SKLB-23bb, A HDAC6-Selective Inhibitor, Exhibits Superior and Broad-Spectrum Antitumor Activity via Additionally Targeting Microtubules

Consisting of 11 members, the classical histone deacetylase (HDAC) enzyme family has emerged as an attractive target in cancer therapy (1). In recent years, 5 HDAC pan-inhibitors (vorinostat, romidepsin, belinostat, panobinostat, and chidamide) were approved for clinical application (2–4). The reported side effects of these drugs include fatigue, nausea, thrombocytopenia, and cardiotoxicity (5). To avoid these undesired responses, the development of isoform-selective inhibitors would be appreciated (6).

HDAC6 has long been referred to as a mysterious member of the HDAC family, for it harbors two catalytic domains and one unique C-terminal zinc finger domain that bind ubiquitin (7). Located in the cytoplasm, HDAC6 mainly interacts with nonhistone proteins (8). Functionally, the inhibition of HDAC6 could potentially disrupt tumor protein homeostasis, block tumor migration, and suppress prosurvival signaling (9). Thus, mostly applied in combination with other chemotherapeutic agents in clinical trials, HDAC6 inhibitors were considered to be antitumor candidates (10, 11). Especially in multiple myeloma, which depends highly on abnormal protein clearance, it was well reported that the inhibition of HDAC6 could synergize with proteasome inhibitors to achieve significant clinical benefits (12–14).

Recently, the structure of the two catalytic domains of HDAC6 was solved, which led to more understanding of the mechanism of this enzyme and elevated directions toward the design of selective inhibitors (15, 16). Tubacin and tubastatin A (Fig. 1) were the first reported HDAC6-selective inhibitors (17, 18). Regretfully, these agents did not have the therapeutic window for cancer treatment due to unfavorable pharmacokinetic profiles. In this decade, several new agents have been announced, and some individual compounds have displayed antitumor potential, alone or synergistically with other drugs (19–22). The most clinically successful examples were ACY1215 (rocilinostat) and its analogue ACY241 (refs. 23, 24; Fig. 1). ACY1215 has entered phase II for multiple myeloma treatment. ACY1215 exhibited potential when administrated in combination with bortezomib (23). Nevertheless, published data implied that the efficiency of ACY1215 as a single agent was limited and its application was restricted to the tumor types susceptible to protein homeostasis disruption (23, 25–27).

Our group previously reported an orally bioavailable HDAC6-selective inhibitor, SKLB-23bb, with antitumor capability (ref. 28; Fig. 1). SKLB-23bb showed therapeutic potential against a panel of tumor types representing solid or hematologic malignancies both in vitro and in vivo. Interestingly, the efficiency of this agent appeared to be superior against solid tumor models. In this study, we found that although SKLB-23bb selectively inhibited HDAC6, its antitumor activity was independent of this enzyme. Dramatically, it was revealed that SKLB-23bb could impact the cellular microtubule system by additionally targeting tubulin. These discoveries explained the superior and broad-spectrum antitumor efficiency of SKLB-23bb. This work also suggested that HDAC6 inhibitors would optimally be administrated in the tumor setting in combination with other chemotherapeutic agents. Exploitation of compounds with additional targets apart of HDAC6 would also be encouraged (29).

The antibodies against HDAC6, Ac-α-tubulin, Ac-H₃, α-tubulin, p-H₃, and PARP were purchased from Santa Cruz Biotechnology. The antibodies against caspase-9, p-AKT, and Bax were purchased from Cell Signaling Technology. The antibodies against caspase-3 and AKT were purchased from Abcam. The antibodies against Bcl-2 and Bcl-X_L were purchased from Sigma Chemical Co. The antibody against β-tubulin was purchased from Genetex. The antibodies against GAPDH and β-actin were purchased from Origene. The reagents of ACY1215, SAHA, colchicine, paclitaxel, and vinblastine were purchased from Meilunbio. SKLB-23bb was synthesized in laboratory as described previously (28).

