Development of a Novel Multi-Isoform ALDH Inhibitor Effective As an Antimelanoma Agent Saketh S

Published OnlineFirst November 21, 2019; DOI: 10.1158/1535-7163.MCT-19-0360

MOLECULAR CANCER THERAPEUTICS | SMALL MOLECULE THERAPEUTICS

Development of a Novel Multi-Isoform ALDH Inhibitor Effective as an Antimelanoma Agent Saketh S. Dinavahi1,2,3, Raghavendra Gowda1,2,3,4, Krishne Gowda1, Christopher G. Bazewicz2,3,5, Venkat R. Chirasani1, Madhu Babu Battu6, Arthur Berg7, Nikolay V. Dokholyan1,8, Shantu Amin1, and Gavin P. Robertson1,2,3,4,5,9,10

ABSTRACT ◥ The aldehyde dehydrogenases (ALDH) are a major family of KS100 was mitigated by development of a nanoliposomal formu- detoxifying enzymes that contribute to cancer progression and lation, called NanoKS100. NanoKS100 had a loading efficiency of therapy resistance. ALDH overexpression is associated with a poor approximately 69% and was stable long-term. NanoKS100 was prognosis in many cancer types. The use of multi-ALDH isoform or 5-fold more selective for killing melanoma cells compared with isoform-specific ALDH inhibitors as anticancer agents is currently normal human fibroblasts. NanoKS100 administered intravenously hindered by the lack of viable candidates. Most multi-ALDH at a submaximal dose (3-fold lower) was effective at inhibiting isoform inhibitors lack bioavailability and are nonspecific or toxic, xenografted melanoma tumor growth by approximately 65% with- whereas most isoform-specific inhibitors are not effective as mono- out organ-related toxicity. Mechanistically, inhibition by KS100 therapy due to the overlapping functions of ALDH family members. significantly reduced total cellular ALDH activity to increase reac- The present study details the development of a novel, potent, multi- tive oxygen species generation, lipid peroxidation, and accumula- isoform ALDH inhibitor, called KS100. The rationale for drug tion of toxic aldehydes leading to apoptosis and autophagy. Col- development was that inhibition of multiple ALDH isoforms might lectively, these data suggest the successful preclinical development be more efficacious for cancer compared with isoform-specific of a nontoxic, bioavailable, nanoliposomal formulation containing a inhibition. Enzymatic IC50s of KS100 were 207, 1,410, and 240 novel multi-ALDH isoform inhibitor effective in the treatment of nmol/L toward ALDH1A1, 2, and 3A1, respectively. Toxicity of cancer.

Introduction they often leave behind therapy-resistant cancer cells with a stem cell– like phenotype, which serve as a reservoir for disease recurrence and Malignant melanoma is an aggressive neoplasm accounting for the metastasis (3). majority of skin cancer–related deaths (1). The outlook for metastatic Cancer cells with stem cell characteristics comprise a small subset of disease remains poor, with current 5-year survival rates of 20% (1). undifferentiated cells that initiate tumor formation and generate However, treatment strategies for malignant melanoma have vastly multipotent progenitors (3). They promote tumor aggressiveness, improved with the discovery of targeted therapies to BRAF and MEK repopulation after injury, and metastasis, having intrinsic resistance along with the development of immune checkpoint inhibitors (2). to radiotherapy, chemotherapy, and targeted therapies (4). A major Although current treatment strategies may kill the bulk of tumor cells, mechanism by which these cells develop resistance is through upregulation of the aldehyde dehydrogenases (ALDH), which has impaired the response to preoperative chemotherapy and radiotherapy in 1Department of Pharmacology, The Pennsylvania State University College of esophageal carcinoma (5), conventional chemotherapy, erlotinib and Medicine, Hershey, Pennsylvania. 2The Melanoma and Skin Cancer Center, The gefitinib in lung carcinoma (6), olaparib in breast carcinoma (7), and Pennsylvania State University College of Medicine, Hershey, Pennsylvania. 3The cyclophosphamide in a myriad of carcinomas (8, 9). Melanoma Therapeutics Program, The Pennsylvania State University College of The 19 human ALDH isozymes are broadly defined as a superfamily 4 þ Medicine, Hershey, Pennsylvania. Foreman Foundation for Melanoma Research, of NAD(P) -dependent enzymes that participate in aldehyde metab- The Pennsylvania State University College of Medicine, Hershey, Pennsylvania. 5Department of Dermatology, The Pennsylvania State University College of olism, catalyzing the oxidation of toxic aldehydes into carboxylic Medicine, Hershey, Pennsylvania. 6Laboratory of Molecular Cell Biology, Centre acids (10–13). ALDH activity within cells is generally a composite of for DNA Fingerprinting and Diagnostics, Uppal, Hyderabad, India. 7Department the activities of multiple ALDH isoforms, which have overlapping of Public Health Sciences, The Pennsylvania State University College of Medicine, substrate specificity (14, 15). The ALDHs confer a survival advantage 8 Hershey, Pennsylvania. Department of Biochemistry and Molecular Biology, to metabolically active cancer cells, by oxidizing aldehydes that The Pennsylvania State University College of Medicine, Hershey, Pennsylvania. accumulate and cause oxidative damage, into less toxic, more soluble 9Department of Pathology, The Pennsylvania State University College of Med- icine, Hershey, Pennsylvania. 10Department of Surgery, The Pennsylvania State carboxylic acids (16, 17). Accordingly, ALDH overexpression is linked University College of Medicine, Hershey, Pennsylvania. to poorer survival in gastric, breast, lung, pancreatic, and prostate carcinomas, as well as in head and neck squamous cell carcinomas Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/). (HNSCC; refs. 8, 11, 14, 18, 19). The ALDH1A1, 1A2, 1A3, 3A1, and 3A2 isozymes are particularly important in cancer progression and Corresponding Author: Gavin P. Robertson, The Pennsylvania State University College of Medicine, 500 University Drive, Hershey, PA 17033. Phone: 717-531- resistance to anticancer therapies (8, 15, 20, 21). 8098; Fax: 717-531-0480; E-mail: [email protected] Current ALDH inhibitors can be categorized into multi-ALDH isoform inhibitors and isoform-specific inhibitors, which primarily Mol Cancer Ther 2020;19:447–59 inhibit one isoform (11). Limitations of multi-ALDH isoform inhi- doi: 10.1158/1535-7163.MCT-19-0360 bitors, such as N,N-diethylaminobenzaldehyde (DEAB), which targets 2019 American Association for Cancer Research. ALDH1A1, 1A2, 1A3, 1B1, 2, and 5A1, 4-dimethylamino-4-methyl–

AACRJournals.org | 447

Dinavahi et al.

