Transglutaminase Type 2-MITF axis regulates phenotype switching in skin cutaneous melanoma

Abstract

Skin cutaneous melanoma (SKCM) is the deadliest form of skin cancer due to its high heterogeneity that drives tumor aggressiveness. Melanoma plasticity consists of two distinct phenotypic states that co-exist in the tumor niche, the proliferative and the invasive, respectively associated with a high and low expression of MITF, the master regulator of melanocyte lineage. However, despite efforts, melanoma research is still far from exhaustively dissecting this phenomenon. Here, we discovered a key function of Transglutaminase Type-2 (TG2) in regulating melanogenesis by modulating MITF transcription factor expression and its transcriptional activity. Importantly, we demonstrated that TG2 expression affects melanoma invasiveness, highlighting its positive value in SKCM. These results suggest that TG2 may have implications in the regulation of the phenotype switching by promoting melanoma differentiation and impairing its metastatic potential. Our findings offer potential perspectives to unravel melanoma vulnerabilities via tuning intra-tumor heterogeneity.

Fig. 1. Analysis of the TG2 significant clinical value in TCGA cancer datasets.

a–e Overall survival based on TG2 expression level in SKCM (Skin cutaneous melanoma), LUSC (Lung squamous cell carcinoma), GBM (Glioblastoma multiforme), KIRC (Kidney renal clear cell carcinoma), and LGG (Brain lower grade glioma) was obtained through Kaplan-Meier analysis by sorting samples for high and low TG2 expression groups according to the quartile (Cutoff-High=25%; Cutoff-Low=75%) on GEPIA. Percent survival was plotted, and p-values were shown as per figure specification, respectively. f Schematic representation of the impact of TG2 expression on LUSC, GBM, KIRC, LGG, and SKCM. TG2 expression is a worst prognostic signature in LUSC, GBM, KIRC, and LGG (represented in blue), whereas it has a positive clinical value in SKCM only (in red). g Forest plot showing the detailed table of the Univariate Cox-Regression Survival Analysis of TG2 expression in LGG, LUSC, GBM, KIRC, and SKCM, retrieved using the Survival Genie Software. The plot shows the hazard ratio and 95% confidence intervals associated with the two considered groups of patients (high and low expression of TG2), along with Walt test and log-rank p-values. Cut-off values applied to the two subsets of patients and the sample number in each group are also shown. To assess the Hazard Ratio (HR) based on TG2 expression in LGG, GBM, KIRC, LUSC, and SKCM primary tumors, we used the Cutp option for the Cut-off establishment (the cut-point is estimated based on martingale residuals using the survMisc package to stratify patients into high and low groups). Squares represent the Hazard Ratio (HR), while the horizontal lines depict the upper and lower limits of the HR 95% confidence interval (arrow pointing to the upper limit indicates that the interval is higher than the maximum shown). Confidence Interval (CI). Likelihood Ratio (LR). Negative significant prognostic values are represented in blue squares, while positive associations in red.

Fig. 2. Generation and multi-omics characterization of TG2 knock-out melanoma B16F10 cells.

a Schematic representation of the employed strategy. B16F10 TG2 KO clones were generated by means of the CRISPR/Cas9 genomic editing tool. After obtaining the clones, they were subjected along with the B16F10 WT cell line to Proteomics profiling and RNA-seq analyses. b Immunoblot analyses showing the obtained TG2 KO clones, namely TG2 KO 1 and TG2 KO 2. Actin was used as loading control. c TGM2 expression evaluated by qRT-PCR analysis in B16F10 WT cells and TG2 KO clones (number of independent biological replicates = 8). Statistical significance was calculated with One-Way ANOVA and specified with asterisks (^****p < 0.0001). Data are represented as mean ± SEM. d Venn diagrams showing the differentially expressed proteins from comparative Proteomic analyses of the TG2 KO clones. Comparisons were divided in Up and Down-regulated Proteomics targets (hits and candidate proteins). Areas of overlap indicate shared protein targets. Statistically significant targets were defined based on adj.p < 0.05 (adjusted p < 0.05). Proteins were annotated as “hits” with FDR < 5% and a fold change of at least 100% and as “candidates” with FDR < 20% and a fold-change of at least 50%. e Bar plot representative of the GO enrichment analyses of the top 11 downregulated Biological Processes (BPs). Bar color represents the adj.p-value (dark blue= most significant). Bar lengths refer to the proportion of enriched proteins for each term. f Heat map of comparative proteomic analysis of the melanogenesis related proteins was generated using pheatmap R package. g Bar plot representative of the GO enrichment analyses of the top 15 upregulated Biological Processes (BPs). Bar color represents the adj.p-value (dark blue= most significant). Bar lengths refer to the proportion of enriched proteins for each term. h Heat map of comparative proteomic analysis of the migration and adhesion proteins was generated using pheatmap R package. Dot plots representative of down (i) and up (l) regulated GO of Biological Processes analyses performed on significantly differentially expressed genes (DEGs) obtained from RNAseq profiling of the TG2 KO 2 clone (FDR < 0.01). Bubble colors represent the adj.p-value (red=most significant). The rich factor refers to the proportion of enriched genes for each term.

