C57BL/6J mice were obtained from Charles River Laboratories (Sulzfeld, Germany). FoxP3^DTR-eGFP (DEREG) mice were kindly provided by Tim Sparwasser, Hannover, Germany (5). Heterozygous DEREG mice were set up for mating with wild-type littermates. Gender-matched littermates aged 8–12 weeks were used for the experiments. The mice were fed with standardized mouse chow and acidified drinking water ad libitum. All clinical examinations, biopsies and bleedings were performed under anesthesia using intraperitoneal (i.p.) administration of a mixture of ketamine (100 mg/g, Sigma-Aldrich, Taufkirchen, Germany) and xylazine (15 mg/g, Sigma-Aldrich). All animal experiments were conducted according to the European Community rules for animal care, approved by the respective governmental administration [V242.29833/2016(49-4/16), V245-46582/2015(78-5/12), V312-7224.122-5(30-2/13)], Ministry for Energy, Agriculture, the Environment and Rural Areas and performed by certified personnel.

Total rabbit anti-mouse COL7 IgG and rabbit anti-mouse COL17 IgG were prepared as previously described (16, 20). Rabbits were immunized with recombinant proteins of the non-collagenous (NC)-1 domain of murine COL7 or the extracellular portion of murine COL17 (NC15A), which were supplied commercially (Eurogentec, Seraing, Belgium). IgG from immune and normal rabbit sera was purified by affinity chromatography using protein G. The reactivity of all IgG fractions was analyzed by immunofluorescence microscopy of murine skin.

Antibody transfer-induced studies for the induction of experimental EBA and BP followed published protocols with minor modifications (16, 20). To induce Treg depletion, mice were treated with 1 µg diphtheria toxin (DT, Sigma-Aldrich)/100 μl PBS/mouse on days 1, 2, 5, 8, and 11 after initial IgG injections. For experimental EBA induction, mice received a total of four i.p. injections of 1 mg rabbit anti-mCOL7 IgG on days 0, 3, 6, and 9 (in total 4 mg rabbit anti-mCOL7 IgG). For experimental BP induction, mice received a total of six i.p. injections of 5 mg rabbit anti-mCOL17 IgG on days 0, 2, 4, 6, 8, and 10 (in total 30 mg rabbit anti-mCOL17 IgG). Different body parts were individually scored by the appearance of crust, erythema, lesions and/or alopecia by blinded personal. Control animals were injected with a total of four i.p. injections of 1 mg normal rabbit IgG on days 0, 3, 6, and 9 (in total 4 mg IgG). The total body score is a composite of 2.5% per ear, snout and oral mucosa; 0.5% per eye; 9% for head and neck (excluding eyes, ears, oral mucosa, and snout); 5% per front limb; 10% per hind limb and tail; and 40% for the remaining trunk (21). Blood and tissue samples were collected on days 5 and 12. Serum was collected from the blood samples by centrifugation and was analyzed for cytokine release using the LEGENDplex™ Mouse Inflammation Panel (BioLegend) as described by the manufacturer’s protocol. Tissue samples were snap frozen in liquid nitrogen for the analysis of mRNA and for immunostainings.

All experiments were repeated with a minimum of two independent experiments using different batches of purified IgG.

Direct immunofluorescence microscopy was performed to detect rabbit IgG and murine C3 in experimental PD as described (16, 20). Briefly, frozen sections were prepared from tissue biopsies and incubated with FITC-labeled goat anti-rabbit IgG antibody (Dako Deutschland GmbH, Hamburg, Germany). For histology, skin samples were fixed in 3.7% paraformaldehyde. The 8-μm-thick sections from paraffin-embedded tissues were stained with hematoxylin and eosin (H&E) according to standard protocols. For immunohistochemistry, paraffin sections from lesional skin were stained for T cells and granulocytes as previously described (16). Briefly, mAbs against CD4 (BD Biosciences, Heidelberg, Germany) and Gr-1 (BD Biosciences) were used as primary antibodies, and biotin rabbit anti-rat IgG (Dako) was used as the secondary Ab, followed by detection with ExtrAvidin^® alkaline phosphatase (Sigma-Aldrich). Alkaline phosphate activity was visualized with Fast Blue (BB Salt, Sigma-Aldrich). Samples were stained by hematoxylin according to standard protocols.