pent-2-ynthioic acid-S-methylester (DIMATE), which targets purchased from Tocris Biosciences. Isatin and the multi-ALDH ALDH1A1 and 3A1, and citral, which targets ALDH1A1, 1A3, and isoform inhibitor DEAB were purchased from Sigma Aldrich. 2, lack bioavailability or have toxicity (11). DIMATE has tumor- inhibitory efficacy when injected i.p. but will require further preclinical ALDH structure preparation evaluation (22). More recently, the ALDH inhibitors (aldis) -1, -2, -3, The structures of ALDH1A1, 2, and 3A1 bound to the inhibitors -4, and -6 have been developed, which target ALDH1A1, 2, and 3A1, CM037, psoralen, and CB7, respectively (4X4L, 5L13, and 4L2O), were and show efficacy in killing cultured cancer cells, particularly as retrieved from the protein data bank (PDB). The three-dimensional combinatorial therapy (23–25). However, these compounds have structures of the protein complexes were prepared using the protein mainly been tested in vitro and thus require further validation in preparation wizard tool (Schrodinger, LLC, 2017); water molecules preclinical models (23–25). were deleted except those in the inhibitor-binding pocket, bond orders Isoform-specific inhibitors, such as Cpd 3 and CM037 (targeting were assigned, hydrogen atoms were added, and metal ions were ALDH1A1), CVT10216 (targeting ALDH2), and CB7 and CB29 treated as described previously (35–37). Next, the orientation of the (targeting ALDH3A1), have limited efficacy in killing cultured cancer side chain structures of Gln and Asn was flipped, if necessary, to cells, particularly when used as monotherapy, and have not been tested provide the maximum degree of H-bond interactions. The charge state in animal cancer models (18, 26, 27). NCT-501, which targets of His residues was optimized. Finally, a restrained minimization of the ALDH1A1, has been shown to be effective in inhibiting HNSCC protein structure was performed using the OPLS force field with growth in animals via intratumoral injection, suggesting poor systemic backbone atoms being fixed. The minimized protein was used for the bioavailability (28). Other more bioavailable ALDH1A1-specific inhi- docking analysis. The structure was validated using Gaia webserver bitors have been developed, such as the orally bioavailable compounds (http://chiron.dokhlab.org). NCT-505 and NCT-506, and the i.p. available compounds 13 g and 13 h, but have not yet been evaluated in preclinical animal models (29, 30). ALDH grid generation and ligand preparation Therefore, ALDH inhibitors are needed that inhibit the multiple, Prepared protein structures were used to generate scoring grids for functionally overlapping ALDH isoforms, with an acceptable phar- subsequent docking calculations as described previously (35–37). To macologic profile. each protein crystal structure, a grid box of default size (20 20 The current study describes the design and development of a novel, 20A0) was centered on the corresponding active site position. Default potent, multi-isoform ALDH inhibitor, called KS100. KS100 was parameters were used, and no constraints were included during grid developed because ALDH1A1, 2, and 3A1 overexpression was generation. The ligand preparation was then performed using the observed in a cell line melanoma progression model, and targeting ligprep module in Schrodinger (35–37). these individual ALDH isoforms did not affect cultured cell growth. KS100 potently inhibited multiple ALDH isoforms with negligible Molecular docking modeling of ALDH with inhibitors toxicity when administered in a nanoliposomal form, called The starting conformations of ligands were minimized using the NanoKS100. NanoKS100 was bioavailable and inhibited melanoma OPLS 2005 force field until the energy difference between subsequent tumor growth by approximately 65% at submaximal (3-fold lower) structures was 0.001 kJ/mol-A0. The docking study was performed doses. Most importantly, KS100 significantly reduced total cellular using GLIDE 6.6 in Maestro 10.1 (35–37). The GLIDE (Grid Ligand ALDH activity compared with several ALDH inhibitors leading to Docking with Energetics) algorithm estimates a systematic search of enhanced reactive oxygen species (ROS) generation, lipid peroxida- positions, orientations, and conformations of the ligand in the tion, and accumulation of toxic aldehydes causing increased apoptosis enzyme-binding pocket via a series of hierarchical filters. The shape and autophagy. and properties of the receptor are symbolized on a grid by various dissimilar sets of fields that furnish precise scoring of the ligand pose. The potential hit compounds with good fitness score were considered Materials and Methods for docking analysis. All the hits were subjected to the extra precision Cell lines, culture conditions, and chemicals (XP) mode of GLIDE. Default values were accepted for Van der Waals Normal human fibroblasts (FF2441) were provided by Dr. Craig scaling, and input partial charges were used. During the docking Myers, Penn State College of Medicine (Hershey, PA). The human process, the GLIDE score was used to select the best conformation melanoma cell lines WM35, WM115, WM278, WM3211, 1205 Lu, for each ligand (35–37). and A375M and normal melanocytes (NHEM) were provided by Dr. Meenhard Herlyn Wistar Institute (Philadelphia, PA). The human Modeling to assess specificity of ALDH inhibitor binding melanoma cell line UACC 903 was provided by Dr. Mark Nelson, All bound crystal water molecules and ligands were stripped out of University of Arizona (Tucson, AZ). The wild-type BRAF melanoma the crystal structures of ALDH1A1, 2, and 3A1 prior to docking. cell line C8161.Cl9 was provided by Dr. Danny Welch, University of Simultaneously, the structure of KS100 was built and optimized in Kansas (Kansas City, KS), and MelJuSo was provided by Dr. Judith Marvin sketch workspace. Because ALDH1A1, 2, and 3A1 are depos- Johnson, Institute for Immunology, Germany. Cell lines were main- ited in oligomeric states in the PDB database, monomeric conforma- fi tained in a 37 C humidi ed 5% CO2 atmosphere incubator and tions of respective structures were extracted, and missing atoms or periodically monitored for phenotypic and genotypic characteristics residues were relocated through homology modeling. The structures and tumorigenic potential to validate and confirm cell line identity. were optimized using DMD software suite, and subsequently, molec- Cell lines were authenticated and tested for mycoplasma contamina- ular docking using Medusadock suite (http://medusadock.dokhlab. tions periodically and are used within 15 passages after authentication. org/) was employed, which is known for its rapid sampling efficiency The ALDH1A1- and 3A1-specific inhibitors, Cpd 3 (31, 32) and and high prediction accuracy (38). Initially, molecular docking of CB7 (18, 33) respectively, were synthesized in-house according to KS100 to the active site of ALDH1A1 alone was attempted as the previously published procedures. The ALDH1A1-specific inhibitor, conformations of ALDH1A1, 2, and 3A1 are structurally identical CM037 (34), and ALDH2-specific inhibitor, CVT10216 (34), were (Supplementary Fig. S1). From the ALDH1A1–KS100 docked

448 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

complex, it was evident that the KS100 binding pocket in ALDH1A1 and release the drug into the solution. The precipitated lipids were was lined by the residues: Ser-121, Phe-171, Val-174, Met-175, separated via centrifugation at 10,000 rpm for 15 minutes. The Trp-178, Glu-269, Phe-290, His-293, Gly-294, Tyr-297, Cys-302, supernatant was then used to measure KS100 concentration, Cys-303, Ile-304, Tyr-457, Gly-458, Val-460, and Phe-466. calculated from a standard curve of KS100 from 0.01 to 1 mg/mL. To identify the off-target effects of KS100, the binding scaffold of A 1:1 solution of chloroform to methanol was used as the reference KS100 as a substructure was extracted and employed in Erebus (http:// blank. The percentage of KS100 incorporated into nanoliposomes erebus.dokhlab.org), a webserver that searches the entire PDB database was calculated as (incorporated KS100/total KS100) X 100 (40, 41). for a given substructural scaffold (39). Erebus identiﬁes off-target structures from the PDB database by matching substructures with the (b) Stability. Stability of NanoKS100 stored at 4C was assessed weekly same amino acids and atoms segregated by identical distances (within by comparing size and zeta potential using the Malvern Zetasizer ﬁ a given tolerance) as the atoms of the query structure (39). Finally, the and measuring IC50 ef cacy for killing UACC 903 melanoma cells prediction accuracy of a match was evaluated by the root-mean-square by MTS assay and comparing these values with that of freshly deviation (RMSD) or by the normal weight with a given variance. manufactured NanoKS100 (40, 41).

Synthesis of KS100 (c) In vitro drug release kinetics. Drug release kinetics of KS100 from KS100 was synthesized as shown in Scheme 1 of Fig. 2B. Briefly, 5,7- the liposomes were measured using 1 mL of purified NanoKS100 dibromoisatin (1) (10 mmol) was dissolved in anhydrous DMF by dialysis in 1 L of 10 mmol/L reduced glutathione at room (30 mL) and cooled on ice with stirring. Solid K2CO3 (11 mmol) temperature through a molecular weight cutoff of 25 kDa mem- was added, and the dark-colored suspension was brought to room brane (Spectra Por). NanoKS100 (0.05 mL) in the dialysis bag was temperature and stirred for 1 hour. 1,4-Bis(bromomethyl)benzene removed at 0.5, 1, 2, 4, 8, 12, 24, 36, 48, and 72 hours, and the (40 mmol) was added slowly with constant stirring until the starting amount of KS100 released at each time point was estimated using material had been consumed (monitored by TLC). The reaction UV-visible spectrophotometry as detailed previously (40, 41). mixture was poured into cold water and extracted with ethyl acetate. The ethyl acetate layer was washed with water, brine, and dried (d) Hemolytic activity. The hemolytic activity assay was performed as fl over MgSO4. The solvent was removed, and the crude product was described previously (41, 42). Brie y, fresh mouse and rat blood purified by silica gel column chromatography using 80:20 hexanes/ were drawn and placed into an EDTA test tube. Erythrocytes were ethyl acetate as the eluent to yield the intermediate 5,7-dibromo-1-(4- separated from plasma by centrifugation at 1,500 rpm for 10 bromomethylbenzyl)-1H-indole-2,3-dione (2) as orange-red crystals. minutes at 4C using PBS. Erythrocyte pellets were diluted with To the intermediate compound (1.02 mmol), thiourea (1.02 mmol) 50 mL PBS in centrifuge tubes to give a 5% v/v solution and and ethanol (25 mL) were added and heated to reflux until the then treated with 5 mmol/L KS100 in DMSO, NanoKS100 (10– starting material had been consumed (monitored by TLC). The 40 mmol/L) in PBS, empty liposome, or 1% Triton X-100 (positive solvent was removed under vacuum. The final product (2-[4-(5,7- control). Samples were incubated at 37C for 60 minutes and then dibromo-2,3-dioxo-2,3-dihydro-indol-1-ylmethyl)benzyl]isothiourea) centrifuged at 12,000 rpm for 10 minutes. Next, supernatants were (3) was recrystallized in ethanol-ethyl acetate to afford KS100 (yield transferred to a 96-well plate and absorption measured at 540 nm. 70%). The identity of KS100 was confirmed by nuclear magnetic The amount of hemoglobin released in the presence of 1% Triton resonance as well as mass spectra analysis, and purity (>99%) was X-100 was set as 100% lysis, and % hemolysis was calculated as: quantified by high-performance liquid chromatography analysis. (absorbance of the samples at 540 nm/absorbance of the positive control) X 100. Preparation of NanoKS100 KS100 was encapsulated into a nanoliposome by first combining siRNA transfections L-a-Phosphatidylcholine and 1,2-Dipalmitoyl-sn-Glycero-3- Duplex stealth siRNA sequences for scrambled and ALDH1A1, Phosphoethanolamine-N-[Methoxy(Polyethylene glycol)-2000] 1A2, 1A3, 1L1, 2, 3A1, 5A1, 18A1 and BRAF were obtained from ammonium salt in chloroform at 80:20 mol % for a final lipid Invitrogen. Individual siRNAs were introduced into UACC 903 cells concentration of 25 mg/mL (Avanti Polar Lipids; refs. 40, 41). Note via nucleofection using an Amaxa nucleofector with solution that 5 mg of KS100 (in methanol) was then added to the lipid mixture, R/program K-17. Nucleofection efficiency was >90% with 80% to dried under nitrogen gas, and resuspended in saline at 60 C. The 90% cell viability. siRNA knockdown was confirmed either by mixture was then sonicated at 60 C for 30 minutes followed by qRT-PCR or by Western blots. extrusion through a 100-nm polycarbonate membrane using Avanti Mini-Extruder (Avanti Polar Lipids Inc.). (40, 41). qRT-PCR analysis Total RNA was extracted by Trizol (Sigma), and cDNA was Characterization of nanoparticle-based KS100 called generated by reverse transcription kit (Applied Biosystems). qRT-PCR NanoKS100 was performed using a SYBR green kit (Qiagen). Expression of each isoform in melanoma cell lines was normalized to corresponding (a) Drug encapsulation. Efficiency of encapsulation of KS100 in the expression in fibroblasts. Primers for the ALDH isoforms were used nanoliposomal formulation was estimated by UV-visible spectro- as described previously (15). photometry (SPECTRAmax M2 plate reader; Molecular devices; refs. 40, 41). Specifically, 1 mL of NanoKS100 solution was added ALDH isoform–specific enzyme assays to a 10 kDa Centricon filter tube (Millipore) and centrifuged at ALDH1A1, 2, and 3A1 enzyme assays were performed as described 3,750 rpm for 30 minutes to remove free KS100. Next, 0.5 mL of by the manufacturer (R & D systems). Isoform-specific aldehydes were purified NanoKS100 was combined with 0.5 mL of a 1:1 solution of converted to their respective carboxylic acids along with the conver- þ chloroform to methanol to destroy the nanoliposomal structure sion of NAD to NADH (absorbance at 340 nm). Specifically, 1 mg/mL

