Sphingolipid metabolism determines the therapeutic efficacy of nanoliposomal ceramide in acute myeloid leukemia

Therapeutic advances for the treatment of acute myeloid leukemia (AML) have been limited in part because of an incomplete understanding of its underlying biology.^1-4  AML that arises out of myelodysplastic syndrome (MDS) highlights this challenge and represents a disease subcategory with a decidedly poor clinical outlook.^1-6  We studied sphingolipid biology in different forms of AML and evaluated the anti-AML efficacy of nanoliposomal C6-ceramide (Lip-C6). Sphingolipids are an extensive classification of lipids that play roles in cell survival and proliferation as well as stress and death.^7-12  The most well-studied sphingolipid is ceramide, a proapoptotic sphingolipid, which serves as a hypothetical center of sphingolipid metabolism.^7-12  Lip-C6 delivers the short-chain C6-ceramide analog and has entered phase 1 clinical trials for patients with solid tumors (ClinicalTrials.gov identifier: NCT02834611).¹³  Lip-C6 exerts anticancer efficacy across a spectrum of malignancies.^7,13-17  This study evaluated the preclinical efficacy of Lip-C6 for AML with MDS-related changes (AML-MRC), which was uniquely sensitive to Lip-C6 because of its propensity to convert C6-ceramide to proapoptotic sphingolipid metabolites. In contrast, resistance to Lip-C6 by de novo AML (DN-AML) was overcome by using cotreatment with vinblastine, which restored this proapoptotic sphingolipid phenotype.

Patients were classified as having either AML-MRC according to World Health Organization (WHO) criteria^1,2  or as having DN-AML if they did not have MDS-related changes or an antecedent hematologic disorder or dysplasia (supplemental Table 1). MDS-related changes were noted if there was a history of prior MDS or cytogenetic abnormalities associated with MDS as defined by Vardiman et al³  in an update to the WHO classification. We also performed next-generation mutational profiling as previously described,^4,15  to subclassify many patients based on the occurrence of MDS-related mutations per Papaemmanuil et al.¹  All clinical samples and information were collected at the Penn State College of Medicine under Institutional Review Board–approved informed consent. Additional sample information and clinical data can be found in the supplemental Methods.

The Nup98-HoxD13 transgenic mouse model fully recapitulates the cellular features of MDS and predictably evolves to AML.¹⁸  Therefore, we used this mouse as a model of AML-MRC. In contrast, we used the MLL-AF9 and Flt3-ITD transgenic mice as models of DN-AML. In all cases, transgenic mouse leukemia was confirmed by the investigative team and the New Hampshire Veterinary Diagnostic Laboratory.

Liposome generation,^14-17  apoptosis assays,¹⁶  colony-forming assays,¹⁶  lipidomics,¹⁹  and animal therapeutic studies^15,20  were carried out as previously described. Additional study-specific information for these methods, as well as details on cell culture and statistical analyses can be found in the supplemental Methods. For original data, please contact the corresponding authors. All animal studies were approved by the Institutional Animal Care and Use Committees of the University of New Hampshire and the Penn State College of Medicine.

Apoptosis was evaluated in patient AML samples exposed to 20 µM Lip-C6 for 48 hours. AML-MRC samples were uniquely sensitive to Lip-C6 (Figure 1A; supplemental Table 2). In contrast, variable sensitivity to Lip-C6 was observed from patient DN-AML samples (Figure 1B; supplemental Table 2). Lip-C6 substantially interfered with the colony-forming capacity of isolated bone marrow cells from transgenic AML-MRC mice,¹⁸  but not from transgenic DN-AML mice and wild-type controls (Figure 1C; supplemental Figures 1-3). Similarly, in vivo treatment with Lip-C6 significantly reduced detectable leukemia in AML-MRC but not DN-AML transgenic mice (Figure 1D; supplemental Figures 4 and 5). In addition, the colony-forming capacity for human AML-MRC patient samples was significantly lower in the presence of Lip-C6 than for DN-AML patient samples (Figure 1E; supplemental Table 2; supplemental Figures 1-3). Overall, these results showed that samples of AML-MRC are uniquely sensitive to Lip-C6 monotherapy.