The following cell lines were obtained from the ATCC: human colon cancer cell lines hct116 and HT29; human ovarian cancer cell lines A2780s and SKOV3; human lung cancer cell lines H460 and A549; human breast cancer cell lines MCF-7 and MDA-MB-231; human melanoma cell line A375; human liver cancer cell line HepG2; human multiple myeloma cell lines ARD, U266, and RPMI-8226; human lymphoma cell lines Ramos, HBL-1, and Jeko-1; human leukemia cell lines MV4-11, K562 and LAMA84s. The multiple myeloma cell line MM1S was kindly provided by Dr. Li Zhang in West China Hospital (Chengdu, China). Cells were cultured in DMEM or RPMI1640 medium according to the instructions from ATCC, with the medium containing 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin at 37 °C in an atmosphere of 5% CO₂. All the cells were tested and authenticated by an AmpFlSTR Identifiler PCR Amplification Kit purchased from Thermo Fisher Scientific in the year of 2016 in our laboratory. All the cell lines were confirmed to be mycoplasma negative via a PCR method. The cell lines were used for experiments or implanted into immunodeficient mice between passages 6 and 14.

An MTT assay was performed to evaluate the drug cytotoxicity against the tumor cell lines. MTT (3- (4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2-H-tetrazolium bromide) was purchased from Sigma Chemical Co. Cells were treated with various concentrations of SKLB-23bb or ACY1215 in 96-well culture plates for 72 hours in final volumes of 200 μL (5,000 cells per well for adherent cells and 20,000 cells per well for suspension cells). Then, 20 μL of MTT (5 mg/mL in PBS) was added to each well and incubated for an additional 1–4 hours. The plate was finally centrifuged at 1,000 rpm for 5 minutes, and then the medium was removed. MTT formazan precipitate was dissolved in 150 μL of DMSO and shaken mechanically for 5 minutes; then, the absorbance readings at a wavelength of 570 nm were taken on a spectrophotometer (Molecular Devices). Cell viabilities of the tested groups were calculated via comparison with the DMSO (vehicle)-treated group. The IC₅₀ values were calculated by curve fitting with GraphPad Prism 6.0 software.

Cells were collected and washed twice with ice-cold PBS. Snap-frozen tumor samples were pulverized in liquid nitrogen and washed twice with ice-cold PBS. The samples were lysed with RIPA lysis buffer containing 1 mmol/L phenylmethylsulfonyl fluoride (PMSF). RIPA lysis buffer and PMSF were purchased from Beyotime Biotechnology. The protein concentration was determined by the Bradford protein assay. The samples were denatured in sample buffer, and equal amounts of protein were separated according to the molecular weight on an 8%–12% SDS-PAGE gel and transferred onto a polyvinylidene difluoride membrane. Membranes were blocked for 1 hour in 5% dried milk in TBST at room temperature and probed overnight at 4°C with the indicated primary antibodies diluted in “blocking buffer.” Blots were washed thrice for approximately 15 minutes with TBST and incubated with a horseradish peroxidase–conjugated species-specific antibody diluted in blocking buffer for 1 hour at room temperature with rotation. After three additional washes, the enhanced chemiluminescent substrate (Abbkine) was added and the images were captured via a ChemiScope 5300 chemiluminescent detection system (Clinx).

The cellular HDAC6 activity of the cell lines were measured by an HDAC6 Activity Assay Kit (Fluorometric), purchased from Biovision. The harvested cells (1–2 × 10⁶) were washed twice in ice-cold PBS and then lysed with 100 μL HDAC6 lysis buffer supplied in the kit. The protein amount in the lysed samples was quantified via a Bradford protein assay and adjusted to 1 mg/mL. The rest of the experiment was conducted following the protocol of the kit. The fluorescence intensity was measured by a spectrophotometer (Molecular Devices).

The SgRNA sequences designed for HDAC6 knockout were as follows: Sg1 forward: 5′-CACCGTCCCTTGCAGTCCCACGATT-3′; Sg2 reverse: 5′-AAACAATCGTGGGACTGCAAGGGAC-3′; Sg2 forward: 5′-CACCGCCAGTGCTACAGTCTCGCAC-3′; Sg2 reverse: 5′-AAACGTGCGAGACTGTAGCACTGGC-3′. The CRISPR-Cas9–based HDAC6 gene knockout vectors were constructed via cloning the annealed oligonucleotide pair into the plasmid pHKO23. The vectors were further transfected together with plasmid encoding packaging proteins into 293T cells. The viral supernatant was collected 48 hours after transfection.

