Development of SKI-349, a Dual-Targeted Inhibitor of Sphingosine Kinase and Microtubule Polymerization

Abstract

Our sphingosine kinase inhibitor (SKI) optimization studies originated with the optimization of the SKI-I chemotype by replacement of the substituted benzyl rings with substituted phenyl rings giving rise to the discovery of SKI-178. We have recently reported that SKI-178 is a dual-targeted inhibitor of both sphingosine kinase isoforms (SphK1/2) and a microtubule disrupting agent (MDA). In mechanism-of-action studies, we have shown that these two separate actions synergize to induce cancer cell death in acute myeloid leukemia (AML) cell and animal models. Owning to the effectiveness of SKI-178, we sought to further refine the chemotype while maintaining “on-target” SKI and MDA activities. Herein, we modified the “linker region” between the substituted phenyl rings of SKI-178 through a structure guided approach. These studies have yielded the discovery of an SKI-178 congener, SKI-349, with log-fold enhancements in both SphK inhibition and cytotoxic potency. Importantly, SKI-349 also demonstrates log-fold improvements in therapeutic efficacy in a retro-viral transduction model of MLL-AF9 AML as compared to previous studies with SKI-178. Together, our results strengthen the hypothesis that simultaneous targeting of the sphingosine kinases (SphK1/2) and the induction of mitotic spindle assembly checkpoint arrest, via microtubule disruption, might be an effective therapeutic strategy for hematological malignancies including AML.

Graphical Abstract

Herein we report the refinement of a bifunctional sphingosine kinase inhibitor/microtubule disruptor chemotype and identification of a lead compound, SKI-349.

The SphKs, Sphingosine Kinase 1 and 2 (SphK1 and SphK2), are key regulators of the “sphingolipid rheostat”, the intracellular equilibrium between pro-apoptotic and pro-mitogenic/pro-survival sphingolipids¹. Accordingly, the SphKs are highly regarded as potential therapeutic targets for the treatment of cancer owing to the fact that numerous studies have demonstrated that both SphKs 1 and/or 2 are over-expressed/activated in multiple cancer types^2–11. In fact, elevated SphK1 expression/activation levels negatively correlates with decreased 5-year survival and progression-free survival in numerous solid and hematological cancers¹².

These findings prompted us to perform a compound library screen to identify small molecule inhibitors of SphKs 1 and 2. As reported previously, our initial efforts yielded four classes of potent, non-lipid based, “drug-like” small molecule inhibitors of SphK (SKI-I-IV)². Subsequently, we chose to optimize the SKI-I chemotype for further development as an anti-cancer therapeutic strategy due to the fact that SKI-I was the most potently cytotoxic of the four original chemotypes. Through a series of Structure-Activity Relationship studies we identified a refined SKI-I analog (SKI-178) that competes for the Sph binding site in the SphKs, is potently cytotoxic and effectively reduces S1P formation while inducing Cer accumulation against a broad range of cancer cell lines, in-vitro¹³. In subsequent studies, we determined that SKI-178 required cyclin-dependent kinase 1 (CDK1) activity for induction of apoptosis¹⁴. A closer examination of the mechanism-of-action of SKI-178 suggested that SKI-178 may also be a microtubule disrupting agent (MDA) which was confirmed by in-vitro and whole cell tubulin polymerization assays¹⁵. These apoptotic mechanism studies led us to the conclusion that SKI-178 was a Dual-Targeted Inhibitor inhibiting both of the SphK isoforms that simultaneously disrupted microtubule dynamics¹⁵. Importantly, our studies also suggested a unique property of SKI-178, in that the two separate activities (SphK Inhibition and Microtubule Disruption) acted synergistically with each other to enhance the apoptosis of cancer cells. Furthermore, we also demonstrated the therapeutic efficacy of SKI-178 two separate mouse models of acute myeloid leukemia¹⁵.

