Cutaneous kinase activity correlates with treatment outcomes following PI3K delta inhibition in mice with experimental pemphigoid diseases

Abstract

Chronic blistering at the skin and/or mucous membranes, accompanied by a varying degree of inflammation, is the clinical hallmark of pemphigoid diseases that impose a major medical burden. Pemphigoid diseases are caused by autoantibodies targeting structural proteins of the epithelial basement membrane. One major pathogenic pathway of skin blistering and inflammation is activation of myeloid cells following Fc gamma receptor-dependent binding to the skin-bound immune complexes. This process requires activation of specific kinases, such as PI3Kδ, which have emerged as potential targets for the treatment of pemphigoid diseases. Yet, it is unknown if global cutaneous kinase activity present in lesional pemphigoid disease correlates with therapeutic effects following treatment with a given target-selective kinase inhibitor. To address this, we here first determined the kinase activity in three different mouse models of pemphigoid diseases: Antibody transfer-induced mucous membrane pemphigoid (MMP), antibody transfer-induced epidermolysis bullosa acquisita (EBA) and immunization-induced EBA. Interestingly, the kinome signatures were different among the three models. More specifically, PI3Kδ was within the kinome activation network of antibody transfer-induced MMP and immunization-induced EBA, but not in antibody transfer-induced EBA. Next, the therapeutic impact of the PI3Kδ-selective inhibitor parsaclisib was evaluated in the three model systems. In line with the kinome signatures, parsaclisib had therapeutic effects in antibody transfer-induced MMP and immunization-induced EBA, but not in autoantibody-induced EBA. In conclusion, kinase activation signatures of inflamed skin, herein exemplified by pemphigoid diseases, correlate with the therapeutic outcomes following kinase inhibition, demonstrated here by the PI3Kδ inhibitor parsaclisib.

Keywords: animal model; autoimmunity; epidermolysis bullosa acquisita; mucous membrane pemphigoid; neutrophils; pemphigoid; phosphatidylinositol 3-kinase (P13k); signal transduction.

Figure 1

Different kinase pathways are activated in experimental models of pemphigoid diseases (PD). (A) For induction of immunization-induced epidermolysis bullosa acquisita (EBA), B6.s mice were immunized with 120 μg of vWFA2 domain of mCOL7 emulsified (1:1) in Titermax™ and lesional skin was taken for kinome analysis in comparison to skin of mice treated with Titermax™ only. Hence, the data shown indicates differences between the samples. (B) To induce experimental mucous membrane pemphigoid (MMP), C57BL/6J mice were injected every other day with 6 mg of rabbit anti-mLAMα3 IgG or NR IgG for a total of 12 days. (C) Antibody transfer-induced EBA was induced by six injections of 3 mg rabbit anti-mCOL7 IgG or (normal rabbit) NR-IgG for a total of 12 days. In all three models, treatment regimens and STRING database analysis of kinases that are modulated are depicted here for analysis of perilesional skin and (MMP model only) oral mucosa. STING interaction modules are shown at the medium confidence level. Thickness of the lines indicates the strength of the predicted interaction. Colors of kinases are randomly selected. Differently activated kinases that did not have any connections to other kinases were excluded from the networks. n=3/model and group. (D) Venn diagram of overlapping kinase profiles in the respective PD models.

Figure 2

Parsaclisib impairs immune complex-induced neutrophil activation in vitro. Human polymorphonuclear leukocytes (PMNs) were isolated from healthy blood donors and activated by immobilized immune complexes (IC) consisting of recombinant human COL7 (hCOL7E-F) and anti-hCOL7E-F IgG1 in the presence or absence of parsaclisib at an 8-fold concentration range. Relative reactive oxygen species (ROS) release was detected by a luminescence-based assay. (A) Representative example of one donor showing the relative luminescence after IC-stimulation over the time. (B) Area under the curve (AUC, cumulative values) of luminescence, n=12. (C) PMNs were isolated from murine femurs and activated using immobilized ICs consisting of recombinant murine COL7 (mCOL7^C) and anti-mCOL7^C-IgG in the presence or absence of parsaclisib, n=4. (D) ROS were generated enzymatically by myeloperoxidase, catalase, and glucose oxidase, n=4. The ROS scavenger N-acetylcysteine (NAC) was used as assay control. (E) Chemotaxis of freshly isolated PMNs was induced by IL-8 in the presence of parsaclisib using a Boyden chamber assay. The attracted cell number during a time period of 60 minutes is shown as AUC. (F) Human PMNs were stimulated with immobilized ICs in the presence/absence of parsaclisib. To exclude toxicity, the amount of propidium iodide (PI)- and Annexin V-positive cells after IC stimulation was identified. (G) Cryosections of human skin were incubated with IgG isolated from EBA patients, followed by the addition of PMNs, isolated from healthy donors (cryosection assay). Split formation along the dermal epidermal junction (DEJ, dotted line) was analyzed as percentage of the whole DEJ. Asterisks indicate split formation. (A-F) Data was normalized to positive control (either IC- or IL-8-stimulated PMNs). Solvent control was always added to positive and negative controls. (B-G) Data are shown as Tukey’s box-and-whisker plots. ANOVA on ranks (Kruskal-Wallis) was applied followed by a Dunn´s multiple comparison test, (B) n=12, (C) n=2-4, (D, F) n=4, (E) n=5, (G) n=6 (for detailed information see attached raw data table), *p<0.05, **p<0.01.