Hct116 and A2780s cells were infected in the presence of 2 μg/mL polybrene. The next day, the medium was replaced. After 3–5 days for gene editing progress, cells were selected by the addition of 1 μg/mL puromycin. Knockout of HDAC6 was verified by Western blotting.

Wild-type A2780s or HDAC6 knock-out A2780s cells were seeded into 6-well plates, and then treated with compounds at the indicated concentrations for 16 hours. The cells were fixed with 4% paraformaldehyde and then penetrated with PBS containing 0.5% Triton X-100. After blocking for 30 minutes in 5% goat serum albumin at room temperature, the cells were incubated with anti-α-tubulin antibody, or in the colabeling procedure, the cells were incubated with anti-β-tubulin and anti-Ac-α-tubulin antibodies at room temperature for 1 hour. Then, the cells were washed three times with PBS following staining with fluorescent secondary antibodies and labeling of the nuclei with 4,6-diamidino-2-phenylindole (DAPI). The cells were finally washed thrice and visualized using a fluorescence microscope (Olympus).

The EBI competition assay was generally conducted using the same method reported previously (30). The recommended MDA-MB-231 cell line was used. Cells were seeded in 6-well plates at 5 × 10⁵ cells/well. After cell adherence, compounds were added to the cells for 4 hours. Then, EBI (TRC Biomedical Research Chemicals, Canada) was added at 100 μmol/L and incubated for 1.5 hours. Thereafter, the cells were harvested, and cell extracts were prepared. The β-tubulin and EBI: β-tubulin adduct band was detected via Western blotting. GAPDH was also examined to approve the equal protein loading.

Turbidimetric assays of the microtubules were generally performed as described, with some modification (31). Microtubule protein was purchased from Cytoskeleton. Tubulin (3 mg/mL) was incubated in the polymerization buffer (0.1 mol/L PIPES, 2 mmol/L MgCl_2, 0.5 mmol/L EGTA, pH 6.9) at 4°C for 30 minutes while the 96-well plate was prewarmed to 37°C. Before the start of the measurement, various compounds at the indicated concentrations were added to the wells. Then, guanosine triphosphate (GTP) was added to the tubulin to achieve a final concentration of 1 mmol/L and mixed. The tubulin-containing sample was rapidly diluted into the 96-well plate (100 μL per well). Immediately, the change in absorbance at 340 nm was kinetically measured at 37°C by a thermostatically controlled spectrophotometer (Molecular Devices).

Cells were plated on 6-well culture plates. After cell adherence, the cells were treated with the indicated concentrations of SKLB-23bb or ACY1215 for another 24 hours. Cells were washed with PBS two times and then fixed with 75% ethanol overnight. Then, the cells were washed with PBS three times, and stained with PI (50 μg/mL) for 20 minutes. The cells were then subjected to flow cytometry (BD FACSCalibur) for cell-cycle analysis.

A2780s and hct116 cells treated with the indicated concentration of SKLB-23bb were washed with PBS, trypsinized, and reseeded into 6-well plates at 300 cells per well. The HDAC6 knockout cells were seeded into 6-well plates at 150 cells per well. The colonies were allowed to form for 10–14 days. At the end of the culture, cells were washed with PBS, fixed with methanol for 30 minutes, and stained with 0.5% crystal violet for 20 minutes. After careful washing, the images were taken, and colonies were counted manually.

Cells were plated on 6-well culture plates. After cell adherence, cells were treated with the indicated concentrations of SKLB-23bb for 48 hours. Then, the cells were collected and subjected to an AnnexinV/PI Apoptosis Detection kit (Miltenyi Biotec) for staining according to the manufacturer's instructions, and finally analyzed by flow cytometry (BD FACSCalibur).

Hct116 and A2780s cells were seeded into 6-well plates. After cell adherence, the cells were treated with 1000 nmol/L of SKLB-23bb for the indicated time points. Then, the cells were collected, washed, and stained according to the instructions in the JC-1 kit (Keygen Biotech), and finally analyzed by flow cytometry (BD FACSCalibur).