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 449

Dinavahi et al.

of ALDH1A1 was treated with various concentrations of ALDH before measuring fluorescence at 485 nm excitation and 520 nm inhibitors for 15 minutes followed by addition of substrate mixture emission. (10 mmol/L propionaldehyde; 100 mmol/L KCl; 1 mmol/L NAD; 2 mmol/L DTT; 50 mmol/L Tris, pH 8.5), and the absorbance Lipid peroxidation of NADH was measured in kinetic mode for 5 minutes. Similarly, Lipid peroxidation was measured using the thiobarbituric acid 0.5 mg/mL of ALDH2 was used with 2 mmol/L of acetaldehyde as the reactive substances kit according to the manufacturer's instructions substrate, and 0.2 mg/mL of ALDH3A1 was used with 1 mmol/L of (Cayman Chemicals). Briefly, cells were treated with 5 mmol/L of 4-nitrobenzaldehyde as the substrate following the addition of ALDH ALDH inhibitor or DMSO for 24 hours. Cell pellets were lysed in PBS inhibitors. by sonication on ice. Lipids in the lysates were hydrolyzed in the presence of acetic acid and sodium hydroxide. Free malondialdehyde Total cellular ALDH activity assay (MDA) was measured by the reaction to thiobarbituric acid colori- Briefly, total cellular ALDH activity assays were performed on metrically at 530 nm. cell lysates. Note that 20 mg of cell lysate was treated with 1 mmol/L of ALDH inhibitors or DMSO for 15 minutes followed by Apoptosis assay addition of substrate mixture (2 mmol/L acetaldehyde; 100 mmol/L The Annexin-V-PE/7-AAD kit (Millipore) was used to distinguish KCl; 1 mmol/L NAD; 2 mmol/L DTT; 50 mmol/L Tris, pH 8.5), live cells from apoptotic cells as described previously (43). Briefly, cells and the absorbance of NADH was measured in kinetic mode for were incubated with 5 mmol/L of ALDH inhibitor or DMSO for 5 minutes. 24 hours. Cells were washed with PBS and stained with Annexin- V–PE and 7-AAD solution per the manufacturer's instructions. Cells Cell viability assays were acquired by BD Fortessa flow cytometer and gated for four Cell viability assays of UACC 903 cells transfected with siRNA, and distinct regions, namely, live cells (Annexin V-7 AAD ), early apo- þ þ þ melanoma cells (UACC 903, 1205 Lu, C8161.CI9, MelJuSo), normal ptotic (Annexin V-7 AAD ), late apoptotic (Annexin V-7 AAD ), þ human fibroblasts (FF2441), and melanocytes (NHEM) treated with and necrotic (Annexin V-7 AAD ) cells. ALDH inhibitors were performed as described previously (43–45). Briefly, 5,000 cells per well were plated in a 96-well plate and incubated Western blot analysis overnight at 37 C in a 5% CO2 atmosphere. For the siRNA knockdown Cells were harvested by the addition of RIPA lysis buffer, and experiment, cells were incubated for another 72 hours. For the ALDH samples were processed as previously described (46, 47). Briefly, 1 to 6 inhibitor experiments, cells were treated with agents at various con- 2 10 cells were incubated overnight at 37 C in a 5% CO2 atmo- centrations and incubated for 72 hours. Twenty microliter of MTS sphere. For experiments with KS100, the agent was added and protein reagent was then added into each well, and formation of tetrazolium lysates collected following 24 hours of treatment. Blots were probed was measured by absorbance after 1 hour at 492 nm. IC50 values or % with antibodies according to each supplier's recommendations: anti- cells for each experimental group were measured in three independent bodies to cleaved PARP and LC3B from Cell Signaling Technology; experiments. alpha-enolase, ALDH1A1, 2, 3A1, 18A1, BRAF, and secondary antibodies conjugated with horseradish peroxidase from Santa Cruz Toxicity and MTD animal studies Biotechnology. Immunoblots were developed using the enhanced All the animal experiments were conducted according to the guide- chemiluminescence detection system (Thermo Fisher Scientific). lines of Penn State, Hershey Institutional Animal Care and Use Alpha-enolase served as the loading control. Committee. To determine the effective dose for in vivo efficacy studies, KS100 and NanoKS100 were injected i.p. and i.v., respectively, into Tumor efficacy and toxicity assessment Swiss Webster mice (Jackson labs) once daily for 7 days (40, 41). Efficacy and toxicity studies were performed in nude mice as Animals were monitored for changes in body weight, behavior, and described previously (40, 46, 48, 49). Briefly, 1 million cells were physical distress compared with control (DMSO for KS100, empty injected in both flanks of 4- to 6-week-old female nude Balb/c mice liposomes for NanoKS100). Dose escalation was performed to identify (Envigo). After a week, when the tumors were vascularized, animals the MTD. were treated with either NanoKS100 (at various doses) or empty liposomes. Tumor volumes, animal weight and behavior were mon- AldeRed ALDH detection assay itored continuously every other day. Animals were sacrificed after The AldeRed ALDH detection assay (Millipore) was used to tumor volumes in the empty liposome groups exceeded 2,500 mm3 and þ distinguish ALDH cells from the ALDH cells. Briefly, cells were tumors were subsequently collected. incubated with 5 mmol/L of ALDH inhibitor or DMSO for 24 hours. Cells were washed with PBS and stained with AldeRed reagent Assessment of serum biomarkers of major organ toxicity (AldeRed 588-A) for 1 hour as per the manufacturer's instructions. At the end of the UACC 903 xenograft study, serum samples were Cells were acquired by BD Fortessa flow cytometer and gated for analyzed for levels of alanine aminotransferase, alkaline phosphatase, þ ALDH cells using DEAB as a negative control. albumin, globulin, total protein, total bilirubin, blood urea nitrogen, glucose, creatinine, amylase, and calcium (40, 46, 48, 49). Serum ROS assay analysis was performed at the Centralized Biological Laboratory, Penn To quantify ROS levels, the nonfluorescent dye DCFDA (Sigma) State, University Park. Empty liposomes served as the control. was used (43). DCFDA turns to highly fluorescent 20,70-dichlorofluor- escein upon oxidation by ROS generated in cells (43). Briefly, cells were Statistical analysis treated with 5 mmol/L of ALDH inhibitor or DMSO for 24 hours. Statistical analysis was undertaken using the one-way/two-way DCFDA (10 mmol/L) was then added and incubated for 30 minutes ANOVA GraphPad PRISM Version 7.04 software. Dunnett's post hoc

450 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

analysis was performed when there was a significant difference. Results Subsequent analysis of The Cancer Genome Atlas (TCGA) database were considered significant at a P value of < 0.05. to determine the relationship of ALDH overexpression on survival of patients with melanoma yielded variable results. Specifically, overexpression of ALDH1A1 and 2 was associated with slightly improved Results survival (Fig. 1B), whereas high ALDH3A1 expression was associated ALDH overexpression occurs in melanoma and is associated with lower survival (Fig. 1C). However, these data are measurements with disease progression of RNA expression and thus do not take into account overall protein Cancer cell expression of ALDHs often increases with disease levels or ALDH activity in cancer cells. progression, as oxidative stress secondary to high metabolic demands To functionally determine whether targeting ALDH1A1, 2, or 3A1 leads to ROS generation, lipid peroxidation, and the accumulation of in melanoma affects cell proliferation, a rapid siRNA screen was toxic aldehydes, which can inhibit cancer cells (17, 50). Western blot undertaken (Fig. 1D). siRNA for ALDH18A1, a unique ALDH analysis of ALDH1A1, 2, and 3A1 in melanoma cells revealed that isoform that promotes melanoma cell survival through proline syn- ALDH overexpression occurs in melanoma compared with control thesis (51), and V600EBRAF were used as positive controls. Knockdown fibroblast (FF2441) and melanocyte (NHEM) cells (Fig. 1A). Further, of each respective protein by its siRNA is shown in Fig. 1E. Similar to the degree of ALDH expression correlated with melanoma stage the scrambled siRNA, individual siRNA knockdown of ALDH1A1, 2, such that metastatic melanomas exhibited the highest ALDH expres- and 3A1 did not affect UACC 903 cell survival up to 72 hours where sion levels, followed by vertical growth phase and finally radial as the positive control siRNA caused an approximately 50% reduction growth phase melanomas. ALDH expression was not dependent in cell survival (Fig. 1D). Pharmacologic inhibition of ALDH1A1, 2, on BRAF mutational status, as ALDH levels were similar between and 3A1 by isoform-specific inhibitors also had no effect on cell mutant V600EBRAF and wild-type BRAF cells. qRT-PCR analysis of survival, even when 100 mmol/L concentrations were used for 72 hours the ALDH isoforms demonstrated that the isoforms overexpressed in (Fig. 1F). In contrast, DEAB, a multi-ALDH isoform inhibitor, UACC 903 and 1205 Lu melanoma cells compared with control reduced UACC 903 cell survival by 30% at a 100 mmol/L concentration fibroblasts are ALDH1A1, 1A2, 1A3, 1L1, 2, 3A1, 5A1, and 18A1 after 72 hours. These results suggested that targeting multiple ALDH (Supplementary Table S1). Expectedly, when cells were stained with isoforms with overlapping function may be more effective for mela- þ AldeRed dye to isolate the ALDH (cells with high levels of ALDH) noma therapy specifically and anticancer therapy in general. In þ cells and ALDH cells using flow sorting, the ALDH cells had a addition, that inhibiting individual ALDH isoforms may lead to high expression of these isoforms compared with ALDH cells resistance mediated through upregulation of functionally similar (Supplementary Table S1). ALDHs.

