NAMPT and NNMT released via extracellular vesicles and as soluble mediators are distinguished traits of BRAF inhibitor resistance of melanoma cells impacting on the tumor microenvironment

Abstract

Drugs targeting mutant BRAF and MEK oncogenes are effective in melanoma, even though resistance rapidly develops. This complex picture includes acquired intrinsic tumor and tumor microenvironmental-mediated mechanisms. Here we show that melanoma cells resistant to BRAF inhibitors (BRAFi) overexpress the rate-limiting enzymes involved in nicotinamide (NAM) metabolism nicotinamide phosphoribosyltransferase (NAMPT) and nicotinamide N-methyltransferase (NNMT). Remarkably, these cells release NAMPT and NNMT both in the free-form or loaded into extracellular vesicles (EVs). NAMPT is emerging as a key mediator of resistance to BRAFi in melanoma, primarily due to its established role in NAD biosynthesis. Although previously identified as a soluble extracellular factor in this tumor, its presence within EVs released by melanoma cells has not been reported until now, highlighting a previously unrecognized mechanism through which NAMPT may influence the tumor microenvironment (TME). NNMT was revealed to increase in melanoma lesions compared to benign nevi. Here, we report for the first time its overexpression in resistant melanoma cell lines at intracellular and extracellular levels (secreted both as a soluble factor and into EVs). NNMT expression is increased in BRAF-mutated melanoma patients, suggesting a link between its upregulation and the BRAF oncogenic signaling. Moreover, NNMT levels positively correlate with gene signatures associated with pro-inflammatory signaling, immune cell migration, and chemokine-mediated pathways. NNMT pharmacological inhibition and genetic silencing significantly reduce resistant melanoma cell growth. In addition, we found that BRAFi-resistant cells are more sensitive to NNMT inhibition, highlighting a trait of vulnerability of BRAFi-resistant melanomas. Lastly, we proposed for the first time a tetrameric NNMT:TLR4 binding model offering a plausible structural and mechanistic basis for their association. Our functional results indicated that exogenous NNMT treatment is able to trigger NF-κB pathway, one of the main TLR4-dependent signaling, sharing this cytokine-like properties with NAMPT, and opening a future deeper exploration of its functional role in the extracellular space. Overall, the identification of NAMPT and, surprisingly also NNMT, included in EVs and abundantly released from resistant melanoma cells supports the impact of these moonlighting proteins involved in nicotinamide metabolism as mediators of BRAF/MEK inhibitors resistance with tumor intrinsic and potentially tumor microenvironment-mediated mechanisms. Interfering with nicotinamide metabolism could be a valid strategy to counteract drug resistance acting on the multifactorial tumor-host interactions.

Supplementary Information: The online version contains supplementary material available at 10.1186/s12964-025-02361-2.

Keywords: Extracellular vesicles; Metastatic melanoma; NAMPT; NNMT; Resistance; Secretome; Signaling; Tumor microenvironment.

Fig. 1

Characterization of EVs isolated from melanoma cell lines. (A) Representative size distribution profiles from Nanoparticle Tracking Analysis (NTA) of EVs secreted from BRAF-mutated melanoma cell lines A375 and M14 sensitive (S) and BRAF-inhibitor resistant (BiR) are shown. The black curve represents the mean of three measurements, with standard error in red. (B) Mode, (C) mean EVs’ diameters, along with (D) EVs number per mL, are plotted. Error bars represent at least three biological replicates. Repeated measures one-way ANOVA was used to calculate statistical significance (ns- not significant, *p < 0.05) between BiR and S cells. Venn diagrams indicating the number of proteins commonly (E) up and (F) downregulated in A375 and M14 BiR, identified by proteomics, and respectively reactome enrichment analysis of (G) up and (H) downregulated proteins. All the significant modulated proteins (p-value above 0.05) compared with the BRAF inhibitors sensitive A375 and M14 (A375 S; M14 S) cell lines were considered. Four independent biological replicates were considered for statistical analysis

