Figure 1.
Dsg3 and other desmosomal proteins are membrane raft associated.
Primary human keratinocytes were grown to confluence and switched to high calcium media for 16–18 hrs. Following detergent extraction (1% Triton X-100) and ultra-centrifugation on a 5–40% sucrose gradient, 12 fractions were sequentially removed from the gradient and processed via western blot. Dsg3 partitions to the buoyant raft fractions (DRMs, detergent resistant membranes) as indicated by the positive controls flotillin-1 and caveolin-1, and negative control calnexin. Desmosomal components plakoglobin (PG) and plakophilin 2 (pkp-2) were also found to be raft associated. E-cadherin, a classical cadherin of adherens junctions, is not enriched in membrane rafts.
Figure 2.
Dsg3 colocalizes with raft markers at cell-cell borders.
(A) After switching human keratinocytes to high calcium media for 16–18 hrs, surface Dsg3 was labeled live for 10 min with Alexa Fluor 555-conjugated AK15 mAb (top 2 rows) or PV IgG. Dsg3 colocalization with raft markers CD59 (a GPI-anchored protein) and caveolin-1 were compared to colocalization with clathrin, a non-raft membrane component using SIM. Dsg3 colocalized substantially with CD59, moderately with caveolin-1 and very weakly with clathrin. (B) Quantification of Dsg3 colocalization. Mander’s coefficient was used to define the ratio of red fluorescence (Dsg3) found within green (CD59, caveolin-1 or clathrin). Means ± SEM (n = 20–37 border regions); *p<0.05. Scale bar in A, 5 µm.
Figure 3.
Desmosome assembly and adhesion are cholesterol dependent.
(A) Human keratinocytes were treated with 1 mM mβCD (methyl-β-cyclodextrin) during a 3 hr switch from 50 µM to 550 µM calcium. Dsg3 was detected by labeling cells live with AK15 mAb during the last 15 min of the calcium switch. Under control conditions (no mβCD), Dsg3 and DP are recruited to cell borders. Border staining of both Dsg3 and DP is dramatically reduced in cells treated with mβCD, while border staining of adherens junction protein p120 remained similar to control. (B) Dispase-based fragmentation assay after keratinocytes were switched from a 50 µM to 550 µM calcium either in the absence or presence of 1 mM mβCD. Cells switched in the presence of mβCD showed a significant increase in the amount of fragmentation relative to control (no mβCD). Means ± SEM (n = 8 monolayers per group); *p<0.05. (C) Keratinocytes treated with varying concentrations of mβCD during a 6 hr switch from 50 µM to 550 µM were processed by sequential detergent extraction with 1% Triton X-100 and western blot to distinguish between the non-desmosomal and desmosomal pools of Dsg3 and DP. mβCD treatment caused a dose dependent shift of both Dsg3 and DP from Triton insoluble (desmosomal) to soluble (non-desmosomal) pool. (D) Quantification of Dsg3 and DP solubility changes in response to increasing mβCD concentrations. Scale bar in A, 10 µm.
Figure 4.
Dsg3 raft association increases upon calcium addition.
(A) Dsg3 and DP (desmoplakin) remain cytoplasmic when human keratinocytes are cultured in low (50 µM) calcium media. Protein staining increases at regions of cell contact when keratinocytes are cultured in high (550 µM) calcium media and desmosomes assemble. Dsg3 was detected with AK23 mAb post fixation. (B) Confluent keratinocytes cultured in 50 µM or 550 µM calcium media 16–18 hrs prior to solubilization with 1% Triton X-100 and membrane raft fractionation. Western blots were probed for Dsg3 and the raft marker flotillin-1. Dsg3 raft partitioning increases significantly upon shifting cells from low to high calcium conditions. (C) Quantification of relative Dsg3 levels normalized to total Dsg3 across all fractions. Means ± SEM (n = 3); *p<0.05. Scale bar in A, 10 µm.
Figure 5.
Dsg3 does not partition to rafts in cells lacking desmosomal proteins.
(A) In human keratinocytes (HKs) cultured in high calcium media for 16–18 hrs and A431 cells, GFP-tagged Dsg3 (top arrow) partitions to rafts similar to endogenous Dsg3. (B) Dsg3.GFP was expressed in CHO (Chinese hamster ovary) cells and HMEC-1s (immortalized human microvasular endothelial cells), cell types that do not form desmosomes. Dsg3.GFP did not partition to the raft containing fractions in either CHOs or HMEC-1s.
Figure 6.
PV IgG causes redistribution of Dsg3 into raft-containing linear arrays.
(A–C) Dsg3 colocalization with various membrane markers was analyzed using structured illumination microscopy (SIM) in human keratinocytes cultured in high calcium for 16–18 hrs and then treated with PV IgG for 3 hrs. CD59 was detected with FITC conjugated antibody by live labeling for 10 min prior to fixing. Dsg3 was detected with Alexa Fluor 555 conjugated AK15 mAb by live labeling for 10 min prior to fixing for the top two rows. For clathrin colocalization Dsg3 was monitored using PV IgG and secondary antibody detection of human IgG. In response to PV IgG, Dsg3 enters endocytic linear structures (previously termed ‘linear arrays’) that emanate perpendicular from cell-cell borders and extend toward the cell center (Jennings et al., 2011). (A) Raft markers CD59 and caveolin-1 were enriched in linear arrays and colocalizaed significantly with Dsg3 relative to the non-raft marker clathrin. (B) Fluorescence intensity measurements of lines drawn perpendicularly through linear arrays show alignment of Dsg3 (bottom line) and raft marker (top line) fluorescence. (C) Quantification of Dsg3 colocalization in linear arrays as indicated by Mander’s coefficient (ratio of red in green). Means ± SEM (n = 27–36 arrays per group); *p<0.05. (D) SIM was also used to view Dsg3 colocalization with CD59 in linear arrays in excised normal human epidermis injected with PV IgG. Basal keratinocytes are shown. D, dermis. Scale bar in A and D, 5 µm.
Figure 7.
Cholesterol depletion prevents PV IgG-induced Dsg3 redistribution and weakened adhesion.
(A) Human keratinocytes with assembled desmosomes were treated with NH or PV IgG for 3 hrs either in the absence or presence of 1 mM mβCD. Dsg3 was detected by live labeling with AK15 for 30 min on ice. Keratinocytes treated with PV IgG exhibit disrupted Dsg3 staining (surface clustering and linear array formation). mβCD treatment prevented both Dsg3 clustering and linear array formation in response to PV IgG. (B) mβCD treatment protected desmosomes against PV IgG-induced fragmentation. Means ± SEM (n = 4–8 monolayers per group); *p<0.05. Scale bar in A, 10 µm.
Figure 8.
Model for membrane rafts as platforms for desmosome regulation.
Desmosomal protein targeting to membrane rafts is required for the extensive clustering driven by cadherin ectodomain and plaque protein interactions during assembly that yields a mature and tightly packed desmosome. When adhesion is compromised (i.e. in response to PV IgG), clustering in a raft facilitates desmosomal cadherin endocytosis. PG, plakoglobin; pkp, plakophilin; DP, desmoplakin.