A Potent, Metabolically Stable Tubulin Inhibitor Targets the Colchicine Binding Site and Overcomes Taxane Resistance

Microtubules are components of the cytoskeleton that are involved in a multitude of essential cellular functions including mitosis, maintenance of cell shape, intracellular transport, motility, and cell signaling (1). They are composed of α- and β-tubulin heterodimers that readily polymerize and depolymerize in cells. Tubulin polymerization dynamics is an attractive cancer drug target (2). Tubulin inhibitors are generally classified as microtubule-stabilizing or -destabilizing agents based on whether they promote tubulin polymerization or depolymerization. FDA-approved stabilizing agents targeting the taxane -binding site (e.g., paclitaxel, docetaxel, and epothilones) and destabilizing agents targeting the vinca alkaloid binding site (e.g., vinblastine, vincristine, and vinorelbine) are already available for clinical use, while compounds targeting the colchicine binding site (e.g., CA-4P) are being evaluated in preclinical studies (3–6). However, the clinical efficacy of the taxanes and vinca alkaloids is often limited by ATP-binding cassette (ABC) transporter mediated drug efflux pumps, including P-glycoprotein (P-gp), breast cancer–resistant proteins (BCRP), and multidrug-resistant proteins (MRP1 or MRP2; refs. 3, 7–9). Tumor cells exhibiting overexpression of the class III β-tubulin isoform also demonstrate resistance to these agents (10–12). Extensive literature reports indicate that compounds interacting with the colchicine binding site are much less sensitive to these clinically observed mechanisms of resistance (13), suggesting that the development of anti-tubulin agents targeting the colchicine-binding site can overcome limitations associated with existing tubulin inhibitors and improve clinical outcome.

While colchicine itself is an approved drug for gout, it is not approved for cancer therapy due to its toxic side effects which include neutropenia, gastrointestinal upset, bone marrow damage, and anemia (14). Other compounds targeting the colchicine site have been developed and many of them have been or are currently being evaluated in clinical trials for cancer. We previously reported the discovery of diaryl-ketone chemotypes, including a phenyl ring as linker (I-387; ref. 15), 4-substituted methoxybenozyl aryl thiazoles (SMART; ref. 16), phenylaminothiazoles (PAT; ref. 17), arylbenzoylimidazoles (ABI; ref. 9), and reverse arylbenzoylimidazoles (RABI; ref. 18) that interfere with tubulin polymerization by binding to the colchicine domain. While these compounds displayed potent anticancer activity in a number of human melanoma and prostate cancer xenograft models, the ketone moiety in their structure presented a metabolically labile site (19). Further structural optimization to incorporate this metabolic soft spot into a stable ring produced a novel class of indolyl-imidazopyridines (IIP), with DJ101 (structure shown in Fig. 1A) identified as the lead compound (20, 21). DJ101 not only possessed excellent metabolic stability as we designed, it also showed improved anticancer potency and effectiveness in overcoming P-gp–mediated multidrug resistance (MDR) and a number of additional mechanisms responsible for paclitaxel resistance (21–23).

To further preclinically evaluate DJ101 and develop it as the first member of this new generation of tubulin inhibitors, we describe our efforts in this report to determine its molecular interactions by obtaining the high-resolution crystal structure of DJ101 in complex with tubulin proteins; demonstrate the broad in vitro potency of DJ101 against a diverse panel of cancer cell lines; confirm its mechanism of action by examining its effects on microtubule morphology, cell migration, and clonogenic potential; evaluate its preclinical efficacy in suppressing melanoma tumor growth and metastasis using melanoma mouse models; show that DJ101 possesses a good safety profile with off-target primary screening, which suggested it may have negligible off-target effects for major physiologically important targets; and finally, demonstrate its efficacy in vivo for overcoming resistance using parental PC-3 and paclitaxel-resistant PC-3/TxR xenograft models.

