Advertisement

AURKA Overexpression Is Driven by FOXM1 and MAPK/ERK Activation in Melanoma Cells Harboring BRAF or NRAS Mutations: Impact on Melanoma Prognosis and Therapy

Open ArchivePublished:February 07, 2017DOI:https://doi.org/10.1016/j.jid.2017.01.021
      The cell cycle-related genes AURKA and FOXM1 are overexpressed in melanoma. We show here that AURKA overexpression is associated with poor prognosis in three independent cohorts of melanoma patients and correlates with the presence of genomic amplification of AURKA locus and BRAFV600E mutation. AURKA overexpression may also be driven by increased promoter activation through elements such as ETS and FOXM1 found within the 5′ proximal promoter region. Activated MAPK/ERK signaling pathway mediates robust AURKA promoter activation, thereby knockdown of BRAFV600E and ERK inhibition results in reduced AURKA transcription and expression. We show a positive correlation between FOXM1 and AURKA expression in three independent cohorts of melanoma patients. FOXM1 silencing decreases expression of AURKA and late cell cycle genes in melanoma cells. We further found that FOXM1 expression levels are significantly higher in tumors carrying the BRAFV600E mutation compared with the wild-type BRAF (BRAFwt). Accordingly, the knockdown of BRAFV600E also reduces the expression of FOXM1 in BRAFV600E cells. Moreover, Aurora kinase A and FOXM1 inhibition by either genetic knockdown or pharmacologic inhibitors impair melanoma growth and survival both in culture and in vivo, underscoring their therapeutic value for melanoma patients who fail to benefit from BRAF/MEK signaling inhibition.

      Abbreviations:

      CHR (cell cycle gene homology region), ERK (extracellular signal-regulated kinase), m (mutated), MAPK (mitogen-activated protein kinase), MEK (mitogen-activated protein kinase/extracellular signal-regulated kinase), sh (short hairpin RNA), shRNA (short hairpin RNA), TCGA (The Cancer Genome Atlas Network), wt (wild type)

      Introduction

      Malignant melanoma is one of the most aggressive forms of skin cancer. Acquisition of BRAF or NRAS mutations are recurrent events in melanoma (
      Cancer Genome Atlas N
      Genomic classification of cutaneous melanoma.
      ,
      • Davies H.
      • Bignell G.R.
      • Cox C.
      • Stephens P.
      • Edkins S.
      • Clegg S.
      • et al.
      Mutations of the BRAF gene in human cancer.
      ) and trigger the activation of mitogen-activated protein kinase (MAPK), leading to constitutive extracellular signal-regulated kinase (ERK) signaling and enhanced proliferation and survival. Targeted therapies with kinase inhibitors and immunotherapies produce improved rates of overall survival (
      • Chapman P.B.
      • Hauschild A.
      • Robert C.
      • Haanen J.B.
      • Ascierto P.
      • Larkin J.
      • et al.
      Improved survival with vemurafenib in melanoma with BRAF V600E mutation.
      ,
      • Hodi F.S.
      • O'Day S.J.
      • McDermott D.F.
      • Weber R.W.
      • Sosman J.A.
      • Haanen J.B.
      • et al.
      Improved survival with ipilimumab in patients with metastatic melanoma.
      ). Although most BRAFV600 -mutant melanomas are sensitive to Raf and/or MEK inhibitors, a subset of patients fails to respond to such treatments because of an intrinsic resistance (
      • Konieczkowski D.J.
      • Johannessen C.M.
      • Abudayyeh O.
      • Kim J.W.
      • Cooper Z.A.
      • Piris A.
      • et al.
      A melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors.
      ). Moreover, clinical efficacy of these drugs in sensitive melanomas are often limited by the rapid development of resistance (
      • Johnson D.B.
      • Menzies A.M.
      • Zimmer L.
      • Eroglu Z.
      • Ye F.
      • Zhao S.
      • et al.
      Acquired BRAF inhibitor resistance: A multicenter meta-analysis of the spectrum and frequencies, clinical behaviour, and phenotypic associations of resistance mechanisms.
      ,
      • Shi H.
      • Hugo W.
      • Kong X.
      • Hong A.
      • Koya R.C.
      • Moriceau G.
      • et al.
      Acquired resistance and clonal evolution in melanoma during BRAF inhibitor therapy.
      ). Thus, the development of therapies that overcome resistance to MAPK and MEK inhibitors or the identification of alternative targets remains a challenge.
      Melanomas harbor nonrandom and recurrent genomic alterations such as amplification of the 20q13 region (AURKA locus) (
      • Koynova D.K.
      • Jordanova E.S.
      • Milev A.D.
      • Dijkman R.
      • Kirov K.S.
      • Toncheva D.I.
      • et al.
      Gene-specific fluorescence in-situ hybridization analysis on tissue microarray to refine the region of chromosome 20q amplification in melanoma.
      ). AURKA encodes Aurora kinase A, which is a serine/threonine kinase involved in mitotic spindle formation, centrosome separation, and G2/M phase transition during the cell cycle (
      • D'Assoro A.B.
      • Haddad T.
      • Galanis E.
      Aurora-A kinase as a promising therapeutic target in cancer.
      ). Increased AURKA expression has been reported from nevi to primary and metastatic melanoma (
      • Pathria G.
      • Garg B.
      • Borgdorff V.
      • Garg K.
      • Wagner C.
      • Superti-Furga G.
      • et al.
      Overcoming MITF-conferred drug resistance through dual AURKA/MAPK targeting in human melanoma cells.
      ,
      • Wang X.
      • Moschos S.J.
      • Becker D.
      Functional analysis and molecular targeting of Aurora kinases A and B in advanced melanoma.
      ), and its inhibition impairs cell proliferation and migration (
      • Xie L.
      • Meyskens Jr., F.L.
      The pan-Aurora kinase inhibitor, PHA-739358, induces apoptosis and inhibits migration in melanoma cell lines.
      ). Aurora kinase A overexpression is associated with either occurrence of metastasis (
      • Reiter R.
      • Gais P.
      • Jutting U.
      • Steuer-Vogt M.K.
      • Pickhard A.
      • Bink K.
      • et al.
      Aurora kinase A messenger RNA overexpression is correlated with tumor progression and shortened survival in head and neck squamous cell carcinoma.
      ,
      • Weier H.U.
      • Mao J.H.
      Meta-analysis of Aurora kinase A (AURKA) expression data reveals a significant correlation between increased AURKA expression and distant metastases in human ER-positive breast cancers.
      ) or with shorter overall survival in nonmelanoma patients (
      • Zhang J.
      • Li B.
      • Yang Q.
      • Zhang P.
      • Wang H.
      Prognostic value of Aurora kinase A (AURKA) expression among solid tumor patients: a systematic review and meta-analysis.
      ).
      FOXM1 is a member of the forkhead family of transcription factors, the expression of which peaks at the G2/M phase of the cell cycle (
      • Alvarez-Fernandez M.
      • Medema R.H.
      Novel functions of FoxM1: from molecular mechanisms to cancer therapy.
      ). It is frequently deregulated in cancer (
      • Raychaudhuri P.
      • Park H.J.
      FoxM1: a master regulator of tumor metastasis.
      ) and found to correlate with progressive disease in melanoma (
      • Kruiswijk F.
      • Hasenfuss S.C.
      • Sivapatham R.
      • Baar M.P.
      • Putavet D.
      • Naipal K.A.
      • et al.
      Targeted inhibition of metastatic melanoma through interference with Pin1-FOXM1 signaling.
      ,
      • Miyashita A.
      • Fukushima S.
      • Nakahara S.
      • Yamashita J.
      • Tokuzumi A.
      • Aoi J.
      • et al.
      Investigation of FOXM1 as a potential new target for melanoma.
      ). FOXM1 induces the expression of genes involved in the execution of mitosis such as AURKA (
      • Sadasivam S.
      • Duan S.
      • DeCaprio J.A.
      The MuvB complex sequentially recruits B-Myb and FoxM1 to promote mitotic gene expression.
      ,
      • Wierstra I.
      The transcription factor FOXM1 (Forkhead box M1): proliferation-specific expression, transcription factor function, target genes, mouse models, and normal biological roles.
      ). Activation of the Raf-MEK-MAPK pathway is required for nuclear translocation and transactivating activity of FOXM1 (
      • Ma R.Y.
      • Tong T.H.
      • Cheung A.M.
      • Tsang A.C.
      • Leung W.Y.
      • Yao K.M.
      Raf/MEK/MAPK signaling stimulates the nuclear translocation and transactivating activity of FOXM1c.
      ).
      Although AURKA and FOXM1 expression has been reported in melanoma, the understanding of the molecular mechanisms between them and BRAF/NRAS-driven melanoma has remained quite limited. Here, we provide evidence that BRAF and NRAS mutations couple with Aurora kinase A expression by activated ERK and FOXM1. We also investigated the impact on prognosis of melanoma patients, followed by experimental therapies using Aurora kinase A and FOXM1 inhibition.