Figure 1. The ALDH family is collectively important in melanoma. Western blot showing ALDH1A1, 2, and 3A1 expression levels in normal human fibroblasts (FF2441), melanocytes (NHEM), radial growth phase (RGP), vertical growth phase (VGP), and metastatic melanoma cell lines. ALDH expression in general increased during disease progression and was not dependent on BRAF mutational status. Alpha-enolase served as the loading control (A). Data from the TCGA database showing slightly better survival with ALDH1A1 and 2 overexpression (B) and worse survival with ALDH3A1 overexpression (C) in patients with melanoma. The data are available through the UCSC Xena Cancer Browser. Individual siRNA knockdown of ALDH1A1, 2, and 3A1 did not significantly reduce the survival of UACC 903 cells after 72 hours in an MTS assay. siRNA to BRAF and ALDH18A1 served as positive controls. Scrambled siRNA served as the negative control (D). siRNA knockdown of ALDH1A1, 2, 3A1, 18A1, and BRAF in UACC 903 cells was confirmed via Western blot. Alpha-enolase served as loading control (E). Pharmacologic inhibition of ALDH1A1, 2, and 3A1 using ALDH isoform–specific inhibitors (Cpd 3, CVT10216, and CB7, respectively), and the multi-ALDH isoform inhibitor, DEAB, revealed multi-ALDH isoform inhibition was most effective in inhibiting UACC 903 cell survival (F).

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 451

Dinavahi et al.

Identification and development of the novel, potent, would bind and interact more effectively in the ligand-binding pocket multi-isoform ALDH inhibitor, called KS100 of the ALDHs, using the backbones of Isatin and Cpd 3. A series of To create a multi-ALDH isoform inhibitor, an in silico screen was compounds were tested through in silico modeling to determine initially undertaken based on the x-ray crystal structure of ALDH1A1 whether they had optimal docking in the ligand-binding pocket of using various natural products. Isatin was identified during this screen ALDH1A1, and KS100 was selected as the best candidate (Fig. 2A). It as weakly binding to ALDH1A1 compared with the ALDH1A1- was also found to fit well into the ligand-binding pockets of ALDH2 specific inhibitors Cpd 3 and CM037 (Fig. 2A). A medicinal chemistry and 3A1. KS100 had docking scores of 10.247, 8.716, and 13.851 approach was subsequently undertaken to design compounds that for ALDH1A1, 2, and 3A1, respectively (Table 1), compared with

Figure 2. Design, synthesis, and toxicity analysis of the novel, ALDH1A1, 2, and 3A1 inhibitor, called KS100. Based on the structure and binding of Isatin, Cpd 3, and CM037, a medicinal chemistry approach was undertaken to design KS100, which exhibited more effective binding to ALDH1A1, 2, and 3A1 (A). KS100 was synthesized from 5,7-dibromoisatin followed by benzylation as detailed in the Supplementary Materials and Methods (B). A 7-day repeated dose study was conducted for KS100. KS100 was administered i.p. daily, whereas animal body weight, physical and behavioral changes, and mortality were monitored. KS100 was toxic starting at 5 mg/kg (C).

452 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

Table 1. Docking scores and ALDH-inhibitory activity for KS100 and other ALDH inhibitors.

Docking scores IC50s (nmol/L) Compound ALDH1A1 ALDH2 ALDH3A1 ALDH1A1 ALDH2 ALDH3A1

Isatin 5.46 6.398 5.819 15,635 1,821 168,661 28,679 5,047 304 Cpd 3 7.686 9.839 7.695 44 12 72,136 1,640 11,866 548 CM037 11.276 7.137 8.137 98 34 2,278 250 1,774 303 CVT10216 7.892 11.809 8.924 2,427 194 53 2 2,719 608 CB7 8.159 7.846 14.576 139,016 16,934 144,409 11,470 298 29 DEAB 9.154 10.026 11.211 89 23 833 277 15,119 4,091 KS100 10.247 8.716 13.851 207 10 1,410 248 240 50

Note: Isatin, Cpd 3, CM037, CVT10216, CB7, DEAB, and KS100 were docked into the active site pockets of ALDH1A1, 2, and 3A1. Docking scores were calculated using

GLIDE 6.6 in Maestro 10.1. Enzyme inhibition studies for ALDH1A1, 2, and 3A1 were performed as described in the Materials and Methods. KS100 displayed low IC50 values for all three ALDH isoforms tested. Preexisting isoform-speciﬁc ALDH inhibitors (Cpd 3 and CM037 for ALDH1A1, CVT10216 for ALDH2, CB7 for ALDH3A1) and the multi-ALDH isoform inhibitor, DEAB, were used as positive controls.

11.276, 11.809, and 14.576 for CM037 bound to ALDH1A1, Specificity and toxicity of KS100 CVT10216 bound to ALDH2, and CB7 bound to ALDH3A1, To identify potential off-target effects and ALDH-binding speci- respectively. ficity of KS100, the binding scaffold of KS100 as a substructure was Docking scores indicated strong binding of KS100 to ALDH1A1, 2, extracted and employed in Erebus (Supplementary Fig. S2). To and 3A1. KS100 had a p-p interaction with the W178 residue and an precisely determine the most similar binding scaffolds to our query H-bond with the free amine group within the ALDH1A1 ligand– structure, we imposed a cutoff RMSD of 7Å to the query during the binding pocket (Fig. 2A). Similarly, KS100 had p-p interactions with substructural search against the PDB database. The subsequent hits are the F459 and F465 residues along with an H-bond interaction between listed in Supplementary Table S2. The identification of ALDH1A1 as the free amine group and L269 residue within the ALDH2 ligand– the primary hit highlights the accuracy of the Erebus algorithm. The binding pocket. Further, KS100 had a p-p interaction with the R292 RMSD of approximately 2.24 Å between the query and the primary hit residue and an H-bond interaction with the G187 residue in is likely due to the flexible docking approach used during initial ALDH3A1 ligand-binding pocket (Fig. 2A). Due to strong ALDH1A1, docking of KS100 to ALDH1A1. NiFe-hydrogenase from Desulfovibrio 2, and 3A1 binding, KS100 was then synthesized through Scheme 1 fructosivorans was also identified as having a similar substructural shown in Fig. 2B. scaffold. Similar studies were conducted with the active site pockets of Inhibition of ALDH1A1, 2, and 3A1 by KS100 was compared with ALDH2 and 3A1. These isoforms were identified to be the primary Isatin, the ALDH1A1-specific inhibitors Cpd 3 and CM037, the hits, with no observed off-target effects of KS100. Thus, KS100 appears ALDH2-specific inhibitor CVT10216, the ALDH3A1-specific inhib- to have no off-target effects in humans based on the Erebus algorithm, itor CB7, and the multi-ALDH isoform inhibitor, DEAB (Table 1). indicating the specificity of KS100 binding to human ALDHs. Isatin was a relatively ineffective inhibitor of all ALDH isoforms, To evaluate the toxicity of KS100, Swiss Webster mice (n ¼ 3) were m > m having IC50s of 15.6 mol/L for ALDH1A1, 160 mol/L for ALDH2, treated with daily i.p. administration of KS100 at 5, 10, and 15 mg/kg and 5 mmol/L for ALDH3A1. KS100 was an effective inhibitor of and compared with DMSO (Fig. 2C). A 16.6% decrease in animal body ALDH1A1 activity, having an IC50 of 207 nmol/L compared with 44 weight, on average, along with hunched backs and lethargy were and 98 nmol/L for Cpd 3 and CM037, respectively. KS100 was an observed at day 7 in the 5 mg/kg group. All animals treated with fi effective inhibitor of ALDH2 activity, having an IC50 of 1,410 nmol/L 10 and 15 mg/kg KS100 died before day 7, indicating signi cant compared with 53 nmol/L for CVT10216. KS100 effectively inhibited toxicity. Thus, the toxicity associated with multi-ALDH isoform ALDH3A1 activity, having an IC50 of 240 nmol/L compared with 298 inhibition by KS100 necessitated the development of a formulation nmol/L for CB7. DEAB was slightly superior to KS100 in inhibiting with controlled release of the drug to eliminate these effects. ALDH1A1 and ALDH2 activity, having IC50s of 89 and 833 nmol/L, respectively, for these isoforms. However, DEAB was inferior to KS100 Developing a nontoxic, effective, stable nanoliposomal m in inhibiting ALDH3A1, having an IC50 of 15.1 mol/L. formulation of KS100, called NanoKS100 To evaluate the specificity of KS100 to inhibit various ALDH Nanoliposomal formulations can overcome drug toxicity isoforms, the isoforms overexpressed in melanoma cells (Supplemen- (40–42, 52). Therefore, KS100 was loaded into a nanoliposomal tary Table S1) were knocked down in UACC 903 cells using siRNA, formulation, called NanoKS100, and the physiochemical properties and the effect of KS100 on cell survival was evaluated. The potency of of NanoKS100 were analyzed. A schematic representation of KS100 was reduced when there was lower level of ALDH1A1, 1A2, NanoKS100 is shown in Fig. 3A where KS100 is trapped in the 1A3, 2, and 3A1, whereas the potency did not significantly vary when phospholipid bilayer of the nanoliposome. The maximum loading ALDH1L1, 5A1, and 18A1 were knocked down (Supplementary efficiency of KS100 into nanoliposomes was 68.6% (Fig. 3B). The Fig. S1). However, due to the overlapping roles of ALDH1A1, 1A2, size of NanoKS100 was 78.5 nm, with an average charge of þ0.54 eV and 1A3, the precise effect of KS100 to inhibit 1A2 and 1A3 needs to be in saline at the day of manufacture (Fig. 3F and G). Release kinetics of characterized by enzymatic studies. Knockdown of RNA levels was NanoKS100 revealed continuous release of the agent over 48 hours verified by qRT-PCR (Supplementary Fig. S1B). Collectively, these with maximal release of 70% occurring by 48 hours (Fig. 3C). results suggest the successful development of a novel, potent, multi- The efficacy and specificity of NanoKS100 for killing cultured ALDH inhibitor. melanoma cells were examined by MTS assay and compared with

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 453

Dinavahi et al.

