Podoplanin drives dedifferentiation and amoeboid invasion of melanoma

Abstract

Melanoma is an aggressive skin cancer developing from melanocytes, frequently resulting in metastatic disease. Melanoma cells utilize amoeboid migration as mode of local invasion. Amoeboid invasion is characterized by rounded cell morphology and high actomyosin contractility driven by Rho GTPase signalling. Migrastatic drugs targeting actin polymerization and contractility are therefore a promising treatment option for metastatic melanoma. To predict amoeboid invasion and metastatic potential, biomarkers functionally linked to contractility pathways are needed. The glycoprotein podoplanin drives actomyosin contractility in lymphoid fibroblasts and is overexpressed in many cancers. We show that podoplanin enhances amoeboid invasion in melanoma. Podoplanin expression in murine melanoma drives rounded cell morphology, increasing motility, and invasion in vivo. Podoplanin expression is increased in a subset of dedifferentiated human melanoma, and in vitro is sufficient to upregulate melanoma-associated marker Pou3f2/Brn2. Together, our data define podoplanin as a functional biomarker for dedifferentiated invasive melanoma and a promising migrastatic therapeutic target.

Keywords: Cell biology; Molecular biology; Transcriptomics.

Graphical abstract

Figure 1

Podoplanin expression is increased in a subset of human melanomas (A) Brightfield (upper panel) and immunofluorescence (lower panel) imaging of WM983A (left) and WM983B treated with vehicle (DMSO; middle) or ROCK inhibitor (1μM GSK269962A) for 4h (top) or 24h (bottom). Cells are stained for F-actin (white). The scale bars represent 100 (upper panel) or 40 (lower panel) microns. (B) Cell morphology (roundness index) of WM983A (light grey) and WM983B (dark grey) cells as presented in panel (A) Kruskal-Wallis test with Dunn's multiple comparisons, ∗∗p = 0.0036, ∗∗∗∗p < 0.0001. (C) Western blot analysis of ppMLC and MLC in WM983A (left) and WM983B treated with vehicle (DMSO; middle) or 1μM ROCK inhibitor (GSK269962A) for 4h (top) or 24h (bottom). Tubulin and GAPDH are used as loading controls. (D) Podoplanin (PDPN) mRNA expression in WM983A and WM983B cell lines measured by qPCR. PDPN mRNA expression is calculated as relative expression of the housekeeping gene GAPDH (2ˆ-Ct(PDPN)/2ˆ-Ct(GAPDH)). Data shown as mean with dots representing n = 3 biological replicates. Unpaired t test, two-tailed, p = 0.1002. (E) Podoplanin (PDPN) mRNA expression in two datasets (Kabbarah et al.GSE46517 and Riker et al.GSE7553) of human melanoma and appropriate control tissues. Data shown as mean with dots representing melanoma tumors from individual patients. Kabbarah et al.: unpaired t test, two-tailed, ∗p = 0.0250; Riker et al.: Mann Whitney test, two-tailed, p = 0.5052. (F) Podoplanin (PDPN) mRNA expression in human melanoma with BRAF and/or NRAS or no (wt) mutation in TCGA dataset (https://www.cancer.gov/tcga). Data are represented as median ± interquartile range.

Figure 2

Podoplanin controls contractility of melanoma cells in vitro and in vivo (A) Expression of podoplanin (Pdpn) mRNA transcript variant 1 (T1) and 2 (T2) in lymph node (LN) stromal cells cultured ex vivo for 3 days (left), and PDPN⁺ (red) and podoplanin knock-out (PDPN KO; blue) B16F10 murine melanoma cell lines (right). mRNA expression is calculated as fold change and normalized to Pdgfra (LN stromal cells) or Gapdh (B16F10) expression. Data shown as mean with dots representing n = 6 (LN stromal cells) or n = 3 (B16F10) biological replicates. (B) Analysis of podoplanin surface expression in PDPN⁺ (red) and PDPN KO (blue) B16F10 cell lines by flow cytometry. Cells not stained with antibody (no Ab; grey) are used as negative control. gMFI = geometric mean fluorescence intensity. (C) Left: Immunofluorescence of F-actin (white) in PDPN⁺ (top) and PDPN KO (bottom) B16F10 cell lines, labeled with respectively mOrange (red) or CFP (blue). Maximum z stack projections of representative images from n = 3 biological replicates are shown. The scale bars represent 100 microns. Right: Cell area (in μm²; top) and perimeter (in μm; bottom) of PDPN⁺ (red) and PDPN KO (blue) B16F10 cells. Dots represent single cells. n = 77–90 cells collated from 3 biological replicates. Error bars represent median with interquartile range. Mann–Whitney test, two-tailed, ∗∗∗∗p < 0.0001. (D) Left: Immunofluorescence imaging of F-actin (white) in mixed PDPN⁺ (red) and PDPN KO (blue) B16F10 tumor 9 days post-injection. The scale bars represent 100 (top) or 50 (bottom; zoom) microns. Right: Cell area of PDPN⁺ (red) and PDPN KO (blue) B16F10 cells in the tumor. Dots represent single cells (n = 97–197 cells). Error bars represent median with interquartile range. Mann–Whitney test, two-tailed, ∗∗∗∗p < 0.0001.

