An oncogenic role for sphingosine kinase 2

Abstract

While both human sphingosine kinases (SK1 and SK2) catalyze the generation of the pleiotropic signaling lipid sphingosine 1-phosphate, these enzymes appear to be functionally distinct. SK1 has well described roles in promoting cell survival, proliferation and neoplastic transformation. The roles of SK2, and its contribution to cancer, however, are much less clear. Some studies have suggested an anti-proliferative/pro-apoptotic function for SK2, while others indicate it has a pro-survival role and its inhibition can have anti-cancer effects. Our analysis of gene expression data revealed that SK2 is upregulated in many human cancers, but only to a small extent (up to 2.5-fold over normal tissue). Based on these findings, we examined the effect of different levels of cellular SK2 and showed that high-level overexpression reduced cell proliferation and survival, and increased cellular ceramide levels. In contrast, however, low-level SK2 overexpression promoted cell survival and proliferation, and induced neoplastic transformation in vivo. These findings coincided with decreased nuclear localization and increased plasma membrane localization of SK2, as well as increases in extracellular S1P formation. Hence, we have shown for the first time that SK2 can have a direct role in promoting oncogenesis, supporting the use of SK2-specific inhibitors as anti-cancer agents.

Keywords: neoplastic transformation; oncogenesis; proliferation; sphingosine kinase 2; tumorigenesis.

Figure 1. Low-level SK2 overexpression is observed in human cancers, and can promote cell survival and proliferation

A. Heat map showing human cancers where significant (p < 1×10^-4) upregulation of SK2 mRNA levels have been observed in cancerous tissues compared with corresponding normal tissue. Data was extracted from the Oncomine database [28], where each row represents a cancer subtype from an individual dataset. Further detail is presented in Supplementary Figure S1A. B. SK1 and SK2-specific activity upon doxycycline (dox)-induced low- and high-level overexpression in HEK293 Flp-In T-Rex cells. Data shown are mean (± range) of duplicate data points from a representative experiment (of more than three independent experiments). C. Lysates from the HEK293 Flp-In T-Rex cells with doxycycline-induced low- and high-level overexpression of FLAG-tagged SK1 or SK2, or empty vector, were subjected to immunoblot analyses with antibodies against FLAG and α-tubulin. Blots shown are representative of at least three independent experiments. D. Measurement of cell proliferation in HEK293 Flp-In T-Rex cells with doxycycline (dox)-induced low- and high-level overexpression of SK1 or SK2, or empty vector. Data shown are mean ± SEM, n = 3-4. Statistics were performed using an unpaired Student's t-test (two-tailed); ***p < 0.001, ****p < 0.0001. E. Measurement of cell death in HEK293 Flp-In T-Rex cells with doxycycline (dox)-induced low- and high-level overexpression of SK1 or SK2, or empty vector. Data shown are mean ± SEM, n = 4-5. Statistics were performed using an unpaired Student's t-test (two-tailed); **p < 0.01, ***p < 0.001.

Figure 2. Generation of NIH3T3 stable cell lines with varying levels of constitutive SK2 overexpression

A. The NIH3T3 pooled stable cell line expressing SK2 and GFP, or GFP alone (empty vector), were sorted on four separate narrow gates of varying GFP intensity (colored boxes in top panel), to produce new stable lines depicted as ‘very low’, ‘low’, ‘mid’ and ‘high’. These new stable lines were then analyzed by flow cytometry to confirm that the desired narrow GFP-expression levels were obtained as expected. GFP-negative control cells are depicted by a dotted line. B. SK2-specific activity of NIH3T3 cell lines stably expressing ‘very low’ (5-fold), ‘low’ (10-fold), ‘mid’ (20-fold) or ‘high’ (440-fold) levels of SK2 overexpression (above endogenous levels), or empty vector. Data are shown as mean (± range) of duplicate samples from a representative experiment, of at least three independent experiments. C. Lysates from the NIH3T3 vector or SK2 overexpressing cell lines were subjected to immunoblot analyses with antibodies against FLAG, GFP and α-tubulin. Blots shown are representative of at least three independent experiments.

Figure 3. SK2 overexpressed at low levels can elicit oncogenic signaling and drive neoplastic transformation in vitro

A. Lysates from the NIH3T3 vector or SK2-overexpressing cell lines were subjected to immunoblot analyses and probed with antibodies against phospho-AKT, total AKT, phospho-ERK1/2, total ERK1/2, FLAG, GFP, SK1 and α-tubulin. Vect = empty vector with ‘low’ level GFP expression, chosen as a representative control. Densitometry was performed to quantify phospho-AKT and phospho-ERK band intensities, and is presented as a ratio of total AKT and ERK levels, respectively, and is normalized to vector. Blots shown are representative of three independent experiments. B. Contact inhibition of the NIH3T3 vector or SK2-overexpressing cell lines was tested using focus formation assays. Images shown are representative of at least three independent experiments, each performed in duplicate, using at least three independently generated sets of stable lines. C. Number of foci per well from the experiment shown in Figure 3B were quantified and the mean number of foci for duplicate wells was graphed (± range).