Human melanoma cell lines A375, SK-MEL-1, RPMI 7951, and WM-115 (ATCC) were cultured in DMEM supplemented with 10% (v/v) FBS (Atlanta Biologicals) and 1% antibiotic/antimycotic mixture (Sigma-Aldrich). Murine melanoma B16F10 cells (ATCC) were cultured in minimum essential medium (Invitrogen), supplemented with 5% heat-inactivated Hyclone FBS (Thermo Scientific), 1% antibiotic/antimycotic mixture (Sigma-Aldrich), 1% MEM-sodium pyruvate (Invitrogen), 1% MEM-vitamin (Invitrogen), l-glutamine (2 mmol/L final concentration; Invitrogen), and 1% MEM NEAA (Invitrogen). Parental prostate cancer PC-3, its paclitaxel-resistant daughter line PC-3/TxR, parental prostate cancer DU-145, and its docetaxel-resistant daughter line DU-145/TxR were gifts from Dr. Evan Keller at the University of Michigan Medical School (Ann Arbor, MI). PC-3 and DU-145 cell lines were cultured in RPMI1640 media (Gibco by Life Technologies) supplemented with 10% (v/v) FBS (Atlanta Biologicals) and 1% antibiotic/antimycotic mixture (Sigma-Aldrich). Taxane-resistant PC-3/TxR and DU-145/TxR cell lines were cultured in the same media and additionally supplemented with 10 nmol/L paclitaxel or docetaxel, respectively. Paclitaxel or docetaxel was not included in the media for PC-3/TxR or DU-145/TxR for at least one week prior to in vitro and in vivo testing. All cell lines were authenticated by ATCC by short tandem repeat profiling. Cultures were maintained to 80%–90% confluency at 37°C in a humidified atmosphere containing 5% CO₂. Compounds were dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) to make a stock solution of 20 mmol/L. Compound solutions were freshly prepared by diluting stocks with cell culture medium before use.

The clones of the stathmin-like domain of RB3 and TTL were generous gifts provided by Dr. Benoît Gigant (CNRS, France) and Dr. Michel O. Steinmetz (PSI, Switzerland). The expression and purification of proteins were described previously (5, 24, 25). The method of vapor diffusion was applied to generate crystals of T2R-TTL, following the detailed procedure which was reported previously (24, 26, 27). The compound DJ101 was dissolved in DMSO to 10 mmol/L concentration. A small amount of DJ101 solution (0.1 μL) was soaked into crystals under microscope, and the soaked crystals were transferred into an incubator. After overnight incubation at 20°C, the soaked crystals were flash cooled in liquid nitrogen for data collection.

X-ray diffraction data were collected at beamline BL19U1 at SSRF (The Shanghai Synchrotron Radiation Facility, National Center for Protein Science Shanghai, Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, China). The native dataset was collected at a wavelength of 0.97853 Å using MX225 CCD detector. Data were indexed, integrated, and scaled using the program HKL2000. The initial phase was determined by molecular replacement using the apo structure T2R-TTL (PDB code: 4I55) as the searching model. The model was further built manually with COOT and refined with Refmac5. The quality of the final model was checked by PROCHECK and showed good stereochemistry according to the Ramachandran plot (28, 29). The data collection and refinement statistics are summarized in Table 1. PyMol was used to generate all the figures.

A375, RPMI7951, WM115, SK-MEL-1, PC-3, DU-145, PC-3/TxR, and DU-145/TxR cells were seeded in 96-well plates at a concentration of 1,000–5,000 cells per well, depending on growth rate of the cell line. After overnight incubation, the media was replaced and cells were treated with the test compounds at ten concentrations ranging from 0.03 nmol/L to 1 μmol/L plus a vehicle (DMSO) control for 72 hours in four replicates. Following treatment, the MTS reagent (Promega) was added to the cells and incubated in dark at 37°C for at least 1 hour. Absorbance at 490 nm was measured using a plate reader (BioTek Instraments Inc.). IC₅₀ values were calculated by nonlinear regression analysis using GraphPad Prism (GraphPad Software). In addition, DJ101 was evaluated at five concentrations on the NCI-60 cell line panel by the National Cancer Institute Developmental Therapeutics Program (NCI/DTP).

A375 and RPMI7951 cells were seeded on glass coverslips in 6-well plates (5 × 10⁵ cells/well) and incubated overnight. Cells were treated with specified concentrations of DJ101, docetaxel, or vehicle (DMSO) control for 18 hours. Microtubules were visualized after incubating with anti-α-tubulin antibody (Thermo Scientific) and Alexa Fluor 647 goat anti-mouse IgG (Molecular Probes). The coverslips were mounted with Prolong Diamond Antifade mounting media with DAPI (Invitrogen) and images acquired with a Zeiss 710 Confocal microscope and Zen imaging software (Zeiss).

A375 and RPMI7951 cells were seeded in 6-well plates (500 cells/well) in replicates of four and incubated at 37°C overnight. Cells were treated with the compound or equivalent vehicle (DMSO) control and incubated for another week. Cells were then fixed with methanol and stained with 0.5% crystal violet. Images were taken and colony area was quantified with ImageJ software (NIH, Bethesda, MD).