Fig. 3. TG2 expression is required for melanogenesis in B16F10 cells.

a Schematic representation of the employed pigmentation induction protocol, adapted from Sckoniecka et al., 2021 [46]. b Pictures of B16F10 WT and TG2 KOs cell pellets (upper panel) and media color (bottom panel) showing differential melanin (dark color) retainment and secretion between the samples. c Quantitative analyses of extracellular and intracellular melanin content, expressed in (µM)/(µg/cells/mL). Extracellular and intracellular melanin content was normalized on each well protein content. B16F10 WT was used as control during statistical analysis (number of independent biological replicates = 6). d Morphological analysis of B16F10 WT and TG2 KO clones following pigmentation induction by optical microscopy. Melanin granules are indicated with black arrows. Cellular shape is highlighted in blue, red, and green in WT, TG2 KO 1, and TG2 KO 2 respectively. Particularly, in B16F10 WT sample, cells acquire the typical differentiated dendritic shape with protrusions. Conversely, B16F10 TG2 KO cells maintain the typical melanoma spindle-like shape. Scale bar = 200 μm. e Transmission electron microscopy (TEM) images of ultrathin section of B16F10 WT and TG2 KO cells showing melanin-containing granules in the cytosol with relative granules per cell quantification. A higher magnification is reported in the right part of the panel (scale bars indicated in the pictures). Melanin granules (black) are enriched in the perinuclear area of the WT cell line. Isolated and dispersed fewer granules are visible in the KO condition. f Immunoblot analyses and relative densitometry of melanogenesis-related targets (Melan-A, Tyrosinase, and DCT) in B16F10 WT, TG2 KO 1, and TG2 KO 2 cells. Vinculin was used as loading control (number of independent biological replicates = 5). Immunoblot analysis (g) and relative mRNA levels quantified by qRT-PCR analysis (h) of TG2 expression in WT samples, following (PIGM.) or not (N. PIGM.) pigmentation. β-actin was used as a loading control in both immunoblot (number of independent biological replicates = 3) and qRT-PCR (number of independent biological replicates = 5). Statistical analyses of three or more groups were performed with One-Way ANOVA. Two-way ANOVA with Bonferroni’s test was used to compare the data with two variables. Statistical significance is specified with asterisks (^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001). Data are represented as mean ± SEM.

Fig. 4. TG2 expression is required for pigmentation in human melanoma cell lines, and in vivo zebrafish and mouse models.

a Immunoblot analyses and relative densitometry of melanogenesis-related targets (TG2, Tyrosinase, and DCT) in human melanoma cell lines Mel JuSo, IPC-298 and SK-MEL-3. GAPDH was used as loading control (number of independent biological replicates = 3). Statistical significance is specified with asterisks (^*p < 0.05, ^***p < 0.001, ^****p < 0.0001). Data are represented as mean ± SEM. b Relative mRNA levels quantified by qRT-PCR analysis of TG2, DCT and TYRP1 expression in human melanoma cell lines Mel JuSo, IPC-298 and SK-MEL-3. β-actin was used as house-keeping gene in qRT-PCR (number of independent biological replicates = 3). Statistical significance is specified with asterisks (^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001). Data are represented as mean ± SEM. c Photos of zebrafish morphology at 48 hpf comparing melanophores formations in TG2 KD and Ctrl morphants and relative quantification. The zebrafish larvae were injected with 0.1 pmol of zTg2b antisense morpholino/embryo. Ctrl Morpholino (CtrlMO) was used as reference. Images were acquired with the same exposure, at 3.2X magnification. Statistical significance is specified with asterisks (^****p < 0.0001). Data are represented as mean ± SEM. d Histology of mice skin: representative light micrographs of paraffin sections from C57BL/6 WT (A) and KO (B) mice skin, stained with hematoxylin and eosin, where the surface cornified layer and the numerous hair follicle are clearly identifiable; no detectable abnormalities are present in KO (B). C, and D depict magnifications. Melanin granules are visible along the dendritic extensions of melanocytes (arrowheads) are found in WT skin (shown in C). In TG2 KO skin melanin granules are mostly found in the perinuclear region of melanocytes (arrowheads) (shown in D). Scale bar: A, B = 150 µm; C, D = 12.8 µm.

