Serum was obtained by plasmapheresis from a 16-year-old female patient diagnosed as PNP. The criteria we applied included progressive stomatitis, histologic features of acantholysis or lichenoid or interface dermatitis, demonstration of serum antiplakin autoantibodies by immunoblotting or immunoprecipitation, and the presence of an underlying lymphoproliferative neoplasm (11).

This patient presented with painful mucosal and mucocutanous ulcetrations for 4 months before admission. Severe painful erosions involved her tongue, lips, buccal mucosa, eyes, and vulva diffusely. General exfoliative skins were noted on her body. The direct immunofluorescence (DIF) showed deposition of C3 on the basement membrane zone, while indirect immunofluorescence (IIF) showed deposition of IgG on the surface of keratinocytes with the titer at 1:640. IIF using rat bladder as substrate was also positive and the titer was 1:160. IP combined IB detected antibodies against desmoplakin, envoplakin, periplakin, and desmoglein 3. A mass of 5 cm plus 4 cm in the mediastinum was detected by computed tomography. She was diagnosed with PNP and treated with plasmapheresis. Unfortunately, while preparing for the operative status, the patient died of sepsis.

Full-length cDNA fragments encoding human desmoplakin I were synthesized and the target gene of the C-terminal (aa1945-aa2871) of desmoplakin (DP-C) was optimized and cloned into pET-28a (+) by NdeI and XhoI commercially (Genscript, Nanjing, China). The gene of C-terminal of desmoplakin (DP-C) with a C-terminal hexahistidine tag was transformed and expressed in E. coli BL21 (DE3). Briefly, a fresh BL21 (DE3) colony of 3 ml LB medium with 50µg/ml ampicillin was inoculated with 300 ml of LB medium supplemented with 50 μg/ml kanamycin. The LB medium was incubated at 37°C, shaking at 200 rpm until the bacterial suspension reached an optical density (OD) of 0.5 at 600 nm. After 5 h of induction with 0.05 mM/ml isopropyl-b-D-thiogalactoside, the culture was centrifuged. Soluble recombinant DP-C was obtained by sonication and purified by the NTA column (GE, Novagen) with binding buffer, washing buffer, and elution buffer as before (3). Imidazole was removed from the eluted protein by dialysis for 2-3 times at 4°C with phosphate buffered saline (PBS). Then recombinant DP-C was ultrafiltered to concentration and was measured by bicinchoninic acid assay. Finally, sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed and stained with Coomassie brilliant blue.

A certain amount of CNBr-activated Sepharose 4B beads (GE Healthcare) was weighed and suspended in 1 mM HCl. The medium was washed for 15 min with 1 mM HCl following the protocol described in the instruction. Purified DP-C proteins were dialyzed against coupling buffer (0.1 M NaHCO3, 0.5 M NaCl, pH 8.3) and coupled with the medium overnight at 4°C. After washing away excess ligands with at least five times the medium volume of coupling buffer, we transferred the medium to 0.1 M Tris-HCl buffer (pH 8.0) to block any remaining active groups for 2 h, and then washed the medium with three cycles of alternating pH with five medium volumes of each buffer. Each cycle consisted of a wash with 0.1 M acetic acid/sodium acetate, pH 4.0 containing 0.5 M NaCl, followed by a wash with 0.1 M Tris-HCl, pH 8.0 containing 0.5 M NaCl.

Firstly, normal adults’ sera and PNP patients’ sera were diluted with 0.02 M PBS and purified by rProteinA sepharose (GE Healthcare) on AKTA to obtain total IgG. The autoantibodies against DP-C were purified by the AKTA using the recombinant DP-C coupled to the CNBr column. IgG bound to DP-C was eluted (100 mM Glycine, pH 2.7) from the DP-C coupled sepharose columns on AKTA and neutralized with 2 M Tris-HCl, pH 8.0, and then it was concentrated by ultrafiltration with a 0.22 μm filter (Amicon.Millipore, Ireland) against PBS. The IgG of the N-terminus of desmoplakin was purified by the N-terminus of DP and followed the same strategy as the anti-DP-C IgG from the remaining IgG of the same patient after the DP-C affinity chromatography.

