Targeting sphingosine kinase 1 in acute myeloid leukemia: translation to clinic

Acute myeloid leukemia (AML) is an aggressive heterogeneous group of malignancies resulting from various oncogenic genetic lesions that presents as an accumulation of immature myeloid cells in the bone marrow and peripheral blood. Standard induction chemotherapeutics have remained the front line therapy for AML for over 30 years, and despite being often highly effective at achieving initial disease remission, these responses are often short-lived resulting in relapse and death. Indeed, the overall survival in young adults (<60 years) is <30% [1]. Unlike the ever-increasing repertoire of targeted therapies for lymphoproliferative disorders, currently all-trans-retinoic-acid represents the only targeted therapy in routine clinical use for the treatment of one subclass of AML, acute promyelocytic leukemia, with this approach yielding a more favorable prognosis in this AML subtype than presently achievable in any other form of AML. No targeted therapies are in routine clinical use for the remaining subtypes of AML, which constitute about 90% of all AML patients, and thus, there is an enormous unmet need to develop such targeted therapies.

Sphingolipids are bioactive signaling molecules present in all cells that are emerging as potentially important antileukemia targets [2]. Central in the sphingolipid biosynthetic pathway is ceramide, a proapoptotic mediator that is produced de novo by ceramide synthases and efficiently converted to the prosurvival mediator sphingosine 1-phosphate (S1P) through a series of enzyme reactions. One such reaction is the conversion of ceramide to sphingosine by a family of five ceramidases, including acid ceramidase (AC), and then to mitogenic S1P by the sphingosine kinases (SPHK1 and SPHK2) [3]. Targeting this pathway from ceramide to S1P is not in itself a new concept, with numerous preclinical studies in a wide range of cancers demonstrating that increases in cellular ceramide or reduction in S1P have favorable anticancer effects [4]. However, there has been considerable recent interest in targeting this pathway as a means to induce AML cell death, with independent findings in the last year showing AML cells have heightened dependence on this pathway, with AC and SPHK1 found to be upregulated in primary AML blasts, and that targeting these enzymes results in effective induction of apoptotic cell death in these cells [5,6]. Furthermore, other recent studies have demonstrated that the oncogenic internal tandem duplication mutation of the FLT3 receptor tyrosine kinase, common in AML, mediates AML cell survival and resistance to chemotherapeutics via blocking ceramide synthase 1, and thus, reducing the cellular levels of proapoptotic ceramide [7]. Thus, the sphingolipid pathway shows considerable promise as a target for new AML therapies.

Also of increasing importance is the recognition of dysregulated apoptotic pathways in AML with the antiapoptotic myeloid cell leukemia sequence 1 (Mcl-1) protein frequently overexpressed in these cancers [8]. Mcl-1 is one of several prosurvival Bcl-2 family proteins that bind to and inhibit the activation of proapoptotic Bax and Bak, thus preventing mitochondrial cytochrome c release and apoptosis. Potent drugs targeting some of these prosurvival Bcl-2 family members, termed BH3-mimetics, have been developed and are now entering the clinic [9]. Both ABT-737 and ABT-199 (Venetoclax) are examples of these, with the latter highly selective for Bcl-2 and approved in a number of countries for the treatment of chronic lymphocytic leukemia with 17p deletion, and in clinical trials for a range of hematological malignancies including AML, multiple myeloma and lymphomas. AML cells, however, have a high dependency on Mcl-1 for their survival [8], which currently available BH3-mimetic drugs do not inhibit. Thus, Mcl-1 remains the major mechanism of resistance of AML to BH3-mimetic-based therapies.

Notably, recent studies have shown that targeting the sphingolipid pathway in AML can downregulate Mcl-1, and thus enhance AML cell killing by BH3-mimetics and chemotherapeutics [5,6,10]. Indeed, targeting AC in AML cells increased cellular ceramide levels and induced apoptosis that was associated with loss of Mcl-1 [5]. Similarly, we also recently demonstrated that inhibition of SPHK1 is an efficacious strategy in AML [6]. Indeed, SPHK1 inhibition in AML cells resulted in loss of Mcl-1 expression and caspase-dependent apoptosis via a mechanism that involved downstream signaling by S1P receptor 2 [6], a pathway also known to be regulated by AC [11]. Notably, we demonstrated that inhibition of SPHK1 or antagonism of S1P receptor 2 both synergized with the BH3 mimetic ABT-737 to kill AML cells, highlighting the significant potential of this therapeutic strategy [6]. We also demonstrated that targeting SPHK1 reduced leukemic burden and prolonged mouse survival in primary patient-derived mouse xenograft studies [6], further supporting the clinical relevance of this approach. Importantly, while the patient-derived mouse xenograft work in these studies showed clear efficacy of targeting SPHK1 in two independent normal karyotype, FLT3-ITD mutant AMLs, excitingly, this approach may have broad application across most AML subtypes. Indeed, SPHK1 expression appears universally upregulated in all AML subtypes, and analysis of blasts and isolated leukaemic stem/progenitor cells from an extended panel of AML patients, with tumors representative of the various AML subtypes, all appeared sensitive to SPHK1 inhibition.