For FACS analysis, the following antibodies were used: Vio-Green; Brilliant Violet 421™; FITC-, Alexa 647-, PE-, allophycocyanin (APC)-, or APC-Vio770-conjugated anti-mouse CD4 (clone L3T4, or RM4-5); Gr-1 (clone RB6-8C5); CD45 (clone 30F11); CD18 (clone M18/2); CD62L (clone MEL-14); CD11b (clone M1/70); CD25 (clone PC61); Ly6G (clone 1A8); Foxp3 (clone MF23); or appropriate isotype control antibodies (eBioscience via Thermo Fisher Scientific, Dreieich, Germany, Miltenyi Biotec, Bergisch-Gladbach, Germany or BD). After erythrocyte lysis cell suspensions were blocked with anti-mouse CD16/CD32 mAb before staining, and dead cells were excluded from the analysis using propidium iodide (PI). Briefly, for the staining of CD45/Gr-1/CD11b and CD45/CD4 cells, blocked single cell suspensions from spleen and blood of mice suffering from experimental PD were first gated for singlets (FSC-H compared with FSC-A) and leukocytes (SSC-A compared with FSC-A). The leukocyte gates were further analyzed for their uptake of PI to differentiate between live and dead cells and for their expression of CD45, thus, selecting only the live, healthy leukocyte population. To further analyze the purity of isolated Tregs and PMNs for in vitro analysis, cells were stained with CD4/Foxp3/CD25 or Ly6G/CD45/PI, respectively. For PMNs, the purity and viability was ≥90%; for Tregs, the purity was 80%. To determine the activation status of PMNs after treatment w/o ICs and Tregs, the cells were stained with CD45/CD62L/CD18/Ly6G/PI. All stainings were performed using standard protocols for cell surface staining, except for CD4/Foxp3/CD25, where intranuclear staining was performed using FOXP3 Fix/Perm Buffer (BioLegend, San Diego, CA, USA) and BD Perm/Wash™ buffer following manufacturer protocols. FACS analysis was performed using Miltenyi MacsQuant10 or FACSCalibur with MACSQuantify™ (version 2.8) or BD CellQuest Pro (version 5.1) software.

PMNs were isolated from the femurs and tibias of healthy C57BL/6J mice as described in detail elsewhere (16). Tregs were isolated from the spleen of the same animal using a CD4⁺CD25⁺ Regulatory T Cell Isolation Kit, mouse (Miltenyi) following the manufacturer’s protocol. The enrichment of cells was determined by FACS. In total, 2 × 10⁵ PMNs/100 μl were stimulated with ICs formed by 10 µg/ml mCOL7 and 2 µg/ml rabbit anti-mCOL7 IgG as described elsewhere (22) for 60 min at 37°C. Isolated Tregs were then cocultured with the IC-stimulated PMNs for an additional 4.5 h in a ratio of 1:4 (5 × 10⁴ Tregs/2 × 10⁵ PMN/200 μl). To evaluate the activation status, cells were stained for flow cytometry analysis using CD18-FITC, CD62L-PE-Vio770, Ly6G-APC-Vio770, and CD45-VioGreen (Miltenyi) following standard procedures. Dead cells were excluded using PI.

Neutrophil activation was assessed by determining IC-induced ROS release using a previously published protocol (16). Isolated murine neutrophils (2 × 10⁵ cells/100 μl) were preincubated w/isolated murine Tregs (5 × 10⁴ cells/200 μl), for 1 h at 37°C (without ICs), followed by incubation on a 96-well plate (Lumitrac 600, Greiner Bio-One, Frickenhausen, Germany) coated with ICs formed by 10 µg/ml mCOL7 and 2 µg/ml rabbit anti-mCOL7 IgG. ROS release was analyzed using luminol (Sigma-Aldrich) (22). Each plate was analyzed for 99 repeats using a plate reader (GloMax^®-Multi Detection System, Promega GmbH, Mannheim, Germany); the values are expressed as relative luminescence units.

