Immunopathogenesis of hidradenitis suppurativa and response to anti-TNF-α therapy

Abstract

Hidradenitis suppurativa (HS) is a highly prevalent, morbid inflammatory skin disease with limited treatment options. The major cell types and inflammatory pathways in skin of patients with HS are poorly understood, and which patients will respond to TNF-α blockade is currently unknown. We discovered that clinically and histologically healthy appearing skin (i.e., nonlesional skin) is dysfunctional in patients with HS with a relative loss of immune regulatory pathways. HS skin lesions were characterized by quantitative and qualitative dysfunction of type 2 conventional dendritic cells, relatively reduced regulatory T cells, an influx of memory B cells, and a plasma cell/plasmablast infiltrate predominantly in end-stage fibrotic skin. At the molecular level, there was a relative bias toward the IL-1 pathway and type 1 T cell responses when compared with both healthy skin and psoriatic patient skin. Anti-TNF-α therapy markedly attenuated B cell activation with minimal effect on other inflammatory pathways. Finally, we identified an immune activation signature in skin before anti-TNF-α treatment that correlated with subsequent lack of response to this modality. Our results reveal the fundamental immunopathogenesis of HS and provide a molecular foundation for future studies focused on stratifying patients based on likelihood of clinical response to TNF-α blockade.

Keywords: Adaptive immunity; Dermatology; Immunology; Innate immunity; Skin.

Figure 1. Elucidation of the dominant inflammatory pathways in HS skin.

(A) Principal components analysis (PCA) of whole-tissue RNA-Seq data from lesional HS skin (n = 19), nonlesional HS skin (n = 13), and site-matched healthy control skin (n = 16). All samples were taken before the initiation of anti–TNF-α therapy. (B) The top 20 enriched (FDR < 0.05, Fisher exact with Benjamini-Hochberg correction) PANTHER Gene Ontology pathways identified from genes significantly (adjusted P < 0.05, Wald’s test) increased in pretreatment lesional HS skin versus healthy control skin are depicted in red. Fold enrichment of pathways in genes significantly (adjusted P < 0.05, Wald’s test) increased in lesional psoriasis skin (n = 8) versus healthy control skin (n = 9) is depicted in blue. (C) Heatmap depicting the Gene Set Variation Analysis (GSVA) enrichment scores of the top 50 significantly different (adjusted P < 0.05, empirical Bayes test with Benjamini-Hochberg correction) Gene Ontology pathways in whole-tissue RNA-Seq data of pretreatment HS lesional skin versus healthy control skin. Each column depicts an individual patient. Average pathway enrichment scores in HS skin, normal skin, and psoriatic skin, is depicted (left); pathways significantly different (adjusted P < 0.05) comparing HS skin and psoriatic skin are indicated. (D) Ingenuity Pathway Analysis (IPA) of upstream regulators significantly (P < 0.05) different in lesional HS skin versus healthy controls. (E) xCell Scores indicating predicted enrichment of different cell populations in whole-tissue RNA-Seq data from lesional HS (L) and healthy (H) skin. Each dot represents an individual patient. All figure error bars show mean ± SEM. (**P < 0.01, ****P < 0.0001, Mann-Whitney U test.)

Figure 2. Nonlesional skin in HS has defects in immune regulatory pathways.

(A) Representative H&E staining of lesional skin, nonlesional skin, and site-matched healthy control skin. (B) GSVA enrichment scores of the significantly different (adjusted P < 0.05, absolute log fold change > 0.3, empirical Bayes test with Benjamini-Hochberg correction) Gene Ontology immune-related pathways in whole-tissue RNA-Seq data of healthy control skin (n = 16) compared with pretreatment nonlesional HS skin (n = 13). Each column depicts an individual patient. (C) xCell Scores indicating predicted enrichment of different cell populations in whole-tissue RNA-Seq data in nonlesional HS (NL) and healthy (H) skin. Each dot represents an individual patient (Mann-Whitney U test) (D) Normalized counts for selected transcripts in whole-tissue RNA-Seq comparing healthy control skin with pretreatment nonlesional HS skin (adjusted P, Wald’s test). (E) GSVA enrichment scores of the union of Gene Ontology pathways significantly increased (adjusted P < 0.05, empirical Bayes test with Benjamini-Hochberg correction) in pretreatment lesional HS skin (n = 19) versus pretreatment nonlesional HS skin and pretreatment nonlesional HS skin versus healthy control skin. Each column depicts an individual patient.