A375 and RPMI7951 cells were seeded in 24-well plates (2 × 10⁵ cells/well) in replicates of four and incubated overnight. Using a 200-μL pipette tip, a straight line was scratched through the cell monolayer to remove an area of cells, and cells were washed several times to remove any debris and uprooted cells. Media was replaced with media containing vehicle (DMSO), 25 nmol/L of DJ101, or 25 nmol/L colchicine. Images were obtained after 0, 12, and 24 hours with Evos Fl Imaging System (Life Technologies). Analysis was performed with ImageJ software (NIH, Bethesda, MD).

All animal experiments were performed in accordance with the NIH animal use guidelines and protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Tennessee Health Science Center (UTHSC, Memphis, TN). We first estimated the acute maximum tolerable dose (MTD) for DJ101 formulated in PEG300. By progressively increasing injection doses via intraperitoneal route to ICR mice (two mice in a group; Evigo Corporation), the MTD was estimated to be at least 65 mg/kg. To ensure a safety margin during the repeated treatment, the maximum dose was scaled down to 30 mg/kg in the animal experiments.

Nude mice, 6–8 weeks old, were purchased from Evigo. Logarithmic growth phase A375, PC-3, or PC-3/TxR (5 × 10⁷ cells per mL) cells were prepared in phenyl red-free, FBS-free media and mixed with Matrigel immediately before injection into mice. Tumors were established by injecting 100 μL of this mixture subcutaneously in the dorsal flank of each mouse (2.5 × 10⁶ cells). After 2 weeks, mice were randomly divided into control or treatment groups. DJ101 or paclitaxel was dissolved in a 1:1 ratio of PEG300:PBS solution to produce desired concentrations. The vehicle control solution was formulated with equal parts PEG300 and PBS only. Doses (100 μL) of the drug treatment or vehicle control were administered via intraperitoneal injection every other day for the duration of the studies.

Tumor volume was measured three times a week with a caliper and calculated using the formula a × b² × 0.5, where a and b represented the larger and smaller diameters, respectively. Tumor growth inhibition (TGI) at the conclusion of the experiments was calculated as 100 – 100 × ((T − T₀)/(C − C₀)), where T, T₀, C, and C₀ are the mean tumor volume for the specific group on the last day of treatment, mean tumor volume of the same group on the first day of treatment, mean tumor volume for the vehicle control group on the last day of treatment, and mean tumor volume for the vehicle control group on the first day of treatment, respectively (30). Animal activity and body weights were monitored during the entire experiment period to assess potential acute toxicity. At the end of the experiment, mice were sacrificed and the tumors and tissues were dissected out and fixed in 10% buffered formalin phosphate solution prior to pathology staining analysis.

C57BL/6 mice from Charles River Laboratories International, Inc., age 7–8 weeks old, were used to study the inhibition effect of DJ101 on lung metastasis of melanoma cells.

Murine B16F10 melanoma cells growing in a logarithmic growth phase were suspended in the conditioned media at a density of 1 × 10⁶ per mL. Each mouse was inoculated with the tumor cells (100 μL containing 1 × 10⁵ cells) via the lateral tail vein. The treatment started on the third day after the inoculation to ensure the initiation of metastasis before beginning treatment. DJ101 (30 mg/kg) and vehicle were formulated as described above. Doses (100 μL) of DJ101 or vehicle solution were administered via intraperitoneal injection every other day for 2 weeks. Animal activity and body weights were monitored during the entire experiment period to assess acute toxicity. Mice were sacrificed 15 days after the initiation of the experiment, and the lungs were separated, expanded, and preserved in 10% neutral buffered formalin. The number of lung metastasis nodules was recorded. Major organs were also preserved in the same manner for subsequent toxicologic examination.

Fixed tumor xenograft tissues were embedded in paraffin. Serial sections were obtained and stained with hematoxylin and eosin (H&E) and IHC. Staining was performed with rabbit anti-cleaved caspase-3 antibody (Cell Signaling Technology Inc.) and rabbit anti-CD31 (Cell Signaling Technology Inc.) following ABC-DAB methods. Antigen retrieval was performed with H-3300 antigen unmasking solution (Vector Laboratories). Microscopic images were captured with a digital camera at ×20 magnification. Twenty to thirty images from each section were analyzed. Pathologic tissue sections from major organs (heart, liver, kidney, lung and spleen) were examined similarly to identify any potential drug-related effects. Images were obtained with EVOS XL Core microscope (Life Technologies).

Assessment of potentially significant off-target effects of DJ101 to 47 major physiologically important targets was performed by DiscoverX (DiscoverX Corporation) in 78 assays using its Safety47 Panel and standard protocols. DJ101 was assayed at 1 μmol/L concentrations, representing at least 100× the IC₅₀ value as determined in melanoma cell lines by the current and our earlier studies for DJ101 (21).