Neutrophil activation was assessed by determining neutrophil extracellular trap (NET) formation using a previously published protocol (16, 23). Blood collection was conducted with the understanding and written consent of each participant and was approved by the Ethical Committee of the Medical Faculty of the University of Lübeck (09-140). Isolated human neutrophils (2 × 10⁵ cells) were either stimulated with 20 nM PMA or with ICs formed by 10 µg/ml mCOL7 and 2 µg/ml rabbit anti-mCOL7 IgG. The stimulation was performed w/o isolated human Tregs (5 × 10⁴ cells) on a 96-well FLUOTRAC™ 600 plate (Greiner Bio-One) in 10 mM HEPES-buffered medium. NET formation was analyzed for 7 h (IC) or 4 h (PMA) every 5 min at 37°C by using Tecan infinite M200 Pro reader and Tecan i-control 1.7 Software. CO₂ control was achieved during the assay by the use of a Tecan gas module. For statistical analysis, the area under the curve was calculated.

For gene expression analysis in skin sections, 10 cryo-sections (12 µm) were prepared and used for RNA isolation, reverse transcription, and real-time RT-PCR as previously described (24). Briefly, total RNA was isolated according the manufacturer’s protocol (innuPrep RNA Mini Kit, Analytic Jena AG). After reverse transcription, the cDNA was added to either qPCR Master Mix Plus or qPCR Master Mix SYBR Green Plus (Thermo Fisher Scientific Inc., Waltham, USA) and amplified using an SDS ABI 7900 system (Applied Biosystems, Darmstadt, Germany). The amount of cDNA copies was normalized to the housekeeping gene GAPDH using the 2^ΔCT method.