All animal studies were approved by the Animal Care and Use Committee of Sichuan University (Chengdu, Sichuan, China). Five- to 6-week-old female Balb/C athymic nude mice or 6- to 8-week-old female NOD/SCID mice were purchased from the Beijing HFK Bioscience Company. Mice were implanted with the indicated number of cells suspended in 100-μL PBS into the right flank. When the xenografts developed, and the average tumor volume reached 100 mm³, the mice were randomly divided into the indicated groups and treatment began. The length and width of the tumors were measured, and the tumor volume (mm³) was calculated by the formula: π/6 × length × width². Tumor volumes and body weights were measured three times a week. The antitumor activity of the compound was evaluated by tumor inhibition = (1−tumor weight of treat group/tumor weight of control group) × 100%.

Data obtained in the cellular studies were assessed by unpaired Student t test. The tumor weight or tumor volume data were analyzed by ANOVA following a Dunnett multiple comparisons test. A P < 0.05 indicated a significant difference.

The effect of SKLB-23bb on tumor cell viability was tested against a panel of tumor cell lines from different origins. SKLB-23bb exhibited activity with IC₅₀ values under 100 nmol/L, against most of the cell lines checked (Table 1). In comparison, the IC₅₀s of ACY1215 were all above 1000 nmol/L, except for the Jeko-1 cell line (983 nmol/L). It could be implied that some other reported HDAC6-selective inhibitors were inferior compared to SKLB-23bb. According to the literature, the C1A agent inhibited proliferation of a panel of human tumor cell lines at the low micromolar range (19). Another compound, HPOB, had no effect on tumor cell viability alone (20). Recently, a new candidate termed WT161 was reported (21). WT161 significantly enhanced the activity of bortezomib on multiple myeloma, but failed to exhibit an obvious effect on cell viability in single-compound experiments. Hence, the outcome suggested that SKLB-23bb had a superior cytotoxic effect on tumor cell lines compared with ACY1215, or the other known HDAC6-selective inhibitors.

Interestingly, SKLB-23bb seemed to be more efficient against solid tumor cell lines, with IC₅₀s ranging from 31.32 nmol/L to 82.81 nmol/L, while higher data toward hematologic tumors were observed (IC₅₀s ranging from 49.80 nmol/L to 121.28 nmol/L). In contrast, ACY1215 showed better efficiency against hematologic tumor cell lines with IC₅₀s between 983 nmol/L to 4,497 nmol/L, which is consistent with the fact that most publications about this drug were related to hematologic models (23, 25–27). The IC₅₀s of ACY1215 against the solid tumor cell lines were all above 4,000 nmol/L, and the H460, A549, and A2780s cell lines appeared to be robust against ACY1215 treatment. In our earlier report (28), the inhibitory effect of SKLB-23bb against the colon cancer hct116 xenograft model was also superior to that toward the MV4-11 and Ramos xenografts, which represent hematologic malignances. Integrating the preexisting information, it could be inferred that SKLB-23bb possessed a broad antitumor spectrum, especially toward solid tumor models. Thus, it would be important to investigate the underlying mechanism.

The extraordinary advantage of SKLB-23bb compared with ACY1215 triggered our curiosity to further investigate the relationship between the antitumor activity of SKLB-23bb and its cellular target HDAC6. We first screened the basic expression level of HDAC6 in several cell lines. As presented in Fig. 2A, in the panel of the solid tumor cell lines, hct116, A2780s, SKOV3, A549, and MDA-MB-231 highly expressed HDAC6, while the H460 and MCF-7 cell lines exhibited a relatively low HDAC6 expression level. In the six hematologic tumor cell lines investigated, the HDAC6 expression level was high in MM1S, U266, HBL-1, and Jeko-1, and low in the RPMI-8226 and Ramos. The HDAC6 enzymatic activity of the cell lines was also measured. Interestingly, the cellular HDAC6 activity did not strictly correlate with the protein level among the cell lines (Supplementary Fig. S1A).

