metabolic pathways identified in this study; simplified diagram; C, number of carbon atoms; n, number of double bonds; *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001. ACSL3, long-chain-fatty-acid–CoA ligase 3; ACSL1, long-chain-fatty-acid–CoA ligase 1; ACSL5, long-chain-fatty-acid–CoA ligase 5; ACSBG1, long-chain-fatty-acid–CoA ligase ACSBG1; ELOVL1, ELOVL fatty acid elongase 1; ELOVL3, ELOVL fatty acid elongase 3; ELOVL4, ELOVL fatty acid elongase 4; FADS1, fatty acid desaturase 1; FADS2, fatty acid desaturase 2; FADS6, fatty acid desaturase 6.