      Results

      Aurora kinase A overexpression correlates with BRAFV600E mutation and disease prognosis in primary melanomas

      We evaluated Aurora kinase A expression and the BRAFV600E mutation by immunohistochemistry in a tissue microarray containing 189 primary melanomas. Evaluation was performed in 112 of 189 tumors. Overall, 70.5% (79/112) of the tumors harbored BRAFV600E mutation, whereas Aurora kinase A cytoplasmic staining was detected in 56.3% (63/112) of tumors (Figure 1a, left panel). Tumors were scored on a scale of 1–3 according the Aurora kinase A positive-stained cells per field (Figure 1a, middle panel). Aurora kinase A expression was not significantly associated with Breslow index, distant metastasis, or regression (see Supplementary Table S1-S3 online). However, Aurora kinase A expression was more frequently observed in BRAFV600E tumors compared with BRAFwt (54.8% vs. 22.7%, P = 0.013) (Figure 1a, right panel). Although BRAF mutation did not affect patients’ overall survival, patients with no expression of Aurora kinase A showed a significantly increased survival compared with those patients with Aurora kinase A expression (153.8 months, 95% CI = 137.2–170.5 vs. 112.8 months, 95% CI = 97.9–127.7; P = 0.04), indicating that Aurora kinase A expression is a significant prognostic factor in melanoma (Figure 1b). To further confirm these findings, AURKA expression and disease outcome were evaluated in an independent melanoma dataset (GSE22155) (
      • Jonsson G.
      • Busch C.
      • Knappskog S.
      • Geisler J.
      • Miletic H.
      • Ringner M.
      • et al.
      Gene expression profiling-based identification of molecular subtypes in stage IV melanomas with different clinical outcome.
      ), and we found that Aurora kinase A-positive tumors had poor prognosis (hazard ratio = 1.27; P = 0.031) (Figure 1c). Similar findings were observed in the Cancer Genome Atlas (TCGA) Skin Cutaneous Melanoma dataset (
      • Akbani R.
      • Akdemir K.C.
      • Aksoy B.A.
      • Albert M.
      • Ally A.
      • Amin S.B.
      • et al.
      Genomic classification of cutaneous melanoma.
      ), in which, in a multivariate Cox regression analysis adjusted by tumor stage, AURKA overexpression was found to be significantly associated with poorer survival (hazard ratio = 1.29, 95% CI = 1.08–1.56, P = 0.006) (Figure 1d, left panel). A trend in the same direction was observed for disease relapse (hazard ratio = 1.15, 95% CI = 0.98–1.36, P = 0.08) (Figure 1d, right panel).
      Figure 1
      Figure 1BRAF/NRAS mutations and genomic amplification affect AURKA and Aurora kinase A expression levels, which constitute an independent marker for overall survival of melanoma patients. (a) Left panel: Representative BRAFV600E and Aurora kinase A immunostaining in melanoma. Panel 1A: BRAFV600E negative. Panel 1B: Aurora kinase A positive. Panel 2A: BRAFV600E positive. Panel 2B: Aurora kinase A negative. Middle panel: Pie chart depicting proportion of primary melanomas in scored categories of positive-stained cells for Aurora kinase A in five different fields. Right panel: Boxplots depicting the proportion of Aurora kinase A-positive and -negative melanoma tumors according to BRAF status. Numbers represent the percentage of tumors in each category. (b, c) Kaplan-Meier survival curves for melanoma patients according to Aurora kinase A expression. (b) Categorization was made according positive-stained cells for Aurora kinase A and (c) low and high levels of AURKA (NM_003600) expression from gene expression omnibus GSE22155 dataset. (d) High AURKA expression is associated with poor survival in TCGA metastatic melanoma patients. Left panel, Kaplan-Meier curves for the proportion of surviving patients with metastatic melanoma according to AURKA expression (high and low tertiles). Right panel, Kaplan-Meier curves for the proportion of relapsed cases according to AURKA expression (high and low tertiles). (e, f) Boxplots depicting the expression levels of AURKA according to BRAF/NRAS mutations (e) in our cohort of superficial spreading melanoma or (f) from gene expression omnibus GSE22155 dataset. The height of the boxes represents the interquartile range. The central horizontal lines depict the median. The top whiskers represent the 75th percentile + 1.5 × interquartile range, and the bottom whiskers represent the 25th percentile − 1.5 × interquartile range. (g) Pie charts depicting prevalence of BRAFV600E mutation and genomic gains in AURKA loci. (h) AURKA amplification and overexpression in TCGA melanomas. Left panel: Distribution of somatic copy numbers (based on normalized segment mean log2 values). The proportion of tumors classified as possibly amplified is shown for different cutoffs. Right panel: Scatterplot showing the positive correlation (Pearson correlation coefficient) between AURKA copy number and expression in primary and metastatic melanomas. The PCC and corresponding P value are shown. CN, copy number; TCGA, The Cancer Genome Atlas; PCC, Pearson correlation coefficient; WT, wild-type.
      We then assessed whether AURKA gene expression differed by NRAS/BRAF mutation status in an additional set of 50 fresh-frozen superficial spreading melanomas. BRAFV600E and NRASQ61K mutations were detected in 64.4% (29/45) and 4.4% (2/45) of tumors, respectively. (Sequencing was successful in 45 samples). Relative AURKA expression was higher in those BRAFV600E or NRASQ61K melanomas compared with wild-type BRAF/NRAS (mean ΔΔCt = 0.216 vs. mean ΔΔCt = 0.083, P = 0.013) (Figure 1e). Similar findings were detected in GSE22155 data set (Figure 1f).

      Genomic gains in 20q13 correlate with AURKA overexpression

      Next, we evaluated the presence of 20q13 gains and BRAFV600E in 84 fresh-frozen primary melanomas. Gains at 20q13 were present in 35.7% (30/84) of tumors, and BRAFV600E mutation was more frequently detected in tumors without 20q13 gains (Figure 1g). Furthermore, we observed AURKA amplification in 13% (60/470) of TCGA melanomas, 28% if an increase of log2 = 0.3 copy number was considered (Figure 1h, left panel), and a significant positive correlation between AURKA copy number and expression (Figure 1h, right panel) in this series.
      Notably, most of melanoma cell lines harbor AURKA gains and either BRAFV600 or NRASQ61 mutations (see Supplementary Table S4 and Supplementary Figure S1 online), suggesting that cell populations with both aberrations may have a selective advantage for growth in vitro.

      Activated MAPK signaling pathway mediates Aurora kinase A overexpression in melanoma cell lines

      Because analyses in tumors suggest a correlation of mutated BRAF (BRAFm) and AURKA mRNA and protein expression, we extended the analysis to a panel of melanoma cell lines characterized for 20q13 gains and BRAF/NRAS status (see Supplementary Table S4). Variable AURKA mRNA was detected in all tested cell lines, including the BRAFwtNRASwt and AURKAAmp MeWo cells (Figure 2a). Silencing of BRAFV600E expression by small interfering RNAs specific for BRAF decreased both AURKA promoter activity (Figure 2b) and Aurora kinase A levels in BRAFV600E melanoma cells (Figure 2c and d) and reduced phosphorylated ERK levels. The PLX4720 treatment (a selective inhibitor of BRAFV600E) reduced phosphorylated ERK levels and Aurora kinase A in BRAFV600E cells (Figure 2e) and increased Aurora kinase A levels in most NRASQ61 cells. Moreover, PLX4720 stimulated phosphorylated ERK in MeWo cells rather than inhibiting it, as previously described (
      • Hatzivassiliou G.
      • Song K.
      • Yen I.
      • Brandhuber B.J.
      • Anderson D.J.
      • Alvarado R.
      • et al.
      RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth.
      ). In contrast, ERK inhibition by PD98059 decreased AURKA promoter activity and diminished Aurora A protein in all cell lines regardless of BRAF/NRAS status (Figure 2b and e). These results provide support for ERK-mediated Aurora kinase A expression in melanoma.
      Figure 2
      Figure 2Activated MAPK-ERK pathway in melanoma cells drives AURKA expression. (a) Relative expression of AURKA in a panel of melanoma cell lines. NRAS/BRAF mutations and gains in 20q13 are indicated. One representative of three independent experiments is shown and includes the mean ± standard deviation of technical triplicates. (b) Histogram shows the activity of AURKA promoter in melanoma cell lines and in response to either ERK inhibitor treatment or BRAFm knockdown. Cells were harvested 24 hours after transfection with the reporter construct and cell pellets stored at –80 °C. The activity of the hpAURKA-1486luc construct relative to the empty pGL3 vector is shown. All experiments were carried out in triplicate. Data represent mean ± standard error of the mean. P < 0.05 (Student's t-distribution). (c) Representative immunoblot for Aurora kinase A and phosphorylated and total ERK proteins upon transient knockdown of BRAF in the indicated cell lines. α-Tubulin was used as a loading control. (d) Representative images of Aurora kinase A immunofluorescent staining in SK-Mel131 cells. Red signals indicate Aurora kinase A-positive cells, and DAPI staining (blue) was used as a nuclear staining. Scale bar = 50 μm. (e) Representative immunoblots for Aurora kinase A, p-ERK, and total ERK are shown for the indicated cell lines upon treatment with PLX4720 (5 μmol/L), PD98059 (1 μmol/L), or vehicle for 48 hours. α-Tubulin was a loading control. ERK, extracellular signal-regulated kinase; MAPK, mitogen-activated protein kinase; p-, phosphorylated; si, small interfering.

