Caspase-activation powers anti-Desmoglein 3-induced acantholysis in human epidermis

TL;DR

This study shows that caspase activation works with anti-Dsg3 antibodies to cause cell separation in a skin disease called pemphigus vulgaris.

Contribution

The study reveals a dual mechanism involving caspase activation and Dsg3 redistribution in pemphigus vulgaris acantholysis.

Findings

Anti-Dsg3 antibodies cause acantholysis by redistributing Dsg3 to intracellular compartments without caspase activation.

FasL-induced caspase activation synergistically enhances cell adhesion loss by cleaving Dsg3.

The findings suggest a dual mechanism contributing to disease heterogeneity in pemphigus vulgaris.

Abstract

Pemphigus vulgaris (PV) is a life-threatening autoimmune blistering disease caused by circulating autoantibodies against desmoglein (Dsg) 1 and Dsg 3. Whether acantholysis in PV results exclusively from antibody binding to Dsgs or involves additional factors remains controversial. Given that Fas-Ligand (FasL), an activator of apoptotic caspase-8, is increased in the serum and the skin of patients with PV, we investigated the role of caspases in anti-Dsg3-mediated acantholysis using both ex vivo and in vitro models. Our results demonstrated that anti-Dsg3 antibodies induced acantholysis ex vivo in the absence of caspase activation, primarily through the redistribution of Dsg3 to intracellular compartments. FasL-induced caspase activation led to a synergistic amplification of anti-Dsg3-mediated loss of cell adhesion by promoting Dsg3 cleavage. This dual mechanism provides new insights into the disease heterogeneity of PV.

Figures

A). Thus, caspase activation in PV may occur without apoptosis. We next examined: 1. how pathogenic anti-Dsg3 antibodies affected the anchoring of Dsg3 on keratinocyte membranes, and 2. whether FasL-induced caspase activation amplifies these effects. Healthy skin explants were injected with an anti-Dsg3 antibody (AK23), control IgG and/ or FasL into the upper dermis (Fig. [1B](#Fig1)). AK23 is known to induce acantholysis in mice [[20](#CR20)] and loss of cohesion in vitro in keratinocyte monolayers following overnight stimulation [[21](#CR21)]. AK23 alone caused only a subtle loss of cell coh

A, B) and ex vivo (Supplementary Fig. 2C), and widened the interdesmosomal space in vitro (Supplementary Fig. 2D, E), but not ex vivo (Supplementary Fig. 2F). FasL did not provide any additive effect in these structural changes (Supplementary Fig. 2A, B, D, E).

A), indicating that this process is caspase-independent. Dsg3 is known to degrade via endosomal and lysosomal pathways [[25](#CR25), [26](#CR26)]. However, neither the proteasome inhibitor bortezomib (Supplementary Fig. 3B), nor chloroquine (a lysosome inhibitor) (Supplementary Fig. 3C) alone or combined (Supplementary Fig. 3D), prevented the depletion of Dsg3. Altogether, these findings demonstrated that anti-Dsg3 treatment might support the redistribution of endogenous Dsg3 to the TIF or other cell compartments (e.g., lipid rafts [[27](#CR27)]).Fig. 2AK23 disrupts the endogenous distribution

A) and without inducing caspase-8 activation (Supplementary Fig. 4B) or sensitizing cells to apoptosis (Supplementary Fig. 4C). As FasL is released by lesional keratinocytes in PV [[19](#CR19)], it is present in the serum of PV-patients [[11](#CR11)] and enhances antibody-induced acantholysis ex vivo, we investigated its effect on desmosomal detachment. FasL at sublethal levels (Supplementary Fig. 4D) alone had no impact on keratinocyte cohesion, but markedly increased antibody-mediated loss of cell cohesion in a caspase-dependent manner (Fig. [3B](#Fig3)). To specifically assess the role of c

Full text

Introduction

Pemphigus vulgaris (PV) is a chronic, life-threatening blistering disease caused by Immunoglobulin G (IgG) autoantibodies against the desmosomal cadherins desmoglein (Dsg) 3 and Dsg1 [1, 2], essential for cell-cell adhesion in skin and mucous membranes [3, 4]. Antibody binding induces acantholysis, leading to blisters and erosions. While direct inhibition of Dsg3 interaction is a major mechanism, antibody-induced acantholysis also involves intracellular signaling pathways (p38 MAPK, protein kinase C, or tyrosine kinase Src) [5–10] and apoptosis as evidenced by chromatin condensation, nuclear fragmentation and TUNEL positivity in keratinocytes [7, 11–14], along with an altered distribution of Dsg1 or Dsg3 on the cell surface [13, 15].