The synergistic dual-targeted nature of SKI-178 offers numerous potential advantages including the ability to achieve functional equivalence to using a separate SphK inhibitor and an MDA with one compound, to eliminate adverse drug-drug interactions and to circumvent primary and acquired drug resistance. Thus, this novel chemotype and novel mechanism-of-action was of great interest in terms of chemotherapeutic agent development. We have continued to refine the SKI-178 chemotype for enhanced cytotoxic potency and therapeutic efficacy. Through structure guided studies we refined the region between the substituted phenyl rings of SKI-178 (the “linker region”), and identified a congener (SKI-349) with improved cytotoxic, and therapeutic efficacy properties. This refinement enhanced SphK1/2 inhibitory activity while maintaining the microtubule disrupting activity of the original SKI-178 chemotype resulting in an overall log-fold increase in cytotoxicity and therapeutic efficacy in the MLL-AF9 retro-viral transduction model of AML. Together, these studies demonstrate the ability to refine dual-targeted therapeutic agents for enhanced efficacy through structure guided studies.

Our initial optimization of the SKI-I chemotype involved the replacement of the naphthyl-rings of SKI-I with substituted phenyl-rings resulting in the enhanced cytotoxicity of SKI-178¹³. Subsequently, we determined that this modification generated the microtubule disrupting activity of SKI-178, which SKI-I did not possess¹⁵. Importantly, this initial refinement did not appreciably change the dissociation constant of inhibitor binding to the Sph binding site of SphK1 (K_i), suggesting that the external rings of the SKI-class-I chemotype are not determinants of the potency of the inhibitor. Thus, the intervening “linker region” (Figure 1A) of the SKI-178 chemotype was more likely to determine the potency of the inhibitor toward the SphKs and offer the greatest likelihood of success for modification to improve potency.

FIGURE 1: Optimization of the SKI-178 “Linker” region and cytotoxic evaluation of SKI-349.

A). Diagrammatic representation of SKI-178 with the “linker” region denoted. (n=3) B). Fold change in inhibition of SphK1 relative to SKI-178 for compounds 1–27. (n=3) C). Effects of compounds 28–42 on SphK1 activity as compared to vehicle (DMSO). (n=3) D). Diagrammatic representation of SKI-349 E). Expression of myeloperoxidase (MPO) and Pglycoprotein (MDR1) in HL-60 and HL-60 vincristine resistant (HL-60/VCR) cell lines. (n=3) F). Effects of SKI-178 and SKI-349 (36) on cell viability in HL-60 and HL-60/VCR cell lines as determined by MTT assay. (n=3) G). Treespot representation of protein kinase inhibition profiling for SKI-349 at 1μM.

To this end, we conducted a substructure search of the Chembridge Compound Database (Chembridge Inc., San Diego, CA) to identify variants of the SKI-178 “linker region” while maintaining the substituted phenyl-rings of SKI-178. From the resulting 421 potential analogs we chose 27 compounds (Compounds 1–27; Supplemental Figure 1), using a non-guided approach, with features similar to the SKI-178 “linker region” or similar to our other SKI- chemotypes. For instance, compound 1 is similar to our SKI-III chemotype except that it possesses the substituted phenyl-rings of SKI-178 while compound 20 differs from compound 1 by the addition of a –CH2 group². Similarly, compound 2 bears similarity to our SKI-II except that it possesses the substituted phenyl groups of SKI-178². We first determined the inhibitory activity of these 27 compounds toward SphK1 in in-vitro SphK1 activity assays (as described previously¹³). As shown in Figure 1B, 16 of the 27 SKI-178 analogs were more potent toward SphK1 inhibition than SKI-178 and 11 of 27 were less potent. We next determined the cytotoxicity of these compounds and none of the compounds affected cell viability of a panel of human cancer cell lines as determined by MTT assay (data not shown). Thus, an unguided “similarity” approach was not sufficient to identify analogs of SKI-178 with enhanced SKI, MDA and cytotoxic properties.