Figure 3. Development and characterization of the nanoliposomal formulation of KS100, called NanoKS100. NanoKS100 consists of an aqueous core surrounded by a phospholipid bilayer. KS100 is contained within the phospholipid bilayer (A). NanoKS100 was manufactured with a 68.6% loading efﬁciency of KS100 into

nanoliposomes (B). KS100 is released from the nanoliposomal formulation continuously for 48 hours with the maximal release of 70% (C). Cell killing IC50s for KS100 and NanoKS100 against BRAF mutant (UACC 903, 1205 Lu) and wild-type (C8161.CI9, MelJuSo) melanoma cell lines were calculated and compared with that of normal human fibroblasts (FF2441) and melanocytes (NHEM, D). KS100 was approximately 4.5-fold, and NanoKS100 was approximately 5-fold more selective for killing melanoma cells compared with FF2441 and NHEM cells. NanoKS100 is stable for at least 12 months when stored at 4 C with no significant changes in IC50s(E), size (F), or charge (G). NanoKS100 causes significantly lower hemolysis compared with KS100 in both mouse and rat red blood cells. Triton X-100 served as the positive control (H).

fi KS100. The IC50 killing ef cacy of NanoKS100 on FF2441 and NHEM Toxicity of NanoKS100 was further examined in Swiss Webster cells was 11.5 mmol/L compared with 2.3 mmol/L across all melanoma mice (n ¼ 3) treated with i.v. NanoKS100 at 5 to 60 mg/kg for 7 days cells, irrespective of BRAF mutational status, amounting to a killing and compared with empty liposomes. Results revealed negligible selectivity index of approximately 5-fold higher for melanoma cells, weight loss on average (0.6 to 2.5%), with no mortality or abnormal similar to that of KS100 (Fig. 3D). Thus, KS100 maintained its behavioral changes seen in any treatment group (Fig. 4A). The melanoma cell killing efficacy and selectivity in the NanoKS100 maximum dose that could be administered to animals was 60 mg/kg formulation. The cell killing IC50s(Fig. 3E), size (Fig. 3F), and charge as the nanoliposomes were not stable above this loaded concentration. (Fig. 3G) of NanoKS100 did not vary significantly over a 12-month Thus, an MTD of NanoKS100 could not be attained, as doses above period when stored at 4C, indicating stability of the formulation. 60 mg/kg could not be tested. Because i.v. dosing of nanoliposomes can trigger injection site hemolysis (40–42), the effect of NanoKS100 on red blood cell (RBC) NanoKS100 inhibits melanoma tumor development with no lysis was examined. RBCs from mice and rats were incubated with apparent toxicity KS100 or NanoKS100 for 1 hour, and the amount of hemolysis was Having identified the safe dose range of NanoKS100, three sub- quantified. KS100 caused 27% and 19% hemolysis of mouse and rat maximal doses (10, 20, and 30 mg/kg) were selected for tumor- RBCs, respectively, compared with 100% hemolysis with the Triton X- inhibitory studies. UACC 903 melanoma cells were injected into the 100–positive control (Fig. 3H). However, NanoKS100 lysed <5% of flanks of nude mice (n ¼ 6), and once vascularized tumors had formed, RBCs in both groups indicating a protective effect of the nanolipo- mice were treated with daily i.v. NanoKS100 for 20 days. All three somal formulation. treatment groups showed significant inhibition of melanoma

454 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

Figure 4. NanoKS100 inhibited melanoma tumor growth with negligible toxicity. A 7-day repeated dose study was conducted for NanoKS100. NanoKS100 was administered i. v. daily at various doses, whereas animal body weight, physical and behavioral changes, and mortality were monitored (A). NanoKS100 significantly inhibited tumor growth of UACC 903 xenografts compared with empty liposome vehicle control following 20 days of treatment. No significant difference in tumor growth was seen between the NanoKS100 treatment groups (B). NanoKS100 at 20 mg/kg body weight administered daily i.v. led to an approximately 65% reduction in tumor growth in UACC 903 (C) and 1205 Lu (D) xenografts following 20 to 22 days of treatment. NanoKS100 did not significantly affect animal body weight (4C, D-insets) or serum biomarkers of toxicity (E) compared with empty liposome vehicle control. Normal reference ranges for serum biomarkers are included. xenograft growth compared with empty liposomes (Fig. 4B). No KS100 inhibits total cellular ALDH activity to increase ROS, lipid significant differences in toxicity and tumor volumes between groups peroxidation, toxic aldehyde accumulation, apoptosis, and were observed. autophagy Based on these findings, treatment with daily i.v. 20 mg/kg The ALDHs reduce ROS, lipid peroxidation, and toxic aldehyde NanoKS100 was selected for further xenograft experiments (n ¼ 8). accumulation, which can lead to cell damage and apoptosis as shown A >65% reduction in tumor volumes was observed for NanoKS100 in in Fig. 5A (16, 17). To evaluate the effects of KS100 on total cellular both UACC 903 (Fig. 4C) and 1205 Lu (Fig. 4D) xenografts at days 20 ALDH activity, UACC 903 (Fig. 5B) and 1205 Lu (Fig. 5C) cells were to 22 with no significant reduction in animal weights compared with treated with 5 mmol/L of ALDH inhibitor or DMSO for 24 hours, þ the empty liposomes (insets of Fig. 4C and D). The blood of the mice stained with AldeRed reagent, and ALDH cells (cells with high levels with UACC 903 xenografted tumors was collected at day 20, and no of ALDH) were identified using flow cytometry. KS100 was the only þ significant differences in serum biomarkers between NanoKS100 and ALDH inhibitor that significantly reduced ALDH cells in UACC 903 empty liposomes were observed (Fig. 4E). Collectively, these data (52% reduction) and 1205 Lu (57% reduction) cells (representative dot suggest that daily i.v. administration of a submaximal dose of plots in Supplementary Figs. S3 and S4). In addition, UACC 903 NanoKS100 (3-fold lower) is safe and effective in this mouse mela- (Supplementary Fig. S5A) and 1205 Lu (Supplementary Fig. S5B) cell noma model. lysates were treated with 1 mmol/L of ALDH inhibitor or DMSO for 15

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 455

Dinavahi et al.

Figure 5. KS100 reduced total cellular ALDH activity to increase ROS generation, lipid peroxidation, and toxic aldehyde accumulation leading to apoptosis and autophagy. The ALDHs reduce ROS generation, lipid peroxidation, and toxic aldehyde accumulation, the latter of which can lead to cell damage and apoptosis (A). KS100 was the only ALDH inhibitor that significantly reduced ALDHþ cells in both UACC 903 (B) and 1205 Lu (C) cells. ALDHþ cells were analyzed by flow cytometry following staining with AldeRed. DMSO served as the control. UACC 903 (D) and 1205 Lu (E) cells treated with KS100 had increased ROS activity compared with the other ALDH inhibitors tested. DMSO served as control. No ALDH inhibitor significantly increased ROS activity in normal human fibroblasts (FF2441) compared with the DMSO control (F). UACC 903 (G) and 1205 Lu (H) cells treated with KS100 had increased lipid peroxidation and toxic aldehyde accumulation compared with the other ALDH inhibitors tested. DMSO served as the control. Flow cytometric analysis of apoptosis in UACC 903 (I) and 1205 Lu (J) cells treated with 5 mmol/L of ALDH inhibitor for 24 hours showed significantly increased apoptosis with KS100 compared with the other ALDH inhibitors tested in both cell lines. DMSO served as the control. Western blot of increasing concentrations of KS100 (2, 4, and 6 mmol/L) showed increased apoptosis (cleaved-PARP) and autophagy (LC3B) in UACC 903 cells after 24 hours of treatment (K).