Fig. 2

NAMPT and NNMT are enriched in EVs from BiR cells. Volcano plots of significantly modulated proteins in (A) A375 BiR and (B) M14 BiR cell lines, compared with their respective sensitive counterparts (A375 S and M14 S). Black dots with labels for NAMPT and NNMT in each comparison are shown. Proteomics quantification of (C) NAMPT and (D) NNMT relative abundance in EVs isolated from the different cell lines. Repeated measures one-way ANOVA was used to calculate statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001) between BiR and S cells. (E) Representative western blot images of EVs positive (TSG101, Syntenin, CD9) and negative (GM130) markers, as well as (F) NAMPT and NNMT expression in EVs isolated from A375 S, A375 BiR, M14 S and M14 BiR cell lines, and the respective cell lysates. Histograms reporting (G) NAMPT and (H) NNMT relative quantification in EVs, using syntenin as loading control to normalize data [80]. Three independent biological replicates were performed. Unpaired t-test was used to calculate statistical significance (*p < 0.05, **p < 0.01, ^#p = 0.07) between BiR and S cells in each cell line

Fig. 3

NNMT is overexpressed in BiR cell lines. (A) Histograms reporting NNMT mRNA expression levels in S and BiR MM cell lines (M14, A375 and SK-MEL-28). Data from five independent experiments. (B) Analysis of NNMT expression in the same cell line variants by immunoblot. Actin was used as a loading control. Histograms show cumulative data of band quantification (at least n = 4) represented as a ratio of the enzyme/actin levels. Below the graphs are reported representative western blots. (C) NNMT expression evaluated by confocal immunofluorescence staining. Representative images of immunostaining of NNMT (red fluorescence) in (C - left) A375 S and BiR and (D-left) M14 S and BiR cell lines. Hoechst was used to stain the nuclei. Images were acquired with the Operetta instrument and Harmony 3.5.2 software. NNMT mean signal intensity in (C-right) A375 and (D-right) M14 cell lines was calculated with Harmony software. Statistical significance was calculated using unpaired t-test. Data in the graphs are presented as the mean ± SEM. Significance was represented as: **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001

Fig. 4

eNNMT is abundant secreted either in exosomes or as a free protein in the supernatants (SN) derived from BiR MM cell lines. The presence of eNNMT was confirmed by western blot performed on 10× concentrated SN from M14 (A), A375 (B) and SK-MEL-28 (C) S and BiR variants in reducing conditions. Histograms show the cumulative data of n = 4–5 independent replicates. The amount of eNNMT detected was normalized over protein content in each replicate. Below the graphs are reported representative western blots. (D-E) SNs from A375 BiR (D) and SK-MEL28 BiR (E) cells were treated or not with Proteinase K (ProK, 100 ng/ml, 1 h at 37 °C) after ultracentrifugation (100,000 g for 70 min at 4 °C). SNs were concentrated 10x and was included an input condition: SN concentrated 10x before ultracentrifugation. NNMT expression was checked by western blot as reported in the representative images. Vimentin was used as free protein positive control, while Syntenin was used as exosomal protein control. Statistical significance was calculated using unpaired t-test. Data in the Figure are presented as the mean ± SEM. Significance was represented as: *p ≤ 0.05, **p ≤ 0.01

Fig. 5

NAMPT and NNMT are found in the secretome of A375 BiR cell line. (A) Volcano plot of significantly modulated proteins in A375 BiR compared with the respective sensitive A375 S cell line. Black dots with labels for NAMPT and NNMT are shown. Proteomics quantification of (B) NAMPT and (C) NNMT relative abundance in the supernatants of A375 S and BiR cells lines. Repeated measures one-way ANOVA was used to calculate statistical significance (**p < 0.01) between BiR and S cells. Dot plots of gene ontology GO (up), KEGG (middle) and Reactome (bottom) enrichment analysis of secreted proteins (D) up- or (E) downregulated in A375 BiR compared with A375 S cell lines. Dot size indicates gene count, and color represents adjusted p-value. Only significantly modulated proteins (p-value above 0.05) from four independent biological replicates were considered for the statistical analysis

Fig. 6

NAMPT and NNMT are positively correlated and NNMT is more abundant in BRAF mutated MM patients. (A-B) Scatter plot correlating NAMPT and NNMT. Each dot represents a sample of the TCGA SKCM cohort, metastatic samples (A) or melanoma cell lines derived from CCLE dataset (B). Pearson (R) and Spearman (rho) correlations and p-value are shown. (C) NNMT expression (log2(normalized counts + 1)) was compared between BRAF-mutated and wild-type groups using the non-parametric Wilcoxon test. Data distributions were visualized using violin plots, reporting the p-value for statistical significance. (D) Analysis of NNMT expression in M14 BiR and A375 BiR after 24 h of treatment with the combination (combo) of Vemurafenib (10µM) and UO126 (10µM). M14 BiR n = 5, A375 BiR n = 4. Data are reported as n-fold compared to paired untreated condition for each replicate. Statistical significance was calculated using a paired t-test. Data in the Figure are presented as the mean ± SEM. Significance was represented as: *p ≤ 0.05

