Desmoglein 2 modulates extracellular vesicle release from squamous cell carcinoma keratinocytes

Abstract

Extracellular vesicles (EVs) are nanoscale membrane-derived vesicles that serve as intercellular messengers carrying lipids, proteins, and genetic material. Substantial evidence has shown that cancer-derived EVs, secreted by tumor cells into the blood and other bodily fluids, play a critical role in modulating the tumor microenvironment and affecting the pathogenesis of cancer. Here we demonstrate for the first time that squamous cell carcinoma (SCC) EVs were enriched with the C-terminal fragment of desmoglein 2 (Dsg2), a desmosomal cadherin often overexpressed in malignancies. Overexpression of Dsg2 increased EV release and mitogenic content including epidermal growth factor receptor and c-Src. Inhibiting ectodomain shedding of Dsg2 with the matrix metalloproteinase inhibitor GM6001 resulted in accumulation of full-length Dsg2 in EVs and reduced EV release. When cocultured with Dsg2/green fluorescence protein-expressing SCC cells, green fluorescence protein signal was detected by fluorescence-activated cell sorting analysis in the CD90⁺ fibroblasts. Furthermore, SCC EVs activated Erk1/2 and Akt signaling and enhanced fibroblast cell proliferation. In vivo, Dsg2 was highly up-regulated in the head and neck SCCs, and EVs isolated from sera of patients with SCC were enriched in Dsg2 C-terminal fragment and epidermal growth factor receptor. This study defines a mechanism by which Dsg2 expression in cancer cells can modulate the tumor microenvironment, a step critical for tumor progression.-Overmiller, A. M., Pierluissi, J. A., Wermuth, P. J., Sauma, S., Martinez-Outschoorn, U., Tuluc, M., Luginbuhl, A., Curry, J., Harshyne, L. A., Wahl, J. K. III, South, A. P., Mahoney, M. G. Desmoglein 2 modulates extracellular vesicle release from squamous cell carcinoma keratinocytes.

Keywords: EGFR; HNSCC; exosome; fibroblast.

Figure 1.

Intracellular domain of Dsg2 localizes to EVs. A) EVs from cultured medium of A431 cells grown in 10-cm² dishes were isolated by sequential ultracentrifugation or ExoQuick polymer precipitation and subjected to NTA. Screenshot shows light scatter of particles. B) Representative NTA profile showing concentration and size of particles. Both methods resulted in similar particle size (mode: 134 and 127 nm) and count (1.4 × 10⁸ and 1.2 × 10⁸ particles). C) Proteins (1 µg) from A431 total cell lysate (TCL) and EVs were resolved over SDS-PAGE and immunoblotted for markers of EVs (CD63 and CD9), mitochondria (COX IV), cis-Golgi (GM130), GAPDH, and Dsg2 (antibodies 10D2 and H-145). Multiple lower-molecular-weight fragments recognized by H-145 are most likely proteolytic fragments of Dsg2. D) A431 cells were transduced with retroviruses encoding for GFP-labeled Dsg2. EVs were isolated from conditioned medium and visualized under fluorescent microscopy. Scale bars, 50 µm. E) Immunoblotting of A431-Dsg2/GFP TCL and EV for Dsg2 using H-145 and 10D2 antibodies. GAPDH is representative of loading for both Dsg2 antibody blots.

Figure 2.

Dsg2 modulates EV biogenesis. A) A431 and HaCaT cells were transduced with retroviruses encoding for GFP (A431-GFP and HaCaT-GFP) or GFP-labeled Dsg2 (A431-Dsg2/GFP and HaCaT-Dsg2/GFP). Cells were grown to confluency and switched to serum-free medium for 48 h. Cells were lysed for Western blot analysis, showing, in addition to endogenous Dsg2 protein (lower band), expression of GFP-tagged Dsg2 (higher band) in Dsg2/GFP cells. B) EVs were isolated from conditioned medium and subjected to counting by NTA or protein concentration determination by BCA assay; n = 3. C) A431 cells were transduced with lentivirus carrying shRNA targeting GFP or human Dsg2. A431-shGFP and -shDsg2 cells were lysed and proteins immunoblotted for Dsg2 and GAPDH, showing down-regulation of Dsg2. D) EVs were isolated from conditioned medium of A431-shGFP and -shDsg2 cells and subjected to counting by NTA or protein assay (BCA assay), showing that down-regulation of Dsg2 decreased EV protein content; n = 3. E) Total cell lysates (TCL) and EVs collected from conditioned medium of normal primary human keratinocytes were immunoblotted for Dsg2 with intracellular epitope-targeting antibody H-145. Full-length (∼170 kDa) and fragment of ∼110 kDa, most likely from ectodomain shedding, were observed in EVs alongside ∼65-kDa Dsg2-CTF (arrow). F) A431 cells overexpressing Dsg2/GFP were incubated with DMSO or GM6001 (20 µM) for 48 h. EVs were isolated from conditioned medium and immunoblotted using Dsg2 Ab H-145, showing presence of GFP-tagged full-length Dsg2 protein in EVs (arrow). G) A431 SCC cells were incubated with DMSO or GM6001 (20 µM) for 48 h. EVs were isolated from conditioned medium and counted by NTA, showing that inhibition of metalloproteinases decreased EV release. *P < 0.05, **P < 0.01.