Despite the homogenous autoantibody binding in the epidermis of the entire body, blisters develop only at distinct sites and often heal despite the continued presence of antibodies, suggesting that further active signaling contributes significantly to acantholysis [16, 17]. One such pathway involves Fas-Fas ligand (FasL) interactions: FasL levels are elevated in the serum of untreated patients [11, 18], and FasL-deficient mice display reduced PV-related acantholysis [19]. FasL activates the extrinsic apoptotic caspase cascade via Fas, leading to the formation of the Death-Inducing Signaling Complex (DISC) at the intracellular death domain of Fas and activation of caspase-8 and caspase-3, which ultimately results in apoptosis.

In the present study, we investigated the mechanism by which anti-Dsg3 antibody binding leads to loss of cell cohesion in keratinocytes. Specifically, we aimed to elucidate how caspase activation contributes to PV-related acantholysis by employing both in vitro experiments using a keratinocyte cell line and ex vivo experiments using human skin samples. Our findings demonstrate that anti-Dsg3 antibody binding triggers the acantholysis mainly by depleting Dsg3 from the keratinocyte surface. FasL-mediated activation of caspase-8 amplifies this effect by cleaving Dsg3. We provide compelling evidence that cell death is an event that occurs late after the Dsg3-depletion from the keratinocyte surface, therefore not being related to desmosomal detachment.

Results

Caspase-activation enhances anti-Dsg3 antibody-mediated acantholysis and vesicle formation in proximity to desmosomes

To investigate apoptotic features related to PV, we examined human lesional PV skin. Immunohistology revealed active caspase-8 staining in the epidermis (exemplary, Fig. 1A, black arrows), but no TUNEL-positive keratinocytes (Fig. 1A), unlike toxic epidermal necrolysis (TEN), where both were present (Supplementary Fig. 1A). Thus, caspase activation in PV may occur without apoptosis. We next examined: 1. how pathogenic anti-Dsg3 antibodies affected the anchoring of Dsg3 on keratinocyte membranes, and 2. whether FasL-induced caspase activation amplifies these effects. Healthy skin explants were injected with an anti-Dsg3 antibody (AK23), control IgG and/ or FasL into the upper dermis (Fig. 1B). AK23 is known to induce acantholysis in mice [20] and loss of cohesion in vitro in keratinocyte monolayers following overnight stimulation [21]. AK23 alone caused only a subtle loss of cell cohesion in the suprabasal layers (Fig. 1C, green arrows). To explore the role of caspase activation, we utilized a sublethal concentration of FasL (Supplementary Fig. 1B). Co-injection of anti-Dsg3 antibodies and FasL led to discrete loss of cell cohesion in the suprabasal layers within 24 h (Fig. 1C, black arrows) and severe disruption of basal keratinocytes with blister formation and acantholytic cells by 48 h (Fig. 1C, black stars), whereas FasL alone did not affect epidermal cohesion (Supplementary Fig. 1C). Caspase-8 was active either in developing lesion sites or blister roofs, mirroring patient findings (Fig. 1D, red staining; Supplementary Fig. 1D). In summary, these findings indicated that anti-Dsg3 induced acantholysis was enhanced by addition of FasL and occurred together with activation of caspase-8 in acantholytic areas, independently of cell death.Fig. 1. Caspase-activation enhances anti-Dsg3 antibody-induced acantholysis and promotes vesicle formation in proximity to desmosomes.A Representative images of patient skin sections stained with H&E, TUNEL, and IHC for cleaved caspase-8. Cleaved caspase-8 positivity is visualized by pink staining. Exemplarily acantholytic keratinocytes positive for cleaved caspase-8 are highlighted by black arrows. The dotted line indicates the basal membrane. Scale bar: 100 µm. B Schematic illustration of the ex vivo skin model. Stimuli are injected intradermally into 8 mm punch biopsies of healthy skin, which are then maintained in organ culture medium prior to analysis. C Representative H&E staining of ex vivo skin punches following stimulation for the indicated time points. Arrows highlight areas of discrete loss of cell cohesion (green: 24 h; black: 48 h), black stars indicate acantholytic keratinocytes. Scale bar: 50 µm, (n = 3). D Representative staining of stimulated skin punches for indicated time points: cleaved caspase-8 (red) and AK23 (green), DAPI (blue) used to visualize the nucleus. The dotted line marks the basal membrane. Scale bar: 50 µm, (n = 3). E Quantification of vesicles adjacent to desmosomes in transmission electron microscopy (TEM) images. HaCaT monolayers were stimulated for 4 h, while ex vivo skin models were stimulated for 24 h (n = 2). F Representative TEM pictures of stimulated HaCaT monolayers preincubated with AK23 or IgG for 1 h and subsequently stimulated with FasL for 3 h. White arrows indicate intracytoplasmic vesicles in proximity to desmosomes. Scale bar: 200 nm. G Quantification of vesicles adjacent to desmosomes in TEM images. Each dot represents an individual measurement; the horizontal line indicates the mean (n = 2).