For further investigation, the hct116, A2780s, and U266 cell lines were selected, as they displayed high HDAC6 expression levels or enzymatic activity among solid and hematologic cell lines. The effect of the compounds on the acetylation levels of α-tubulin and H₃ were evaluated. As shown in Fig. 2B and Supplementary Fig. S1B, after SKLB-23bb or ACY1215 treatment, the level of Ac-α-tubulin increased significantly at low concentrations of 10 or 100 nmol/L, while that of Ac-H₃ did not obviously increase until higher doses were used, suggesting that both agents selectively inhibited the deacetylase activity of HDAC6, which takes α-tubulin as substrate, not class I HDACs that catalyze the deacetylation of histones. As a reference, the pan-inhibitor, SAHA, triggered the increase of both H₃ and α-tubulin acetylation levels, with no obvious difference (Fig. 2B; Supplementary Fig. S1B).

The superior antitumor activity of SKLB-23bb implied additional mechanisms or targets for several reasons. It could be inferred that, at the concentration of 100 nmol/L, both SKLB-23bb and ACY1215 remarkably elevated the Ac-α-tubulin level, but these two agents showed great difference in the cytotoxic profiles. In the observation of the A2780s cell line, it was evident that ACY1215 induced acetylation of α-tubulin at the concentration of 100 nmol/L, while the IC₅₀ for this agent to affect A2780s cell viability was above 10,000 nmol/L, indicating that inhibition of HDAC6 was not sufficient to achieve antitumor response in this model. Therefore, if HDAC6 was the unique target of SKLB-23bb, this compound should also be inactive towards A2780s. Contradictorily, the cytotoxic IC₅₀ of SKLB-23bb on A2780s was as low as 36.68 nmol/L (Table 1). Furthermore, it could be observed that among the cell lines investigated, the HDAC6 expression level or enzyme activity was poorly correlated with the sensitivity to SKLB-23bb (Fig. 1A; Supplementary Fig. S1A; Table 1).

To clarify whether the cytotoxicity of SKLB-23bb and ACY1215 was dependent on HDAC6, we established HDAC6 stably knocked out cell lines and assessed whether the antitumor activity of SKLB-23bb or ACY1215 differed between the HDAC6-null cell lines and their respective parent cell lines. Via the application of CRISPR-Cas9 system, HDAC6 knock-out hct116 and A2780s (hct116^−HDAC6, A2780s^−HDAC6) cell lines were stably constructed. Losses of cellular HDAC6 and the correlated elevation of Ac-α-tubulin were confirmed (Fig. 3A and B). The effects of SKLB-23bb and ACY1215 on cell viability were subsequently evaluated on these cell lines and their parent wild-type cell lines. SKLB-23bb was still active against the HDAC6 knockout cell lines compared with the parent cell lines, with IC₅₀ values ranging from 41.50 to 53.55 nmol/L in the HDAC6 knockout cells and 39.79 to 45.98 nmol/L in the parent cell lines, respectively. When treating the parent hct116 cell line, the IC₅₀ value of ACY1215 was 5.21 μmol/L. However, the IC₅₀s of ACY1215 in the knockout cell lines were 10.14 and 11.92 μmol/L, respectively, indicating a 1.95- to 2.29-fold increase (Fig. 3A). Similar results were also observed in A2780s cell lines, the IC₅₀ of ACY1215 toward the wild-type cell line was 10.14 μmol/L, while the IC₅₀s in the knockout cell lines were 20.09 and 28.94 μmol/L, respectively, with a 1.98- to 2.85-fold increase (Fig. 3B). These findings indicated that SKLB-23bb could remain efficient in the absence of HDAC6.

The proliferation trends of the HDAC6 knockout cell lines were also compared with their parent cell lines. The results of a clonogenic assay indicated that in the hct116 and A2780s cell lines, the loss of HDAC6 had minimal effect on the rate of cell growth (Supplementary Fig. S2). These findings might explain the reason underlying the lack of efficiency of ACY1215 on solid tumor models, as our observation suggested that in such models, depletion or inhibition of HDAC6 was not sufficient to achieve satisfying therapeutic outcome.