Data was analyzed using Prism Software 5.0 (GraphPad Software, Inc.). Data were provided as mean ± SEM unless otherwise indicated. The statistical significance (P < 0.05) was calculated by one-way ANOVA followed by Dunnett multiple comparison test, comparing each treated group to the corresponding control group for the in vitro colony and migration assays and the in vivo xenograft studies. An unpaired Student t test was used to calculate significance for the lung metastasis study.

We previously reported the design and synthesis of DJ101 (structure shown in Fig. 1A; ref. 21) and a number of potent tubulin inhibitors related to this scaffold (9, 16, 18). While mechanistic studies suggested that these molecules produce their antitumor activities by interacting with the colchicine-binding site in tubulin, the detailed molecular interactions have not previously been fully elucidated. Recently, high-resolution (<3.0 Å) crystal structures of tubulin in complex with several known colchicine site inhibitors were reported (26, 27). Using these newly established procedures, we obtained the high-resolution X-ray crystal structure of DJ101 in complex with αβ-tubulin (deposited to the Protein Databank with PDB code: 5H7O, resolution 2.8 Å). The X-ray parameters and the crystal structure information are summarized in Table 1. This high-resolution crystal structure confirms the direct binding of DJ101 in the colchicine-binding site (Fig. 1B). DJ101 forms three hydrogen bonds with the αβ-tubulin dimer: one hydrogen bond from T179 in loop five in the α-monomer to the imidazole nitrogen, one hydrogen bond from N349 in loop nine in the β-monomer to the indole nitrogen, and one hydrogen bond between the oxygen of the 4-methoxyl moiety and the thiol moiety in C241 in helix seven of the β-monomer (Fig. 1C). A tight hydrophobic “sandwich” formed by sidechains from C241, L255, L248, and N258 wraps the trimethoxyphenyl and the imidazopyridine moieties firmly in the colchicine-binding domain. In addition, the sidechain M259 in helix eight of the β-monomer serves as a “wedge” between sheet 9 and DJ101 to lock its curved conformation (Fig. 1C). It is also evident that DJ101 overlaps with colchicine at its binding site (Fig. 1D). These results provide the first direct evidence of DJ101′s direct interaction with the colchicine-binding site in tubulin to inhibit tubulin polymerization.

Our preliminary studies showed that DJ101 has potent cytotoxic activity against a variety of melanoma and prostate cancer cell lines (21). To expand upon this observation and determine its in vitro efficacy against additional metastatic melanoma cell lines that harbor clinically relevant mutations, we first performed cytotoxicity assays for DJ101 against a panel of human metastatic melanoma cell lines including A375, RPMI7951, WM-115, and SK-MEL-1 (Table 2). These cell lines possess diverse genetic complexity and contain genomic mutations in one or more of the following genes: BRAF, CTNNB1, CDKN2A, PTEN, and TP53. DJ101 demonstrated comparable potency against these lines to colchicine, with IC₅₀ values ranging from 7 to 10 nmol/L in each of the tested cell lines. In addition, DJ101 was tested against NCI-60 cell lines representing diverse types of cancer cells, where its GI₅₀ values were generally less than 10 nmol/L (Supplementary Fig. S1).

We revealed in our previous studies that DJ101 induces a strong depolymerizing effect on purified tubulin protein in a cell-free assay (21). To characterize its effects on microtubule networks and cell morphology, we utilized confocal microscopy to visualize A375 and RPMI7951 cells treated for 18 hours with either DJ101 or docetaxel, a potent microtubule-stabilizing agent that promotes microtubule polymerization. The alteration in microtubule rearrangement in the treated cells can clearly be observed (Fig. 2A). Control cells appeared to exhibit well-organized microtubule networks extending throughout the cell to the cell periphery. Treatment with DJ101 led to reduced and fragmented microtubule networks incorporating polymeric tubulin proteins and emitted a weaker fluorescent signal. This effect in DJ101-treated cells was accompanied by an increase in dispersed cytoplasmic fluorescence, representing a shift to the depolymerized, soluble tubulin form. Conversely, treatment with docetaxel led to hyperpolymerized tubulin and resulted in the formation of extensive microtubule bundles concentrated toward the nucleus. The microtubule fragmentation and disorder observed by DJ101 is consistent with its mechanism of action (i.e., inhibiting tubulin polymerization).

