Tumors carrying BRAF-mutations over-express NAMPT that is genetically amplified and possesses oncogenic properties

FK866 was from Selleckchem (Munich, DE), and GMX1778 was from Sigma (Milan, IT).

Genomic data shown in this paper are in whole or part based upon data generated by The Cancer Genome Atlas (TCGA) Research Network (http://​cancergenome.​nih.​gov/​). Details are included in Additional file 1: Methods.

BRAF V600E-mutated cell lines were M14 and A375 (MM) and 8305C and BHT-101 (THCA), while BRAF WT lines were MEL505 and Mewo (MM) and CAL-62 and ML-1 (THCA), all purchased from DSMZ (Braunschweig, DE) or ATCC (LGC Standards S.r.l. Italy distributor Sesto San Giovanni, Milan, IT), as for NIH-3T3. Cells were cultured as described in Additional file 1: Methods.

The Tween-Myc-BRAF V600E-GFPtag lentivirus vector [21], kindly gift by Prof. Tiacci and Dr. Pettirossi (University and Hospital of Perugia, Perugia, IT), was used to induce over-expression of BRAF V600E in cells that naturally did not harbor BRAF V600E mutation. An empty Tween-GFPtag vector was used as control. GFP⁺ cells were flow sorted (FACSAriaIII, BD Biosciences) and expanded.

The previously obtained lentivirus vector for NAMPT-GFPtag [9] was used to over-express NAMPT in Mewo cell line (BRAF wt) and in NIH-3T3. An empty GFPtag vector was used as control. GFP⁺ cells were flow sorted using FACSAriaIII and expanded for further experiments.

RNA extraction and qRT-PCR were performed as described in [8], using commercially available primers (TaqMan Gene Expression Assays; Thermo Fisher Scientific, Rodano MI, IT) listed in Additional file 1: Methods.

Western blot analysis to evaluate proteins expression was performed as described previously [9]. Antibodies used were listed in Additional file 1: Methods.

Cells were cultured on glass cover slips in 24-well plates for 24 h. Staining with anti-NAMPT antibody (A300-779A, 1:200, Bethyl Laboratories, Montgomery, TX) was performed as described [9]. Slides were analyzed using a TCS SP5 laser scanning confocal microscope with an oil immersion 63× objective and acquired with LAS AF software (both from Leica Microsystems, Milan, IT). Images were processed with Photoshop (Adobe Systems, San Jose, CA). Pixel intensity was calculated using the ImageJ software (http://​rsbweb.​nih.​gov/​ij/​).

Cells (500/well) were seeded into 6-well plates and cultured for 10–12 days in a complete medium. Cells were then fixed with 4% PFA [10 min, room temperature (RT)] and stained with crystal violet (20 min, RT, in the dark). The percentage of occupied colony areas was calculated with ImageJ/Fiji software.

Cell viability was evaluated after 72 h of treatment using an MTT assay kit (Thermo Fisher Scientific). Each sample was measured in triplicate and repeated at least three times.

Statistical comparisons were performed using Graph Pad Prism version 7.0 (Graph Pad Software Inc., La Jolla, CA) Statistical significance was determined by two-tailed Mann–Whitney U or unpaired/paired Student’s t test and one-way ANOVA test. Unless otherwise indicated, data in the Figures are presented as the mean ± SEM.

Statistical TCGA analyses were performed using R v3.5.1 (http://​www.​r-project.​org) and Rstudio v1.1.463 (http://​www.​rstudio.​com/​). Correlation between gene mutation status and NAMPT expression has been calculated by both Kolmogorov–Smirnoff test for difference between the cumulative distributions and Mann–Whitney test for difference in the expression rank.