minutes followed by the addition of aldehyde substrate mixture. KS100 or DMSO for 24 hours. Consistent with the ROS assay, KS100 was the was the most effective at reducing total cellular ALDH activity in both most effective at increasing lipid peroxidation and toxic aldehyde UACC 903 (75% reduction) and 1205 Lu (73% reduction) cells. The accumulation in both cell lines (Fig. 5G and H). DEAB and CM037 remaining ALDH inhibitors significantly reduced total cellular ALDH were the only other inhibitors that significantly increased lipid per- activity compared with controls, particularly CM037 and DEAB, while oxidation and toxic aldehyde accumulation in either cell line. isatin was ineffective. Flow cytometric analysis showed that 5 mmol/L KS100 significantly Levels of ROS were measured in UACC 903 (Fig. 5D) and 1205 Lu increased Annexin-V–positive UACC 903 and 1205 Lu cells compared (Fig. 5E) cells and compared with FF2441 cells (Fig. 5F) following with 5 mmol/L of the other ALDH inhibitors after 24 hours (repre- treatment with 5 mmol/L of ALDH inhibitor or DMSO for 24 hours. sentative dot plots are shown in Supplementary Figs. S6 and S7). No ALDH inhibitor had an effect on ROS levels in FF2441 cells Specifically, KS100 increased the early apoptotic cell fraction þ (Fig. 5F). KS100 was the most effective at increasing ROS levels in both (Annexin-V 7-AAD ) from 9.5% to 22.4% in UACC 903 cells cell lines (Fig. 5D and E). DEAB and CM037 were the only other (Fig. 5I) and from 12.5% to 60.4% in 1205 Lu cells (Fig. 5J). Western agents that significantly increased ROS levels in either cell line. blot analysis of cultured UACC 903 cells following treatment with Subsequently, levels of lipid peroxidation and toxic aldehyde accu- increasing concentrations (2–6 mmol/L) of KS100 for 24 hours mulation were measured in UACC 903 (Fig. 5G) and 1205 Lu (Fig. 5K) showed increased apoptosis and autophagy, exemplified by (Fig. 5H) cells following treatment with 5 mmol/L of ALDH inhibitor elevated levels of cleaved PARP and LC3B, respectively. Collectively,

456 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

these data demonstrate that KS100 significantly reduces total cellular toward ALDH1A1 and 2 compared with DEAB. Importantly, KS100 ALDH activity to increase ROS generation, lipid peroxidation, and exhibited a >60-fold increased potency toward ALDH3A1 compared accumulation of toxic aldehydes leading to increased apoptosis and with DEAB. Further, KS100 has an approximately 24-fold and approx- autophagy. imately 21-fold increased potency toward ALDH1A1 and 3A1, respectively, compared with DIMATE (22, 60, 62). Finally, KS100 has an approximately 3-fold and 5-fold increased potency toward ALDH1A1 Discussion and 3A1, respectively, compared with aldi-6, while having slightly less Increased metabolism of toxic aldehydes through ALDH upregula- potency toward ALDH2 (24). ALDH1A family and 3A1 overexpres- tion can facilitate cancer progression and therapy resistance (9, 17, 53). sion are important in cancer progression and therapy resistance, Thus, numerous ALDH inhibitors have been developed as anticancer whereas ALDH2 inhibition could potentially lead to toxicity through agents and show variable efficacy in the treatment of breast (54–56), disruption of ethanol metabolism (11, 17, 20, 21). Thus, KS100 is likely lung (18, 23, 25, 33), hepatocellular (57), ovarian (26, 29, 30, 58, 59), a more efficacious anticancer ALDH inhibitor than any currently gastric (25), colon (25), prostate (60), and HNSCC (24, 25, 28, 61) as reported multi-ALDH isoform inhibitors. well as glioblastoma (18, 33), leukemia (22), and melanoma (62). KS100 showed efficacy and selectivity for killing cultured mel- Elevated ALDH activity is typically a composite of multiple ALDH anoma cells. Off-target interactions of KS100 as a cause for toxicity isoforms (14, 15). The major isoforms whose overexpression is impli- were minimal based on the Erebus algorithm. However, studies cated in cancer progression and drug resistance include the ALDH1A in mice revealed KS100 to be toxic starting at 5 mg/kg/day due to and 3A family (9, 11, 15, 17, 20, 21, 53). ALDH2 has also been ALDH inhibition. Therefore, a nanoliposomal formulation, called extensively characterized and implicated in various disease states, NanoKS100, was developed to overcome toxicity, as nanoliposomal including alcohol-based cancers (11, 15, 17). Thus, ALDH1A1, 2, and formulations have been shown to decrease toxicity and increase the 3A1 were selected to be the focus of this study. bioavailability of compounds (40–42, 52). A PEGylated liposomal Here, we show that ALDH1A1, 2, and 3A1 overexpression is formulation was chosen as it increases the bioavailability of nano- associated with melanoma progression and that targeting multiple liposomesbyreducingdrugaccumulation in the liver and spleen, ALDH isoforms is more effective for melanoma treatment, likely due thus bypassing reticuloendothelial elimination (40, 42). to the overlapping ability of ALDH isoforms to metabolize toxic NanoKS100 had a 68.6% loading efficiency and remained stable aldehydes (11). These data are consistent with previous reports in for at least 12 months when stored at 4C. NanoKS100 showed which knockdown of ALDH1A1, 2, and 3A1 had minimal effect on similar efficacy and selectivity for killing cultured melanoma cells cancer cell proliferation (63, 64). Subsequently, we undertook a compared with KS100. Importantly, NanoKS100 did not exhibit structure-based drug designing campaign using the backbones of toxicity even at 60 mg/kg/day, the maximum dosage that could be isatin and Cpd 3 to identify a novel, potent, multi-ALDH isoform manufactured based on the loading efficiency and stability of the inhibitor, called KS100, in an attempt to increase anticancer efficacy formulation. and reduce resistance development mediated by the ALDHs. KS100 NanoKS100 was significantly more effective at inhibiting melanoma was evaluated for ALDH-inhibitory activity and found to be a potent tumor growth compared with empty liposomes at 10 to 30 mg/kg/day ALDH1A1, 2, and 3A1 inhibitor. i.v. No significant difference in tumor killing efficacy was observed Currently available multi-ALDH isoform inhibitors include DEAB, among treatment groups, so 20 mg/kg/day i.v. was selected for further DIMATE, citral, and aldis-1, -2, -3, -4, and -6 (11, 23–25). DEAB evaluation in multiple xenograft models. NanoKS100 at 20 mg/kg/day inhibits cultured ovarian cancer and melanoma cells, but minimal i.v. led to a 65% reduction in UACC 903 and 1205 Lu xenografts in vivo fi studies have been undertaken (11, 58). DIMATE has an IC50 of compared with empty liposomes. It also caused no signi cant reduc- 5 mmol/L toward ALDH1A1 and 3A1, inhibits cultured prostate cancer tion in animal weight or increase in serum biomarkers of major organ cells, and administered at 14 mg/kg i.p. daily, reduces growth of function. Collectively, these results indicate efficacy with no apparent melanoma xenografts (22, 60, 62). However, oral bioavailability of toxicity of NanoKS100 in inhibiting melanoma, which are similar to DIMATE and its evaluation in clinical trials has not yet been reported. results of aldi-6 in HNSCC xenograft models at 24 mg/kg/day using Citral inhibits cultured breast cancer cells and xenografts, particularly osmotic mini-pumps (24). However, NanoKS100 was evaluated at a when encapsulated into nanoparticles; however, it has a wide range of submaximal dose (3-fold lower), and the nanoliposomal formulation off-target activities leading to toxicity (55, 56). abrogated the need for osmotic mini-pumps, which are costly and m Aldis-1, -2, -3, and -4 have IC50 values of 2.2 to 7.9 mol/L for invasive. ALDH1A1, 5.4 to 8.6 mmol/L for ALDH2, and 1.7–12 mmol/L for Mechanistically, KS100 was the most effective ALDH inhibitor at ALDH3A1 (23). They inhibit cultured lung cancer and HNSCC cells, reducing total cellular ALDH activity in UACC 903 and 1205 cells. particularly as combinatorial therapy (23, 25). Aldi-6 has superior Specifically, KS100 was the only ALDH inhibitor that significantly þ multi-ALDH isoform potency, with an IC50 of 600 nmol/L for reduced ALDH cells in UACC 903 (52% reduction) and 1205 Lu ALDH1A1, 800 nmol/L for ALDH2, and 1 mmol/L for ALDH3A1 (24). (57% reduction) cell lines, and was at least 2-fold more effective at Aldi-6 inhibits cultured HNSCC cells as monotherapy and in com- reducing total cellular ALDH activity in cell lysates compared with the bination with cisplatin (24). Further, administration at 24 mg/kg/day other ALDH inhibitors. Further, KS100 significantly increased ROS using implantable osmotic mini pumps led to 60% tumor reduction in activity only in melanoma cells. Specifically, KS100 was 2.8-fold and HNSCC xenograft models, an effect enhanced with cisplatin, with no 4.8-fold more effective in increasing ROS levels compared with DEAB toxicity observed (24). However, due to its recent development, it has and CM037, respectively, in both cell lines. The remaining ALDH not been evaluated further. Thus, there is a continued need to develop inhibitors had no significant effect. Similarly, KS100 was 5.6-fold and multi-ALDH isoform inhibitors as anticancer agents, with KS100 11.5-fold more effective in increasing lipid peroxidation and the being a potentially useful drug in this therapeutic field. accumulation of toxic aldehydes compared with DEAB and CM037, Enzymatic IC50s for KS100 were 207, 1,410, and 240 nmol/L toward respectively, in both cell lines, whereas the other ALDH inhibitors had ALDH1A1, 2, and 3A1, respectively. KS100 had similar potency no significant effect.

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 457

Dinavahi et al.