While a large number of sphingosine kinase inhibitors have been developed [4], few have yet translated to human clinical trials. In fact, to date only two sphingosine kinase inhibitors have entered early clinical trials. The SPHK2 inhibitor ABC294640 (YELIVA™) is currently in Phase I/II clinical trials for relapsed/refractory diffuse large B-cell lymphoma (NCT02229981) and multiple myeloma (NCT02757326), and has recently completed Phase I trials in advanced stage solid tumors (NCT01488513, NCT02939807). The SPHK1/2 inhibitor Safingol (l-threo-dihydrosphingosine) has completed Phase I clinical trials for treatment of solid tumors in combination with cisplatin (NCT00084812), and is entering clinical trials for the same indication as a combination therapy with fenretinide (NCT01553071). Both these SPHK inhibitors, however, have poor potency and significant off-targets that impact on interpretation of their clinical use, with ABC294640 shown to be a weak estrogen receptor antagonist [12] and also inhibits the important ceramide biosynthesis enzyme, dihydroceramide desaturase 1 (Des1) [13], while Safingol is well known to inhibit protein kinase C [14].

So, why have not more SPHK1 inhibitors entered clinical trials for cancer therapy? The answer to this question likely resides in the lack of potency and/or selectivity of current SPHK1 inhibitors. The first isoform-selective SPHK1 inhibitor, SK1-I, has been widely used in preclinical in vivo studies, including AML, to reduce cellular S1P, increase ceramide and induce apoptosis. However, its K _i for SPHK1 is 10 μM largely precluding its use in human trials [15]. SKI-178 is a modified version of SK1-I that displays a K _i of 1.3 μM against SPHK1 and has at least 20-fold selectivity over SPHK2 [16]. This agent was able to induce degradation of Mcl-1 in AML cell lines, including one deemed multidrug resistant through overexpression of MDR1 (P-glycoprotein) [10]. While SKI-178 appears promising in vitro, in vivo preclinical studies are required, and its specificity against other enzymes awaits proper evaluation. PF-543 is a much more potent SPHK1 inhibitor with K _i of 3.6 nM that displays 130-fold selectivity over SPHK2. However, despite reducing S1P generation, PF-543 failed to induce death in a range of cancer cell lines [17], an observation that is in opposition to the majority of SPHK1 knockdown studies. The explanation for this may lie in the observation that while PF-543 reduces cellular S1P, it has little effect on ceramide [17], though changes in cellular levels of dihydroceramides upstream of ceramide, ceramide monohexosides and lactosylceramides, suggest that PF-543 may inhibit other enzymes of the sphingolipid pathway that act to rebalance ceramides. A number of other selective SPHK1 inhibitors are being evaluated preclinically with results eagerly awaited [4].

One of the issues in the field appears to be that the majority of sphingosine kinase inhibitors act in a sphingosine-competitive manner, with their similarity to sphingosine raising the potential, and in many cases, realized effects on other sphingolipid metabolic enzymes [4], or other proteins regulated by sphingosine [18]. Notably, ATP-competitive inhibition has remained the backbone of protein kinase inhibitor development for decades, with numerous examples now in the clinic. Thus, the unique structure of the ATP-binding site of SPHK1, which is highly divergent to that of the protein kinases [19], appears to provide an ideal opportunity for the development of ATP-competitive SPHK inhibitors which is only just beginning to be exploited. Indeed, we recently developed an ATP-competitive inhibitor that showed high selectivity to SPHK1 and SPHK2, which demonstrated promising anticancer activity against a range of tumors in vitro and in vivo, including AML [6,20]. While the potency of this SPHK inhibitor, in the low μM range, precludes its use in human trials, these promising preclinical studies provide considerable impetus to develop more potent ATP-competitive SPHK inhibitors that may progress to clinical trials in AML and other cancers.

In summary, most agents that target the sphingosine kinases are yet to be evaluated in human clinical trials. However, the results of an ever-increasing body of preclinical studies examining small molecule sphingosine kinase inhibitors are growing steadily and hold considerable promise for future development of novel therapeutics.