Figure 4. Low-level SK2 overexpression can drive proliferation and tumorigenesis in vivo

NOD/SCID mice were injected with NIH3T3 cell lines stably overexpressing ‘very low’ (5-fold), ‘low’ (10-fold), ‘mid’ (20-fold) or ‘high’ (440-fold) levels of SK2 (above endogenous levels). Empty vector cells with ‘low’ level GFP expression were chosen as a representative control (Vector). A. Table summarizing the number of mice with tumors per cell line, 18 days post-cell injection. B. Images of the excised tumors from each group of NIH3T3 stable SK2 cell lines. Dashed lines indicate where the same image has been spliced together to aid interpretation. C. Weights of the excised tumors from each group of NIH3T3 stable SK2 cell lines. Statistics denote a significant increase in the weights of SK2 ‘very low’ tumors compared to tumors from the SK2 ‘low’, ‘mid’ and ‘high’ groups (* p < 0.05; Student's unpaired two-tailed t-test). D. Equal amounts of total protein from the tumor tissue lysates were subjected to immunoblot analyses with antibodies against FLAG, α-tubulin and GFP. Each lane represents a different tumor sample. Dashed lines indicate where lanes from the same immunoblots have been spliced together to aid interpretation. E. SK2-specific activity from the tumor lysates was measured and graphed as mean (± range) of duplicate samples. These data were plotted alongside data of SK2 activity from the engrafted cell lines, which was transformed from Figure 2B as specific-activity (pmol S1P/min/mg protein) for the purposes of comparison. F. Dual immunofluorescence staining of overexpressed FLAG-tagged SK2 (red) and the proliferation marker Ki-67 (green) on the SK2 tumors. Tumor sections were counterstained with DAPI to indicate nuclei (blue). At least five random fields of view were imaged per tumor and representative images are shown. Scale bar = 50 μm. G. Levels of SK2 overexpression are heterogeneous in the SK2 ‘high’ tumor. Staining of overexpressed FLAG-tagged SK2 was visualized by immunohistological analyses. Multiple images were taken for each tumor and a representative field of view is shown. Arrow denotes a representative area of heterogeneous, intense FLAG (SK2) staining in the SK2 ‘high’ tumor. Scale bar = 100 μm.

Figure 5. Varying levels of SK2 overexpression affect its subcellular localization

A. The subcellular localization of FLAG-tagged SK2 (red) in the NIH3T3 stable ‘low’ and ‘high’ SK2-overexpressing cells was examined by immunofluorescence staining, using FLAG antibody. Nuclei were stained with DAPI (blue) and cell membranes were stained with antibodies against γ-catenin (green). Images are representative of cells observed from three independent experiments. Arrows denote representative plasma membrane localization of ‘low’ SK2. Scale bar = 10 μm. B.-C. Cells from A. were visualized by confocal microscopy and scored based on the presence or absence of either B. distinct nuclear FLAG-tagged SK2 staining or C. plasma membrane-localized FLAG-tagged SK2 staining. A minimum of 200 cells were scored per well, and data were graphed as mean (± SD) of triplicate wells from a single experiment, representative of three independent experiments (** p < 0.01; Student's unpaired two-tailed t-test).

Figure 6. Sphingolipid metabolism is altered when SK2 is overexpressed at varying levels

A. Rate of extracellular S1P formation was determined from intact vector control ‘low’, SK2 ‘low’ and SK2 ‘high’ NIH3T3 stable cell lines. Analyses were performed in triplicate and data are graphed as mean (± SD). Statistics denote significant increases in extracellular S1P compared to vector control cells (* p < 0.05, ** p < 0.01; Student's unpaired two-tailed t-test). B.-E. Intracellular sphingolipid species were analyzed by LC-MS using NIH3T3 vector control ‘low’, SK2 ‘low’ and SK2 ‘high’ stable cell lines. Data are graphed as mean (± SD) of quadruplicate samples for B. individual ceramide species, sphingosine (Sph) and sphingosine 1-phosphate (S1P), C. individual dihydroceramide species, D. total sphingomyelin levels, and E. total dihydrosphingomyelin levels. Statistics denote significant increases in lipids compared to vector control cells (* p < 0.05, ** p < 0.01, *** p < 0.001; Student's unpaired two-tailed t-test).