KS100 also led to increased apoptosis and autophagy, as exemplified Disclosure of Potential Conflicts of Interest by elevated levels of cleaved PARP and LC3B, respectively. KS100- No potential conflicts of interest were disclosed. induced apoptosis was further verified by flow cytometry using Annexin V staining. Specifically, KS100 induced an approximately Authors’ Contributions 2- to 5-fold increase in the early apoptotic cell fraction (Annexin- Conception and design: S.S. Dinavahi, R. Gowda, K. Gowda, S. Amin, G.P. Robertson þ V 7-AAD ) in both cell lines. All other ALDH inhibitors had no effect Development of methodology: S.S. Dinavahi, R. Gowda, K. Gowda, C.G. Bazewicz, on apoptosis. Thus, reduced cellular ALDH activity as well as increased V.R. Chirasani, S. Amin, G.P. Robertson ROS generation, lipid peroxidation, toxic aldehyde accumulation, and Acquisition of data (provided animals, acquired and managed patients, provided apoptosis were substantially higher with KS100 in UACC 903 and 1205 facilities, etc.): S.S. Dinavahi, R. Gowda, K. Gowda, C.G. Bazewicz, G.P. Robertson fi Analysis and interpretation of data (e.g., statistical analysis, biostatistics, cells, con rming the superiority of KS100 at inhibiting multiple ALDH computational analysis): S.S. Dinavahi, R. Gowda, K. Gowda, C.G. Bazewicz, isoforms and its utility as an anticancer agent. V.R. Chirasani, M.B. Battu, A. Berg, N.V. Dokholyan, S. Amin, G.P. Robertson In conclusion, this study demonstrates the association of ALDH Writing, review, and/or revision of the manuscript: S.S. Dinavahi, R. Gowda, overexpression in melanoma progression and that targeting multiple K. Gowda, C.G. Bazewicz, V.R. Chirasani, M.B. Battu, N.V. Dokholyan, S. Amin, ALDH isoforms with overlapping functions may be necessary for G.P. Robertson effective anticancer therapy as well as for preventing resistance. A Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.S. Dinavahi, G.P. Robertson novel, potent, multi-isoform ALDH inhibitor, called KS100, was Study supervision: R. Gowda, S. Amin, G.P. Robertson successfully synthesized and characterized. A nanoliposomal formulation of KS100, called NanoKS100, was developed to minimize Acknowledgments toxicity and exhibited significant tumor-inhibitory activity in mela- G.P. Robertson received research grants from The Foreman Foundation for noma xenograft models. KS100 efficiently inhibited ALDH1A1, 2, and Melanoma Research, The Geltrude Foundation, The Penn State Chocolate Tour 3A1 enzymatic activity, leading to increased ROS generation, lipid Cancer Research Fund, and The Melanoma Research Alliance to support this peroxidation, and toxic aldehyde levels, as well as apoptosis and project. N.V. Dokholyan received grants from Passan Foundation, Bridge V grant from Penn State University, and NIH grant GM114015 to support this autophagy. A limitation of the study was that KS100 could not be project. evaluated for its inhibitory activity against other ALDH isoforms due to lack of commercially available enzymes and x-ray crystal structures. Thus, an siRNA screen to evaluate the specificity of KS100 on other The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance ALDH isoforms was performed showing inhibition of ALDH1A1, with 18 U.S.C. Section 1734 solely to indicate this fact. 1A2, 1A3, 2, and 3A1. Future studies could look at the efficacy of NanoKS100 in other ALDH-overexpressing cancers and its potential as combinatorial therapy with agents having resistance mechanisms Received April 9, 2019; revised August 22, 2019; accepted November 13, 2019; through the ALDHs. published first November 21, 2019.

References 1. Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Cowey CL, Lao CD, et al. 11. Koppaka V, Thompson DC, Chen Y, Ellermann M, Nicolaou KC, Juvonen RO, Combined nivolumab and ipilimumab or monotherapy in untreated melanoma. et al. Aldehyde dehydrogenase inhibitors: a comprehensive review of the N Engl J Med 2015;373:23–34. pharmacology, mechanism of action, substrate specificity, and clinical applica- 2. Ascierto PA, McArthur GA, Dreno B, Atkinson V, Liszkay G, Di Giacomo AM, tion. Pharmacol Rev 2012;64:520–39. et al. Cobimetinib combined with vemurafenib in advanced BRAF(V600)- 12. Sladek NE. Human aldehyde dehydrogenases: potential pathological, pharma- mutant melanoma (coBRIM): updated efficacy results from a randomised, cological, and toxicological impact. J Biochem Mol Toxicol 2003;17:7–23. double-blind, phase 3 trial. Lancet Oncol 2016;17:1248–60. 13. Ma I, Allan AL. The role of human aldehyde dehydrogenase in normal and 3. Nguyen LV, Vanner R, Dirks P, Eaves CJ. Cancer stem cells: an evolving concept. cancer stem cells. Stem cell reviews 2011;7:292–306. Nat Rev Cancer 2012;12:133–43. 14. Jackson B, Brocker C, Thompson DC, Black W, Vasiliou K, Nebert DW, et al. 4. Vlashi E, Pajonk F. Cancer stem cells, cancer cell plasticity and radiation therapy. Update on the aldehyde dehydrogenase gene (ALDH) superfamily. Semin Cancer Biol 2015;31:28–35. Hum Genomics 2011;5:283–303. 5. Ajani JA, Wang X, Song S, Suzuki A, Taketa T, Sudo K, et al. ALDH-1 expression 15. Zhou L, Sheng D, Wang D, Ma W, Deng Q, Deng L, et al. Identification of cancer- levels predict response or resistance to preoperative chemoradiation in resectable type specific expression patterns for active aldehyde dehydrogenase (ALDH) esophageal cancer patients. Mol Oncol 2014;8:142–9. isoforms in ALDEFLUOR assay. Cell Biol Toxicol 2019;35:161–77. 6.HuangCP,TsaiMF,ChangTH,TangWC,ChenSY,LaiHH,etal. 16. Shoeb M, Ansari NH, Srivastava SK, Ramana KV. 4-Hydroxynonenal in ALDH-positive lung cancer stem cells confer resistance to epidermal the pathogenesis and progression of human diseases. Curr Med Chem 2014; growth factor receptor tyrosine kinase inhibitors. Cancer Lett 2013;328: 21:230–7. 144–51. 17. Rodriguez-Zavala JS, Calleja LF, Moreno-Sanchez R, Yoval-Sanchez B. Role of 7. Gilabert M, Launay S, Ginestier C, Bertucci F, Audebert S, Pophillat M, et al. Poly aldehyde dehydrogenases in physiopathological processes. Chem Res Toxicol (ADP-ribose) polymerase 1 (PARP1) overexpression in human breast cancer 2019;32:405–20. stem cells and resistance to olaparib. PLoS One 2014;9:e104302. 18. Parajuli B, Georgiadis TM, Fishel ML, Hurley TD. Development of selective 8. Januchowski R, Wojtowicz K, Zabel M. The role of aldehyde dehydroge- inhibitors for human aldehyde dehydrogenase 3A1 (ALDH3A1) for the nase (ALDH) in cancer drug resistance. Biomed Pharmacother 2013;67: enhancement of cyclophosphamide cytotoxicity. ChemBioChem 2014;15: 669–80. 701–12. 9. Moreb JS, Mohuczy D, Ostmark B, Zucali JR. RNAi-mediated knockdown of 19. Rodriguez-Torres M, Allan AL. Aldehyde dehydrogenase as a marker and aldehyde dehydrogenase class-1A1 and class-3A1 is specific and reveals that each functional mediator of metastasis in solid tumors. Clin Exp Metastasis 2016; contributes equally to the resistance against 4-hydroperoxycyclophosphamide. 33:97–113. Cancer Chemother Pharmacol 2007;59:127–36. 20. Durinikova E, Kozovska Z, Poturnajova M, Plava J, Cierna Z, Babelova A, et al. 10. Marchitti SA, Brocker C, Stagos D, Vasiliou V. Non-P450 aldehyde oxidizing ALDH1A3 upregulation and spontaneous metastasis formation is associated enzymes: the aldehyde dehydrogenase superfamily. Expert Opin Drug Metab with acquired chemoresistance in colorectal cancer cells. BMC Cancer 2018;18: Toxicol 2008;4:697–720. 848.