IP combined IB assay was performed using HaCat cells extract as substrate. The HaCat lysate buffer contained 62.5 mM Tris-buffer (PH = 8.0) with 1% TritonX-100 and the protease inhibitor cocktail tablet (Roche). The HaCat cells extract was precleared with rProtein A Sepharose by incubating it for 45 min at 4°C. Then 25 μg of purified DP-C specific IgG or positive controls (the commercial monoclonal antibodies against DP: Santa Cruz, sc-390975; the commercial monoclonal anti-desmoglein3 IgG: Abcam, ab14416) or healthy donors’ IgG was added to the precleared lysate separately and incubated overnight at 4°C, and then immunoprecipitated with rProtein A Sepharose for 2 h. The immunoprecipitants were washed for six times and separated by SDS-PAGE with a 6% gel, and electrotransferred onto nitrocellulose (NC) membranes. Then, after Ponceau S stained the NC membrane, only the wide and deeply dyed band of 55KDa was visible, most probably corresponding to the Ig heavy chains. In the following IB assay, desmoplakin and desmoglein 3 were detected by other commercial monoclonal antibodies (the monoclonal antibodies against desmoplakin I/II: Abcam, ab247866; the monoclonal antibodies against desmoglein 3: Abcam, ab128927).

The Desmoglein 3 ELISA kit (MBL, Japan) was used to confirm whether the purified IgG contained the desmoglein3-specific IgG. Normal control IgGs from six healthy volunteers, positive control IgGs from five PNP patients, and negative control from one PNP patient were included to determine the proper dilution ratio. IgG purified by rProteinA was diluted at 1 mg/ml and the assay was performed afterward. The proper dilution ratio of the IgG was determined when the cut-off value (mean + 3SD) could distinguish between positive and negative control. IgG was diluted by dilution buffer in the kit. The proper diluted ratio was 1:15. Other steps were strictly performed as per the manufacturer’s instructions.

HEK (PromoCell) cells were seeded in triplicate on 12-well plates containing 0.1 mM calcium, then switched to the medium containing 1.2 mM calcium after reaching confluence, as described by Saleh et al. Cells were incubated in the medium containing 80 μg/ml of purified anti-DP-C autoantibodies or normal human IgG at 37°C overnight. Following washing in PBS, cells were incubated in situ with 0.3 ml of dispase II (> 2.4 U/ml; Roche) for 30 min to detach the epidermal sheet from the wells. The released sheets were carefully washed in PBS and then subjected to mechanical stress by pipetting with a 1 mL pipette. Fragments were fixed by adding formaldehyde at a final concentration of 3% and stained by adding crystal violet (Sigma-Aldrich) (12). At least three independent experiments were done in duplicates for each autoantibody.

The concentration of the IgG was calculated by measuring OD280 (OD280 of 1 mg/ml IgG =1.43). IgG was affinity-purified on a recombinant DP-C column. Neonatal BALB/C mice (Beijing Vital River Laboratory Animal Technology, China) were obtained at 12–24 h of age (body weight, 1.4–1.6 g). The affinity-purified anti-DP-C IgG was injected at 3.3, 1.7, and 0.6 mg of total protein per gram of body weight into the subcutaneous of neonatal mice. Normal IgG was injected as a control with the same dose. Parts of the neonatal mice were sacrificed at 3 h and skins were harvested for direct immunofluorescence by freezing in optimal cutting temperature compound (OCT). The other mice were sacrificed at 24 h and skins were also obtained for HE staining, DIF, TUNEL assay, by either fixation in 10% formalin, freezing in OCT, or putting in the glutaraldehyde fixed solution prepared for electronic microscopy accordingly.