Ultrastructural transmission electron microscopy (TEM) analysis of HaCaT monolayers and ex vivo skin models revealed that anti-Dsg3 antibody treatment increased intracellular vesicle formation (Fig. 1E). In vitro studies uncovered an intriguing synergistic effect between AK23 and FasL on vesicle formation. Co-stimulation with AK23 and FasL significantly increased the number of vesicles compared to both FasL IgG and AK23 alone (Fig. 1F, G). Furthermore, anti-Dsg3 treatment significantly reduced desmosome length in vitro (Supplementary Fig. 2A, B) and ex vivo (Supplementary Fig. 2C), and widened the interdesmosomal space in vitro (Supplementary Fig. 2D, E), but not ex vivo (Supplementary Fig. 2F). FasL did not provide any additive effect in these structural changes (Supplementary Fig. 2A, B, D, E).

In summary, the anti-Dsg3 monoclonal antibody induced acantholysis, which was enhanced and accelerated by co-stimulation with FasL in the presence of caspase-8 activation. Importantly, these processes occurred without any signs of cell death. Ultrastructurally, the binding of anti- Dsg3 antibodies induced vesicle formation, which was significantly enhanced by the addition of FasL. This points to distinct mechanisms controlling the fate of Dsg3 in PV.

Anti-Dsg3 treatment leads to an altered cellular distribution of endogenous Dsg3

The depletion of desmosomal Dsg3 is a key driver in the loss of cell cohesion [22, 23]. In our ex vivo skin model, we investigated the effect of anti-Dsg3 antibody treatment on the distribution of Dsg3 in the epidermis. Antibody binding was confirmed by IgG staining (Fig. 2A, green staining). In control- treated samples, endogenous Dsg3 was evenly distributed across the cell surface, with occasional cytoplasmic localization (Fig. 2A red staining, upper row). Anti-Dsg3 antibody treatment decreased the expression of Dsg3 in the suprabasal layer with concomitant enrichment in the apical region of certain suprabasal keratinocytes (Fig. 2A, red staining, lower row). Additionally, granular structures were formed in the cytoplasm (Fig. 2A, inset, white arrows), with most granules containing both Dsg3 and AK23, suggesting their co-translocation. Within 30 min of anti-Dsg3 antibody treatment of epidermal monolayers, endogenous Dsg3 levels in the total cell lysate decreased (Fig. 2B). The observed reduction correlated with a time-dependent decline in Dsg3 levels in the Triton-soluble fraction (TSF), which comprises cytosolic and membrane-associated proteins that are not tightly bound to the cytoskeleton or lipid rafts [24]. This occurred alongside the translocation of the AK23 heavy chain (Fig. 2C). Notably, both Dsg3 and AK23 accumulated in the Triton-insoluble fraction (TIF), which contains proteins associated with intracellular vesicles, in a time-dependent manner (Fig. 2C). This likely happened at a later stage, as the translocated fraction of Dsg3 is comparatively smaller in TIF compared to TSF. These findings strongly confirmed our histological observation demonstrating the formation of AK23 and Dsg3-positive granules upon anti-Dsg3 treatment (Fig. 2A). Importantly, caspase inhibition by pan-caspase inhibitor Z-VAD-FMK (zVAD) did not prevent the reduction of Dsg3 in the lysate (Supplementary Fig. 3A), indicating that this process is caspase-independent. Dsg3 is known to degrade via endosomal and lysosomal pathways [25, 26]. However, neither the proteasome inhibitor bortezomib (Supplementary Fig. 3B), nor chloroquine (a lysosome inhibitor) (Supplementary Fig. 3C) alone or combined (Supplementary Fig. 3D), prevented the depletion of Dsg3. Altogether, these findings demonstrated that anti-Dsg3 treatment might support the redistribution of endogenous Dsg3 to the TIF or other cell compartments (e.g., lipid rafts [27]).Fig. 2AK23 disrupts the endogenous distribution of Dsg3 in a time-dependent manner.A Immunofluorescence staining of healthy skin punches stimulated with AK23 or control IgG for 48 h, showing IgG (green) and Dsg3 (red) localization, (n = 5). Arrows indicate intracellular granules. The dotted line marks the basal membrane. Scale bar: 50 µm. WB analysis of Dsg3 in HaCaT keratinocyte monolayers under different treatment conditions: B Time-course of Dsg3 expression following AK23 treatment, (n = 3). C Comparison of Dsg3 presence in Triton soluble (TSF) and Triton insoluble fraction (TIF) of HaCaT monolayers treated with AK23 for the indicated time points (n = 3). Caspase-8 and calveolin-1 confirm the purity of the respective TIF and TSF. Equal protein amounts were analyzed in all WB experiments.