Figure 3

Podoplanin dependent transition to amoeboid dissemination in melanoma cells (A) Proportions of PDPN⁺ (red) and PDPN KO (cyan) cell populations within mixed B16F10 tumors. 7 individual sections across 2 different tumors were analyzed using QuPath software (Figure S5). (B) Enrichment graph showing proportions of PDPN⁺ B16F10 cells in the invasive front (IF) compared to overall proportions within mixed PDPN⁺/PDPN KO whole tumors. Data is shown as median with interquartile range. Multiple t-tests, ∗p = 0.024, ∗∗p = 0.003. (C) Immunofluorescence imaging of podoplanin (white) in mixed PDPN⁺ (red) and PDPN KO (blue) B16F10 tumor 9 days post-injection. Arrow heads indicate disseminated PDPN⁺ B16F10 cells. The scale bars represent 100 (top) or 50 (bottom; zoom) microns. (D) Time-lapse imaging of mixed PDPN^lo (mCherry-labeled; magenta) and PDPN-CFP (green) transfected 5555 murine melanoma tumor (see also Video S1). Dashed line indicates PDPN-CFP⁺ 5555-mCherry cell tracked over time, with zooms of surface-rendering showing rounded protrusions of PDPN-CFP⁺ 5555-mCherry cells. The scale bar represents 30 microns (top) 5 microns (lower panels 1 and 2). Time of each frame indicated in minutes. (E) Left: Velocity of PDPN^lo (magenta) and PDPN-CFP (green) transfected 5555-mCherry cells in vivo. Mann Whitney test, two-tailed, ∗∗p = 0.0013. Right: Percentage of static or motile PDPN^lo (magenta) and PDPN-CFP (green) transfected 5555-mCherry cells.

Figure 4

Overexpression of podoplanin drives amoeboid blebbing (A) Time-lapse imaging of 5555 cells transfected with PDPN-CFP (magenta) and LifeAct (green). The yellow dashed line is representative for the line plot used to quantify PDPN-CFP and LifeAct intensity as shown in panel (B) The scale bar represents 2 microns. (B) PDPN-CFP (left) and LifeAct (right) expression (as mean fluorescence intensity (MFI)) at timepoints indicated in panel (A) Expression is measured at three lines (length = 3.35 microns; example in panel A (yellow dashed line)) through the bleb. Analysis of one representative bleb is shown. Data shown as mean ± SD. (C) Podoplanin and pMLC (left) or pERM (right) expression (as mean fluorescence intensity (MFI) in 5555 untreated cells (control; pink triangles), or 5555 cells treated with DOX (to induce podoplanin expression) and transfected with pMLC or pERM (transfected; black dots). (D) Podoplanin (magenta), pMLC (top) or pERM (bottom) (yellow), F-actin (cyan) expression in 5555 cells treated with DOX (to induce podoplanin expression) and transfected with pMLC or pERM. Nuclei are stained with DAPI (blue). The scale bars represent 20 and 10 (zoom) microns. (E) Representative histograms and quantification of podoplanin surface expression following induction with doxycycline. Data shown as mean ± SD, N = 3 biological replicates. (F and E) Representative contour plot and quantification of percentages of apoptotic cells following podoplanin induction as shown in (E) Data shown as mean ± SD, N = 3 biological replicates.

Figure 5

Loss of podoplanin restores pigmentation and melanocyte differentiation (A and B) Tumors (A) and cells pellets (B) of PDPN⁺ and PDPN KO B16F10 cell lines. (C) Imaging of pigmentation (brightfield; left) of PDPN⁺ and PDPN KO B16F10 cell lines, labeled with mOrange (red) or CFP (blue) respectively (middle). The scale bar represents 50 microns. (D) Heatmaps showing expression (Z score) of podoplanin (PDPN) and eight dedifferentiation-associated genes in datasets of primary tumor samples of melanoma patients (top; Riker et al.GSE7553) and metastatic melanoma cultures (bottom; Mannheim cohort, GSE4843; from Hoek et al., 2006). For the Mannheim cohort, each number indicates a separate metastatic melanoma culture, and the BRAF and/or NRAS mutation status is indicated for each. (E) mRNA expression of five melanocyte-associated (Mlana/Mart-1 (p = 0.2045), Tyr (p = 0.9276), Dct (p = 0.9859), Gpnmb (p = 0.1308), Pmel (p = 0.9994); left) and three invasion-associated (Kit (p > 0.9999), Mitf (∗∗∗∗p < 0.0001), Pou3f2 (encodes Brn2; ∗∗∗∗p < 0.0001; right) genes in PDPN⁺ B16F10 cells. mRNA expression is calculated as fold change of Gapdh expression and normalized to expression in PDPN KO B16F10 cells (set at 1 as indicated by dashed line). Data shown as mean with dots representing n = 3 biological replicates. Two-way ANOVA with Sidak's multiple comparisons; p values indicated above. (F) Western blot analysis of Podoplanin, Melan-A, Tyrosinase and Brn2 (encoded by Pou3f2 gene) in PDPN KO and PDPN⁺ B16F10 cells. Tubulin and histone H3 are used as loading controls.