Fig. 5. TG2-MITF interaction is required for MITF nuclear translocation.

Relative mRNA levels quantified by qRT-PCR analysis of MITF expression in B16F10 WT, TG2 KO 1 and TG2 KO 2 (a) and in human melanoma cell lines Mel JuSo, IPC-298 and SK-MEL-3 (b). β-actin was used as house-keeping gene in qRT-PCR (number of independent biological replicates = 3–5). c Immunoblot analysis of MITF in B16F10 WT, TG2 KO 1 and TG2 KO 2. Vinculin was used as loading control (number of independent biological replicates = 3). d Cytosolic-nuclear fractionation assay and relative densitometric analyses evaluating the expression and localization of MITF and TG2 in WT and KO clones, following (PIGM.) or not (N. PIGM.) pigmentation induction. Vinculin and Lamin C were used as loading controls, respectively marking the cytosolic and the nuclear fractions (number of independent biological replicates = 3). e In situ Proximity Ligation Assay (PLA) showing the interaction between MITF and TG2 in B16F10 WT and TG2 KO conditions, following or not pigmentation induction. Each red spot represents a single interaction. DNA was stained with DAPI (in blue). Quantification of dots per cells is represented in the graph on the right. Statistical analyses were performed with One-Way ANOVA and specified with asterisks (^**p < 0.01, ^***p < 0.001, ^****p < 0.0001). Data are represented as mean ± SEM.

Fig. 6. Characterization of TG2 KO melanoma tumorigenic potential in vitro and in vivo primary tumors and metastatic formations.

A Colony formation assay with quantification of the number of colonies per sample (number of independent biological replicates = 5). The number of colonies was assessed with ImageJ (One-Way ANOVA, ^*p < 0.05). B Growth curve comparing the proliferation rate (expressed in growth percentage/hours) between B16F10 WT and the two TG2 KO clones. Parental B16F10 and Cas9-transfected cells displayed the same proliferation rate (not shown, number of independent biological replicates = 3). C–E Analysis of in vivo primary tumor growth from C57BL/6 WT orthotopic mice models after injection with the indicated cell lines. 4 animal models were injected for each group. Excised tumors are reported (C), showing a difference in pigmentation relatable to the different melanin content between TG2 WT and KO clones. Tumors were measured daily to assess the growth volume (D) and were weighted after the excision (E). F Analysis of lung experimental metastasis formation in C57BL/6 mice induced from B16F10 WT and TG2 KO tail vein injection. A picture of the front and back of mice lungs is reported for each condition. White arrows point to the experimental metastatic processes. G Multiplex IHC on lung experimental metastasis tissue of C57BL/6 mice induced from B16F10 WT and TG2 KO tail vein injection and relative tumor area quantification. Melanoma cells infiltration in tissues was visualized by anti-Melan-A staining (in yellow). DNA was stained with DAPI (in blue). Multiple 4 µm sections from 4 mice per conditions were used for the statistical analyses. Statistical analyses of three or more groups were performed with One-Way ANOVA. Two-way ANOVA with Bonferroni’s test was used to compare the data with two variables. Statistical significance is specified with asterisks (^*p < 0.05). Data are represented as mean ± SEM.

Fig. 7. Schematic representation of the hypothesized working model.

a TG2 expression increases during pigmentation of melanoma cells. Also, following pigmentation, TG2 interaction with MITF allows the transcription factor nuclear translocation and its subsequent transcriptional activation by synthesis of the melanogenesis-related genes (Tyr, Dct, Melan-A, etc), which allow intracellular melanin synthesis and extracellular secretion. Thanks to TG2, MITF^High levels enable the maintenance of the melanocytic/differentiated state. b Loss of TG2 inhibits correct MITF nuclear translocation, contributing to a downregulation of melanogenesis and melanoma de-differentiation. Loss of TG2 increases MITF^Low/AXL^High ratio, switching melanoma cells to the mesenchymal/invasive phenotype, increasing its metastatic capacity by promoting cell motility, alteration of cell-adhesion molecules, and extracellular remodeling. Schematic representation was created with BioRender.com.