We therefore next employed a structure guided approach by docking the SKI-178 “linker region” analogs into the Sph binding site of the homology model of SphK1 generated by Kennedy et al.¹⁶ using the docking program GLIDE (Schrodinger LLC). Using the docking approach, we identified a second cohort of SKI-178 “linker region” analogs which were predicted to bind to SphK1 with a greater affinity than SKI-178. We obtained 15 (Compounds 25–42; Supplemental Figure 1) of these compounds from Chembridge Inc. (San Diego CA) and determined their SphK1 inhibition profiles in in-vitro SphK1 activity assays¹³. As shown in Figure 1C, compounds 36, 38 and 40 were efficient SphK1 inhibitors as compared to vehicle only controls. Initial cytotoxicity screens in HL-60 cells revealed that compound 36 was potently cytotoxic, whereas, compounds 38 and 40 were not (data not shown). Thus, compound 36 was selected for further characterization as it had both improved SphK1 inhibitory activity and cytotoxicity. This compound was reclassified as SKI-349 (Figure 1D).

We next compared the effects of SKI-349 on cell viability to those of SKI-178, in the human acute myeloid leukemia (AML) cell line, HL-60, and their multi-drug resistant variants HL-60/VCR. AML cells acquire resistance to MDAs such as Vincristine (Vinc) by up-regulating myeloperoxidase (MPO) and multi-drug transporters (MDR-1, P-glycoprotein)^17–19. As shown in Figure 1E, we confirmed upregulation of MPO and MDR-1 in HL-60/VCR cells relative to the Vinc sensitive HL-60 parent cell line. SKI-349 is significantly more potent than SKI-178, as determined by MTT assays (as described previously¹⁴) in both HL-60 and HL-60/VCR cells (Figure 1F). Indeed, the calculated IC₅₀s for cell viability of SKI-349 are 22±6nM in HL-60 cells and 23±2nM in HL-60/VCR cells, as compared to 470±60nM and 1100±200nM for SKI-178, respectively. Importantly, the fact that HL-60/VCR cells were sensitive to both SKI-178 and SKI-349, indicates that neither compound are substrates of the drug efflux pumps (i.e. MDR1, MRP1).

We previously demonstrated that the SKI-178 chemotype competes for Sph binding within the active site of SphK1¹³. This fact decreases the likelihood of cross-inhibition of protein kinases, as compared to the prototypical protein kinase inhibitor that is ATP competitive. We confirmed that SKI-349, like SKI-178, was Sph competitive (data not shown) implying that it too would be unlikely to display “off-target” inhibition of protein kinases. To verify this assumption, we subjected SKI-349 (1 μM) to a protein kinase inhibitor profiling screen (Eurofins Discoverx, Inc.) of 468 protein kinases. Remarkably, as shown in Figure 1G, SKI-349 significantly abrogated the binding of only two protein kinases (GSKα and GSKβ; 2.5 and 27% remaining respectively) to an immobilized affinity ligand. SKI-349 also bound to the pseudokinase domains of Jak1 and Tyk2 (5.4% and 14% remaining, respectively). No other protein kinases were significantly inhibited by SKI-349.

To determine the SphK isoform selectivity of SKI-349, we conducted in-vitro SphK Activity Assays against both SphK1 and SphK2 using purified recombinant SphK proteins and isotype specific buffers¹³. As shown in Figure 2A, SKI-349 significantly inhibits SphK1 at concentrations as low as 2μM. Consistent with the results of docking predictions, SKI-349 is more potent towards SphK1 than SKI-178 with an IC₅₀ of ~ 3 μM, whereas the IC₅₀ of SKI-178 was ~30 μM. We next examined whether SKI-349 inhibited SphK2. We did not observe inhibition of SphK2 at the concentrations tested (Figure 2B). However, we suspect that the artificial buffer system employed in SphK1 (1% Triton X-100) and SphK2 (1M KCl) might lead to false negative results under the standard SphK activity assay conditions similar to what we observed with SKI-178^13–15,20.

FIGURE 2: In-vitro inhibition of SphK1 and SphK2 and in-vivo inhibition of microtubule polymerization by SKI-349.