Sphingolipids such as sphingosine-1-phosphate (S1P) and glucosylceramide oppose the proapoptotic activity of ceramide by stimulating proliferation and survival.^7-12  Therefore, AML resistance to Lip-C6 was anticipated to follow C6-ceramide metabolism to neutral/prosurvival sphingolipids (Figure 1F-H). Patient AML samples were exposed to 10 µM Lip-C6 for 24 hours before lipidomic analysis.¹⁹  Patient AML-MRC primarily converted C6-ceramide to proapoptotic metabolites, with sphingosine being predominately elevated (Figure 1F-G; supplemental Figures 6 and 7). In contrast, patient DN-AML converted C6-ceramide to a wide range of both proapoptotic and neutral or prosurvival metabolites (Figure 1F; supplemental Figures 6-8).

We previously demonstrated that vinblastine can synergize with Lip-C6 in models of colorectal cancer and hepatocellular carcinoma.¹⁴  As a microtubule destabilizing agent, vinblastine interferes with vesicular transport between subcellular compartments and may therefore impede sphingolipid metabolism. We used lipidomics to evaluate patient AML samples exposed to a combination of 10 µM Lip-C6 and 2.5 nM vinblastine for 24 hours.¹⁹  Vinblastine significantly directed the metabolism of C6-ceramide to the proapoptotic metabolites sphingosine and physiological/endogenous-ceramides (Figure 2A; supplemental Figure 7). We corroborated these findings using human AML cell lines (supplemental Figure 9). Vinblastine shifts sphingolipid metabolism so that C6-ceramide is converted to proapoptotic metabolites.

Patient AML samples were exposed to 20 µM Lip-C6, 5 nM vinblastine for 48 hours, or both, and apoptosis was evaluated. Lip-C6 was much more effective for AML-MRC than for DN-AML, but the combination of Lip-C6 with vinblastine induced robust apoptosis of either AML-MRC or DN-AML (Figure 2B-C). In addition, the combination of Lip-C6 and liposomal vinblastine significantly prolonged the lifespan of C57BL/6J mice engrafted with C1498 cells, a rapidly lethal model of AML (Figure 2D).²⁰  Likewise, in NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/SzJ (NSG) mice xenografted with MV4-11-LUC-YFP human AML cells, vinblastine monotherapy mediated a transient cytoreductive effect, which was overcome, whereas Lip-C6 exerted no standalone efficacy in this model of FLT3-ITD AML (Figure 2E). In contrast, the combination of vinblastine with Lip-C6 abrogated detectable leukemia and induced a complete remission in 50% of the mice lasting >150 days after discontinuation of therapy (Figure 2E-F; supplemental Figures 10 and 11). Altogether, these studies demonstrated robust anti-AML efficacy for the combination of Lip-C6 with vinblastine.

This study has identified varied ceramide metabolism in MDS and AML, informing the development of a novel application of therapeutic Lip-C6 and vinblastine. AML-MRC is uniquely sensitive to Lip-C6 because C6-ceramide is preferentially converted to proapoptotic metabolites, including sphingosine and endogenous ceramides. In contrast, DN-AML was less sensitive to Lip-C6 because of an increased propensity to metabolize C6-ceramide to neutral or prosurvival metabolites. Combining Lip-C6 with vinblastine shifted the metabolism of C6-ceramide back to sphingosine and endogenous ceramides. By disrupting intracellular trafficking, vinblastine may restrict C6-ceramide delivered by Lip-C6 to plasma membrane-localized metabolic pathways. Ceramide species with endogenous fatty acid chain lengths may be more cytotoxic to leukemia cells.^7-12,21  Dany et al²¹  recently showed that elevations of C18:0 ceramide can lead to lethal mitophagy in AML. Furthermore, ceramide accumulation may counterbalance S1P-mediated Mcl-1 expression, which is a pathway that can be targeted in certain cases of AML.^22,23  Alternatively, sphingosine buildup, and subsequent lysosomal instability, may be responsible for the therapeutic efficacy observed in this study. Intriguingly, AML progenitor cells have been shown to be sensitive to lysosomal disruption.²⁴  Finally, we recently showed that sphingosine-based molecules could disrupt endosomal-lysosomal membrane trafficking leading to a disruption in lysosomal biogenesis and autophagy.²⁵  Overall, this study demonstrated that nanoliposomal delivery strategies for ceramide and vinblastine in combination that target sphingolipid metabolism have the potential to robustly treat the heterogeneous nature of AML.