The finding that SKLB-23bb was active toward cell lines regardless of their HDAC6 status inferred that this agent has an additional target. As the similar quinazoline structure was applied in the microtubule-targeting agent MPC6827 (32), we speculated that SKLB-23bb might affect the microtubule system and experiments were conducted for validation. We assessed whether SKLB-23bb could impact the cellular microtubule system in situ via immunofluorescence staining in the A2780s cell line. As illustrated in the left of Fig. 4A, when treated with 100 nmol/L of SKLB-23bb, the normal arrangement of the microtubules was disrupted, and a higher concentration of SKLB-23bb (500 nmol/L) caused significant disruption of microtubule assembly. Moreover, from the properties of the observed microtubule morphologic changes triggered by SKLB-23bb, we could predict that SKLB-23bb acted as a microtubule polymerization inhibitor, as both SKLB-23bb and the reference compound colchicine resulted in relatively diffused cellular microtubules (33). Cellular microtubules appeared in a more aggregated fashion after treatment with paclitaxel, a known tubulin polymerization enhancer (34). Furthermore, the modality of the microtubule arrangement in the ACY1215-treated group tended to be similar to that of the vehicle control group. We subsequently investigated the effects of these compounds on microtubules in the HDAC6 knockout A2780s cells and similar results were observed (Fig. 4A, right). These observations indicated that SKLB-23bb additionally targeted microtubules, and this effect was independent of HDAC6.

In distinct approaches, we discovered that SKLB-23bb could induce acetylation of α-tubulin and inhibit tubulin polymerization. It would be important to clarify the relationship between the two effects of this compound, as HDAC6 is a microtubule-associated deacetylase (35, 36). Thus, a microtubule and acetylated microtubule costaining immunofluorescence experiment was conducted. It could be observed that SKLB-23bb simultaneously inhibited microtubule polymerization and induced α-tubulin acetylation at 100 nmol/L and 500 nmol/L (Supplementary Fig. S3). For comparison, the HDAC6 inhibitor ACY1215 enhanced Ac-α-tubulin but had no observed effect on microtubule morphology. Colchicine inhibited microtubule polymerization and paclitaxel enhanced microtubule polymerization, but the two tubulin targeting agents had no obvious effect on the microtubule acetylation status. The phenotype of the HDAC6 knock-out cells was similar to that of the ACY1215 group with enhanced Ac-α-tubulin, which is consistent with the previous finding that the loss of HDAC6 was correlated with the elevation of α-tubulin acetylation level (Supplementary Fig. S3). Thus, the outcome suggested that the effects of SKLB-23bb on microtubule acetylation and microtubule polymerization were both independent.

An N,N′-ethylene-bis(iodoacetamide; EBI) competition assay was conducted to investigate the binding site of SKLB-23bb on tubulin. In this method, the reference agent colchicine could engage the colchicine site in β-tubulin, preventing the EBI probe from forming a secondary β-tubulin band (EBI-tubulin adduct), which could be detected in Western blotting (30). As shown in Fig. 4B, SKLB-23bb dose-dependently reduced the quantity of the EBI-tubulin adduct, with a maximum effect achieved at 50 μmol/L, confirming that SKLB-23bb binds to the colchicine site. Vinblastine was used as a negative control to illustrate that no false results were obtained. Vinblastine binds to the vinblastine site, that is nonoverlapping with the colchicine site (37, 38).

An in vitro tubulin polymerization assay was subsequently conducted to further evaluate the direct interaction between SKLB-23bb and tubulin. As displayed in Fig. 4C, it was confirmed that SKLB-23bb inhibited tubulin polymerization. Although SKLB-23bb impacted tubulin assembly to a weaker extent compared with colchicine, a lag at the starting period of tubulin polymerization was obvious at the concentration of 10 μmol/L. These findings illustrated that SKLB-23bb was a microtubule polymerization inhibitor.

The effect of SKLB-23bb on the cell-cycle distribution was assessed in the hct116 and A2780s cell lines. The results indicated that SKLB-23bb dose-dependently blocked the cell cycle at G₂–M phase (Fig. 5A). Further examination of p-H₃, a well-reported mitotic marker, by Western blotting confirmed the cell-cycle arrest caused by SKLB-23bb treatment (Fig. 5B). Moreover, no obvious effect on cell-cycle distribution was shown when investigation was made on ACY1215 (Supplementary Fig. S4), which is consistent with its report on lymphoma models (25). These data inferred that the G₂–M phase cell-cycle arrest induced by SKLB-23bb was due to the tubulin binding effect, as triggering of G₂–M phase cell-cycle arrest is a critical feature of microtubule-binding agents (39).