Figure 3.

Cav1 suppresses EV release. A) A431- and A431-overexpressing Dsg2/GFP cells were lysed; proteins were resolved over SDS-PAGE and immunoblotted for Dsg2, Flot1, Cav1, and actin (loading control). Signals were normalized to respective actin immunoblots and averaged; bars are expressed as percentage of Flot1/Cav1 in A431 cells. **P < 0.01; n = 3. B) A431-GFP and A431-Dsg2/GFP cells were seeded, immunostained for Dsg2 and Cav1, and imaged by confocal microscopy. Note reduced Cav1 level in cells overexpressing Dsg2. Scale bars, 10 μm. C) A431 cells were transiently transfected with siCtrl or pool targeting Cav1 (siCav1). Total cell lysates were immunoblotted for Cav1 and GAPDH to illustrate Cav1 knockdown. D) EVs were isolated from 48 h conditioned medium after confluence of siRNA-transfected A431s. EVs were subjected to counting by NTA or protein concentration determination (BCA assay); values were normalized to number of cells on plate and are expressed as percentage of control; n = 3. Transient knockdown of Cav1 was sufficient to increase both number and protein associated with secreted EVs; n = 3. *P < 0.05.

Figure 4.

Dsg2 is highly expressed in HNSCCs and partitions into EVs in sera of patients with HNSCC. A) Consecutive sections of normal human oral mucosa (OM) and HNSCC were immunostained for Dsg2 (left), normal IgG control (middle), and hematoxylin and eosin (right). Scale bars, 50 μm. B) EVs were isolated from sera of healthy individuals and patients with HNSCC (n = 3 each) by polymer precipitation and subjected to immunoblotting for Dsg2 (H-145 and DG3.10), EGFR, and actin for loading control. Signals corresponding to Dsg2 CTF using Dsg2-specific antibodies H-145 and DG3.10 were normalized to actin immunoblots and presented as percentage control.

Figure 5.

Internalization of Dsg2-containing EVs by fibroblasts. A) Primary human fibroblasts were incubated with PKH26-labeled A431 EVs (red), immunostained for β tubulin (green), and imaged by confocal microscopy. Z stack demonstrates EV uptake by recipient cell; insets show enlarged labeled EV, delineated by crosshairs. Cells treated with PBS-treated PKH26 dye did not have any intracellular accumulation of dye (Supplemental Fig. S3A). Nuclei were stained with DAPI (blue). Scale bar, 20 μm. B) Fibroblasts were cocultured with either A431-GFP or A431-Dsg2/GFP keratinocytes for 72 h to allow passive transfer of EVs. Cells were incubated with phycoerythrin-conjugated CD90 antibody to label fibroblasts. FACS analysis gated live (top left) CD90⁺ cells (top right)—the fibroblast population. GFP signal intensity was measured in gated fibroblasts to show uptake of GFP or Dsg2/GFP from keratinocytes (bottom). Median fluorescence intensities (MFI) for GFP of fibroblasts and A431-Dsg2/GFP, when cultured alone, were 2 and ∼1300, respectively. Fibroblasts in coculture with A431-Dsg2/GFP had MFI of 130.

Figure 6.

Dsg2-EVs enhance fibroblast cell growth. A) Primary human fibroblasts were treated for 48 h with EVs (10 μg/ml) isolated from conditioned medium of A431-GFP and A431-Dsg2/GFP cells. Cell proliferation was measured by WST-1 assay and normalized to untreated control; n = 6. Two-way ANOVA, Sidak multiple comparisons correction, Tukey’s HSD post hoc test. B) Fibroblasts were treated for 48 h with EVs (10 µg/ml) isolated from conditioned medium of A431-shGFP and A431-shDsg2 cells. Cell proliferation was measured by WST-1 assay and normalized to untreated control; n = 6. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 7.

SCC EVs contain panel of mitogenic proteins and stimulate mitogenic pathways in recipient fibroblasts. A) Total cell lysate (TCL) and EVs from A431-GFP and A431-Dsg2/GFP cells were immunoblotted for Dsg2, EGFR, c-Src, GM130, COXIV, CD63, CD9, and GAPDH. c-Src blot is overexposed to show equal amounts of cellular c-Src. B) A431-GFP and A431-Dsg2/GFP EV-associated EGFR and c-Src were quantified and normalized to GAPDH; data are expressed as percentage of control; n = 3. *P < 0.05, **P < 0.01. C) Primary human fibroblasts were treated with EVs from A431-GFP or A431-Dsg2/GFP cells for up to 24 h. Cells were washed, lysed, and immunoblotted for total and phosphorylated EGFR (Y1173), c-Src, ERK1/2, and AKT, along with GAPDH for equal loading. D) P-AKT signal was quantified at each time point for A431-GFP and A431-Dsg2/GFP EV-treated fibroblasts, normalized to GAPDH, and expressed as percentage of signal at 0 h time point. *P < 0.05; n = 4.