      Aurora kinase A depletion leads to decreased cell proliferation and survival

      AURKA knockdown efficiently decreased Aurora kinase A protein but induced both the expression and activity of ERK in most cell lines (Figure 3a).
      Figure 3
      Figure 3Inhibition of Aurora kinase A reduces cell proliferation and survival. (a) Representative immunoblots showing expression of Aurora kinase A and ERKs after transient knockdown of AURKA in the indicated melanoma cell lines. α-Tubulin was a loading control. (b) Representative immunofluorescent images from sensitive and vemurafenib-resistant melanoma cell lines upon AURKA siRNA or negative control siRNA, showing positive-cell staining for Aurora kinase A (red) and the mitotic marker PH3 (green). DAPI staining (blue) was used as a nuclei reporter. (c) Indicated melanoma cell lines were treated with 50 nmol/L MLN8237 or vehicle for 48 hours followed by immunostaining for PH3. Green signals indicate cells undergoing mitosis. Scale bars = 50 μm. (d) Cellular senescence assessed by ß-galactosidase staining (SA-ß-Gal) in the indicated melanoma cell lines upon incubation with 50 nmol/L MLN8237 or vehicle for 3 days. The total number of positive and negative cells were counted in five different fields (×10 magnification). Histograms show the fold increase ± standard error of the mean of SA-ß-Gal–positive cells relative to vehicle (SK-Mel 173 = 10.94 ± 0.77, SK-Mel 147 = 12.65 ± 1.83, WM 1366 = 10.69 ± 3.06, SK-Mel 131 = 7.96 ± 1.2, WM 1552C = 4.33 ± 0.86). Bottom panel shows a representative staining of WM-1552 senescent cells (blue) upon treatments. Scale bars = 50 μm. (e) Melanoma cells were incubated in the presence of 20–40 nmol/L MLN8237 or DMSO (vehicle) for 60 hours. Cells were lysed, and caspase-3 activity was measured. Data are mean ± standard error of the mean of three independent experiments. P < 0.05 (Student t test analysis). ERK, extracellular signal-regulated kinase; h, hour; M, mol/L; p-, phosphorylated; si, small interfering.
      We analyzed the in vitro effects of AURKA depletion on cell proliferation using NRASQ61 cells and an extended panel of BRAFV600E cell lines (here including WM-35, WM-793, and 1205LU). We found a significant reduction of cell numbers upon AURKA knockdown in all cell lines, regardless of BRAF/NRAS mutation status (see Supplementary Figure S2a online).
      BRAFV600E melanomas frequently develop resistance to BRAF signaling inhibitors; thus, we tested the benefit of Aurora kinase A inhibition on vemurafenib-resistant BRAFV600E cells. Remarkably, AURKA siRNA knockdown reduced number of mitotic cells (achieved by immunofluorescent staining with the phosphorylated histone H3 marker (
      • Hsu J.Y.
      • Sun Z.W.
      • Li X.
      • Reuben M.
      • Tatchell K.
      • Bishop D.K.
      • et al.
      Mitotic phosphorylation of histone H3 is governed by Ipl1/aurora kinase and Glc7/PP1 phosphatase in budding yeast and nematodes.
      ) in sensitive M#238 and M#249 and the vemurafenib-resistant M#238AR and M#249AR melanoma cell lines (
      • Sondergaard J.N.
      • Nazarian R.
      • Wang Q.
      • Guo D.
      • Hsueh T.
      • Mok S.
      • et al.
      Differential sensitivity of melanoma cell lines with BRAFV600E mutation to the specific Raf inhibitor PLX4032.
      ).
      A pharmacological approach was conducted using MLN8237, which is a second generation Aurora phosphorylation inhibitor with higher specificity for Aurora kinase A over Aurora kinase B. Treatments resulted in a significant reduction of cell numbers regardless of BRAF/NRAS status (up to 60% reduction at the highest concentration) (see Supplementary Figure S2b online). We observed a significant reduction of PH3-positive cells in NRASQ61H (TRP), BRAFV600E (SK-Mel131), and NRASwt/BRAFwt (MeWo) cells after Aurora kinase A inhibition (Figure 3c).
      After MLN8237 treatment, all cell cultures exhibited increased ß-galactosidase activity, indicative of senescence (Figure 3d), and caspase-3 activity (Figure 3e). These results indicate that Aurora kinase A inhibition by MLN8237 reduces melanoma cell proliferation and induces senescence and apoptosis in both BRAF and NRAS contexts as well as in vemurafenib-resistant cells (see Supplementary Figure S2 online).