Neonatal mice skins used as substrate for DIF were biopsied 3 h after injection. The mice skins were embedded in OCT and made frozen sections with a thickness of 4-5 um. Fixed with acetone at 4°C for 10 min, and immersed with 0.01 mol/L PBS (pH7.4) in 4°C for 1 min, the specimen was always kept at certain humidity. Each section was blocked with 100ul of 10% goat serum at room temperature for 30 min and washed with 0.01 mol/L PBS (pH 7.4) three times. Rabbit anti-human IgG (H+L)-FITC antibodies were diluted at 1:200, and added to each section, and then kept away from light and incubated at 37°C for 30 min, and washed at 1 min each time for four times. Lastly, a drop of buffered glycerin was added to cover the glass slide. The images were collected by a laser confocal microscope at 488 nm.

The skin specimens were cut into small pieces and placed in half-strength Karnovsky fixative. Samples were fixed while on a rotator at room temperature for 2 h and then washed in PBS at 4°C. Postfixation was in osmium tetroxide for 2 h at a 4°C temperature. Specimens were dehydrated in a graded ethanol series (15 min each), stained en bloc in uranyl acetate, and embedded in Eponate 812 resin via propylene oxide. Semithin sections (0.5 mm) cut on an ULTRACUT UCT/UC6 ultramicrotome (Leica, UK) were stained with azure II and methylene blue. Ultrathin sections (80-90 nm) were stained with 50% alcoholic uranylacetate (15 min) and lead citrate (10 min) and examined by a JEM-1230 transmission electron microscope (Tokyo, Japan).

The experimental mouse skin samples were embedded in OCT. Cryopreserved tissue sections were fixed in 4% paraformaldehyde in PBS for 20 min and subjected to TUNEL assay using the in situ cell death detection kit (Roche) according to the manufacturer’s instructions. Following permeabilization and wash, sections were incubated with a reaction mixture containing TdT and fluorescence-conjugated dUTP for 1 h at 37°C. The labeled DNA was examined under a confocal microscope.

The t-test was performed for comparisons between two groups by SPSS (SPSS, Inc., Chicago, IL, USA). GraphPad Prism 5.0 was used to compare differences and drew a graph. P < 0.05 was considered statistically significant.

Written informed consent was obtained from all patients and healthy donors. The studies, including the animal study, were performed according to the Declaration of Helsinki and were approved by the medical ethical committee of Peking University First Hospital.

Recombinant DP-C from aa1945 to aa2781 of desmoplakin was expressed by E. coli BL21. It contained plakin A, B, and C domains ( Figure 1A ). Between the B and C domains was the Linker subdomain (not shown in Figure 1A ). The mass weight turned out to be about 106 kDa. An isolated band was detected by Coomassie-staining on the SDS-PAGE after purification by the Ni-NTA column ( Figure 1B ).

Total IgG was obtained from the PNP patient’s serum by rProteinA column (GE). Autoantibodies against DP-C were purified by the recombinant DP-C coupled CNBr column from the total IgG. Before the functional experiments, the specificity of the DP-C affinity purified IgG was confirmed by IP-IB with HaCat cells extract as substrate and ELISA (4). The commercial monoclonal antibodies against DP, the purified anti-DP-C IgG, N-terminus of desmoplakin affinity purified IgG, and IgG of the healthy donors were incubated with the extract of HaCat cells for IP, and then the IB assay was performed using another monoclonal antibodies against desmoplakin I/II. As shown in the left panel of Figure 1 C, desmoplakin I and II (protein bands mass weighted 250 kDa and 210 kDa) were detected in the IP complex of DP-C affinity purified IgG (C3), the same as the IP complex of commercial monoclonal anti-desmoplakin antibody (C1), the extract of HaCat cells (C2), and the IP complex of N-terminus of desmoplakin affinity purified IgG (C4). To exclude the possibility of pathogenic interference of autoantibodies against the antigen of pemphigus vulgaris presented in our preparation, the IP-IB assay with the monoclonal antibodies against desmoglein 3 was also performed. As shown in the right panel of Figure 1C , desmoglein 3 was detected in the extract of HaCat cells (C6) and the IP complex of monoclonal antibodies against desmoglein 3 (C7) worked as the positive control. The complex of DP-C affinity IgG (C9) was negative, the same as the healthy donor control (C11). To further determine whether the DP-C affinity purified IgG can react with desmoglein 3 or not, ELISA was performed with IgG from PNP patients and healthy volunteers. Five PNP sera, all with antibodies against Dsg3, were performed as the positive controls. One PNP serum without antibodies to desmoglein 3 and six healthy adults served as controls. The cut-off OD value was 0.137 (Mean+3SD). The negative PNP and anti DP-C autoantibodies were both tested negative (both of their OD =0.135), while the mean OD value of the positive control is 0.272, 0.623, 2.175, 0.962, 0.221, respectively ( Figure 1D ).

