Sphingosine 1-phosphate (S1P) is currently one of the most intensely studied lipid mediators. Interest in S1P has been propelled by the development of fingolimod, an S1P receptor agonist prodrug, which revealed both a theretofore unsuspected role of S1P in lymphocyte trafficking and that such modulation of the immune system achieves therapeutic benefit in multiple sclerosis patients. S1P is synthesized from sphingosine by two SphKs (sphingosine kinases) (SphK1 and SphK2). Manipulation of SphK levels using molecular biology and mouse genetic tools has implicated these enzymes, particularly SphK1, in a variety of pathological processes such as fibrosis, inflammation and cancer progression. The results of such studies have spurred interest in SphK1 as a drug target. In this issue of the Biochemical Journal, Schnute et al. describe a small molecule inhibitor of SphK1 that is both potent and selective. Such chemical tools are essential to learn whether targeting S1P signalling at the level of synthesis is a viable therapeutic strategy.
The sphingosine kinases catalyse the transfer of phosphate from ATP to sphingosine to generate S1P (sphingosine 1-phosphate). S1P is a pleiotropic lipid mediator that, acting through a set of cell-surface GPCRs (G-protein-coupled receptors) and less-understood intracellular targets, signals a wide variety of cells and tissues. Prominent cellular responses to S1P are survival and migration; the S1P mimetic fingolimod (FTY720) implicated S1P in lymphocyte trafficking and control of heart rate . In mammals two genes encode SphKs (sphingosine kinases), which are closely related (~50% sequence identity) members of the diacylglycerol kinase superfamily [2,3]. The SphKs are fundamentally redundant in that mice null for either isotype are viable and fertile, whereas the lack of any functional SphK alleles results in mid-gestation lethality in mice . Both SphK1 and SphK2 recognize naturally occurring (D-erythro) sphingosine and sphinganine (the reduced form of sphingosine) as substrates. However, closer examination reveals striking differences between the isotypes. Although alterations to the D-erythro sphingoid structure are generally not tolerated by SphK1 (K. Lynch and T. Macdonald, unpublished work), SphK2 phosphorylates a variety of sphingoid-like compounds, as well as the unnatural D,L-threo-sphinganine . Indeed, the first drug that ‘targeted’ a SphK isotype is FTY720, which is converted into the active FTY720-P by SphK2 [5,6].
The anti-apoptotic and pro-migratory actions of S1P on cultured cells has encouraged thinking that blocking S1P action might be a useful anti-neoplastic strategy. Interfering RNA strategies to ‘knock down’ SphK1 levels have been documented to evoke apoptosis in several cancer cells types. Although there is no evidence that SphK1 undergoes activating mutations in cancers, SphK1 expression has been found to correlate with reduced survival in patients with high-grade astrocytomas . These and other studies have led to the hypothesis that SphK1 is a drug target for pathologies involving hyperproliferation. However, the SphK1 inhibitors announced to date are not up to the task of testing that hypothesis. Most of the existing inhibitors are cationic amphiphilic compounds, ‘long-chain bases’ in the yeast biologist's lexicon, that in most cases have not been documented to ‘cover’ their target selectively and are often cytotoxic when applied at micromolar concentrations.
Thus the compound PF-543 described by Schnute et al.  in this issue of the Biochemical Journal is a welcome addition to the SphK1 chemical biology tool kit. The route to their inhibitor is a testament to the power of modern drug discovery – efficient screening of a small-molecule chemical library against recombinant human SphK1 in a 384-well format to identify hits, followed by medicinal chemistry optimization. In one screening assay format, the substrate (FITC–sphingosine) was separated from FITC–S1P by microfluidic capillary electrophoresis. In a second format, a homogeneous assay based on a change in fluorescence polarization on antibody capture of the fluorescent-tagged ADP product was used. Together these assays were used to screen more than half a million compounds. The hits from these assays were vetted in a cell-based assay wherein cells were provided with C17-sphingosine (naturally occurring mammalian sphingosine is C18) and the C17-S1P formed was quantified by ESI (electrospray ionization)–LC (liquid chromatography)–MS. Medicinal chemistry optimization, which involved combining aspects of hits from both screens, yielded PF-543.
Although the PF-543 structure is clearly different from that of previous SphK1 inhibitors, it shares the 1,2-amino alcohol motif of sphingosine. Rather than the primary amine in sphingosine, the nitrogen in PF-543 is a tertiary amine, forming part of a pyrrolidine ring. Thus PF-543 is, like sphingosine, a base, but the 13-carbon alkyl ‘tail’ of sphingosine is replaced in PF-543 by a phenyl and two benzyl groups linked by oxygen and sulfonyl groups. PF-543 is competitive with sphingosine at SphK1; the inhibitory constant (Ki) is 3.6 nM. Indeed, the compound is phosphorylated by SphK1, albeit inefficiently, but the phosphorylated form of PF-543 does not bind the S1P receptors to an appreciable extent.