      AURKA overexpression is associated with FOXM1 and BRAFV600E

      In 67 primary melanomas from tissue microarray, nuclear FOXM1 protein was observed in 63.9% of cases and was more frequently detected in Aurora kinase A-positive melanomas (80% vs. 52%; P = 0.019) (Figure 4a). To further assess the crosstalk between AURKA overexpression and FOXM1 activity, AURKA expression profiles were analyzed for an association with FOXM1 targets. Computing the Pearson’s correlation coefficient between AURKA and each other gene in the TCGA melanoma dataset, and using the Gene Set Enrichment Analysis (
      • Subramanian A.
      • Tamayo P.
      • Mootha V.K.
      • Mukherjee S.
      • Ebert B.L.
      • Gillette M.A.
      • et al.
      Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles.
      ) tool, identified FOXM1 targets (defined either by drug responses in cell cultures [
      • Gormally M.V.
      • Dexheimer T.S.
      • Marsico G.
      • Sanders D.A.
      • Lowe C.
      • Matak-Vinković
      • et al.
      2015 Suppression of the FOXM1 transcriptional program via novel small molecule inhibition.
      ] or by canonical binding sites [
      • Matys V.
      • Fricke E.
      • Geffers R.
      • Gössling E.
      • Haubrock M.
      • Hehl R.
      • et al.
      TRANSFAC: transcriptional regulation, from patterns to profiles.
      ]) as significantly coexpressed (Figure 4b). A positive correlation was also found between AURKA and FOXM1 expression by data mining in Genomics of Drug Sensitivity datasets (
      • Garnett M.J.
      • Edelman E.J.
      • Heidorn S.J.
      • Greenman C.D.
      • Dastur A.
      • Lau K.W.
      • et al.
      Systematic identification of genomic markers of drug sensitivity in cancer cells.
      ) of different cancer cell lines (Figure 4c).
      Figure 4
      Figure 4FOXM1 mediates AURKA transcriptional activation and Aurora kinase A expression. (a) Immunostaining of FOXM1 in 67 paraffin-embedded melanoma biopsy samples. FOMX1 expression was compared between those tumors showing positive staining for Aurora kinase A versus those with negative staining, as described. Significance was determined using the chi-square two-sided test. (b) Positive association between AURKA and FOXM1 target expression. Gene Set Enrichment Analysis output plots showing significant positive correlation (based on PCCs) between AURKA and FOXM1 targets: upper panel, targets based on FOXM1 inhibition and bottom panel, targets based on conserved transcription factor binding sites. The Gene Set Enrichment Analysis scores and the nominal P-values are shown. (c) Correlation of AURKA and FOXM1 mRNA expression in human cell lines from different cancer types. Data was obtained from gene expression omnibus datasets. Each point represents the log2 expression of a single cell line. The solid line presents an estimated trend of the expression of genes. The selected probe set is indicated. (d) FOXM1 mRNA expression in superficial spreading melanoma biopsy samples. Individual values of normalized expression in BRAFV600E or BRAFwt tumors are plotted. The central horizontal lines depict the median. Significance was determined using the chi-square two-sided test. (e) Representative immunoblots for Aurora kinase A, FOXM1, and phosphorylated and total ERK proteins upon transient knockdown of FOXM1 in the indicated cell lines. α-Tubulin was a loading control. (f) Left panel: schematic representation of AURKA promoter 5′ proximal (-90 nucleotides to the transcription initiation site) containing the indicated putative sites. Right panel: the AURKA reporter constructs containing either the wild type promoter sequence or the mutated sites driving the reporter gene luciferase are shown. pGL3 empty vector (control) and reporter constructs were co-transfected with phRL-CMV vector into SK-Mel131 cells, and activity was measured after 24 hours using the Dual-Luciferase Reporter Assay (Promega Corporation, Madison, WI). The activity of each construct was expressed relative to hpAURKA-90wtluc construct. Data are mean ± standard error of the mean of three independent experiments. P < 0.05, ∗∗P < 0.01 (Student t- test analysis). CDE, cell cycle-dependent element; CHR, cell cycle gene homology region; ERK, extracellular signal-regulated kinase; IHQ, immunohistochemistry; luc, luciferase; PCC, Pearson correlation coefficient; si, small interfering.
      FOXM1 expression was measured in 40 out of 50 superficial spreading melanoma primary samples, and higher FOXM1 mRNA levels were found in BRAFV600E melanomas compared with BRAFwt melanomas (Figure 4d).
      FOXM1 silencing in melanoma cells decreased Aurora kinase A regardless of BRAF/NRAS mutational status, but not phosphorylated ERK levels (Figure 4e). Moreover, silencing BRAF significantly decreased its target FOSL1 (FRA-1) and FOXM1 levels in BRAFV600E cells (Figure 5b).
      Figure 5
      Figure 5Targeting FOXM1 affects Aurora kinase A expression and melanoma growth. (a) AURKA promoter activity in human melanoma cell lines and in response to inhibitors. Cells were transfected with hpAURKA-90wtluc construct and incubated overnight with mithramycin (100 nmol/L), FDI-6 (20 μmol/L), PD 98059 (1μmol/L), or DMSO (vehicle). Histogram shows levels of luciferase activity relative to vehicle. All experiments were carried out in triplicate. Data represent mean ± standard error of the mean (Student t test analysis). Decreased levels of luciferase were significant for all inhibitors in comparison with vehicle. Remarkable differences were obtained with FDI-6 treatment (∗∗P = 0.005). (b) Histogram shows relative mRNA expression for FOXM1 and FRA-1 (FOXL1) genes in six different melanoma cell lines upon knockdown of BRAF. Data are referred to siRNA control. One representative of three independent experiments is shown and includes the mean ± standard deviation of technical triplicates (Student t test analysis). P-values < 0.05 are indicated by an asterisk. (c) Representative immunoblot showing expression of Aurora kinase A and FOXM1 in MeWo, SK-Mel147, and WM 1366 melanoma cells upon silencing FOXM1 by shFOXM1#0 sequence. α-Tubulin was a loading control. (d) Ectopic expression of Aurora kinase A upon FOXM1 silencing in SK-Mel 131 melanoma cells by transient transfection with either the plasmid pEB_AURKA-GST carrying AURKAwt or the empty vector. Cells were treated with either FDI-6, bortezomib, or vehicle for 48 hours. Cell proliferation was evaluated by water-soluble tetrazolium salts assay (WST-1; Roche Diagnostics GmbH, Mannheim, Germany) and absorbance at 450 nm was referred to vehicle-treated control cells. Shown is a representative of one out of three independent experiments, bars and standard deviations of technical triplicates (Student t test analysis). ∗P < 0.05. Upper panel: SK-Mel 131 pLKO.1. Middle panel: SK-Mel 131 shFOXM1 #5. Bottom panel: cell extracts were analyzed for the expression of GST and Aurora kinase A. ß-Actin was used as loading control. (e, f) Relative expression of FOXM1 and AURKA, respectively, in a panel of human melanoma cell lines exposed to 10 nmol/L bortezomib for 48 hours. Histograms show relative expression compared with the vehicle. All experiments were carried out in triplicate. Bars represent mean ± standard deviation. P < 0.05 was calculated by Student t test. (g) Representative immunoblots for the detection of cleaved PARP and Aurora kinase A. Cell lysates were collected from cultures exposed to or vehicle (DMSO) for 48 hours. The arrow indicates the position of cleaved fragments. α-Tubulin was used as a loading control. M, mol/L; si, small interfering; sh, short hairpin RNA; TUB, tubulin.

      The AURKA proximal promoter contains ETS- and FOXM1-binding sites

      The AURKA promoter sequence and transcription initiation site have been determined (
      • Tanaka M.
      • Ueda A.
      • Kanamori H.
      • Ideguchi H.
      • Yang J.
      • Kitajima S.
      • et al.
      Cell-cycle-dependent regulation of human aurora A transcription is mediated by periodic repression of E4TF1.
      ). Potential ETS-family–, FOXM1-, and cell cycle gene homology region (CHR)-binding sites have been predicted through in silico analysis (see Supplementary Figure S3b online). We used the hpAURKA-1486luc construct containing 1,486 base pairs of the 5′ untranslated region cloned into reporter vector pGL3 (generously provided by Y. Ishigatsubo, Yokohama, Japan) as a template to generate reporter constructs containing single, double, and triple mutations in these sites. The constructs were transfected into melanoma cells to test the importance of these putative sites in the activation of AURKA promoter.
      The hpAURKA-90luc construct retained similar activity to that of the hpAURKA-415luc, which was higher than the largest hpAURKA-1486luc (data not shown). The treatment of melanoma cells with either PD98059; FDI-6, a specific inhibitor of FOXM1 (Gormally et al., 2000); or mithramycin, an inhibitor of Sp-1 binding, resulted in reduced activity of the promoter construct (Figure 5a) and Aurora kinase A protein levels (see Supplementary Figure S3b). The proximal region spanning from -90 nucleotides to the major transcription initiation site contains a consensus sequence (GGAA) for the ETS family of transcription factors embedded in the binding sequence of the E4TF1 element (CCTTAAG), as described previously (
      • Furukawa T.
      • Kanai N.
      • Shiwaku H.O.
      • Soga N.
      • Uehara A.
      • Horii A.
      AURKA is one of the downstream targets of MAPK1/ERK2 in pancreatic cancer.
      ). The activity of the reporter in BRAFV600E cells was dramatically decreased by either the mutation or deletion of this element (Figure 4f). Just downstream of the E4TF1/ETS, a tandem repressor element containing a cell cycle-dependent element and a CHR was detected previously (
      • Tanaka M.
      • Ueda A.
      • Kanamori H.
      • Ideguchi H.
      • Yang J.
      • Kitajima S.
      • et al.
      Cell-cycle-dependent regulation of human aurora A transcription is mediated by periodic repression of E4TF1.
      ). Mutations in the CHR sequence resulted in a significant increase of the reporter activities, mainly observed in the short construct hpAURKA-75luc and the hpAURKA-90- ETSmluc in SK-Mel 131 cells harboring BRAFV600E (Figure 4f).
      Furthermore, we identified a consensus sequence (AAACA) for the FOXM1-binding site located between -10 and -15 nucleotides to the transcription initiation site. In melanoma cells, mutation of this element caused a modest but significant decrease of the reporter activities compared with ETS. To decipher the contribution of this site, we transduced the HEK 293 cells (BRAFwt and devoid of FOXM1) with a lentivirus vector expressing FOXM1wt. The exogenous FOXM1 expression increased the activity of the wild-type construct as much as 2-fold compared with Mock cells (data not shown). We found that mutation of this site caused a significant decrease of luciferase activity in HEK 293v5-FOXM1 cells compared with control cells. Furthermore, the double mutant construct (hpAURKA-75CHRm FOXM1mluc) showed lower activity in HEK 293-FOXM1 than in control cells (see Supplementary Figure S3c online), suggesting that both CHR and FOXM1 sites might be required for appropriate genomic binding by FOXM1.