DJ101 was tested in a colony-forming assay to evaluate its long-term growth-inhibitory effects on two different metastatic melanoma cell lines (Fig. 2B). In A375 cells, treatment with a low concentration (4 nmol/L) of DJ101 resulted in colonies covering only 30.6% ± 0.8% of the total surface area, which was significantly lower than the cells exposed to only the vehicle (DMSO; 75.6% ± 3.7%; Fig. 2C). Colchicine was used as a reference control because its ability to potently and persistently inhibit colony formation is well-documented (31, 32). The effect of DJ101 on A375 cell colony formation was similar to colchicine at the same concentration, which had colonies covering a total area of 34.7% ± 1.5% (Fig. 2C). Comparable reductions in colony formation were measured in the RPMI7951 cell line treated with either DJ101 (47.1% ± 5.7%) or colchicine (50.9% ± 3.4%), whereas control RPMI7951 cells formed colonies occupying a surface area of 83.2% ± 2.1%. Higher concentrations (20 nmol/L) of DJ101 or colchicine resulted in complete colony obliteration. Dunnett multiple comparison tests after one-way ANOVA analysis gave an overall P value of less than 0.0001 for each treated group compared with the control, indicating that they are statistically different.

To demonstrate that DJ101 interferes with cell migration through microtubule destabilization, we performed a scratch migration assay (Fig. 2D). After 24 hours, untreated A375 and RPMI7951 cells had nearly achieved complete closure of the wound by migrating into 86.2% ± 2.2% and 93.0% ± 3.6% of the scratch area, respectively (Fig. 2E). A375 and RPMI7951 cells treated with DJ101 demonstrated impaired cell migration, occupying only 40.8% ± 6.0% and 24.6% ± 2.4% of the scratch area, respectively. Similar but less inhibition of the scratch area was observed colchicine-treated cells, achieving 52.0% ± 2.7% closure for the A375 cell line and 52.3% ± 3.4% for the RPMI7951 cell line. The one-way ANOVA P values in both cell lines were less than 0.0001, and Dunnett multiple comparison test against the control indicated P values of no more than 0.001 (Fig. 2E). Taken together, our in vitro studies demonstrate that DJ101 strongly reduces aberrant cancer cell proliferation at low concentrations and hinders cell migration at least as efficiently as colchicine in metastatic melanoma.

The antitumor efficacy of DJ101 was first tested in an A375 xenograft model in nude mice. After 2 weeks of treatment, the group of mice receiving 15 mg/kg doses of DJ101 had an average tumor growth inhibition (TGI) of 66.4%, while the group receiving 30 mg/kg doses of DJ101 averaged 92.8% inhibition compared with vehicle control (Fig. 3A). Dunnett multiple comparison test indicated that treatment with 15 mg/kg and 30 mg/kg doses of DJ101 was significantly better than the vehicle (P < 0.001 and 0.0001, respectively) based on the percent increase in final tumor volume. One-way ANOVA analysis gave a P value of < 0.0001 suggesting a significant difference among all groups (Fig. 3A). Mice exhibited normal physical activity during the study and body weights increased slightly (Fig. 3B), indicating negligible acute toxicities. H&E stains of representative tumor sections in the vehicle control group showed aggressive growth with numerous mitotic cells in different stages (Fig. 3C). The increase of metaphase cells and decrease of anaphase cells in the DJ101 treatment groups indicate mitotic blocking in the G₂–M phase, consistent with our previous cell-cycle analysis (21). Tumor sections from both low- and high-dose treatment groups showed significantly less proliferation, and there were abundant apoptotic cells showing dense nuclear pyknosis and cytoplasmic karyorrhexis. There was also extensive central tumor necrosis in the treatment groups. In addition, CD31-labeled endothelial cells in the vehicle control tumor sections exhibited well-developed networks of capillaries or small blood vessels around tumors, while the DJ101 treatment groups displayed severely distorted blood vessels or absence of CD31-positive stains, suggesting potential vascular disrupting properties of DJ101. Finally, caspase-3 staining of tumor sections in the control group showed very few positive regions, but tumor sections from the DJ101 treatment groups showed an increase in caspase-3–positive labeled cells, confirming enhanced apoptosis due to DJ101 treatment.

As a major obstacle in treating melanoma is tumor metastasis, we next assessed the efficacy of DJ101 in suppressing melanoma metastasis using a B16F10 experimental lung metastasis model in mice. After two weeks of treatment with 30 mg/kg doses of DJ101, the average number of tumor nodules that developed on the lungs was 1.7 ± 0.3, whereas the vehicle-treated group averaged 10.5 ± 2.0 nodules, demonstrating a 6.2 times decrease in lung metastasis for those receiving DJ101 treatment (Fig. 3D). An unpaired Student t test gave a P value of <0.001, representing a significant decrease in lung metastasis for the treatment group (Fig. 3D). Body weights (Fig. 3E) and physical activities of mice were normal in both groups. These results show that DJ101 is well-tolerated for doses up to 30 mg/kg in mice and can efficiently reduce the potential for lung metastasis of murine melanoma.