To highlight a possible association between NAMPT expression and BRAF mutations in human cancers, we analyzed the TCGA database showing that BRAF-mutated tumors (MUT, in red) display significantly higher levels of NAMPT, as reported by cumulative distribution function (P = 3.94 × 10^–12, Fig. 1A). This observation was obtained globally considering all cancers, notwithstanding intrinsic tumor heterogeneity (Additional file 1: Fig. S1A). We then focused our analysis on melanoma [Skin Cutaneous Melanoma (SKCM) TCGA cohort] and THCA, as tumors with significant percentages of BRAF mutations (Additional file 1: Fig. S1B) [25, 26]. All annotated BRAF mutations were considered, with V600E being the most frequently represented (131/186, 70% for SKCM and 242/254, 95% in THCA). In the case of SKCM 20 cases (11%) were characterize by the V600K, a second activating mutation, while the remaining 35 cases (19%) presented with other BRAF mutations (Additional file 1: Fig. S1C). In some instances (12/254, 5% for THCA and 16/186, 9% for SKCM), the tumors displayed additional mutations in the RAS pathway (NRAS, NF1, and CKIT) and were discarded from the analyses presented in Fig. 1A. In both these tumor types, NAMPT expression was significantly higher in patients harboring BRAF mutations [SKCM: 170 BRAF-MUT vs. 63 BRAF-WT, P = 0.0006, and THCA: 242 BRAF-MUT vs. 127 BRAF-WT, P = 10^–15 (Fig. 1B)].

The positive correlation between NAMPT expression and BRAF mutations in patients was then confirmed at protein level by comparing NAMPT staining in BRAF-WT and BRAF-mutated samples in MM tissue microarrays (TMA) and THCA whole slide sections.

In a MM cohort including 68 BRAF-WT and 97 BRAF-MUT tissue samples, a statistically significant increase of NAMPT staining, expressed as H score, was highlighted in BRAF-mutated cases (P < 0.05, Fig. 2A, left graph). Notably, when considering only the true WT cases (n = 28, i.e., excluding those with NRAS and c-KIT mutations), statistical significance markedly increased in all comparisons [28 WT cases vs. 97 BRAF-MUT (n = 97, P = 0.0008), or vs. NRAS-MUT (n = 37, P = 0.01), or vs. c-KIT-MUT (n = 3, P = 0.005), or vs. a combination of all mutations in BRAF, NRAS and c-KIT (n = 137, P = 0.0007)] (Fig. 2A, graph on the right). IHC analysis of NAMPT expression highlighted a strong reactivity for NAMPT in cases harboring these mutations (Fig. 2A, IHC representative images).

In THCA, only BRAF V600E mutation was analyzed, as it is the most commonly observed genetic abnormality in this tumor [27]. Consistent with MM samples, NAMPT staining was significantly increased in BRAF-MUT (n = 56) compared to BRAF-WT (n = 28) THCA cases (P = 0.0007, Fig. 2B).

Overall, these results demonstrate a positive and statistically significant correlation between BRAF mutational status and NAMPT mRNA and protein expression levels in patients’ tumors. In addition, mutations in other genes that insist on the MAPK/ERK pathway, such as NRAS and c-KIT, also lead to NAMPT over-expression.

To have a formal validation of the direct relationship between over-activation of the BRAF oncogenic pathway and NAMPT expression, two BRAF-WT MM cell lines (MEL505 and Mewo, Fig. 3A, B) and two BRAF-WT THCA cell lines (CAL-62 and ML-1, Fig. 4A, B) were infected with a lentivirus carrying the BRAF V600E construct, leading to the expected constitutive activation of the pathway, with increased baseline ERK phosphorylation (p-ERK) levels (Figs. 3 and 4). As demonstrated by RNA and protein analysis, NAMPT expression in these cells was significantly increased compared to control cells infected with an empty vector (Figs. 3 and 4). Up-modulation was specific for NAMPT, while the other three NBEs, i.e., NAPRT, NMRK1, and QPRT, were unaffected by infection with BRAF V600E (Figs. 3 and 4, heatmaps on the right). These findings indicate that the BRAF oncogenic pathway is directly responsible for NAMPT transcriptional and protein over-expression.