      Targeted inhibition of FOXM1 reduces Aurora kinase A expression and survival in melanoma

      We silenced FOXM1 in five human melanoma cell lines, which have differing tumorigenic properties. We used three independent short hairpin RNA (shRNA) sequences for FOXM1 (hereafter called #0, #4, and #5) cloned into lentiviral vectors. Independent pools of puromycin-resistant cells were generated in each case, and the off-target effects on the promoter activity (see Supplementary Figure S4a online), mRNA, and protein were analyzed (Figure 5c, and see Supplementary Figure S4b online). The expression of the target was reduced by all shRNA sequences (sh) for FOXM1 compared with pLKO.1 control cells. Moreover, the expression of five other late cell cycle genes, previously identified as transcriptional targets of FOXM1 (
      • Aytes A.
      • Mitrofanova A.
      • Lefebvre C.
      • Alvarez M.J.
      • Castillo-Martin M.
      • Zheng T.
      • et al.
      Cross-species regulatory network analysis identifies a synergistic interaction between FOXM1 and CENPF that drives prostate cancer malignancy.
      ,
      • Müller G.A.
      • Wintsche A.
      • Stangner K.
      • Orohaska S.J.
      • Stadler P.F.
      • Engeland K.
      The CHR site: definition and genome-wide identification of a cell cycle transcriptional element.
      ), was significantly reduced for most of them (see Supplementary Figure S5 online). Ectopic expression of AURKA could restore the effects on cell proliferation induced by FDI-6 and bortezomib in both SK-Mel 131pLKO.1 and SK-Mel 131shFOXM1#5 (Figure 5d).
      We treated melanoma cultures with bortezomib, which is a proteasome inhibitor reported to target FOXM1 (
      • Bhat U.G.
      • Halasi M.
      • Gartel A.L.
      FoxM1 is a general target for proteasome inhibitors.
      ). Bortezomib treatment (1–40 nmol/L) reduced cell survival in a dose-dependent manner, although with different sensitivities across cell lines: SK-Mel173, TRP, SK-Mel147, and WM1552C cell lines showed low half maximal inhibitory concentration (IC50) values (4.9–6.3 nmol/L) compared with MeWo (IC50 = 8.7 nmol/L), WM 1366 (IC50 = 21 nmol/L), and SK-Mel131 (IC50 = 19 nmol/L) cell lines (see Supplementary Figure S6 online). Cell lines with higher IC50 values were those harboring 20q13 gains, suggesting that 20q13 gains can influence bortezomib sensitivity in melanoma.
      Moreover, bortezomib treatment (10 nmol/L) diminished the expression of FOXM1 and AURKA in all melanoma cell lines except for MeWo and WM-1366 (Figure 5e and f). Interestingly, both cell lines exhibited a high phosphorylated protein kinase B expression (data not shown). Finally, we detected that bortezomib treatment induced apoptosis by triggering the cleavage of PARP in all cell lines examined (Figure 5g).
      We engrafted melanoma cells expressing the silencing vector shFOXM1#0 (or pLKO.1 as controls) into immunodeficient mice and observed a reduction in tumor growth compared with controls. (Figure 6a, left and middle panels). Moreover, staining for Aurora A and the proliferation marker PH3 showed a decrease of positive cells in shFOXM1#0 tumors compared with controls (Figure 6a, right panel).
      Figure 6
      Figure 6Drug targeting Aurora kinase A and FOXM1 significantly reduces tumor burden. (a) Stable knockdown of FOXM1 attenuates the growth of melanoma xenografts. Left panel: Histogram depicting tumor volumes after 7 days of orthotopic implantation of shFOXM1 melanoma cells and the respective pLKO.1 Control: shFOXM1-MeWo (BRAFwt/NRASwt), SK-Mel147 (NRASm), WM 1366 (NRASm), and WM 1552C (BRAFm). Bars show mean ± standard error of the mean of six mice per group. Significance was determined using Mann-Whitney U test. P < 0.05. Middle panel shows representative images of excised tumors 14 days after injection. Right panel shows representative immunostaining for Aurora kinase A and PH3, indicating cells undergoing mitosis in shFOXM1 and controls from WM 1366 and WM 1552C xenografts. DAPI staining (blue) was used as a nuclei reporter. Scale bars = 50 μm. Diagrams show (b) tumor volumes of melanoma xenografts harboring NRAS, (c) BRAF mutation, and (e) BRAF mutation sensitive M#238 and the vemurafenib-resistant M#238 AR. Tumor-bearing mice received 30 mg/kg of MLN8237, 1 mg/kg of bortezomib, or vehicle. Treatments were initiated when tumors reached 100 mm3. Results are presented as mean ± standard error of the mean of tumor volumes (mm3) from six treated mice. P values are calculated by Student t test (P < 0.05, ∗∗P < 0.01) in TRP and WM 1552C or Mann-Whitney U test in SK-Mel147 and SK-Mel 131 xenografts comparing drug-treated versus vehicle at each point. (d) Representative sections of WM 1552C tumors of mice receiving bortezomib or vehicle stained for PH3 to depict cells undergoing mitosis. Scale bar = 50 μm. m, mutated; sh, short hairpin.

      Therapeutic benefit of Aurora kinase A and FOXM1 inhibition on melanoma xenografts

      Both MLN8237 and bortezomib treatments of immunodeficient mice bearing melanoma tumors impaired tumor growth compared with vehicle. The effects were observed in melanoma xenografts regardless of BRAF/NRAS mutations and vemurafenib resistance (Figure 6b, c, and e). Moreover, we detected a significant reduction in number of proliferative cells in bortezomib-treated tumor xenografts (Figure 6d). No signs of discomfort were observed during the treatment of mice.