FasL synergistically enhances anti-Dsg3-induced loss of cell cohesion via caspase-8

While Dsg3 depletion via sequestration in TIF contributes to anti-Dsg3 antibody-induced acantholysis, the protein can also be altered by proteolytic cleavage. It has previously been shown that induction of apoptosis causes Dsg3 cleavage, as observed with staurosporine treatment [28]. We found that FasL stimulation induced time-dependent cleavage into ~100 kDa and ~75 kDa fragments (Fig. 3A). In a dispase-based keratinocyte dissociation assay [29], anti-Dsg3 treatment alone reduced keratinocyte cohesion within 4 h (Fig. 3B), unaffected by the pan-caspase inhibitors zVAD or QVD-OPH (QVD) (Supplementary Fig. 4A) and without inducing caspase-8 activation (Supplementary Fig. 4B) or sensitizing cells to apoptosis (Supplementary Fig. 4C). As FasL is released by lesional keratinocytes in PV [19], it is present in the serum of PV-patients [11] and enhances antibody-induced acantholysis ex vivo, we investigated its effect on desmosomal detachment. FasL at sublethal levels (Supplementary Fig. 4D) alone had no impact on keratinocyte cohesion, but markedly increased antibody-mediated loss of cell cohesion in a caspase-dependent manner (Fig. 3B). To specifically assess the role of caspase-8, we utilized HaCaT cells overexpressing the short isoform of cFLIP (cFLIP_S_ OE), which blocks caspase-8 activation (Supplementary Fig. 4E) [30]. In HaCaT cells overexpressing cFLIP_S_, in opposite to control cells, FasL did not further augment the anti-Dsg3-mediated loss of cell cohesion (Fig. 3C). These findings confirm the pivotal role of targeted caspase-8 activation in amplifying this acantholytic mechanism. Notably, the cleavage of Dsg3 occurred mainly in the TSF and to a lesser extent in the TIF. This cleavage was completely prevented by zVAD, indicative of caspase-dependency (Fig. 3D). In conclusion, the full-length Dsg3 protein level was reduced by two different mechanisms. Anti-Dsg3 treatment promoted a reduction of full-length Dsg3 level, while FasL decreased full-length Dsg3 through cleavage. This unravels a synergistic effect of anti-Dsg3 monoclonal antibodies and FasL by promoting loss of cell-adhesion (Fig. 3E). Under physiological conditions and at the time point when we observed a loss of cell cohesion ex vivo (Fig. 1C), anti-Dsg3 antibody treatment caused a minor reduction of full-length Dsg3 in the TSF after 48 h. Notably, a decrease of full-length Dsg3 in the TIF was already evident after 24 h (Supplementary Fig. 4F). Consistent with our in vitro findings, AK23 progressively accumulated in the TIF over time (Fig. 2C and Supplementary Fig. 4F). Furthermore, FasL treatment induced a minor progressive cleavage in the TSF but not in TIF, likely reflecting underlying biological differences as well as differences in the technical procedures (Supplementary Fig. 4F). In conclusion, FasL enhanced anti-Dsg3-induced loss of cell cohesion by caspase-8-induced cleavage of Dsg3 in the presence of anti-Dsg3 antibody.Fig. 3. FasL synergistically enhances AK23-induced loss of cell adhesion via a caspase-8-dependent mechanism.A WB analysis of HaCaT monolayers stimulated with FasL. (n = 3). B HaCaT monolayers were pre-stimulated with zVAD for 1 h, then stimulated with AK23 or control IgG for further 1 h, followed by FasL for 3 h, before conducting a dispase-based keratinocytes dissociation assay (DDA), (n = 6). Error bars represent the SEM. C HaCaT cFLIP_S_ OE or control monolayers were pre-stimulated with AK23 or IgG, followed by FasL stimulation, and analyzed by DDA (n = 4). Error bars represent the SEM. D–E WB analysis of HaCaT monolayers pre-stimulated with zVAD 1 h, followed by FasL incubation for 3 h, (n = 3) (D) or pre-stimulated with AK23 or IgG for 1 h, followed by FasL stimulation for 3 h, (n = 3) (E). Equal protein amounts were analyzed in all WB experiments. fr: fragment; h.c.: heavy chain; TIF: Triton insoluble fraction; TSF: Triton soluble fraction.