To assess whether DJ101 treatment produces potential organ toxicities, pathologic analysis was performed on major organs collected from both melanoma in vivo studies including the heart, kidney, liver, lung and spleen. Sections stained with hematoxylin and eosin revealed no apparent drug-related injury or pathologic changes in the cellular structure of the various tissues in both xenograft mouse models (Fig. 4A) and the experimental lung metastasis mouse model (Fig. 4B).

In vitro pharmacologic profiling is increasingly being used to identify undesirable off-target activity profiles early in the drug discovery process. To further test for potentially significant off-target effects and determine the safety profile, DJ101 was evaluated at 1 μmol/L in the Safety47 Panel using functional assays with all human targets for safety screening (Fig. 4C; Supplementary Table S1). This in vitro pharmacologic profiling service includes the assessment of the functional response of 47 major physiologically important targets across 78 assays to a test compound and normalized to respective controls. Only a value higher than 70% response is considered significant. DJ101 only showed significant responses in 2 of the 78 assays, namely norepinephrine transporter (NET) blocking and glucocorticoid receptor (GR) antagonism. As the concentration tested was more than 100-fold of the IC₅₀ value for DJ101, which is expected to be well above its practical physiologic concentration, results from this Safety47 Panel screening strongly suggest that DJ101 has minimal potential off-target effects. Collectively, these in vivo and in vitro findings support a good safety profile for DJ101.

We first confirmed our previous results that DJ101 maintains efficacy in paclitaxel resistant cell lines in vitro (21), by determining the potency of DJ101, paclitaxel, docetaxel, and colchicine in the parental PC-3, DU-145, paclitaxel-resistant PC-3/TxR, and docetaxel-resistant DU-145/TxR cell lines (Supplementary Table S2). Paclitaxel and docetaxel outperformed DJ101 and colchicine in the drug-sensitive PC-3 and DU-145 cell line. However, in the PC-3/TxR- and DU-145/TxR–treated cells, DJ101 was the most efficacious among the four and was roughly equipotent in its cytotoxicity to the parental cell lines, while the other three tubulin inhibitors have a significantly large drug-resistant index. We next compared the efficacy of DJ101 and paclitaxel in vivo using both PC-3 and PC-3/TxR xenograft models. Both paclitaxel (15 mg/kg) and DJ101 (30 mg/kg) inhibited tumor growth as determined by tumor volume growth (Fig. 5A). The TGI for the group receiving paclitaxel treatment was 101.1% and 78.8% for the DJ101-treated group in PC-3 xenografts. ANOVA analysis followed by Dunnett multiple comparison test of final volumes resulted in P values of <0.0001, suggesting significant differences. Their efficacy against tumors was also evident based on final tumor weight (Fig. 5B) and representative tumor images (Fig. 5C), where the DJ101- and paclitaxel-treated groups showed a 47.1% (P < 0.001) and 79.3% (P < 0.0001) reduction in tumor weight compared with the control group, respectively. No acute toxicities were observed on the basis of physical activity and body weights (Fig. 5D). In contrast, using the same dosing schedule and frequency as we did in the parental PC-3 model, DJ101 caused a TGI of 104.0%, remarkably demonstrating an overall reduction in tumor volume in the PC-3/TxR xenograft model, whereas paclitaxel only modestly inhibited tumor growth by 37.8% (Fig. 5E). These results were corroborated by final tumor weights, where DJ101 caused a 77.3% reduction and paclitaxel showed only a 35.6% reduction compared with the control group (Fig. 5F). Representative tumor images are shown in Fig. 5G. For both the tumor volumes and tumor weights, one-way ANOVA analysis demonstrated a difference amongst groups of P < 0.0001, and Dunnett multiple comparison test showed a much greater significance between control versus DJ101 (P < 0.0001) and vehicle control versus paclitaxel (P < 0.05). Similarly, no acute toxicities were observed in this model (Fig. 5H).

Interfering with tubulin dynamics is a validated approach for anticancer treatment and many antimitotic microtubule stabilizers and destabilizers are widely used clinically or are undergoing clinical development (33). However, many of these agents cause neurotoxicity, exhibit chemical instability, or have elevated metabolic clearance (19). In addition, the clinical efficacy of many FDA-approved tubulin inhibitors is often limited by drug efflux pumps or overexpression of certain tubulin subtypes, most notably the βIII tubulin isotype (34). Design and development of new tubulin inhibitors targeting the colchicine binding site represent an attractive approach for improving and advancing tubulin inhibitors. We previously reported several indole derivatives interacting with the colchicine-binding site that are highly potent in vitro and metabolically stable (21). Herein, we focused on the most promising compound in this class, DJ101, and further demonstrated the high potency of DJ101 against a broader panel of metastatic melanomas with varying degrees of genetic complexity based on genomic mutations (BRAF, CTNNB1, CDKN2A, PTEN, or TP53) and the NCI-60 panel, providing additional evidence to support its anticancer activities.