The authors thank Don Wojchowski of the University of New Hampshire for his review of the manuscript and helpful recommendations. In addition, the authors would like to acknowledge the Drug Discovery, Development, and Delivery core laboratory at the Penn State College of Medicine for helping to prepare Lip-C6 for this study.

This work was funded in part by grants from the National Institutes of Health, National Cancer Institute (P01 CA171983 [M.K., T.P.L., and H.-G.W.] and K22 CA190674 [B.M.B.]), a Memorial Sloan Kettering Cancer Center Support Grant/Core Grant (P30 CA0087480) (R.L.L.), a National Institute of General Medical Sciences, University of New Hampshire COBRE Pilot Project Grant (P20 GM113131) (B.M.B.), the Bess Family Charitable Fund (T.P.L.), a generous anonymous donor (T.P.L.), the Penn State University Kiesendahl Family Endowed Leukemia Research Fund and the Kenneth Noel Memorial Fund (D.F.C.), and Pennsylvania Tobacco Settlement funds (M.K. and T.P.L.).

Contributions: B.M.B. designed and supervised the study, supervised, performed, and analyzed data generated from cellular and animal experiments, analyzed all sphingolipidomic data, performed statistical analysis, generated figures, and wrote the manuscript; W.W. and P.T.T. performed experiments using transgenic mouse models and C1498-engrafted mouse models, conducted colony-forming assays using transgenic murine models, helped analyze respective data, and generated figures; T.E.F. generated raw sphingolipidomic data using LC-MS³; C.A., N.R.K., and V.G.D. performed cellular and NSG mouse experiments, helped analyze respective data, and generated respective figures; D.B.N. performed histopathologic work after the animal experiments; R.M.O. performed cellular experiments; T.J.B. performed cellular experiments and revised the manuscript; E.C.S., A.L.C., and V.P. performed experiments using transgenic mouse models; S.-F.T. performed cellular experiments and extractions for lipidomics; S.S.S. developed and prepared nanoliposomes; S.T.S. helped develop anionic nanoliposomes; T.G.D. developed and prepared nanoliposomes and performed cellular lipidomic experiments; J.Z. and J.L. analyzed data and performed statistical analyses; A.D.V. and R.L.L. performed mutational profiling and advised the study; H.-G.W. provided modified MV4-11 cells and advised NSG animal experiments; D.J.F. and T.P.L. advised and revised the manuscript; A.S. designed, supervised, and performed cellular and NSG animal experiments, analyzed respective data, and helped generate respective figures; M.K. supervised the development of nanoliposomes, designed and supervised the study, and revised the manuscript; and D.F.C. designed and supervised the study, acquired patient materials, and revised the manuscript.

Conflict-of-interest disclosure: Several patents have been issued and licensed from the Penn State Research Foundation to Keystone Nano, Inc. (State College, PA) for proprietary nanoscale formulations that encapsulate therapeutic bioactive lipids, such as the ceramide nanoliposome. M.K. is the chief medical officer and cofounder of Keystone Nano, Inc. T.P.L. is on the scientific advisory board and has stock options for both Keystone Nano, Inc. and Bioniz Therapeutics. R.L.L. is on the supervisory board of Qiagen; is a scientific advisor to Loxo, Imago, C4 Therapeutics, and Isoplexis, which each include an equity interest; receives research support from and has consulted for Celgene and Roche; has received research support from Prelude Therapeutics; has consulted for Incyte, Novartis, Astellas Pharma, Morphosys, and Janssen Pharmaceutica; and has received honoraria from Eli Lilly and Amgen for invited lectures and from Gilead Sciences for grant reviews. D.F.C. has received research support for clinical studies conducted by Daiichi Sankyo, Ambit Biosciences, Astellas Pharma, Novartis, Incyte, Cyclacel Pharmaceuticals, Celgene, MedImmune, Merck, and Gilead Sciences. The remaining authors declare no competing financial interests.

Correspondence: Brian M. Barth, Department of Molecular, Cellular and Biomedical Sciences, University of New Hampshire, 46 College Rd, Durham, NH 03824; e-mail: brian.barth@unh.edu; and David F. Claxton, Penn State Hershey Cancer Institute, Division of Hematology and Oncology, Department of Medicine, Pennsylvania State University College of Medicine, 500 University Dr, Hershey, PA 17033; e-mail: dclaxton@pennstatehealth.psu.edu.