Discussion

Mechanisms underlying antibody-related acantholysis remain a matter of debate; however, surface Dsg3 levels are considered critical for preserving epidermal integrity. Reduction of Dsg3 [15, 22, 31, 32] is attributed to protein degradation [33], cleavage [34, 35] or translocation [25, 36]. We were now able to show that anti-Dsg3 antibody rapidly internalized complexes consisting of endogenous Dsg3 and anti-Dsg3 antibodies, depleting Dsg3 from the cell surface and causing loss of keratinocyte cohesion without initial degradation. Internalized Dsg3/antibody complexes may later co-localize with lysosomal markers [26, 33]. Consistent with these observations, we detected a degradation of Dsg3 in suprabasal skin layers coinciding with late-stage acantholysis, which is in agreement with the findings of Jolly et al. [37]. However, this late Dsg3 degradation may be modulated by additional signaling pathways activated by other PV-IgG or serum components. We have identified intracellular caspase-related Dsg3 cleavage as one possible mechanism resulting in increased acantholysis. Previous studies have shown that Dsg3 can translocate into lipid rafts [38] or intracellular vesicles [39]. Our TEM studies revealed that anti-Dsg3 treatment increased the number of intracellular vesicles near desmosomes and led to desmosome shortening in vitro and ex vivo, consistent with findings from Egu et al. [40]. It is tempting to speculate that the Dsg3-containing granules observed in our immunofluorescence studies might correspond to the vesicles that we detected adjacent to desmosomes in the TEM analyses. Moreover, reduced cell cohesion and increased acantholysis were evident, consistent with other reports [22, 41]. Simultaneously, we detected an increase of both Dsg3 protein and anti-Dsg3 antibodies in fractions enriched for endosomes and lipid rafts, indicating co-translocation of Dsg3 and anti-Dsg3 antibody. Lipid rafts play a key role in PV pathogenesis, as desmosome disassembly and endocytosis triggered by PV-IgG depend on these membrane microdomains [27, 36, 42]. Of note, the active Fas-DISC is known to localize in lipid rafts [43], and elevated FasL levels have been detected in PV sera, originating from keratinocyte- or cytotoxic T-cell driven sources [11, 44, 45]. Blocking soluble FasL has been shown to reduce blister formation in an ex vivo PV model [46]. Fas-signaling synergizes with anti-Dsg3 antibodies, promoting endocytosis [47]. Here, we demonstrate that combined treatment with FasL and anti-Dsg3 antibodies produced a pronounced synergistic effect, markedly promoting endocytosis and blister formation ex vivo.

The relevance of apoptosis in PV-related blister formation remains controversial: while some studies link PV-IgG to caspase-8/–3 activation, resulting in apoptosis [13, 32, 48], others claim acantholysis occurs independently of cell death, considering it a secondary, irrelevant side effect or a parallel event [49]. In our study, we demonstrated that binding of an anti-Dsg3 antibody alone did not activate caspases or induce apoptosis. However, the activation of apoptotic caspases significantly enhanced the effect of anti-Dsg3 antibody both in vitro and ex vivo. Notably, we observed active caspase-8 in lesional skin without cell death, which contrasts with reports of TUNEL-positive keratinocytes [11, 12, 48, 50, 51] and suggests mechanistic or temporal heterogeneity. The relevance of the apoptotic signaling pathway was emphasized recently by Pacheco-Trovar et al. [52]. These data, together with the results of our ex vivo model, demonstrated that caspase-8 activation preceded or coincided with blister formation. The presence of TUNEL-positive cells may indicate cell death occurring at a later stage. Interestingly, a recent study from Peng et al. reported increased Fn14 (Fibroblast Growth Factor-Inducible 14, a TNFR superfamily surface receptor) expression in both lesional and perilesional skin of PV patients [53]. While Fn14 activation per se does not lead to high rates of apoptosis, it can markedly enhance FasL-induced apoptosis, which may explain our findings. Antibody binding may induce expression of pro-apoptotic factors such as Fn14, thereby amplifying the apoptotic effect of FasL [54]. Consistent with this, our ex vivo experiments demonstrated the presence of caspase-8 activity only when the skin was treated with both anti-Dsg3 antibody and FasL, while no cell death was detected under the same conditions in vitro. Recent findings identified herpes simplex virus infection as a trigger for pemphigus vulgaris [55]. Given the established association of herpes simplex infection with apoptotic pathways, apoptosis-related signaling likely plays a role in the pathogenesis of PV [56]. Moreover, blockade of FasL represents a therapeutic target, as demonstrated by the use of intravenous immunoglobulins containing the Fas-receptor in severe, refractory cases of pemphigus [57, 58]. In addition, more experimental approaches targeting this pathway, such as the monoclonal antibody PC111, are currently under investigation [46, 59].