Tubulin inhibitors interacting with the colchicine-binding site have diverse structures. Thus it has been very challenging to reliably decipher the molecular interactions between an inhibitor and tubulin using molecular modeling with crystal structures containing a different class of tubulin inhibitors. Another challenge is that available X-ray crystal structures often have low resolution which further impedes reliable molecular modeling studies. In this report, we obtained the high resolution X-ray crystal structure of DJ101 in complex with αβ-tubulin. The crystal structure confirmed the binding of DJ101 at the colchicine-binding site in tubulin, overlapping well with that of colchicine. The three hydrogen bonds formed between DJ101 and the tubulin dimer and additional strong hydrophobic interactions from surrounding residues firmly anchored DJ101 in this colchicine binding site. Interestingly, our previous molecular modeling studies using the crystal structure of 1SA0 showed a different binding pose for DJ101, in which the top portion of the DJ101 has a 180 degree flip (21). Examination of the crystal structures of 1SA0 and the current DJ101 complex clearly revealed significant conformational changes for the T4 loop in the tubulin α-subunit and T7 loop in the tubulin β-subunit. It is apparent that the “closing up” of the T4 to the colchicine-binding pocket in the 1SA0 structure prevented the top moiety of DJ101 in adopting its true binding pose, and thus forced its top moiety to rotate 180 degrees. It is well known that while the colchicine site can accommodate a wide range of structurally distinct molecules (34, 35), seemingly insignificant changes to many of these molecules can result in a total loss of activity. Results from this study underline the variation of conformations accommodated for by α-T4 and β-T7 and their critical contributions to the optimal bindings of different scaffolds at the colchicine site in tubulin, as well as the importance of high-resolution crystal structures over molecular modeling in guiding structure optimizations for tubulin inhibitors interacting at this site.

Taxanes, such as docetaxel, are microtubule-stabilizing agents that target the taxane site within the lumen of polymerized microtubules and alter their conformation to the more stable GTP-bound β-tubulin structure, thereby locking them in the polymerized state (36). Microtubule-destabilizing agents, on the other hand, exert their antimitotic effects by interfering with tubulin dynamics as opposed to simply reducing polymerized tubulin, leading to mitotic arrest and eventual apoptosis (36). To further confirm the mechanism of action of DJ101 and evaluate the physical change in microtubule networks, we performed immunofluorescent imaging studies of tubulin structure using melanoma cells that had been treated with our microtubule destabilizing compound DJ101 or docetaxel which potently stabilizes microtubules. In the control cells, the microtubule networks displayed an arrangement of organized, fibrous microtubules extending throughout the elongated cells. Cells treated with DJ101 displayed dispersed and disordered tubulin fragments and greater cytoplasmic fluorescence consistent with depolymerization and elevated free tubulin. Docetaxel-treated cells displayed dense and abundant arrays of tubulin accompanied by a strong fluorescence signal. The immunofluorescent images clearly differentiated the stabilizing and destabilizing effects on microtubule structure and cellular framework by these two opposing classes of compounds.

In addition to their well-defined roles in supporting cellular structure and their involvement in mitosis, microtubules are also implicated in cell migration and motility. It has been suggested that dynamic instability imposed on the microtubules by suppressing and interfering with tubulin behavior causes cell immobility, as the cell is less able to remodel and respond to change the cell shape demanded for cell migration (32). The antimitotic and antimigratory effects of DJ101 were demonstrated by its ability to inhibit anchorage-dependent colony propagation and suppress cell motility in a wound healing assay. The cessation of colony formation and cell movement was attributed to disruption of microtubule dynamics, although additional studies are needed to further elucidate the mechanism of microtubule-binding drugs on different facets of cellular function.

As a further validation of its anticancer potential, DJ101 effectively inhibited melanoma tumor growth and lung metastasis in vivo in mouse models. Because antimitotic drugs exploit different sites on tubulin and microtubules, synergistic combinations could be investigated to optimize their clinical usefulness. This assertion is supported by a number of reported studies that combined paclitaxel with other tubulin targeting drugs (37–40). It will also be important to combine a novel tubulin inhibitor such as DJ101 with a different targeted agent. Toward this end, we previously demonstrated that strong synergy can be achieved by combining ABI-274 (a less metabolically stable analogue of DJ101) and vemurafenib in a BRAF inhibitor–resistant melanoma xenograft model (30). Our ongoing efforts will continue to investigate combination therapies to maximize the clinical potential of DJ101.