458 Mol Cancer Ther; 19(2) February 2020 MOLECULAR CANCER THERAPEUTICS

ALDH Inhibition for Melanoma

21. Hartomo TB, Van Huyen Pham T, Yamamoto N, Hirase S, Hasegawa D, Kosaka chondria-mediated apoptosis in non-small cell lung cancer A549 cells. Y, et al. Involvement of aldehyde dehydrogenase 1A2 in the regulation of cancer Tumour Biol 2016;37:12579–87. – – stem cell properties in neuroblastoma. Int J Oncol 2015;46:1089 98. 44. Rao PC, Begum S, Sahai M, Sriram DS. Coptisine-induced cell cycle arrest at G2 22. Venton G, Perez-Alea M, Baier C, Fournet G, Quash G, Labiad Y, et al. Aldehyde M phase and reactive oxygen species-dependent mitochondria-mediated apo- dehydrogenases inhibition eradicates leukemia stem cells while sparing normal ptosis in non-small-cell lung cancer A549 cells. Tumour Biol 2017;39: progenitors. Blood Cancer J 2016;6:e469. 1010428317694565. 23. Khanna M, Chen CH, Kimble-Hill A, Parajuli B, Perez-Miller S, Baskaran S, et al. 45. Dinavahi SS, Nyayapathy S, Perumal Y, Dharmarajan S, Viswanadha S. Com- Discovery of a novel class of covalent inhibitor for aldehyde dehydrogenases. bined inhibition of PDE4 and PI3Kdelta modulates the inflammatory compo- J Biol Chem 2011;286:43486–94. nent involved in the progression of chronic obstructive pulmonary disease. 24. Kim J, Shin JH, Chen CH, Cruz L, Farnebo L, Yang J, et al. Targeting aldehyde Drug Res (Stuttg) 2014;64:214–9. dehydrogenase activity in head and neck squamous cell carcinoma with a novel 46. Dinavahi SS, Noory MA, Gowda R, Drabick JJ, Berg A, Neves RI, et al. small molecule inhibitor. Oncotarget 2017;8:52345–56. Moving synergistically acting drug combinations to the clinic by comparing 25. Okazaki S, Shintani S, Hirata Y, Suina K, Semba T, Yamasaki J, et al. Synthetic sequential versus simultaneous drug administrations. Mol Pharmacol 2018; lethality of the ALDH3A1 inhibitor dyclonine and xCT inhibitors in glutathione 93:190–6. deficiency-resistant cancer cells. Oncotarget 2018;9:33832–43. 47. Dinavahi SS, Prasanna R, Dharmarajan S, Perumal Y, Viswanadha S. A novel, 26. Morgan CA, Hurley TD. Characterization of two distinct structural classes of potent, small molecule AKT inhibitor exhibits efficacy against lung cancer cells selective aldehyde dehydrogenase 1A1 inhibitors. J Med Chem 2015;58:1964–75. in vitro. Cancer Res Treat 2015;47:913–20. 27. Overstreet DH, Knapp DJ, Breese GR, Diamond I. A selective ALDH-2 inhibitor 48. Gowda R, Inamdar GS, Kuzu O, Dinavahi SS, Krzeminski J, Battu MB, et al. reduces anxiety in rats. Pharmacol Biochem Behav 2009;94:255–61. Identifying the structure-activity relationship of leelamine necessary for inhibit- 28. Kulsum S, Sudheendra HV, Pandian R, Ravindra DR, Siddappa G, R N, et al. ing intracellular cholesterol transport. Oncotarget 2017;8:28260–77. Cancer stem cell mediated acquired chemoresistance in head and neck cancer 49. Kuzu OF, Gowda R, Sharma A, Noory MA, Dinavahi SS, Kardos G, et al. can be abrogated by aldehyde dehydrogenase 1 A1 inhibition. Mol Carcinog Improving pharmacological targeting of AKT in melanoma. Cancer Lett 2017; 2017;56:694–711. 404:29–36. 29. Yang SM, Martinez NJ, Yasgar A, Danchik C, Johansson C, Wang Y, et al. 50. Uchida K. 4-Hydroxy-2-nonenal: a product and mediator of oxidative stress. Discovery of orally bioavailable, quinoline-based aldehyde dehydrogenase Prog Lipid Res 2003;42:318–43. 1A1 (ALDH1A1) inhibitors with potent cellular activity. J Med Chem 2018; 51. Kardos GR, Wastyk HC, Robertson GP. Disruption of proline synthesis in 61:4883–903. melanoma inhibits protein production mediated by the GCN2 pathway. 30. Huddle BC, Grimley E, Buchman CD, Chtcherbinine M, Debnath B, Mehta P, Mol Cancer Res 2015;13:1408–20. et al. Structure-based optimization of a novel class of aldehyde dehydrogenase 1A 52. Brys AK, Gowda R, Loriaux DB, Robertson GP, Mosca PJ. Nanotechnology- (ALDH1A) subfamily-selective inhibitors as potential adjuncts to ovarian cancer based strategies for combating toxicity and resistance in melanoma therapy. chemotherapy. J Med Chem 2018;61:8754–73. Biotechnol Adv 2016;34:565–77. 31. Vine KL, Locke JM, Ranson M, Pyne SG, Bremner JB. An investigation into the 53. Allison SE, Chen Y, Petrovic N, Zhang J, Bourget K, Mackenzie PI, et al. cytotoxicity and mode of action of some novel N-alkyl-substituted isatins. J Med Activation of ALDH1A1 in MDA-MB-468 breast cancer cells that over- Chem 2007;50:5109–17. express CYP2J2 protects against paclitaxel-dependent cell death mediated by 32. Krishnegowda G, Prakasha Gowda AS, Tagaram HR, Carroll KF, Irby RB, reactive oxygen species. Biochem Pharmacol 2017;143:79–89. Sharma AK, et al. Synthesis and biological evaluation of a novel class of isatin 54. Matsunaga N, Ogino T, Hara Y, Tanaka T, Koyanagi S, Ohdo S. Optimized analogs as dual inhibitors of tubulin polymerization and Akt pathway. dosing schedule based on circadian dynamics of mouse breast cancer stem cells Bioorg Med Chem 2011;19:6006–14. improves the antitumor effects of aldehyde dehydrogenase inhibitor. Cancer Res 33. Parajuli B, Fishel ML, Hurley TD. Selective ALDH3A1 inhibition by benzimid- 2018;78:3698–708. azole analogues increase mafosfamide sensitivity in cancer cells. J Med Chem 55. Thomas ML, de Antueno R, Coyle KM, Sultan M, Cruickshank BM, Giaco- 2014;57:449–61. mantonio MA, et al. Citral reduces breast tumor growth by inhibiting the cancer 34. Dinavahi SS, Bazewicz CG, Gowda R, Robertson GP. Aldehyde dehydrogenase stem cell marker ALDH1A3. Mol Oncol 2016;10:1485–96. inhibitors for cancer therapeutics. Trends Pharmacol Sci 2019;40:774–89. 56. Zeng S, Kapur A, Patankar MS, Xiong MP. Formulation, characterization, and 35. Pulla VK, Alvala M, Sriram DS, Viswanadha S, Sriram D, Yogeeswari P. antitumor properties of trans- and cis-citral in the 4T1 breast cancer xenograft Structure-based drug design of small molecule SIRT1 modulators to treat cancer mouse model. Pharm Res 2015;32:2548–58. and metabolic disorders. J Mol Graph Model 2014;52:46–56. 57. Kreuzer J, Bach NC, Forler D, Sieber SA. Target discovery of acivicin in cancer 36.PullaVK,SriramDS,SoniV,ViswanadhaS,SriramD,YogeeswariP. cells elucidates its mechanism of growth inhibitiondaggerElectronic supplemen- Targeting NAMPT for therapeutic intervention in cancer and inflammation: tary information (ESI) available: Synthesis, cloning, protein expression, purifi- structure-based drug design and biological screening. Chem Biol Drug Des cation and biochemical assays. Chem Sci 2014;6:237–45. 2015;86:881–94. 58. Morgan CA, Parajuli B, Buchman CD, Dria K, Hurley TD. N,N- 37. Pulla VK, Sriram DS, Viswanadha S, Sriram D, Yogeeswari P. Energy-based diethylaminobenzaldehyde (DEAB) as a substrate and mechanism-based inhib- pharmacophore and three-dimensional quantitative structure–activity relation- itor for human ALDH isoenzymes. Chem Biol Interact 2015;234:18–28. ship (3D-QSAR) modeling combined with virtual screening to identify novel 59. Condello S, Morgan CA, Nagdas S, Cao L, Turek J, Hurley TD, et al. beta- small-molecule inhibitors of silent mating-type information regulation 2 homo- Catenin-regulated ALDH1A1 is a target in ovarian cancer spheroids. Oncogene logue 1 (SIRT1). J Chem Inf Model 2016;56:173–87. 2015;34:2297–308. 38. Ding F, Yin S, Dokholyan NV. Rapid flexible docking using a stochastic rotamer 60. Fournet G, Martin G, Quash G. a, b-Acetylenic amino thiolester inhibitors of library of ligands. J Chem Inf Model 2010;50:1623–32. aldehyde dehydrogenases 1&3: suppressors of apoptogenic aldehyde oxidation 39. Shirvanyants D, Alexandrova AN, Dokholyan NV. Rigid substructure search. and activators of apoptosis. Curr Med Chem 2013;20:527–33. Bioinformatics 2011;27:1327–9. 61.YangSM,YasgarA,MillerB,Lal-NagM,BrimacombeK,HuX,etal. 40. Gowda R, Dinavahi SS, Iyer S, Banerjee S, Neves RI, Pameijer CR, et al. Discovery of NCT-501, a potent and selective theophylline-based inhib- Nanoliposomal delivery of cytosolic phospholipase A2 inhibitor arachi- itor of aldehyde dehydrogenase 1A1 (ALDH1A1). J Med Chem 2015;58: donyl trimethyl ketone for melanoma treatment. Nanomedicine 2018;14: 5967–78. 863–73. 62. Perez-Alea M, McGrail K, Sanchez-Redondo S, Ferrer B, Fournet G, Cortes J, 41. Gowda R, Kardos G, Sharma A, Singh S, Robertson GP. Nanoparticle-based et al. ALDH1A3 is epigenetically regulated during melanocyte transformation celecoxib and plumbagin for the synergistic treatment of melanoma. Mol Cancer and is a target for melanoma treatment. Oncogene 2017;36:5695–708. Ther 2017;16:440–52. 63. Yue L, Huang ZM, Fong S, Leong S, Jakowatz JG, Charruyer-Reinwald A, et al. 42. Gowda R, Madhunapantula SV, Sharma A, Kuzu OF, Robertson GP. Nanoli- Targeting ALDH1 to decrease tumorigenicity, growth and metastasis of human polee-007, a novel nanoparticle-based drug containing leelamine for the treat- melanoma. Melanoma Res 2015;25:138–48. ment of melanoma. Mol Cancer Ther 2014;13:2328–40. 64. Luo Y, Dallaglio K, Chen Y, Robinson WA, Robinson SE, McCarter MD, et al. 43. Rao PC, Begum S, Jahromi MA, Jahromi ZH, Sriram S, Sahai M. Cytotoxicity of ALDH1A isozymes are markers of human melanoma stem cells and potential – – withasteroids: withametelin induces cell cycle arrest at G2 M phase and mito- therapeutic targets. Stem Cells 2012;30:2100 13.

AACRJournals.org Mol Cancer Ther; 19(2) February 2020 459

Development of a Novel Multi-Isoform ALDH Inhibitor Effective as an Antimelanoma Agent

Saketh S. Dinavahi, Raghavendra Gowda, Krishne Gowda, et al.

Mol Cancer Ther 2020;19:447-459. Published OnlineFirst November 21, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-19-0360

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2019/11/21/1535-7163.MCT-19-0360.DC1

Cited articles This article cites 64 articles, 7 of which you can access for free at: http://mct.aacrjournals.org/content/19/2/447.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/19/2/447. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.