Overall, we suggest that antibody-mediated loss of cell cohesion occurs independent of caspase activation and is driven by the redistribution of full-length Dsg3 from the cell surface into lipid-enriched compartments. Additional caspase-activation, such as FasL-induced caspase-8 activation, enhances this process by promoting Dsg3 cleavage prior to cell death and probably enhances the internalization of Dsg3 (Fig. 4). This mechanism simplifies anti-Dsg ab-mediated acantholysis but solely does not lead to keratinocyte cohesion loss.Fig. 4. The role of caspase-dependent signaling in anti-Desmoglein 3-induced acantholysis.A Anti-Dsg3 antibodies (such as AK23) binding to Dsg3 within the desmosomes triggers the translocation of Dsg3 to lipid rafts and its internalization, ultimately leading to acantholysis. B FasL interacts with the Fas receptor, initiating Death-Inducing Signaling Complex (DISC) formation and caspase activation. Active caspases cleave Dsg3, further enhancing antibody-induced acantholysis, independently of cell death. Dsg3 Desmoglein 3, FasL Fas ligand.

In conclusion, this dual mechanism provides a novel insight into the manifestation of PV at distinct body sites and disease heterogeneity in PV.

Materials and methods

Antibodies and reagents

The following antibodies were used for Western Blot (WB): β-actin (A2103) and LC3 (L8918) (both Sigma-Aldrich, St. Louis, USA), Dsg3 (sc-23912, Santa Cruz, California, USA); Caveolin-1 (#3267) and NIK (#4994) (both Cell Signaling, Danvers, USA), cFLIP (NF6) and caspase-8 (C-15) were kindly gifted by P.H. Krammer; caspase-3 (cpp32/19, BD Bioscience, San Jose, USA) and caspase 3 active (cleaved caspase-3, AF 835, R&D, Minneapolis, USA); HRP-conjugated goat anti-rabbit (4030-05) and HRP-conjugated goat anti-mouse IgG1 (1070-05), IgG2a antibody (1080-05) (all Southern Biotechnology Associates, Birmingham, USA).

The following antibodies were used for IF or IHC: Polyclonal rabbit anti-human IgG (F0315, Agilent Technologies, Santa Clara, California, USA), Mouse IgG1 Alexa Fluor® 488-conjugated Antibody (IC002G, R&D, Minneapolis, USA), Cleaved caspase-8 (#9496, Cell Signaling, Danvers, USA), anti-Desmoglein 3 antibody (ab183743, Abcam, Cambridge, United Kingdom). Secondary antibodies were purchased from Thermo Fisher Scientific Inc., Waltham, USA: # A-11010 for staining Dsg3 and # A-11001 for staining AK23. DAPI was used for nuclear staining (1198406, AppliChem GmbH, Darmstadt, Hessen).