Figure 3. Tissue-infiltrating myeloid cells are dysfunctional in HS skin.

(A) Uniform manifold approximation and projection (UMAP) plots of scRNA-Seq data of myeloid cells sorted from 2 end-stage HS skin donors versus 2 healthy controls. Plots are equally sampled to 2943 cells per sample. Population identification was manually assigned. (B) Percentages of myeloid cell subsets identified in scRNA-Seq data depicted in A. (C) UMAP plots of myeloid cells immunophenotyped by CyTOF. Cells were pregated on live, singlet, CD45⁺CD3^–CD19^– events and represent 20,442 cells from 7 healthy donors, 38,820 cells from 5 active inflammatory HS lesions, and 8492 cells from 5 end-stage HS surgical resections. (D) Percentages of myeloid subsets identified via manual gating in 14 healthy donors, 12 biopsies from active HS lesions, and 18 end-stage HS surgically resected specimens. (*P < 0.05, **P < 0.01, ****P < 0.001, 1-way ANOVA.)

Figure 4. IL-1 signaling is increased in myeloid cells of HS skin.

(A) Dot plot depicting expression of IL-1 family genes in scRNA-Seq of the top 4 most abundant myeloid clusters. Data averaged from 2 HS skin samples are depicted in red and from 2 healthy skin samples in blue. Dot size indicates percentage expression within clusters, while color intensity indicates degree of expression. Boxed dots indicated an adjusted P < 0.01 for the comparison between healthy and HS skin (Wilcoxon’s rank sum test). (B) UMAP plots showing intensity of expression of IL-1B and 3 genes downstream of IL-1 signaling (CXCL8, PTGS2, and NFKBIA). Data are combined for 2 HS donors and for 2 healthy controls, and plots are equally sampled to 5886 cells per donor type. Violin plots indicating expression level over the entire population are inset in the lower right for each gene (Wilcoxon’s rank sum test).

Figure 5. HS progression is associated with a transition from skin-infiltrating memory B cells to plasma cells.

(A) Total normalized counts of immunoglobulin genes in whole-tissue RNA-Seq data from lesional (L, n = 19) and nonlesional (NL, n = 13) HS patients before anti–TNF-α therapy and healthy (H, n = 16) controls. (**P < 0.01, 1-way ANOVA.) (B) Normalized counts for selected B cell chemokine, chemokine receptor, and B cell survival factors in whole-tissue RNA-Seq before anti–TNF-α therapy (*P < 0.05; ***P < 0.005; ****P < 0.001, Wald’s test, DESeq). (C) Representative CyTOF plot of CD19 versus CD3 expression in cells from healthy donor skin, biopsied active inflammatory HS lesions, or end-stage HS skin (n = 17). Cells are pregated on live, singlet, CD45⁺ events. (D–G) Percentages of total B cells (D), naive B cells (E), plasma cells and plasmablasts (F), and memory B cells (G) among total live, singlet, CD45⁺ events in CyTOF data sampled from healthy donors (n = 14), active HS lesion biopsies (n = 10), and end-stage HS surgical resections (n = 18). (*P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, 1-way ANOVA.)

Figure 6. Type 1 T cell responses are dominant in HS skin.

(A) Flow cytometric quantification of Treg percentages within the CD4⁺ T cell compartment within lesional (n = 14) and nonlesional skin (n = 8) of patients with HS before anti–TNF-α treatment compared with healthy controls (n = 7) and lesional skin from patients with psoriasis (n = 7). (B) Flow cytometric quantification of ratios of CD4⁺ Th1 cells to Tregs of patients described in A. (C) Flow cytometric quantification of IFN-γ production within the CD8⁺ T cell compartment of patients described in A. (D) Flow cytometric quantification of IFN-γ production within the CD4⁺ Tcon compartment of patients described in A. (E) Flow cytometric quantification of IL-17A production within the CD8⁺ T cell compartment of patients described in A. (F) Flow cytometric quantification of IL-17A production within the CD4⁺ Tcon compartment of patients described in A. (G) Flow cytometric quantification of TNF-α production within the CD8⁺ T cell compartment of patients described in A. (H) Flow cytometric analysis of TNF-α production within the CD4⁺ Tcon compartment of patients described in A. (I) UMAP plots of CD3⁺ T cells immunophenotyped by CyTOF illustrating intensity of T-bet expression. Cells were pregated on live, singlet, CD45⁺CD3⁺ events and represent 34,432 cells from 7 healthy donors, 75,400 cells from 5 active inflammatory HS lesions, and 21,557 cells from 5 end-stage HS surgical resections. Percentage of T-bet⁺ events within CD4⁺ and CD8⁺ T cells. Data were manually gated for 14 healthy donors, 12 active inflammatory HS lesions, and 18 end-stage surgical resections. (J) UMAP plots of scRNA-Seq data of all cells obtained from a myeloid-enriching sort from 2 end-stage HS skin donors versus 2 healthy controls. Intensity of expression of IFN-γ, IL-17A, and TNF-α in all populations is depicted in each plot. (*P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001, 1-way ANOVA.)