A). Effects of SKI-178 and SKI-349 on SphK1 catalytic activity determined by in-vitro SphK1 activity assays at the indicated concentrations relative of vehicle (DMSO). B). Effects of SKI-178 and SKI-349 on SphK2 catalytic activity determined by in-vitro SphK2 activity assays at the indicated concentrations relative of vehicle (DMSO). C). Effects of SKI-349 (500nM) on tubulin polymerization in MOLM-13 cells treated for 18h, as determined by whole-cell in-vivo tubulin polymerization assays. Vincristine (200nM) paclitaxel (100 nM), and PF-543 (10 μM) were included as controls. (*p=<0.05)

Of the ~40 SKI-178 analogs tested, only SKI-349 was significantly cytotoxic at nanomolar concentrations. We have determined that the cytotoxicity of SKI-178 arises from its dual-targeted nature as a microtubule disrupting agents (MDA)¹⁴. We next examined whether SKI-349 also possessed MDA activity. Thus, we conducted a whole cell microtubule polymerization assay (as described previously¹⁵) in MOLM-13 cells. As shown in Figure 2C, SKI-349 significantly disrupts the polymerization of microtubules in intact cells similar to Vinc and opposite to the microtubule stabilizing agent Paclitaxel.

We have previously determined that SKI-178 bound to the Sph binding site of SphK1¹³ and to the colchicine binding site of the α/β tubulin dimer (unpublished observations). To further confirm the enhanced target selectivity of SKI-349 relative to SKI-178, we performed ligand-docking studies of SKI-178 and SKI-349 to the crystal structures of both SphK1 (3VZD.pdb) and the α/βtubulin dimer (1SA0.pdb). The predicted docking scores (GLIDE; Schrodinger LLC) of SKI-178 and SKI-349 into the solved X-ray crystal structure of SphK1 were −7.1 and −7.9 Kcal/mol, respectively, suggesting that SKI-349 binds to SphK1 with a greater affinity than SKI-178. Similarly, SKI-349, like SKI-178, docks to the colchicine binding site of the α/β tubulin dimer. GLIDE docking scores suggest that both SKI-349 (−6.4 Kcal/mol) and SKI-178 (−6.7 Kcal/mol) bind to tubulin with similar affinity. These results are consistent with the observation that SKI-349 is a more potent SphK inhibitor than SKI-178 and that SKI-178 and SKI-349 are equipotent microtubule disruptors.

We have recently demonstrated that SKI-178 did not inhibit SphK2 in in-vitro activity assays, but did directly target engage SphK2 using the Cellular Thermal Shift Assay (CETSA)^21,22. Unlike activity assays that rely on recombinant proteins and artificial buffer systems, CETSA measures the ability of a compound to directly bind to and stabilize/destabilize its target protein(s) when whole cells are exposed to a thermal denaturation gradient. To demonstrate that SKI-349 does, indeed, directly engage both SphK1 and SphK2, in whole cells, HEK-293 cells overe-xpressing either SphK1 or SphK2 were treated with vehicle or SKI-349 (1μM), harvested and subjected to a thermal denaturation gradient, lysed, and the soluble proteins, after thermal treatment, were analyzed by western blot analysis for the remaining amount of soluble SphK1 or SphK2 at each temperature, compared to vehicle treated cells (as described previously¹⁵). If SKI-349 directly binds to SphK1 or SphK2 we would expect to see a thermal stability shift in the presence of SKI-349 relative to vehicle treated cells. As shown in Figure 3A, SKI-349 directly target engages both SphK1 and SphK2 and induces their stabilization, relative to vehicle.

FIGURE 3: Direct target engagement of SphK1, SphK2 and β-tubulin by SKI-349.

A). CETSA analysis of direct target engagement of SphK1 and SphK2 by SKI-349 (1μM) in HEK-293 cells over-expressing either SphK1 or SphK2. (n=3) B). ITDRF analysis of the IC₅₀ of thermal stabilization of SphK1 and SphK2 by SKI-349 (1μM) in HEK-293 cells over-expressing either SphK1 or SphK2. (n=3) C). CETSA analysis of direct target engagement of β-Tubulin by SKI-349 (1μM) in HEK-293 cells. (n=3) D). ITDRF analysis of the IC₅₀ of thermal stabilization of β-Tubulin by SKI-349 (1μM) in HEK-293 cells. (n=3)