The following stimuli and reagents were used: the IgG human isotype control (# 02-7102, Thermo Fisher Scientific Inc., Waltham, USA), Dulbecco´s Phosphate Buffered Saline (DPBS, P04-36500, PAN-Biotech GmbH, Aidenbach, Germany), pan-caspase inhibitors Z-Val-Ala-DL-Asp-fluoromethylketone (Z-VAD-FMK) (4026865, Bachem GmbH, Bubendorf, Switzerland) and Quinoline-Val-Asp-Difluorophenoxymethylketone (Q-VD-OPh) (SML0063, Sigma-Aldrich, St. Louis, USA). For expression of Fc-FasL, we used constructs published previously [60]; kindly provided by P. Schneider, Epalinges, Switzerland. One unit of Fc-FasL was determined as a 1:1000 dilution of the stock Fc-FasL supernatant, and one unit/ml of Fc-FasL supernatant was sufficient to kill 50 percent (LD50) of HaCaT cells, seeded at 50% confluence and stimulated overnight. Chloroquine diphosphate salt (C6628, Sigma Aldrich, St. Louis, Missouri, United States). Bortezomib (BTZ) (5.04314, Sigma Aldrich, St. Louis, Missouri, United States). Poly(I: C) (#tlrl-pic, InvivoGen, San Diego, USA). For expression of His-FLAG-TRAIL (HF-TRAIL), we used constructs published previously [61].

IgG-production and purification

Hybridoma cells for the mouse monoclonal anti-Dsg3 antibody AK23 were kindly provided by M. Amagai. The cells were cultured in suspension using RPMI 1640 medium at 37 °C in a humidified atmosphere at 5% CO_2_. The culture medium was supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% non-essential amino acids, 1% sodium pyruvate (all from Gibco/Life Technologies, Carlsbad, CA, USA), and 55 µM β-mercaptoethanol (Sigma-Aldrich, Munich, Germany). For antibody production, the cells were grown in a medium with 60% of ISF-I hybridoma medium (Sigma-Aldrich) and 40% culture medium in 1 L volume in EZ flasks (KDBIO, Berstett, France), and grown for 30 days before harvesting by centrifugation [20]. AK23 purification was performed following Beckert et al. [62].

Generation of cell lines

cFLIP_S_ OE HaCaT cells were generated using a pCFG5-IEGZ retroviral vector, as described previously [54]. The expression of cFLIP_S_ was confirmed by WB.

Cell culture

The spontaneously immortalized HaCaT human keratinocyte cell line (kindly provided by P. Boukamp, formerly DZFK Heidelberg), HaCaT cFLIP_S_ OE and control cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (P04-04515, PAN-Biotech GmbH, Aidenbach) supplemented with 10% Fetal Bovine Serum (FBS) standard (P30-3306, PAN-Biotech GmbH, Aidenbach, Germany) at 37 °C in 5% CO_2_ atmosphere.

Conditions for cell stimulation

HaCaT cells were grown to confluent monolayers. The cells were used up to passage 48.

The following stimulation conditions were used: pre-stimulation with zVAD (10 µM) or QVD (10 µM) for 1 h. Pre-stimulation with BTZ (1 µM) and/ or chloroquine (100 µM) was conducted for 5 h. 1 Unit of Fc-FasL was determined as a 1:1000 dilution of the stock Fc-FasL supernatant, and 1 Unit/ml of Fc-FasL supernatant was sufficient to kill 50% (LD50) of A375 melanoma cells, as described previously [54].

For Dispase-based keratinocyte dissociation assay (DDA) and electron microscopy, parental or transduced HaCaT cells were pre-stimulated for 1 h with IgG or AK23 (20 µg/mL for parental HaCaT and 30 µg/mL for transduced HaCaT cells), followed by FasL (0.8 U/mL) stimulation for 3 h.

For WB, IgG or AK23 (30 µg/mL) and FasL (0.8 U/mL). Poly(I: C) (10 µg/mL) and HF-TRAIL (500 ng/mL) were added for the indicated time points. Cells were pre-incubated with IgG or AK23 for 4 h, followed by FasL stimulation for 3 h.

For transmission electron microscopy, HaCaT cells were grown to confluence on 8-well chamber slides and were pre-incubated with IgG or AK23 (20µg/mL) for 1h, followed by FasL (0.8U/mL) stimulation for a further 3h.

Stimulation was performed using DMEM supplemented with chelated FBS (1.6 mM final concentration of Ca^2+^) at 37 °C in 5% CO_2_ atmosphere. Chelated FBS was produced and purified as described previously [29].

Propidium iodide staining

A 100% confluent layer of HaCaT cells on 96-well cell culture plates was stimulated for the indicated time points as described above. The cells were trypsinized, washed with DPBS and stained with PI (10 µg/mL) for 15 min. BD Accuri C6 flow cytometer (BD Bioscience, Franklin Lakes, New Jersey, U.S.) was used for FACS analysis.

Dispase-based keratinocyte dissociation assay

The method was performed under previously established conditions [29]. Resulting fragments were quantified using ImageJ software (1708195; Bio-Rad Laboratories Inc., Hercules, CA, USA) [63] or alternatively counted manually.