      Discussion

      This study shows that Aurora kinase A/AURKA overexpression correlates with melanoma patient survival in our cohort of patients and in two independent datasets. These data highlight the role of AURKA in melanoma prognosis, as recently described in other solid tumors (
      • Zhang J.
      • Li B.
      • Yang Q.
      • Zhang P.
      • Wang H.
      Prognostic value of Aurora kinase A (AURKA) expression among solid tumor patients: a systematic review and meta-analysis.
      ). The prognostic value of Aurora kinase A in melanoma encourages further study of the molecular mechanisms responsible for Aurora kinase A overexpression in melanoma. The genomic amplification is a recurrent mechanism for Aurora kinase A overexpression in cancer cells (
      • Zhou H.
      • Kuang J.
      • Zhong L.
      • Kuo W.L.
      • Gray J.W.
      • Sahin A.
      • et al.
      Tumour amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transformation.
      ). We detected genomic AURKA overrepresentation in 35.7% of cases and in 13% of melanomas from TCGA dataset, which is similar to the previously reported frequency (
      • Koynova D.K.
      • Jordanova E.S.
      • Milev A.D.
      • Dijkman R.
      • Kirov K.S.
      • Toncheva D.I.
      • et al.
      Gene-specific fluorescence in-situ hybridization analysis on tissue microarray to refine the region of chromosome 20q amplification in melanoma.
      ). Moreover, TCGA dataset showed a significant positive correlation between AURKA copy number and gene expression, suggesting that genomic amplification results in Aurora kinase A overexpression in melanoma. However, transcription activation and posttranscriptional mechanisms can also induce AURKA expression in the absence of gene amplification (
      • D'Assoro A.B.
      • Haddad T.
      • Galanis E.
      Aurora-A kinase as a promising therapeutic target in cancer.
      ,
      • Tanaka M.
      • Ueda A.
      • Kanamori H.
      • Ideguchi H.
      • Yang J.
      • Kitajima S.
      • et al.
      Cell-cycle-dependent regulation of human aurora A transcription is mediated by periodic repression of E4TF1.
      ). We found a positive correlation between Aurora kinase A-positive tumors and BRAFV600E mutation, suggesting that activation of MAPK signaling by BRAF/NRAS mutations affects AURKA expression. BRAF inhibition by either PLX4720 or small interfering RNAs targeting BRAF decreased Aurora kinase A levels in BRAFV600E cells, and ERK inhibition by PD98059 reduced the AURKA promoter activity and Aurora kinase A levels in both BRAFV600E and NRASQ61 melanoma cells. Together, these results indicate that Aurora kinase A induction occurs through an ERK-dependent mechanism in melanoma cells. Melanoma cells also induce AURORA B expression via the MAPK pathway (
      • Bonet C.
      • Giuliano S.
      • Ohanna M.
      • Bille K.
      • Allegra M.
      • Lacour J.P.
      • et al.
      Aurora B is regulated by the mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) signaling pathway and is a valuable potential target in melanoma cells.
      ).
      We found that either mutations in or deletion of the putative ETS site in the proximal region dramatically decreased the activity of the promoter in BRAFm melanoma cells, which have Ras/MAPK signaling constitutively activated, as observed in pancreatic tumors (
      • Furukawa T.
      • Kanai N.
      • Shiwaku H.O.
      • Soga N.
      • Uehara A.
      • Horii A.
      AURKA is one of the downstream targets of MAPK1/ERK2 in pancreatic cancer.
      ). Indeed, the constitutive Ras/MAPK signaling mediates the transcription of ETS-containing genes through the phosphorylation of the ETS proteins, as previously reported (
      • Foulds C.E.
      • Nelson M.L.
      • Blaszczak A.G.
      • Graves B.J.
      Ras/mitogen-activated protein kinase signaling activates Ets-1 and Ets-2 by CBP/p300 recruitment.
      ,
      • Furukawa T.
      • Kanai N.
      • Shiwaku H.O.
      • Soga N.
      • Uehara A.
      • Horii A.
      AURKA is one of the downstream targets of MAPK1/ERK2 in pancreatic cancer.
      ). Similarly, activated ERK may facilitate FOXM1 transactivation and binding to its target promoters (
      • Lok G.T.
      • Chan D.W.
      • Liu V.W.
      • Hui W.W.
      • Leung T.H.
      • Yao K.M.
      • et al.
      Aberrant activation of FOXM1 signalling cascade triggers the cell migration/invasion in ovarian cancer cells.
      ;
      • Ma R.Y.
      • Tong T.H.
      • Cheung A.M.
      • Tsang A.C.
      • Leung W.Y.
      • Yao K.M.
      Raf/MEK/MAPK signaling stimulates the nuclear translocation and transactivating activity of FOXM1c.
      ;
      • Wierstra I.
      The transcription factor FOXM1 (Forkhead box M1): proliferation-specific expression, transcription factor function, target genes, mouse models, and normal biological roles.
      ). Accordingly, we show that BRAF knockdown effectively diminished FOXM1 levels in BRAFm cells. We detected high FOXM1 expression in BRAFV600E primary melanomas in agreement with a recent study (
      • Miyashita A.
      • Fukushima S.
      • Nakahara S.
      • Yamashita J.
      • Tokuzumi A.
      • Aoi J.
      • et al.
      Investigation of FOXM1 as a potential new target for melanoma.
      ). However, in contrast to the Miyashita et al. study, FOXM1 expression was not associated with poor survival in our cohort of patients, probably because of the limited number of melanoma tumors analyzed.
      We analyzed the connection between AURKA overexpression and FOXM1 activity in TCGA melanoma dataset and found a significant positive correlation between AURKA and FOXM1 targets on conserved transcription factor binding sites.
      We identified a consensus sequence (AAACA) for a FOXM1-binding site located between the tandem repressive element cell cycle-dependent element/CHR and the transcription initiation site of the AURKA promoter. In melanoma cells, mutation of this FOXM1 element promotes the opposite effect to that of CHR, causing a modest but significant decrease of the reporter activities compared with the wild type. A similar decrease was observed with treatment with FDI-6, a specific inhibitor of FOXM1. Given the proximity of CHR and FOXM1 elements in the AURKA promoter and the effects on transcription after mutation of their respective sites, we suggest that both may cooperate in the regulation of AURKA transcription. For instance, FOXM1 binding to the promoters does not always require forkhead-binding sites, but rather depends on CHR elements, as reported previously (
      • Chen X.
      • Müller G.A.
      • Quaas M.
      • Fischer M.
      • Han N.
      • Stutchbury B.
      • et al.
      The forkhead transcription factor FOXM1 controls cell cycle-dependent gene expression through an atypical chromatin binding mechanism.
      ).
      Because in melanoma FOXM1 is overexpressed and positively correlated with Aurora kinase A expression levels, we explored the therapeutic benefits of Aurora kinase A or FOXM1 inhibition in melanoma cells. Here we show that AURKA inhibition by MLN8237 significantly reduced cell proliferation and induced apoptosis and senescence, as previously observed (
      • Liu Y.
      • Hawkins O.E.
      • Su Y.
      • Vilgelm A.E.
      • Sobolik T.
      • Thu Y.M.
      • et al.
      Targeting aurora kinases limits tumour growth through DNA damage-mediated senescence and blockade of NF-kappaB impairs this drug-induced senescence.
      ). Notably, MLN8237 treatment impaired tumor growth of melanoma cells independently of BRAF/NRAS status and of vemurafenib-resistant melanoma cells as well. However, treatment was unable to fully impair tumor growth, indicating the convenience of using Aurora kinase A inhibitors in combination with other therapies (not explored here) (
      • Matulonis U.A.
      • Sharma S.
      • Ghamande S.
      • Gordon M.S.
      • Del Prete S.A.
      • Ray-Coquard I.
      • et al.
      Phase II study of MLN8237 (alisertib), an investigational Aurora A kinase inhibitor, in patients with platinum-resistant or -refractory epithelial ovarian, fallopian tube, or primary peritoneal carcinoma.
      ). Accordingly,
      • Vilgelm A.E.
      • Pawlikowski J.S.
      • Liu Y.
      • Hawkins O.E.
      • Davis T.A.
      • Smith J.
      • et al.
      Mdm2 and aurora kinase a inhibitors synergize to block melanoma growth by driving apoptosis and immune clearance of tumor cells.
      have shown that combined MDM2 and AURKA inhibition prolongs p53 activation and results in proliferation arrest, increased apoptosis, and immune clearance of melanoma cells. A promising therapeutic approach is the dual targeting of AURKA and MAPK inhibitors that leverages antiproliferative and apoptotic activities overcoming microphthalmia-associated transcription factor (MITF)–conferred resistance to AURKA inhibitors (
      • Pathria G.
      • Garg B.
      • Borgdorff V.
      • Garg K.
      • Wagner C.
      • Superti-Furga G.
      • et al.
      Overcoming MITF-conferred drug resistance through dual AURKA/MAPK targeting in human melanoma cells.
      ).
      FOXM1 inhibition caused cell-proliferation arrest and impaired melanoma growth. Thus, we conducted in vivo treatments with bortezomib, which inhibits FOXM1 transcriptional activity and expression (
      • Bhat U.G.
      • Halasi M.
      • Gartel A.L.
      FoxM1 is a general target for proteasome inhibitors.
      ,
      • Gartel A.L.
      A new target for proteasome inhibitors: FoxM1.
      ). Bortezomib fully blocks tumor growth in xenografts, regardless of BRAF/NRAS alterations. It is previously shown that bortezomib causes melanoma cell death by induction of the proapoptotic protein NOXA and the subsequent release of proapoptotic factors from the mitochondria (
      • Fernandez Y.
      • Verhaegen M.
      • Miller T.P.
      • Rush J.L.
      • Steiner P.
      • Opipari A.W.
      • et al.
      Differential regulation of noxa in normal melanocytes and melanoma cells by proteasome inhibition: therapeutic implications.
      ). However, bortezomib could not counteract the intrinsically high levels of Bcl-2 and Bcl-xL expressed by some melanoma cells, and thus, the combination with other drugs is somewhat indicated (
      • Wolter K.G.
      • Verhaegen M.
      • Fernandez Y.
      • Nikolovska-Coleska Z.
      • Riblett M.
      • de la Vega C.M.
      • et al.
      Therapeutic window for melanoma treatment provided by selective effects of the proteasome on Bcl-2 proteins.
      ). We have previously described that phosphatidylinositol 3 kinase/protein kinase B (PI3K/AKT) inhibitors and bortezomib combination induces a synergistic increase in melanoma cell death (
      • Yeramian A.
      • Sorolla A.
      • Velasco A.
      • Santacana M.
      • Dolcet X.
      • Valls J.
      • et al.
      Inhibition of activated receptor tyrosine kinases by Sunitinib induces growth arrest and sensitizes melanoma cells to Bortezomib by blocking Akt pathway.
      ). A phase Ib study combining Aurora kinase A inhibitor and bortezomib in relapsed multiple myeloma is now in progress, and results may unveil the advantages of the combined therapy also for melanoma patients (
      • Rosenthal A.
      • Kumar S.
      • Hofmeister C.
      • Laubach J.
      • Vij R.
      • Dueck A.
      • et al.
      A phase Ib study of the combination of the Aurora kinase inhibitor alisertib (MLN8237) and bortezomib in relapsed multiple myeloma.
      ).
      In summary, the study indicates that the ERK pathway-FOXM1-Aurora kinase A axis is crucial in melanoma development and suggests that Aurora kinase A expression affects disease prognosis. Moreover, data from NRASm cell lines or from vemurafenib-resistant cells suggests that Aurora kinase A and FOXM1 inhibition should be considered for melanoma patients who fail to benefit from BRAF/MEK inhibition.

      Materials and Methods

      Experimental procedures are described in Supplementary Materials online. See also Supplementary Tables S5 through S7 online.

      Melanoma samples

      Three sets of primary melanoma were included: (i) a tissue microarray with 189 formalin-fixed, paraffin-embedded samples from the same period of time, (ii) 50 fresh-frozen superficial spreading melanoma with a Breslow index less than 4 mm, and (iii) 84 melanomas including 70 consecutively fresh-frozen samples and 14 formalin-fixed, paraffin-embedded samples derived from patients independently of disease stage. Clinical and histological characteristics are provided in Supplementary Tables S1 through S3. Samples included in sets (i) and (iii) were collected at the Melanoma Unit–Hospital Clinic Barcelona, and set (ii) was collected at the IDIBELL–University Hospital of Bellvitge. Written informed consent was obtained for use of the tumor sample in these studies. Studies were approved by the institutional review boards of the Hospital Clinic and Comité Etic d'Investigació Clínica (CEIC) of IDIBELL–University Hospital of Bellvitge (Barcelona, Spain).