Primer sequences and concentrations of the analyzed genes were previously published (24, 25) or depicted below: Ccr5 (for: 5′-CCC ACT CTA CTC CCT GGT ATT C-3′; rev: 5′-GCA GGA AGA GCA GGT CAG AG-3′; 0.5 µM each), Ccr7 (for: 5′-TGG TGG TGG CTC TCC TTG-3′; rev: 5′-GGC CTT AAA GTT CCG CAC ATC-3′; 0.5 µM each), Cd11c (for: 5′-CCA CTG TCT GCC TTC ATA TTC-3′; rev: 5′-GAC GGC CAT GGT CTA GAG-3′; 0.5 µM each), Cd19 (for: 5′-GAA AAT GCA GAT GAG GAG CTG G-3′; rev: 5′-GCT GCA TAG AGG ATC CCT CTC-3′; probe: 5′-CAA CCA GTT GGC AGG ATG ATG GAC TTC CT-3′), Cd3 (for: 5′-ATA GGA AGG CCA AGG CCA AG-3′; rev: 5′-TCA GGC CAG AAT ACA GGT C-3′; probe: 5′-CCA GAC TAT GAG CCC ATC CGC AAA GG-3′), Cxcl-1/KC (for: 5′-CAG ACC ATG GCT GGG ATT C-3′; rev: 5′-GAA CCA AGG GAG CTT CAG-3′; probe: 5′-CCT CGC GAC CAT TCT TGA GTG TGG CTA TGA C-3′), Cxcl-10/IP-10 (for: 5′-GAG GGC CAT AGG GAA GCT TG-3′; rev: 5′-CGG ATT CAG ACA TCT CTG CTC-3′; probe: 5′-CAT CGT GGC AAT GAT CTC AAC ACG TGG-3′), Cxcl-2/MIP-2 (for: 5′-AGT GAA CTG CGC TGT CAA TG-3′; rev: 5′-GCT TCA GGG TCA AGG CAA AC-3′; 0.125 µM each), Cxcl-9/MIG (for: 5′-TTG GGC ATC ATC TTC CTG GAG-3′; rev: 5′-GCA GGA GCA TCG TGC ATT C-3′; probe: 5′-CTT ATC ACT AGG GTT CCT CGA ACT CCA CAC-3′), Gapdh (for: 5′-GAC GGC CGC ATC TTC TTG T-3′; rev: 5′-CAC ACC GAC CTT CAC CAT TTT-3′; probe: 5′-CAG TGC CAG CCT CGT CCC GTA GA-3′), Gr-1 (for: 5′-GCG TTG CTC TGG AGA TAG AAG-3′; rev: 5′-CTT CAC GTT GAC AGC ATT ACC-3′; 0.25 µM each), Ifn-g (for: 5′-GCA AGG CGA AAA AGG ATG C-3′; rev: 5′-GAC CAC TCG GAT GAG CTC ATT G-3′; probe: 5′-TGC CAA GTT TGA GGT CAA CAA CCC ACA G-3′), Il-10 (for: 5′-TCC CTG GGT GAG AAG CTG AAG-3′; rev: 5′-CAC CTG CTC CAC TGC CTT G-3′; probe: 5′-CTG AGG CGC TGT CAT CGA TTT CTC CC-3′), Il-13 (for: 5′-GGA GCT TAT TGA GGA GCT GAG-3′; rev: 5′-CAG GGA ATC CAG GGC TAC AC-3′; probe: 5′-CAT CAC ACA AGA CCA GAC TCC CCT GTG C-3′), Il-17a (for: 5′-TCA GAC TAC CTC AAC CGT TCC-3′; rev: 5′-CTT TCC CTC CGC ATT GAC AC-3′; probe: 5′-CAC CCT GGA CTC TCC ACC GCA ATG AAG-3′), Il-33 (for: 5′-GTG ATC AAT GTT GAC GAC TCT GG-3′; rev: 5′-GGG ACT CAT GTT CAC CAT CAG-3′; 0.5 µM each), Il-4 (for: 5′-GAG ACT CTT TCG GGC TTT TCG-3′; rev: 5′-AGG CTT TCC AGG AAG TCT TTC AG-3′; probe: 5′-CCT GGA TTC ATC GAT AAG CTG CAC CAT G-3′), Itgam/Mac-1 (for: 5′-CTT CAC GGC TTC AGA GAT GAC-3′; rev: 5′-CTG AAC AGG GAT CCA GAA GAC-3′; 0.5 µM each), Tnf-a (for: 5′-CCC TCA CAC TCA GAT CAT CTT CTC-3′; rev: 5′-TGG CTC AGC CAC TCC AG-3′; probe: 5′-CTG TAG CCC ACG TCG TAG CAA ACC AC-3′).

The data were analyzed using SigmaPlot, version 12 (Systat Software Inc., Chicago, IL, USA). Applied tests and confidence intervals are indicated in the respective text and figure legends. A p-value < 0.05 was considered statistically significant.

To investigate the impact of Tregs on skin inflammation and blistering in PDs, we assessed the impact on Treg depletion in antibody transfer models of BP (20) and EBA (26). In brief, experimental PD was induced by repetitive injections of rabbit anti-mouse COL7 (Figure 1A) or anti-mouse COL17 IgG (Figure 2A) in wild-type or DEREG mice (5), which were both injected with DT. Subsequently, clinical disease manifestation, expressed as body surface area affected by PD skin lesions, was assessed as the primary endpoint.

In antibody transfer-induced EBA, Treg depletion in DEREG mice led to a significant (more than twofold) increase in skin inflammation and blistering (Figure 1B,C), accompanied by an increase in leukocyte dermal infiltration, while IgG binding and C3 deposition at the dermal-epidermal junction were not affected (Figure 1D) compared with DT injected wild-type controls. In the dermal infiltrate, we could detect large numbers of Gr-1-positive cells (neutrophils and macrophages) and CD4-positive cells (Figure 1D). To verify whether these cell types are influenced by Treg depletion, we analyzed them in the spleens of DEREG and wild-type mice after 12 days of experimental EBA (Figure 1E). Here, in DEREG mice, the percentages of CD4 T cells and Gr-1^int/CD11b neutrophils are significantly increased, but the number of Gr-1^hi/CD11b macrophages is stable. These data implicate an effect of Treg depletion on CD4 T cells and neutrophils that will be investigated in more detail. Importantly, an injection of the same amount of normal rabbit IgG into DT-treated DEREG mice was not sufficient to induce a skin blistering phenotype, but an increase of Gr-1-positive cells in the dermis was detectable (Figure S1 in Supplementary Material). To control the depletory effect of the DT treatment on DEREG mice, we examined the number of Foxp3/CD25-positive CD4 T cells and showed that the depletion is valid throughout the whole experimental procedure in blood and spleen. The number of Tregs is reduced by half (Figure 1F).