Figure 7. Anti–TNF-α therapy preferentially attenuates the B cell response.

(A) PCA of RNA-Seq data from lesional skin taken from HS patients before initiation of anti–TNF-α therapy (PreTx, n = 19) and on anti–TNF-α therapy (On Tx, n = 16), as well as site-matched healthy controls (n = 16). (B) The top 20 enriched (FDR < 0.05, Fisher exact with Benjamini-Hochberg correction) PANTHER Gene Ontology Pathways identified from genes significantly (adjusted P < 0.05, Wald’s test) increased in pre–anti–TNF-α lesional HS skin compared with lesional skin of patients on anti–TNF-α treatment. B cell–related pathways are highlighted in red. (C) xCell scores comparing predicted enrichment of B cell subsets in whole-tissue RNA-Seq data of pretreatment patients compared with patients on anti–TNF-α therapy. Each dot represents an individual patient. (*P < 0.05, **P < 0.01, Mann-Whitney U test.) (D) Normalized counts of CXCL13 transcripts in whole-tissue RNA-Seq of lesional skin before anti–TNF-α therapy versus on anti–TNF-α therapy (*P < 0.05, Wald’s test). (E) Total counts of immunoglobulin genes in whole-tissue RNA-Seq from lesional HS patients before anti–TNF-α therapy versus patients on treatment (*P < 0.05, unpaired t test).

Figure 8. Anti–TNF-α therapy has minimal effects on IL-1 and type 1 T cell inflammation.

(A) Heatmap depicting the GSVA enrichment scores of top 40 Gene Ontology pathways in whole-tissue RNA-Seq data that were significantly (adjusted P < 0.05, empirical Bayes test with Benjamini-Hochberg correction) increased or decreased when comparing pretreatment HS lesional skin (n = 19) with healthy controls (n = 16) and were also significantly (adjusted P < 0.05) increased or decreased in comparison of on-treatment HS lesional skin (n = 16) with healthy controls. Each column depicts an individual patient. (B) Flow cytometric analysis of indicated cytokines within the CD4⁺ Tcon or CD8⁺ compartments comparing lesional skin of patients before anti–TNF-α treatment (PreTx, n = 14) versus patients on anti–TNF-α treatment (On Tx, n = 9). (ns, nonsignificant, unpaired t test.) (C) Percentage of Ki67⁺ of CD19⁺ B cells and HLA-DR median metal intensity following 3 days of culture with either isotype control antibody or increasing doses of anti–TNF-α antibody. Each dot represents a culture well for a single donor sample. Ki67 data are representative of 3 separate donors. (*P < 0.05, **P < 0.01, ***P < 0.005, 1-way ANOVA.)

Figure 9. Response to anti–TNF-α therapy correlates with lower B cell and leukocyte chemotaxis immune signatures.

(A) PCA of RNA-Seq data from lesional skin of patients with HS before initiation of anti–TNF-α therapy comparing patients who later responded to therapy (n = 7) and those who did not (n = 7). (B) The top 20 enriched (FDR < 0.05, Fisher exact with Benjamini-Hochberg correction) PANTHER Gene Ontology pathways identified from genes significantly (adjusted P < 0.05, Wald’s test) increased (top) or decreased (bottom) in lesional HS skin of nonresponders (NR) versus responders (R) to anti–TNF-α therapy. (C) Normalized counts for selected transcripts in whole-tissue RNA-Seq comparing responders to anti–TNF-α therapy with nonresponders (Wald’s test). (D) Normalized counts for selected B cell–related transcripts (left, Wald’s test) and total counts of immunoglobulin genes (right, unpaired t test) and comparing responders to anti–TNF-α therapy with nonresponders.