      Tumor xenografts and histological studies

      Mouse studies were carried out with the approval of the Bellvitge Biomedical Research Institute Animal Ethics Committee (procedure 4587 AFF) and in accordance with the Spanish National Health and Medical Research Council’s guidelines and AAALAC for the care and use of laboratory animals. Details are provided in the Supplementary Materials.

      Meta-analysis and statistics

      Simple cross-tabulations and Pearson χ2 or Fischer exact test were used to analyze correlations between clinical-histological and molecular variables. Mann-Whitney and/or Kruskal-Wallis tests were used to analyze Breslow index and age of onset and gains of AURKA gene. Patient survival was estimated by Kaplan-Meier and compared by log-rank test. Multivariate analysis was carried out using the Cox regression model to assess the independence of prognostic factors.
      To validate the prediction model, independent datasets were analyzed. Preprocessed and normalized gene expression (RNA-seq) and copy number (segment mean log2 values) data from primary and metastatic melanomas of TCGA (n = 470) were obtained from the Genomic Data Commons portal. The FOXM1 target sets were compiled from previous datasets (
      • Gormally M.V.
      • Dexheimer T.S.
      • Marsico G.
      • Sanders D.A.
      • Lowe C.
      • Matak-Vinković
      • et al.
      2015 Suppression of the FOXM1 transcriptional program via novel small molecule inhibition.
      ) and TRANScription FACtor database (TRANSFAC; geneXplain GmbH, Wolfenbüttel, Germany) (
      • Matys V.
      • Fricke E.
      • Geffers R.
      • Gössling E.
      • Haubrock M.
      • Hehl R.
      • et al.
      TRANSFAC: transcriptional regulation, from patterns to profiles.
      ). The rank-based analysis used the Gene Set Expression Analysis tool with standard parameters and PCCs computed between AURKA and all other genes across all primary and metastatic melanomas. The multivariate Cox regression analysis was performed with 358 metastatic melanomas, which included 193 events of death. In this analysis, stages IIB-C to IV were found to be associated with poorer survival.
      Graphpad prism software 6.0 (Graphpad, La Jolla, CA) and SPSS 18.0 software (SPSS, Inc., Chicago, IL) were used for the rest of the statistical analyses.

      Conflict of Interest

      MAP is recipient of unrestricted research grants from Hoffmann-La Roche and Astellas for the support of the ProCURE research program (Catalan Institute of Oncology).

      Acknowledgments

      This work was supported by the “Fundació La Marató de TV3” (201331-30 to SP and 201331-32 to AF), the “Asociación Española Contra el Cáncer” (GCB15152978SOEN), ISCIII (PI13/2272 to AF and PI15/00854 to MAP), RTICC RD12/0036/0008 and FEDER PIE13/00022-ONCOPROFILE to MAP, and AGAUR (2014SGR0334 and 2014SGR0364). Grants from ISCIII (PI15/00716 and PI15/00956), AGAUR (2014SGR603), EU 6thFP (LSHC-CT-2006-018702, GenoMEL), EU 7thFP (Diagnostics), and NIH (CA83115) were to the Melanoma Unit Hospital Clinic Barcelona.