The standard CETSA assay can also be modified to generate an Isothermal Dose-Response fingerprint (ITDRF_CETSA) curve for the protein of interest at a temperature where the test compound is known to stabilize/destabilize its targets^{21, 22}. To determine IC₅₀ of direct target engagement of SKI-349 toward both SphK1 and SphK2, HEK-293 cells over-expressing either SphK1 or SphK2 were treated with vehicle or a concentration range of SKI-349 for 24h, harvested and subjected to a thermal denaturation at 61°C, lysed, and the remaining soluble proteins, after thermal treatment, were analyzed by western blot analysis for SphK1 or SphK2 protein (as described previously¹⁵). Employing the ITDRF_CETSA method, we have determined that the IC₅₀ for “direct target engagement” of both SphK1 and SphK2 are ~60 nM (Figure 3B). These data are evidence of “direct target engagement” of both SphK1 and SphK2, in whole cells, by SKI-349.

We further examined the MDA activity of SKI-349 by CETSA assays, in HEK-293 cells, to demonstrate direct targeting of β-tubulin by SKI-349¹⁵. As shown in Figure 3C, SKI-349 directly target engages β-tubulin in intact cells as judged by its ability to destabilize β-tubulin relative to vehicle treated control cells. Further, examination of the IC₅₀ of β-tubulin target engagement by ITDRF_CETSA suggests that SKI-349 disrupts microtubule polymerization as low as 25 nM in whole cells (Figure 3D).

To further demonstrate that SKI-349 was a bona fide SphK inhibitor, we determined the effects of SKI-349 on sphingolipid metabolite levels in HL-60 cells (as described previously¹⁴). As expected of SphK inhibitors, both SKI-178 at 5μM and SKI-349 at 500 nM dosages increased levels of pro-apoptotic Cers (C16:0 and C18:0) and Sph after 24 hours of treatment (Table 1). We also observed a non-significant decrease in S1P levels with SKI-349 suggesting that SKI-349 may inhibit both SphK1 and SphK2. Importantly, the fact that 500 nM SKI-349 achieved similar effects on sphingolipid levels as 5 μM SKI-178 suggests that SKI-349 is a more potent inhibitor of the SphKs than SKI-178.

Table 1:

Sphingolipid Analysis of HL-60 Cells Treated with SKI-178 and SKI-349 24h)

Treatment	Sph	S1P	C16:0 Cer	C18:0 Cer	Units
DMSO	18.3±3.5	3.1±0.3	33.5±4.1	1.6±0.2	pmoles/106 cells
SKI-178 (5 μM)	28.7±6.9**	3.2±0.7*	84.0±2.1**	4.8±0.2*	pmoles/106 cells
SKI-349 (500 nM)	27.0±6.1*	2.3±0.1*	87.0±5.7**	4.7±0.4*	pmoles/106 cells

^*p=<0.05 relative to DMSO control

We have previously demonstrated that SKI-178 induces apoptosis through the synergistic combination of its SKI and MDA activities resulting in G2/M cell cycle arrest and apoptosis¹⁴. This effect requires the sustained activation of CDK1 during mitotic arrest leading to phosphorylation/inactivation of anti-apoptotic Bcl-2 at Ser70 and Caspase 3/7 activation. To further confirm that SKI-349 induces cell death in a manner similar to SKI-178, we next examined the apoptotic mechanism-of-action of SKI-349 in HL-60 and HL-60/VCR cells (as described previously¹⁴). As shown in Figure 4A, SKI-349 induces Bcl-2 phosphorylation at Ser70 and Caspase 7 activation in human AML cell lines after 24h of treatment. Importantly, as was observed for SKI-178, a CDK-1 inhibitor, Rö-3306 (10μM), completely blocked the SKI-178/SKI-349 induced phosphorylation of Bcl-2 and activation of Caspase 7, demonstrating that CDK-1 activity is required for these effects.

FIGURE 4: Examination of the apoptotic mechanism of action of SKI-349.