PF-543 is more than 100-fold less potent in inhibiting SphK2 than SphK1, providing an excellent margin of selectivity. The authors also counter-screened against a panel of almost 50 human protein kinases and found PF-543 to be inactive. This result is not unexpected given that PF-543 binds to the sphingosine site of SphK1. PF-543 also fails to block a more relevant lipid kinase, diacylglycerol kinase α, but its activity against the even more closely related ceramide kinase was not reported.
To assess their inhibitor in cells, the authors worked mostly with human head and neck cancer 1483 cells. This cell line was chosen because SphK1 activity predominates over SphK2 activity. PF-543 treatment of 1483 cells blocked the conversion of exogenous C17-sphingosine into C17-S1P, and after 1 h lowered the endogenous S1P 10-fold while raising sphingosine, but not ceramide, levels in a concentration-dependent manner. The rapidity whereby the 1483 cells metabolize S1P is reminiscent of a recent report using amidine-based SphK1 inhibitors applied to human leukaemia U937 cells . The IC50 values for the responses to PF-543 were single digit nanomolar, which is in agreement with the inhibitory constant of the compound determined recombinant SphK1. Furthermore, PF-543 was found to inhibit the conversion of C17-sphingosine into C17-S1P in human blood ex vivo. Taken together, these results clearly indicate that PF-543 is a reliable tool compound that rapidly and potently inhibits SphK1 activity in human cells.
The literature is replete with reports that S1P is a survival factor for cultured cells, leading to suggestions that reduction in S1P by inhibiting SphKs might lead to cell senescence or death. Many of these reports involve suppression of SphK1 levels with interfering RNA or the use of SphK1-null mice. The concept of a ‘sphingolipid rheostat’, whereby SphK inhibitors would shift the balance between S1P and sphingosine and its biosynthetic precursor ceramide, has been put forward . According to this thinking, the opposing actions of ceramides/sphingosine and S1P, the ratio of which would be increased by SphK inhibition, would alter the fate of a cell. In a test of this idea, Schnute et al.  applied PF-543 to 1483 cell cultures at a concentration (1 μM) that effectively blocks SphK1 (and lowers S1P while raising sphingosine), but they could not detect differences in the growth of the cell populations for a period of 7 days. This result was recapitulated with an additional five human cancer cell lines.
Taken together with similar results reported recently using a different SphK1 inhibitor , a reasonable conclusion is that mounting an effective sustained blockade of SphK1, with a concomitant increase in the cellular sphingosine/S1P ratio, is not notably cytotoxic to cultured cancer cells. The corollary to this conclusion being that the cytotoxicity observed with earlier generation, less potent and selective SphK1 inhibitors is due to off-target mechanisms. Although this is a facile explanation for differences in responses to different inhibitors, the disconnect between results obtained with interfering RNA strategies or SphK1-null mice and SphK1 inhibitors is less readily understood. That the slow reduction, or absence, of SphK1 protein does not equate to the rapid onset inhibition of enzyme activity indicates that SphK1 might have a role that is independent of its catalytic activity.
The lack of growth inhibition as a consequence of selective SphK1 inhibition is only one aspect of SphKs as potential drug targets. The effects of selective SphK1 inhibitors, such as PF-543, in combination with traditional anti-neoplastic cytotoxins or with a SphK2 inhibitor need to be determined. Likewise, a role for SphK1 in the production of pro-inflammatory cytokines deserves investigation.
Although PF-543 appears to be a nearly ideal tool for mounting a blockade of SphK1 in vitro, its effects in vivo were not reported. Of immediate interest is learning whether administration of PF-543 to mice or rats drives a rapid decrease in circulating S1P, as was observed with the amidine-base compound . If this response proves to be a general property of SphK1 inhibitors, then a biomarker to index enzyme inhibition is readily accessible. If PF-543 persists in vivo, its value would be increased because deployment of drug-like SphK1 inhibitors in animal models of disease is the next step towards deciding whether SphK1 inhibitors should be advanced to the clinic. If PF-543 fails to provide adequate target coverage in rodents, one hopes that synthetic manipulation of its scaffold will provide molecules with the desired properties.
Kevin Lynch is a co-inventor on several patent applications claiming sphingosine kinase inhibitors and uses thereof. This IP (intellectual property) has been licensed to a commercial entity in which the author holds equity.