      Supplementary Material

      References

        • Akbani R.
        • Akdemir K.C.
        • Aksoy B.A.
        • Albert M.
        • Ally A.
        • Amin S.B.
        • et al.
        Genomic classification of cutaneous melanoma.
        Cell. 2015; 161: 1681-1696
        • Alvarez-Fernandez M.
        • Medema R.H.
        Novel functions of FoxM1: from molecular mechanisms to cancer therapy.
        Front Oncol. 2013; 3: 30
        • Aytes A.
        • Mitrofanova A.
        • Lefebvre C.
        • Alvarez M.J.
        • Castillo-Martin M.
        • Zheng T.
        • et al.
        Cross-species regulatory network analysis identifies a synergistic interaction between FOXM1 and CENPF that drives prostate cancer malignancy.
        Cancer Cell. 2014; 25: 638-651
        • Bhat U.G.
        • Halasi M.
        • Gartel A.L.
        FoxM1 is a general target for proteasome inhibitors.
        PLoS One. 2009; 4: e6593
        • Bonet C.
        • Giuliano S.
        • Ohanna M.
        • Bille K.
        • Allegra M.
        • Lacour J.P.
        • et al.
        Aurora B is regulated by the mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) signaling pathway and is a valuable potential target in melanoma cells.
        J Biol Chem. 2012; 287: 29887-29898
        • Cancer Genome Atlas N
        Genomic classification of cutaneous melanoma.
        Cell. 2015; 161: 1681-1696
        • Chapman P.B.
        • Hauschild A.
        • Robert C.
        • Haanen J.B.
        • Ascierto P.
        • Larkin J.
        • et al.
        Improved survival with vemurafenib in melanoma with BRAF V600E mutation.
        N Engl J Med. 2011; 364: 2507-2516
        • Chen X.
        • Müller G.A.
        • Quaas M.
        • Fischer M.
        • Han N.
        • Stutchbury B.
        • et al.
        The forkhead transcription factor FOXM1 controls cell cycle-dependent gene expression through an atypical chromatin binding mechanism.
        Mol Cell Biol. 2013; 33: 227-236
        • D'Assoro A.B.
        • Haddad T.
        • Galanis E.
        Aurora-A kinase as a promising therapeutic target in cancer.
        Front Oncol. 2016; 5: 295
        • Davies H.
        • Bignell G.R.
        • Cox C.
        • Stephens P.
        • Edkins S.
        • Clegg S.
        • et al.
        Mutations of the BRAF gene in human cancer.
        Nature. 2002; 417: 949-954
        • Fernandez Y.
        • Verhaegen M.
        • Miller T.P.
        • Rush J.L.
        • Steiner P.
        • Opipari A.W.
        • et al.
        Differential regulation of noxa in normal melanocytes and melanoma cells by proteasome inhibition: therapeutic implications.
        Cancer Res. 2005; 65: 6294-6304
        • Foulds C.E.
        • Nelson M.L.
        • Blaszczak A.G.
        • Graves B.J.
        Ras/mitogen-activated protein kinase signaling activates Ets-1 and Ets-2 by CBP/p300 recruitment.
        Mol Cell Biol. 2004; 24: 10954-10964
        • Furukawa T.
        • Kanai N.
        • Shiwaku H.O.
        • Soga N.
        • Uehara A.
        • Horii A.
        AURKA is one of the downstream targets of MAPK1/ERK2 in pancreatic cancer.
        Oncogene. 2006; 25: 4831-4839
        • Garnett M.J.
        • Edelman E.J.
        • Heidorn S.J.
        • Greenman C.D.
        • Dastur A.
        • Lau K.W.
        • et al.
        Systematic identification of genomic markers of drug sensitivity in cancer cells.
        Nature. 2012; 483: 570-575
        • Gartel A.L.
        A new target for proteasome inhibitors: FoxM1.
        Expert Opin Investig Drugs. 2010; 19: 235-242
        • Gormally M.V.
        • Dexheimer T.S.
        • Marsico G.
        • Sanders D.A.
        • Lowe C.
        • Matak-Vinković
        • et al.
        2015 Suppression of the FOXM1 transcriptional program via novel small molecule inhibition.
        Nat Commun. 2014; 12: 5165
        • Hatzivassiliou G.
        • Song K.
        • Yen I.
        • Brandhuber B.J.
        • Anderson D.J.
        • Alvarado R.
        • et al.
        RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth.
        Nature. 2010; 464: 431-435
        • Hodi F.S.
        • O'Day S.J.
        • McDermott D.F.
        • Weber R.W.
        • Sosman J.A.
        • Haanen J.B.
        • et al.
        Improved survival with ipilimumab in patients with metastatic melanoma.
        N Engl J Med. 2010; 363: 711-723
        • Hsu J.Y.
        • Sun Z.W.
        • Li X.
        • Reuben M.
        • Tatchell K.
        • Bishop D.K.
        • et al.
        Mitotic phosphorylation of histone H3 is governed by Ipl1/aurora kinase and Glc7/PP1 phosphatase in budding yeast and nematodes.
        Cell. 2000; 102: 279-291
        • Johnson D.B.
        • Menzies A.M.
        • Zimmer L.
        • Eroglu Z.
        • Ye F.
        • Zhao S.
        • et al.
        Acquired BRAF inhibitor resistance: A multicenter meta-analysis of the spectrum and frequencies, clinical behaviour, and phenotypic associations of resistance mechanisms.
        Eur J Cancer. 2015; 51: 2792-2799
        • Jonsson G.
        • Busch C.
        • Knappskog S.
        • Geisler J.
        • Miletic H.
        • Ringner M.
        • et al.
        Gene expression profiling-based identification of molecular subtypes in stage IV melanomas with different clinical outcome.
        Clin Cancer Res. 2010; 16: 3356-3367
        • Konieczkowski D.J.
        • Johannessen C.M.
        • Abudayyeh O.
        • Kim J.W.
        • Cooper Z.A.
        • Piris A.
        • et al.
        A melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors.
        Cancer Discov. 2014; 4: 816-827
        • Koynova D.K.
        • Jordanova E.S.
        • Milev A.D.
        • Dijkman R.
        • Kirov K.S.
        • Toncheva D.I.
        • et al.
        Gene-specific fluorescence in-situ hybridization analysis on tissue microarray to refine the region of chromosome 20q amplification in melanoma.
        Melanoma Res. 2007; 17: 37-41
        • Kruiswijk F.
        • Hasenfuss S.C.
        • Sivapatham R.
        • Baar M.P.
        • Putavet D.
        • Naipal K.A.
        • et al.
        Targeted inhibition of metastatic melanoma through interference with Pin1-FOXM1 signaling.
        Oncogene. 2016; 35: 2166-2177
        • Liu Y.
        • Hawkins O.E.
        • Su Y.
        • Vilgelm A.E.
        • Sobolik T.
        • Thu Y.M.
        • et al.
        Targeting aurora kinases limits tumour growth through DNA damage-mediated senescence and blockade of NF-kappaB impairs this drug-induced senescence.
        EMBO Mol Med. 2013; 5: 149-166
        • Lok G.T.
        • Chan D.W.
        • Liu V.W.
        • Hui W.W.
        • Leung T.H.
        • Yao K.M.
        • et al.
        Aberrant activation of FOXM1 signalling cascade triggers the cell migration/invasion in ovarian cancer cells.
        PLoS One. 2011; 6: e23790
        • Ma R.Y.
        • Tong T.H.
        • Cheung A.M.
        • Tsang A.C.
        • Leung W.Y.
        • Yao K.M.
        Raf/MEK/MAPK signaling stimulates the nuclear translocation and transactivating activity of FOXM1c.
        J Cell Sci. 2005; 118: 795-806
        • Matulonis U.A.
        • Sharma S.
        • Ghamande S.
        • Gordon M.S.
        • Del Prete S.A.
        • Ray-Coquard I.
        • et al.
        Phase II study of MLN8237 (alisertib), an investigational Aurora A kinase inhibitor, in patients with platinum-resistant or -refractory epithelial ovarian, fallopian tube, or primary peritoneal carcinoma.
        Gynecol Oncol. 2012; 127: 63-69
        • Matys V.
        • Fricke E.
        • Geffers R.
        • Gössling E.
        • Haubrock M.
        • Hehl R.
        • et al.
        TRANSFAC: transcriptional regulation, from patterns to profiles.
        Nucleic Acids Res. 2003; 31: 374-378
        • Miyashita A.
        • Fukushima S.
        • Nakahara S.
        • Yamashita J.
        • Tokuzumi A.
        • Aoi J.
        • et al.
        Investigation of FOXM1 as a potential new target for melanoma.
        PLoS One. 2015; 10: e0144241
        • Müller G.A.
        • Wintsche A.
        • Stangner K.
        • Orohaska S.J.
        • Stadler P.F.
        • Engeland K.
        The CHR site: definition and genome-wide identification of a cell cycle transcriptional element.
        Nucleic Acids Res. 2014; 42: 10331-10350
        • Pathria G.
        • Garg B.
        • Borgdorff V.
        • Garg K.
        • Wagner C.
        • Superti-Furga G.
        • et al.
        Overcoming MITF-conferred drug resistance through dual AURKA/MAPK targeting in human melanoma cells.
        Cell Death Dis. 2016; 7: e2135
        • Raychaudhuri P.
        • Park H.J.
        FoxM1: a master regulator of tumor metastasis.
        Cancer Res. 2011; 71: 4329-4333
        • Reiter R.
        • Gais P.
        • Jutting U.
        • Steuer-Vogt M.K.
        • Pickhard A.
        • Bink K.
        • et al.
        Aurora kinase A messenger RNA overexpression is correlated with tumor progression and shortened survival in head and neck squamous cell carcinoma.
        Clin Cancer Res. 2006; 12: 5136-5141
        • Rosenthal A.
        • Kumar S.
        • Hofmeister C.
        • Laubach J.
        • Vij R.
        • Dueck A.
        • et al.
        A phase Ib study of the combination of the Aurora kinase inhibitor alisertib (MLN8237) and bortezomib in relapsed multiple myeloma.
        Br J Haematol. 2016; 174: 323-325
        • Sadasivam S.
        • Duan S.
        • DeCaprio J.A.
        The MuvB complex sequentially recruits B-Myb and FoxM1 to promote mitotic gene expression.
        Genes Dev. 2012; 26: 474-489
        • Shi H.
        • Hugo W.
        • Kong X.
        • Hong A.
        • Koya R.C.
        • Moriceau G.
        • et al.
        Acquired resistance and clonal evolution in melanoma during BRAF inhibitor therapy.
        Cancer Discov. 2014; 4: 80-93
        • Sondergaard J.N.
        • Nazarian R.
        • Wang Q.
        • Guo D.
        • Hsueh T.
        • Mok S.
        • et al.
        Differential sensitivity of melanoma cell lines with BRAFV600E mutation to the specific Raf inhibitor PLX4032.
        J Transl Med. 2010; 8: 39
        • Subramanian A.
        • Tamayo P.
        • Mootha V.K.
        • Mukherjee S.
        • Ebert B.L.
        • Gillette M.A.
        • et al.
        Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles.
        Proc Natl Acad Sci USA. 2005; 102: 15545-15550
        • Tanaka M.
        • Ueda A.
        • Kanamori H.
        • Ideguchi H.
        • Yang J.
        • Kitajima S.
        • et al.
        Cell-cycle-dependent regulation of human aurora A transcription is mediated by periodic repression of E4TF1.
        J Biol Chem. 2002; 277: 10719-10726
        • Vilgelm A.E.
        • Pawlikowski J.S.
        • Liu Y.
        • Hawkins O.E.
        • Davis T.A.
        • Smith J.
        • et al.
        Mdm2 and aurora kinase a inhibitors synergize to block melanoma growth by driving apoptosis and immune clearance of tumor cells.
        Cancer Res. 2015; 75: 181-193
        • Wang X.
        • Moschos S.J.
        • Becker D.
        Functional analysis and molecular targeting of Aurora kinases A and B in advanced melanoma.
        Genes Cancer. 2010; 1: 952-963
        • Weier H.U.
        • Mao J.H.
        Meta-analysis of Aurora kinase A (AURKA) expression data reveals a significant correlation between increased AURKA expression and distant metastases in human ER-positive breast cancers.
        J Data Mining Genomics Proteomics. 2013; 4: 127
        • Wierstra I.
        The transcription factor FOXM1 (Forkhead box M1): proliferation-specific expression, transcription factor function, target genes, mouse models, and normal biological roles.
        Adv Cancer Res. 2013; 118: 97-398
        • Wolter K.G.
        • Verhaegen M.
        • Fernandez Y.
        • Nikolovska-Coleska Z.
        • Riblett M.
        • de la Vega C.M.
        • et al.
        Therapeutic window for melanoma treatment provided by selective effects of the proteasome on Bcl-2 proteins.
        Cell Death Differ. 2007; 14: 1605-1616
        • Xie L.
        • Meyskens Jr., F.L.
        The pan-Aurora kinase inhibitor, PHA-739358, induces apoptosis and inhibits migration in melanoma cell lines.
        Melanoma Res. 2013; 23: 102-113
        • Yeramian A.
        • Sorolla A.
        • Velasco A.
        • Santacana M.
        • Dolcet X.
        • Valls J.
        • et al.
        Inhibition of activated receptor tyrosine kinases by Sunitinib induces growth arrest and sensitizes melanoma cells to Bortezomib by blocking Akt pathway.
        Int J Cancer. 2012; 130: 967-978
        • Zhang J.
        • Li B.
        • Yang Q.
        • Zhang P.
        • Wang H.
        Prognostic value of Aurora kinase A (AURKA) expression among solid tumor patients: a systematic review and meta-analysis.
        Jpn J Clin Oncol. 2015; 45: 629-636
        • Zhou H.
        • Kuang J.
        • Zhong L.
        • Kuo W.L.
        • Gray J.W.
        • Sahin A.
        • et al.
        Tumour amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transformation.
        Nat Genet. 1998; 20: 189-193