Western blot analysis

Cells were washed with DPBS and lysed as described previously [64]. Sonication was performed to further analyze the Triton-insoluble fraction in the cell pellet [65]. For lysates generated from stimulated skin explants (ex vivo model), most of the dermis was removed with a scalpel, and the skin was lysed using Lysematrix D (1169130-CF; MP Biomedicals, Santa Ana, California, United States). The TIF was subsequently obtained by sonication in 2% SDS. Five µg of the protein were separated on a 4-12% gradient gel (NP0329BOX; Thermo Fisher Scientific Inc., Waltham, MA, USA) with the SDS-PAGE technique after heat denaturation of the proteins (95 °C, 5 min) and then transferred to membranes (IB24001X3; Thermo Fisher Scientific Inc., Waltham, MA, USA). Membranes were then blocked in TPBS containing 5% milk powder (70166-500 G; Sigma Aldrich, St. Louis, Missouri, United States) for 2 h at room temperature (RT) and washed with TPBS. Blots were then incubated with primary antibodies overnight at 4 °C, followed by incubation with an appropriate secondary antibody for 1 h at RT. Protein bands were visualized with an Immobilon Forte Western HRP substrate (WBLUF0500, Merck, Darmstadt, Germany).

Electron microscopic analysis

Stimulated cells were fixed in 3% glutaraldehyde in 0.1 M Soerensen’s phosphate buffer (Roth, Karlsruhe, Germany). Samples were post-fixed in 1% OsO_4_ (E19100, Science Services, Munich, Germany) in 25 mM sucrose buffer (1.07651.1000, Merck, Darmstadt, Germany), dehydrated by ascending ethanol series, and embedded in Epon. Ultrathin sections were cut in the horizontal plane. Contrast was enhanced by staining with 0.5% uranyl acetate (E22499-05, Science Services, Munich, Germany) and 1% Sato’s lead citrate. Samples were examined using a Hitachi HT7800 transmission electron microscope (Hitachi, Japan) operating at an acceleration voltage of 100 kV. Analysis was performed by ImageJ. In each independent experiment (n = 2), at least 50 images per condition were captured, each containing at least one desmosome. The person acquiring the images was unaware of the corresponding sample condition. Desmosomes were measured using ImageJ with the following criteria: desmosomal length was determined as the longest continuous visible segment (50k magnification); interdesmosomal space was measured three times at different points along the desmosome (100k magnification); and the number of vesicles was counted as the total vesicles per microscopic field (50k magnification) containing at least one desmosome. The mean of the measurements from each experiment was used for statistical analysis.

Ex vivo skin models

The skin was obtained from safety margins after surgeries at the Department of Dermatology, University Hospital RWTH Aachen. Excess skin from excised safety margins was used on the same day. Before further use, the tissue was washed three times for 15 min each in DPBS. The subcutis was dissected, 8 mm pieces of skin (~50 mm²) were taken by punch biopsy and placed in 6-well culture plates with cell culture inserts (353091, Corning (Corning Inc.), Somerville, USA). 2 ml of medium/well (equal amounts of DMEM + 1% antibiotics/antimycotics (15240096, Thermo Fisher Scientific Inc., Waltham, USA) + 10% FBS and KBM-2 Keratinocyte Growth media (CC31-03, Lonza, Basel, Switzerland)) was added. Human IgG (40 µg), AK23 (40 µg) and/ or FasL (1:10) were diluted in PBS in a volume of 50 µl/punch and injected by a needle (0.4 mm diameter) into the upper dermis. The culture medium was changed daily. The samples were frozen in liquid nitrogen.

Histology and immunohistochemistry

Cryosections or formalin-fixed and paraffin-embedded tissues were used for histopathology (H&E staining) and immunohistochemistry. Image processing was applied identically to all samples and controls.

Immunofluorescence

Cryosections were used for immunofluorescence. Image processing was applied identically to all samples and controls.

TUNEL

TUNEL (TdT-mediated dUTP-biotin nick end labeling) staining was performed according to the manufacturer’s instructions (TUNEL-kit C10617, Thermo Fisher Scientific Inc., Waltham, USA).

Statistics

All data were expressed as the mean ± SEM (standard error of the mean). For pairwise comparison, a two-tailed Student’s t test was performed, whereas multiple group comparison was evaluated using a one-way ANOVA. We used Graphpad PRISM (Version 10) to conduct statistical analyses. ns p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.

Supplementary information

Supplementary Text Original Western Blots pdf