A). HL-60 and HL-60/VCR cells were treated with SKI-178 (5μM) or SKI-349 (300nM) for 24 h in the presence or absence of Rö-3306 (10μM). The phosphorylation status of Bcl-2 was determined by western blot analysis using phospho-specific antibodies (Ser70). Caspase activation was determined by western blot with antibodies specific for cleaved Caspase 7. GAPDH was included as a loading control. (n=3) B). HL-60 cells were treated as indicated for 16 h. Western blot analysis was performed with the indicated antibodies. SKI-349 (1μM) PF-543 (10μM), ABC294640 (20μM), C = Combretastatin A4 (2 nM), V = Vola (20 nM) and I = Ispinesib (3 nM) (n=3).

As we reported previously, an SKI that does not affect microtubule polymerization (PF-543), synergistically enhances cell death in combination with multiple MDAs that induce spindle assembly checkpoint (SAC) arrest including Vinc and combretastatin A4¹⁵. As shown in Figure 4B, we extend this observation, in HL-60 cells, to non-MDA agents that also induce SAC arrest including, the polo kinase inhibitor (volasertib) and kinesin spindle protein inhibitor (ispinesib). Furthermore, the SKI (ABC294640; Apogee Biotechnology) also displays synergistic induction of apoptosis in combination with combretastatin A4, volasertib and ispinesib suggesting that targeting the SphKs is responsible for the synergy rather than “off-target” effects of the SKIs. Interestingly, SKI-349, as a single agent, was equally effective to the combination of SKIs (PF-543 or ABC294640) and combretastatin A4, volasertib or ispinesib, and displayed a similar mechanism of action suggesting that SKI-349 effectively recapitulates the combinatorial actions of selective SKIs + MDAs (Figure 4B).

We previously demonstrated the therapeutic efficacy of our dual-targeted agent SKI-178 in multiple mouse models of AML¹⁵. One of these models is a murine retro-viral transduction model of human MLL/AF9 driven AML. We next employed the MLL/AF9 model system to address the central question of whether SKI-349 is capable of blocking the development/progression of AML in-vivo. As previously described¹⁵, Kit⁺Sca1⁺MLL/AF9-GFP⁺ LSCs were sorted by flow-cytometry and implanted into sub-lethally irradiated recipients with recipient mice developing AML (as demonstrated by WBC counts of >10⁴/μl) approximately 4–6 weeks after transplantation. 4 weeks after initial transplantation, AML was confirmed and groups of 5 mice were randomized and treated with either vehicle or SKI-349 (1–5 mg/Kg) every other day. Remarkably, all SKI-349 doses significantly extended survival relative to vehicle (Figure 5A). The impressive survival in the SKI-349 treated mice suggested attainment of complete remission in this aggressive model of AML.

FIGURE 5: Determination of the therapeutic efficacy of SKI-349 in the retro-viral transduction mouse model of MLL/AF9 leukemia.

A). SKI-349 treatment promotes survival of leukemic mice. Four weeks after transplant of bone marrow cells infected with MLL-AF9, groups of 5 mice were treated with either vehicle control or SKI-178 (1, 2.5 and 5mg/Kg), every other day and survival curves were established. B). Representative flow cytometric analysis of bone marrow (BM) and spleen upon termination of experiment. MLL/AF9 donor (CD45.1+)/CD34+ leukemia stem cells were quantified in the vehicle (red boxes) and SKI-349 treated (green boxes) BM and spleen samples respectively.

After two weeks of SKI-349 treatment as indicated above, PB was isolated and CBC analysis was performed. As shown in Table 2, SKI-349 treatment resulted in a dose-dependent normalization of total WBC counts comparable to naïve C57BL/6J control mice without affecting RBC or platelet (PLT) counts. To further evaluate the effects of SKI-349 on normal hematopoiesis, we examined the BM and spleen of the SKI-349 treated mice at the termination of the experiment (6 weeks). In the MLL/AF9 mouse model, recipient mice are sub-lethally irradiated allowing repopulation of the recipient (CD45.2⁺) immune system in the absence of donor (CD45.1⁺) MLL/AF9 transduced AML cells. As shown in Figure 5B (Red boxes), vehicle treated mice have a robust CD45.1⁺ population of MLL/AF9 transduced AML cells in their BM and Spleen. Importantly, normal mouse hematopoietic cells (CD45.2⁺/CD45.1⁻) had repopulated the BM and spleen at all SKI-349 doses further indicating that SKI-349 does not affect normal hematopoesis (Figure 5B).