To examine the pathogenicity of the anti-DP-C-specific IgG in vitro, we performed a dispase-based keratinocytes dissociation assay with normal human epidermal keratinocytes (HEK) ( Figure 2A ). IgG of healthy adults were use as negative control. The numbers of dissociated fragments caused by anti-DP-C IgG were significantly higher than normal control (P-value=0.0207, Figures 2 B, C).

To further determine whether anti-DP-C IgG in PNP sera was pathogenic in vivo, the purified autoantibodies were injected subcutaneously into 16 neonatal mice at different doses. Before the injection experiments, the titer of the original anti-DP-C IgG before injection was 1:640 tested by the IIF on rat bladder. The maximal dose was then set at 5 mg, the middle dose was set at 2.5 mg, and the minimal dose was set at 1 mg. The corresponding concentrations were about 3.3 mg/g, 1.7 mg/g, and 0.6 mg/g, respectively. The sera of the mice were all positive tested by IIF on the rat bladder, and the titers of the 3.3 mg/g dose group were 1:20 to 1:40, the titers of the 1.7 mg/g dose group were both 1:10, and the titers of the 0.6 mg/g group were 1:10, which were in the trend of dose-dependent. Normal human IgG was injected as a control at the same dose 24 h later. Obvious blisters on the back of neonatal mice accepting 3.3 mg/g anti-DP-C IgG can be seen ( Figures 3A, B ) but not the ones injected with normal human IgG ( Figure 3C ). One of the mice injected with 1.7 mg/g anti-DP-C IgG formed blisters on the back after two pinches. But 0.6 mg/g anti-DP-C antibodies didn’t induce visible blisters in neonatal mice even after being pinched three times. The mean pinch scores of different dose groups were calculated ( Table 1 ). The highest dose group (3.3 mg/g) got the highest score, and the score of the lowest group (0.6 mg/g) and the healthy control group were zero, which is dose-dependent. Hematoxylin and eosin staining of skin biopsy showed intraepidermal blisters formation and significantly accelerated acantholysis in the spionous cell layer in the 3.3 mg/g group ( Figures 3D–G ). Acantholysis was found in one of the two mice injected with a lower concentration of IgG (1.7 mg/g ( Figure 3H ), 0.6 mg/g). In contrast, mice injected with 3.3 mg/g normal human IgG does did not show such histological changes ( Figure 3I ). In DIF, positive human IgG deposition was observed on the keratinocyte cell surfaces in the spinous layers of the biopsy skin 3 h after injection with 3.3 mg/g anti-DP-C IgG ( Figure 4A ). No deposition of IgG on the epidermis of normal control was found ( Figure 4B ).

To further observe the changes causing acantholysis, the electronic microscopy examination was performed and showed the disconnection of keratin intermediate filaments from desmosomes (arrows) in the mice injected with anti-DP-C IgG ( Figure 4C ) compared with the control group ( Figure 4D ).

A TUNEL assay was performed to further investigate whether apoptosis occurred in the mice skin, and positivity was observed in the epidermis 24 h after anti-DP-C-IgG injection ( Figure 4E ), but not in the skin after being injected with normal human IgG ( Figure 4F ).