We then asked whether the direct impact of the BRAF oncogenic pathway on NAMPT expression could reflect a different sensitivity to NAMPTi between WT and BRAF-mutated cell lines. To understand if BRAF-mutated tumors rely on NAMPT activity and could therefore respond to NAMPTi, we compared the effects of two known NAMPTi (i.e., FK866 and GMX1778) in the BRAF-mutant vs BRAF-WT cell lines. Dose–response proliferation assays in BRAF-mutated MM cell lines (M14 in orange, A375 in red) and in BRAF-mutated THCA (BHT-101 in blue, 8305C in light blue), as shown in Additional file 1: Fig. S2A, demonstrated a significant response in term of reduction of cell viability in response to both NAMPTi (at doses < 10 nM) after 72 h in all BRAF-mutated cell lines compared to control cells. Selecting an intermediate dose of both NAMPTi (i.e., 25 nM), also previously employed in MM model [9], we obtained more sensitivity to NAMPTi in MM cell lines (M14, A375) carrying BRAF mutations than in the WT ones (MEL-505, Mewo, Fig. 5A, upper graphs). The same results were obtained in THCA cell lines (8305C, BHT-101 vs. ML-1). Unexpectedly CAL-62 cell line showed an high sensitivity to both NAMPTi (Fig. 5A, lower graphs), even if it is BRAF-WT. Analyzing differences in the mutational status of this THCA cell line compared to the other we noticed that CAL-62 cells carry an activating mutation in the KRAS (G12R, c.34G>C) oncogene, the third most common KRAS mutation in pancreatic ductal adenocarcinoma (PDAC) [28]. This kind of mutation belongs to the genetic alterations in epidermal growth factor receptor (EGFR) pathway, like BRAF and NRAS, suggesting that also different mutations in upstream genes that insist on the same oncogenic signaling pathway may influence responses to NAMPTi, making these cells highly sensitive to NAMPT targeting.

Consistently, using BRAF-WT MM cells infected with a lentivirus carrying the BRAF V600E construct, sensitivity to NAMPTi is markedly increased (red vs. white bars, Fig. 5B, C).

NAMPT over-expression is a feature of several cancers, not only the BRAF-mutated ones. Moreover, the observation that NAMPT-regulated NAD metabolism impact many tumors’ cellular pathways is well established by several studies in these years [3, 5]. However, the role of NAMPT in tumorigenesis is not fully understood, prompting the question of whether NAMPT itself has oncogenic properties.

Gain-of-function mutations or genetic amplification convert proto-oncogenes into oncogenes. Therefore, we looked for NAMPT mutations and copy number alterations (CNA) interrogating the TCGA datasets. Very few mutations in NAMPT were found in all tumors, less than 1%, confirming previous results [29]. On the contrary, we revealed a significant NAMPT CNA at the 7q22 locus where NAMPT mapped in all tumor samples, assessed by array CGH and measured as Segments of Gain Or Loss (SGOL) score (Fig. 6A, P < 2.2e−16). NAMPT CNA positively correlates with its expression, across 33 different tumor types. Notably, the average NAMPT SGOL score across all tumor types in the TCGA cohort is always positive (Additional file 1: Fig. S3A), suggesting a general tendency toward amplification. NAMPT expression levels (Additional file 1: Fig. S3A, lower panels) parallel its CNA levels (Additional file 1: Fig. S3A, higher panel).

Focusing again on melanoma, we observed amplification of the whole chromosome 7 in melanoma samples, at variance with normal subjects (black line) showing no CNA in that chromosome, in line with previous results [30]. Interestingly, lower SGOL scores were more frequently observed in metastatic samples (red line), and higher scores were more frequently found in primary melanoma samples (blue line, P < 2.2e−16), as depicted in Fig. 6B, suggesting that amplification of chromosome 7 is an early event in melanoma development. Analyzing only NAMPT gene, it is evident that it is more frequently subjected to gain/amplification in melanoma compared to control-matched samples (Fig. 6C). No statistical differences were observed in the NAMPT SGOL score between primary and metastatic samples (data not shown). These observations are consistent with the possibility of a gain-of-function role for NAMPT in tumors, and specifically in melanoma.

To directly determine the pro-tumorigenic effect of NAMPT we over-expressed the enzyme in the melanoma Mewo cell line (BRAF-WT) and in NIH-3T3 (murine immortalized fibroblasts), very commonly exploited in this kind of experiments [31]. When testing the impact of NAMPT over-expression on their growth capacity, we observed that both Mewo and NIH-3T3 cells with NAMPT over-expression grew faster, as measured using the MTT assay (Fig. 7A). Loss of contact inhibition and high saturation density leading to uncontrolled proliferation are well-known features of transformed or malignant cells. In addition, by using colony-forming assays, we demonstrated that NAMPT over-expression leads to copious focus (colony-forming units, CFU) formation in Mewo and NIH-3T3 cells (Fig. 7B).

Together, these data suggest that NAMPT over-expression positively influences the capacity of melanoma cells to grow even independently from the mutations on the key oncogene BRAF. Additionally, our results on fibroblasts, paved the way to further studies to consolidate the notion to consider NAMPT a novel proto-oncogene in tumor.