Figure 3

Parsaclisib improves the clinical outcome in immunization-induced EBA. B6.s mice were immunized with 120 μg of mCOL7^vWFA2 emulsified in Titermax™. If 2% or more of the body surface area was covered with lesions in individual mice, these animals were treated with different concentrations of parsaclisib or methylprednisolone (MP), n=9/group. (A) Relative magnitude of affected body surface area in relation to the time of allocation over the time and representative clinical images of the mice 4 weeks after allocation. Data are shown as mean ± SD. The representative clinical images for the indicated treatments were obtained at the end of the 4-week treatment period. (B) To visualize IgG binding to the dermal-epidermal junction, ear skin was stained for anti-mouse IgG (green) and nuclei (DAPI, blue). Overall, no difference among the groups was noted. Representative direct IF pictures and mean fluorescent intensity of IgG binding. (C) Representative H&E-stained ear skin at the end of the experiment. The cumulative histological score index (amount of skin infiltration, epidermal thickening and split formation at the DEJ) was analyzed in lesional back skin. (B-C), data are shown as Tukey’s box-and-whisker plots. ANOVA on ranks (Kruskal-Wallis) was applied followed by a Dunn´s multiple comparison test. (D) PamGene kinome analysis of lesional skin treated with solvent vs. 3 mg/ml parsaclisib (n=3/group) was performed. STRING database analysis of kinases that are modulated by parsaclisib; contrasting kinase activation in lesional skin from solvent- to parsaclisib-treated mice. STING interaction modules are shown at the medium confidence level. Thickness of the lines indicates the strength of the predicted interaction. Colors of kinases are randomly selected. Differently activated kinases that did not have any connections to other kinases were excluded from the networks. *p<0.05, **p<0.01, scale bar=100 µm.

Figure 4

Parsaclisib improves the clinical outcome in antibody transfer-induced MMP. C57BL/6J mice were injected every other day with 6 mg of rabbit anti-mLAMα3 IgG (n=15/group) or normal rabbit (NR) IgG (n=6) for a total of 12 days. Administration of MP, solvent or parsaclisib (0.3, 1, or 3 mg/kg) was initiated on day 0 of the experiment and continued until day 11. (A) Percentage of affected body surface area was determined at days, 0, 4, 8 and 12. Representative clinical images of the mice at day 12. Data are shown as mean ± SD (B) High-resolution endoscopy of mouth oral cavity (i.e., pharyngeal mucosa, tongue, and right/left buccal) was conducted to determine the extent of oral lesions at day 12. Data are shown as scatter blot, including mean ± SEM. Representative pictures of oral cavity. Arrows indicate mucosal lesions, asterisks indicate teeth. PamGene kinome analysis of lesional skin treated with solvent vs. 3 mg/ml parsaclisib (n=3). (C) The mean fluorescence intensity of IgG binding to the dermal-epidermal junction was analyzed in ears or cheek (oral mucosa), respectively at the final end point. (D) The cumulative histological score index (amount of skin infiltration, epidermal thickening and split formation at the DEJ) of ears, or mucosa was analyzed at day 12. (C, D) Data are shown as Tukey’s box-and-whisker plots. ANOVA on ranks (Kruskal-Wallis) was applied followed by a Dunn´s multiple comparison test. String database analysis of kinases that are modulated by parsaclisib in (E) perilesional skin and (F) mucosa; contrasting kinase activation from solvent- to parsaclisib-treated mice. STING interaction modules are shown at the medium confidence level. Thickness of the lines indicates the strength of the predicted interaction. Colors of kinases are randomly selected. Differently activated kinases that did not have any connections to other kinases were excluded from the networks. *p<0.05, ***p<0.001, ****p>0.0001.

Figure 5

Parsaclisib does not affect antibody transfer-induced EBA. C57BL/6J mice were injected every other day with rabbit anti-mCOL^vWFA2 IgG (n=17/group) or normal rabbit (NR) IgG (n=5) for a total of 12 days. Administration of methylprednisolone (MP), solvent or parsaclisib (0.3, 1, or 3 mg/kg) was initiated on day 0 of the experiment and continued until day 11. (A) Percentage of affected body surface area was measured at days, 0, 4, 8 and 12. Representative clinical images of the mice at day 12. Data are shown as mean ± SD. (B) To visualize IgG binding to the dermal-epidermal junction, ear skin was stained for anti-mouse IgG (green) and nuclei (DAPI, blue). Representative direct IF pictures and mean fluorescent intensity of IgG binding. (C) Representative H&E-stained ear skin at the end of the experiment. The cumulative histological score index (amount of skin infiltration, epidermal thickening and split formation at the DEJ) was analyzed in lesional back skin. (B, C), data are shown as Tukey’s box-and-whisker plots. ANOVA on ranks (Kruskal-Wallis) was applied followed by a Dunn´s multiple comparison test. (D) PamGene kinome analysis of lesional skin treated with solvent or 3 mg/ml parsaclisib (n=3), respectively. String database analysis of kinases that are modulated by parsaclisib; contrasting kinase activation in lesional skin from solvent- to parsaclisib-treated mice. STRING interaction modules are shown at the medium confidence level. Thickness of the lines indicates the strength of the predicted interaction. Colors of kinases are randomly selected. Differently activated kinases that did not have any connections to other kinases were excluded from the networks, Scale bar=100 µm.