Off-target adverse drug reactions (ADR) often contribute to the high attrition rate in the drug discovery and development process. Reducing potential off-target ADRs early on will help to mitigate the risk of expensive failure in late stages. To assess the safety of DJ101 and its potential off-target effects, the inhibition of DJ101 was evaluated against 47 major physiologically important GPCRs, nuclear hormone receptors, transporters, ion channels, kinases, and nonkinase enzymes with Safety47™ Panel. This pharmacologic profile screening includes the assays recommended by major pharmaceutical companies including AstraZeneca, GlaxoSmithKline, Novartis and Pfizer (41). Of the 78 assays tested, DJ101 at a very high concentration only elicited a functional response to NET and GR antagonism, suggesting a good safety profile for this scaffold.

Finally, while taxanes are some of the most clinically effective chemotherapy drugs available, their clinical success is often limited by the emergence of intrinsic and acquired drug resistance (42). The most common mechanisms of resistance to taxanes are derived from the overexpression of ATP-binding cassette proteins such as P-gp and alterations in tubulin–isoform expression, particularly of βIII-tubulin (13). There is evidence that suggests that tubulin-binding agents that specifically target the colchicine-binding site may circumvent these resistance mechanisms (43). In the paclitaxel-resistant PC-3/TxR prostate cancer cell line, more than 200 genes are upregulated in addition to P-gp overexpression, which represent a large number of paclitaxel resistance mechanisms (22, 23). We demonstrated that DJ101 maintains its potency in both paclitaxel sensitive PC-3 and paclitaxel-resistant PC-3/TxR prostate cancer xenograft models in mice. Other groups have also reported elevated sensitivity of paclitaxel-resistant cancer cells to colchicine-binding site drugs. One such study investigated a variety of microtubule targeting agents that bind to the taxane site, vinca alkaloid binding site, and colchicine binding site for docetaxel resistant MCF-7 breast cancer cells. They reported that while MCF-7_TXT cells demonstrated cross-resistance to vinca alkaloids, they were more sensitive to colchicine-binding site agents including 2MeOE2, ABT-751, CA-4P, and colchicine than the nonresistant counterpart MCF-7 cells (44). This also supports the notion that colchicine-binding agents such as DJ101 may be an alternative treatment when tumors acquire resistance to treatment by taxanes.

In summary, we have obtained the high-resolution X-ray crystal structure of DJ101 which represents a novel class of tubulin inhibitors targeting the colchicine-binding site. DJ101 depolymerizes microtubules in vitro and disrupts microtubule morphology distinctly from agents that stabilize tubulin polymerization. It is effective against a broad panel of metastatic melanomas representing different gene mutations as well as other types of cancers as revealed by the NCI-60 screening. DJ101 demonstrates strong antitumor efficacy and decreases metastasis in two melanoma mouse models without causing apparent toxicity to major organs and possesses a good safety profile. Furthermore, DJ101 maintains potency and efficacy in the paclitaxel-resistant PC-3/TxR xenograft model. Collectively, this preclinical evaluation and our previous studies of DJ101 strongly support its further development as a new generation of tubulin inhibitor targeting the colchicine-binding site.

No potential conflicts of interest were disclosed.

The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the NIH/NCI.

Conception and design: K.E. Arnst, D.-J. Hwang, J.T. Dalton, D.D. Miller, W. Li

Development of methodology: K.E. Arnst, D.-J. Hwang, Y. Xue, J. Yang, J.T. Dalton, W. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.E. Arnst, Y. Xue, T. Costello, D. Hamilton, Q. Chen, J. Yang, W. Li

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.E. Arnst, Y. Wang, D.-J. Hwang, Y. Xue, J. Yang, F. Park, J.T. Dalton, D.D. Miller, W. Li

Writing, review, and/or revision of the manuscript: K.E. Arnst, Y. Wang, D.-J. Hwang, Y. Xue, F. Park, J.T. Dalton, D.D. Miller, W. Li

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.E. Arnst, Y. Xue, J. Yang, W. Li

Study supervision: W. Li

This work is supported by NIH/NCI grants R01CA148706 (to W. Li and D. D. Miller) and National Natural Science Foundation of China (81572995, 81703553 to J. Yang and Y. Wang).

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 with 18 U.S.C. Section 1734 solely to indicate this fact.