Fig. 7

NNMT correlates with inflammatory and immune-related categories. Gene set enrichment analysis (GSEA) for indicated pathways of gene expression data from TCGA melanoma transcriptomes. Samples were then stratified into two groups (NNMT_high and NNMT_low) based on the median NNMT expression. All the represented gene sets positively correlate with NNMT expression at a false discovery rate (FDR < 0.05). Enrichment score (ES), normalized enrichment score (NES)

Fig. 8

Pharmacological inhibition or genetic knockdown of NNMT reduces BiR cells growth. (A-C) Short‑term proliferation assay (CCK-8) evaluating the sensitivity of M14 BiR (A), A375 BiR (B) and SK-MEL-28 BiR (C) to the indicated increasing doses of NNMTi (72 h of treatment). NAMPT inhibitor FK866 (50 nM) was used as positive control of cell growth arrest. Data are from 3 independent experiments (each performed in triplicate) and are represented as % of proliferation (fold change over untreated condition). (D) Colony‑forming ability (colony‑forming units CFU) of A375 BiR treated with NNMTi (50–100 µM) in comparison with untreated cells for 12 days. Cells were stained with crystal violet and representative images are shown. On the right, histograms show the cumulative quantification of the percentage (%) area with colonies at the end of the 12‑days period (2 independent experiments performed in triplicates). (E-F) CCK-8 assay comparing cell growth rate between M14 BiR transfected with siCRT vs. siNNMT (E) and between stably NNMT silenced SK-MEL-28 BiR and control cells (F). Data are from 2 independent experiments (each performed in triplicate) and are represented as absorbance 450 nm optical density (OD). (G) Representative images of CFU derived from stably shNNMT SK-MEL-28 BiR vs. shSCR cells maintained in culture for 12 days. On the right the cumulative graph representing the quantification of the percentage (%) area with colonies at the end of the 12‑days period (3 independent experiments). (H-I) CCK-8 assay comparing sensitivity to NNMTi (50–100 µM, 72 h of treatment) of BiR vs. S cell lines. Data are represented as % of proliferation (fold change over untreated condition). Statistical significance was calculated using one-way ANOVA multiple comparisons (treatment groups compared with the untreated group) for A-D, unpaired t-test for E-I. Data in the Figure are presented as the mean ± SEM. Significance was represented as: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001

Fig. 9

Putative NNMT: TLR4 complex model and eNNMT signaling function. NNMT-TLR4 complex was modeled with AI (AlphaFold Multimer) and compared with TLR4:MD2. (A) Crystal structure of TLR4-MD2 complex (PDB: RFXI) shown in worm representation. TLR4 subunits are colored blue and yellow, while MD-2 molecules are purple. (B) Superposition of five predicted NNMT-TLR4 complex models, illustrating structural convergence. (C) The highest- ranked NNMT-TLR4 model is depicted in a surface representation. NNMT-TLR4 interaction sites are highlighted in pink, while the TLR4 homodimerization interface is marked in red. NNMT molecules are green. The upper and lower panels depict the top and front views, respectively. (D) Western blot analysis of TLR4 and NF-kB p-p65 in differentiated THP-1 macrophages upon treatment (30 min) with increasing doses of rNNMT (50-100-200 ng/ml). (E) Representative western blot analysis of NF-kB p-p65 in differentiated THP-1 macrophages (n = 4) upon treatment (30 min) with rNNMT and rNAMPT (both at 50 ng/ml). Cumulative graph on the right represents quantification of p-p65/p65, one-way ANOVA multiple comparisons. (F) Representative western blot analysis of basal expression of TLR4 in MM cells BiR/S variants. (G) Representative western blot analysis of NF-kB p-p65 in A375 BiR (n = 4) upon treatment (30 min) with rNNMT and rNAMPT (both at 50 ng/ml). Cumulative graph on the right represents quantification of p-p65/p65, one-way ANOVA multiple comparisons. Data in the Figure are presented as the mean ± SEM. Significance was represented as: *p ≤ 0.05