Corresponding observations were made in antibody transfer-induced BP (Figure 2). Again, the depletion of Tregs in DEREG mice led to significant aggravation of skin inflammation and blistering (Figures 2A–C) and an increase in leukocyte dermal infiltration (Figure 2D), while IgG binding and C3 deposition (Figure 2D) at the dermal-epidermal junction were not affected. Taken together, we demonstrate the crucial role of Tregs in dampening (auto) antibody-induced, myeloid cell-driven skin inflammation and blistering.

In principle, Tregs could modulate myeloid-driven skin inflammation by two mechanisms that are not mutually exclusive. First, Tregs could alter the migratory behavior of myeloid cells. Second, Tregs might dampen IC-induced myeloid cell activation. Both myeloid migration and activation are essential for inflammation and blistering in PD (12). To address the first possibility, chemokine and cytokine expression in lesional skin of wild-type or DEREG mice after experimental BP and EBA was evaluated (for comparison to healthy skin see Table S1 in Supplementary Material). Interestingly, the expression of innate cytokines, such as TNF, known as prominent cytokines in PD skin lesions (13, 27), did only slightly differ between wild-type and DEREG mice and was not significant in BP skin lesions (Tables 1 and 2). Of note, the Th2 cytokines IL-4, IL-10, and IL-13 were significantly higher expressed in lesional DEREG skin (Tables 1 and 2). The presence of these Th2 cytokines would implicate a rather anti-inflammatory cytokine milieu. By contrast, the expression of the inflammatory Th1 cytokine IFN-γ and the chemokine CXCL-9 was evaluated in DEREG lesions. Therefore, the increased dermal infiltrate observed in DEREG mice after PD induction is most likely driven by IFN-γ and CXCL-9 (28, 29), whereas the anti-inflammatory properties of the other differentially expressed cytokines, especially IL-10 (30), are not sufficient to prevent blistering. The ratio of Th2-specific IL-10 to Th1-specific IFN-γ is significantly shifted to a more Th1-specific phenotype in DEREG mice in EBA and BP (Figure 3). In accordance with this finding, we evaluated pro-inflammatory cytokine expression in the serum of wild-type or DEREG mice after 12 days of PD; here, the most prominent cytokine is IFN-γ, which is released in DEREG mice but nearly undetectable in wild-type mice in experimental EBA and BP (Tables 3 and 4). In addition, in experimental BP IL-1β that is slightly reduced. By contrast, IL-10 shows no significant change in the serum, which is in line with the assumption that the increased expression of anti-inflammatory cytokines in the skin is ineffective.

To investigate a potential direct effect of Tregs on PMN function in more detail, we performed in vitro assays and stimulated PMNs with ICs consisting of COL7/anti-COL7 in the presence or absence of Tregs (Figure 4). Whereas the expression of β2 integrin (CD18) is strongly reduced by Tregs (Figure 4A), the shedding of CD62L in Ly6G-positive neutrophils is unchanged (Figure 4B). This finding is in accordance with previous data that clearly show the importance of CD18 in experimental PD (31). An effect on the survival of PMNs in the observed time period could not be detected (Figure 4C). In contrast to the effects on the cell surface marker expression, Tregs do not influence the IC-induced ROS release of PMNs after 1 h preincubation of both cell types (Figures 5A,B) and had no effect on the production of NETs, which are released during a programmed cell death, the so-called NETosis induced by IC or PMA (Figures 5C–E).