Table 2:

CBC Data for SKI-349 in MLL-AF9 mice

	Naïve Controls (C57BL/6J)	MLL/AF9+ Vehicle Treated	SKI-349 (1.0 mg/Kg)	SKI-349 (5.0 mg/Kg)
WBC (K/mL)	6.5 ± 0.4	25.6 ± 4.5	8.7 ± 1.0	6.1 ± 1.2
RBC (M/mL)	9.8 ± 0.3	8.0 ± 0.3	9.2 ± 0.4	9.0 ± 0.4
Hb (g/dL)	13.4 ± 0.5	20.9 ± 1.2	12.8 ± 0.6	12.0 ± 0.3
HCT (%)	45 ± 1.4	33 ± 2.4	44 ± 1.6	41 ± 1.5
PLT (K/mL)	687 ± 53	1207 ± 394	1193 ± 22	1089 ± 328

Herein, we report the refinement of our novel, dual-targeted SKI-178 SKI/MDA chemotype. Through a combination of studies, we were able to identify a congener (SKI-349) that was an improved SKI while maintaining its MDA ability. SKI-349 has been previously described as a highly effective analog of the MDA combretastatin A4²³. Our studies have demonstrated that, like SKI-178, SKI-349 is a dual-targeted SKI/MDA agent with a potency toward AML cell lines regardless of their intrinsic multi-drug resistance and was effective in an immunocompetent mouse model of MLL-AF9 driven AML.

It has become clear that “targeted” therapeutics are often much more promiscuous than originally intended. However, this promiscuity is not always deleterious and can actually be beneficial in many cases. Such is the cases for the protein kinase inhibitors, where multi-kinase inhibitors have become commonplace and, in some cases have made their way into clinical practice. Perhaps one of the best examples is imatinib mesylate (Gleevec). Originally developed to target the Bcr/Abl fusion protein, imatinib mesylate was later found to inhibit other tyrosine kinases including c-kit²⁴. While inhibition of c-kit can be considered an “off-target” effect in term of Bcr/Abl targeting, the ability to inhibit c-kit led to the FDA approval of imatinib mesylate for gastrointestinal stromal tumors (GIST). Thus, the concept of “off-target” effects as a deleterious factor that lessens the enthusiasm for a particular compound is incorrect. In acknowledgement of this fact, the term “polypharmacology” has become common in the drug development community²⁵.

The SKI-178/349 chemotype is a unique “polypharmacological agent” in that inhibition of the two separate targets of the compounds, the SphKs and microtubule dynamics, synergistically induces apoptosis. We have previously shown that the SKI-178 chemotype was very effective in the MLL-AF9 mouse model of AML at concentrations greater than 10 mg/kg¹⁵. With SKI-349, we achieved 100% survival with doses as low as 1 mg/kg (Figure 5A), thus we have not yet determined the minimal effective dose (MED) of SKI-349. In addition to determining the MED, the next step in the development of this lead compound would be to determine whether a single agent strategy (SKI-349) is more efficacious than a dual agent strategy such as volasertib+PF-543. Both approaches have their advantages and disadvantages. A single agent has the advantage of eliminating the possibility of adverse drug-drug interactions, whereas, the dual agent approach allows for individual titration of the separate targets²⁵. Furthermore, given the propensity of MDAs to induce peripheral neuropathies, it may be advantageous to use a combination of SKIs with a non-MDA SAC inducing agent such as volasertib. Ongoing studies will elucidate the optimal dual-targeted strategy to move forward to the clinic.

In conclusion, our results with two separate SKI/MDA agents (SKI-178 and SKI-349) strengthen the hypothesis that SphK inhibition may be an effective therapeutic strategy to block the development and/or progression of MLL/AF9 driven AML. The exact role of the SphKs in the development/progression of MLL-AF9 AML remains to be elucidated, similarly, whether other MLL-fusion driven AMLs or any other subtype of the heterogeneous population of AML rearrangements is susceptible to an SKI/MDA based therapeutic strategy remains to be determined.