Homology Modeling and Docking Simulations
Modeling, simulations and visualizations were performed using MOE Version 2009.10 (Chemical Computing Group). Multiple sequence alignment was performed using BioEdit v7.0.9 [37]. The human amino acid sequences for SK1 and 2 were aligned individually with the sequence of the x-ray crystal structure of Staphylococcus aureus DAG kinase (PDB 2QV7). The kinase domain of SK is recognized by the NCBI conserved domains database as a DAG kinase domain (COG1597: LCB5). For computationalSK Activity Assays
The enzymatic activities of recombinant human SK1 or SK2 (purity $74% by coomassie staining for each isozyme, BPS
Figure 1. Structures and potencies of SK inhibitors. Recombinant human SK1 and SK2 were assayed in the presence of varying concentrations of the indicated compounds, and Kis were calculated as described in the Materials and Methods section. Data are mean 6 SEM of three independent experiments.
docking of S1P to SK1 and SK2, homology models with ADP bound were generated using PDB accession code 2QV7 as the input. Prior to the analyses and simulations, the DAG kinase protein was protonated at pH 7.5 and the structure was energyminimized. Two-phased docking consisted of a primary dock calculating 50 poses using triangle matcher placement and London dG scoring. The top 30 poses for each compound were refined using forcefield placement and Affinity dG scoring.Phenotypic Assays
To assess proliferation, cells were treated with the SK inhibitors for the indicated times, and cell numbers were quantified using the sulforhodamine assay [38]. Expression levels of SK1 and SK2 were measured by quantitative PCR using GAPDH for normalization, as previously described [17]. For sphingolipid mass measurements, cells were treated with inhibitors at their respective IC50 for 48 hr, and then washed with PBS. The cell pellets were subjected to sphingolipid profiling by HPLC-MS by the Lipidomics Core Facility at MUSC as described elsewhere [34]. In the migration and invasion assays, cells were harvested after treatment with inhibitors at their respective IC50 for 48 hr, washed and suspended in serum-free media with the same concentration of the inhibitor. For migration assays, 50,000 cells were placed in each cell culture insert (24-well format, 8 mm pore size, BD Biosciences) for 1 hr at room temperature. For invasion assays, matrigel pre-coated inserts were used as previously described [17]. In both cases, inserts were placed into 24-well plates with 0.3 ml of 10%-serum-containing medium having the same concentration of the inhibitor as the insert. Cells were cultured for 4 hr for migration or 24 hr for invasion assays, and the number of cells that migrated to the underside of the inserts was determined by crystal violet staining as previously described [17].
Western Blot Analyses
Following treatment with the SK inhibitors, cells were washed, harvested and lysed, and supernatants were prepared and protein concentrations were determined as previously described [17]. Immunoblotting was carried out with the following primary antibodies (Cell Signaling): AKT, pAKT, ERK1/2, pERK1/2, STAT3, pSTAT3, p21, p53, FAK, pFAK(Y397), Beclin1, LC3, bactin; and corresponding HRP-conjugated anti-rabbit or antimouse secondary antibodies, and immunocomplexes were visualized by the chemiluminescence. Protein expression was quantified by densitometric scanning of the BIOMAX XAR films after normalization to b-actin using the program NIH ImageJ.
Results Molecular Pharmacology of SK Inhibitors
SK kinetic parameters. It was first necessary to determine the Michaelis-Menten constant (Km) for sphingosine for recombinant human SK1 and SK2. Varying the sphingosine concentration under initial velocity assay conditions provided Kms of 1060.45 mM and 560.36 mM for SK1 and SK2, respectively. The maximal rate (Vmax) catalyzed by SK1 was approximately double that of SK2 (Figure 2A), suggesting a higher catalytic turnover number for SK1. SK inhibition. Five small molecule inhibitors (Figure 1) were evaluated for their inhibitory effects and selectivity toward recombinant SK1 and SK2 using sphingosine concentrations of 10 and 5 mM, respectively. Dose-response curves for inhibition of SK1 and SK2 are shown in Figure 2B and indicate that DMS inhibits SK1 and SK2 with similar potencies (IC50,60 and 20 mM, respectively). SKI-II demonstrated slightly higher potency toward SK2 (IC50 = 20 mM) than toward SK1 (IC50 = 35 mM). This compound has been widely cited as an SK1 inhibitor, but these data indicate that it is a SK1/2-dual inhibitor which favors SK2 inhibition. We have previously described phenyladamantanebased compounds that inhibit SKs [34,36,39]. ABC294735 is one of the most potent compounds in this series causes dose-dependent inhibition of both SK1 and SK2 (IC50s,10 mM), and is therefore considered to be a SK1/2-dual inhibitor. ABC294640 demonstrates dose-dependent inhibition on SK2 with an IC50,40 mM without impacting the activity of SK1 at concentrations up to at least 100 mM consistent with our previous report [34]. CB5468139 was identified by screening the ChemBridge Corporation Diverset collection for SK1 inhibitors and caused dose-dependent inhibition of SK1, with a high potency (IC50,2 mM) without affecting SK2 activity up to at least 100 mM. Therefore, we now report CB5468139 to be a commercially-available SK1-selective inhibitor. To further investigate the inhibitory mechanisms of the SK inhibitors, we conducted substrate competition assays with sphingosine or ATP. As expected [23], DMS demonstrated competitive inhibition with respect to sphingosine (data not shown). Lineweaver-Burk plots demonstrate that ABC294735 is a sphingosine-competitive inhibitor for both SK1 and SK2 (Figure 3A), and that the SK2-selective compound, ABC294640, is also a sphingosine-competitive inhibitor (Figure 3B). Because of this, the IC50 values depend on the concentration of sphingosine used in the assays and so Kis for each compound were calculated as a more appropriate means for comparing their inhibitory potencies. As tabulated in Figure 1, DMS is the least potent dualinhibitor with Kis of 16 and 14 mM for SK1 and SK2, respectively, which are consistent with previous reports [8,40]. SKI-II and ABC294735 have similar Kis towards SK1 and SK2, with ABC294735 being substantially more potent than SKI-II. Although ABC294640 (Ki = 9.3 mM) is not as potent as ABC294735, its SK2-selectivity and pharmacological properties in vivo made it a superior clinical candidate [34]. From its structure, CB5468139 was suspected to be an ATP mimetic, and this was confirmed through Lineweaver-Burk analyses (Figure 3C). The targeting of the ATP binding site by CB5468139 was confirmed by profiling its effect on a panel of protein kinases, which demonstrated that 2 mM resulted in $50% inhibition of 12 of 65 protein kinases tested (Figure S1). Nevertheless, the fact that CB5468139 is much more potent toward SK1 than SK2 indicates that the catalytic sites of the two isoenzymes are sufficiently different to allow selective pharmacologic inhibition. Because ABC294640 is selective toward SK2 it appears that the sphingosine binding site also significantly differs between SK1 and SK2. Therefore, we sought structural insight to these selective binding properties through computational modeling of the catalytic domains of the SK isoenzymes.