The clonogenic assay is a well approbated method of testing the antitumor efficiency of chemotherapeutic agents (40). As shown in Fig. 6A, SKLB-23bb could inhibit the clonogenic potential of both hct116 and A2780s cells in a concentration-dependent manner. To evaluate whether SKLB-23bb could trigger tumor cell death, hct116 and A2780s cells were treated with the indicated concentrations of SKLB-23bb and dual-stained by Annexin V and PI. The apoptotic cell populations were analyzed by flow cytometry. The outcomes displayed that after incubation of increasing doses of SKLB-23bb, the Annexin V–positive population increased, inferring that SKLB-23bb induced tumor cell apoptosis (Fig. 6B). For further demonstration, the effects of SKLB-23bb on caspase-9, caspase-3, and the PARP axis were investigated. The Western blotting results indicated that the intrinsic apoptosis activator caspase-9, the effector caspase-3 and the cell death biomarker PARP were activated by SKLB-23bb in both time and concentration-dependent manners in hct116 and A2780s cells (Fig. 6C). The activation of caspase-9 could be related to the loss of mitochondrial membrane potential (MMP). To verify, a JC-1 probe detection was conducted. The outcome of the flow cytometry analysis indicated that the lowering of the red fluorescence intensity was accompanied by the increase of the green fluorescence intensity, which labeled MMP loss (Supplementary Fig. S5A). The status of some key proteins related to the intrinsic apoptosis pathway was also investigated via Western blotting. SKLB-23bb reduced the level of oncoprotein AKT as well as its phosphorylated form in a time-dependent manner (Supplementary Fig. S5B). The protein levels of Bax, Bcl-2, and Bcl-X_L were also analyzed. The results indicated that SKLB-23bb treatment upregulated the proapoptotic protein Bax and suppressed the antiapoptotic member Bcl-X_L. Another antiapoptotic protein, Bcl-2, was not affected by SKLB-23bb treatment (Supplementary Fig. S5B). The upregulation of Bax and suppression of Bcl-X_L suggested the breaking of the balance between the two groups of MMP-regulating factors, the elevation of the Bax/Bcl-X_L ratio and the activation of the intrinsic apoptosis pathway.

SKLB-23bb had good pharmacokinetic profiles and was orally bioavailable (28). HBL-1 xenografts, representing a B-cell lymphoma model, were established utilizing NOD/SCID mice. As reports suggested that ACY1215 could exhibit therapeutic potential against lymphomas, the activity of SKLB-23bb and ACY1215 were compared at the same dose and routine in this model. Both administrated three times a week orally at the dose of 40 mg/kg, SKLB-23bb resulted in significant tumor growth inhibition with a 58.22% tumor-inhibitory rate, while ACY1215 showed only a 26.75% tumor-inhibitory rate (Supplementary Fig. S6; Supplementary Table S1).

To thoroughly evaluate the therapeutic potential of SKLB-23bb against solid tumor models, hct116, A2780s, and MCF-7 xenografts were established on Balb/c nude mice and SKLB-23bb was scheduled for oral administration at various doses three times a week. SKLB-23bb treatment significantly inhibited tumor growth in the three models (Fig. 7A–C). As presented in Table 2, the hct116 model representing colon cancer had tumor inhibition rates of 51.28%, 60.37%, and 66.05% at the doses of 12.5 mg/kg, 25 mg/kg, and 50 mg/kg, respectively. In ovarian cancer A2780s xenografts, the tumor-inhibitory rate of the lowest dose, 3 mg/kg, was 30.65%, and the data of the 6 mg/kg, 12.5 mg/kg, and 25 mg/kg groups were 62.19%, 75.50%, and 77.39%, respectively. Similar results were also observed in the breast cancer MCF-7 model, the inhibitory rates of the 12.5 mg/kg, 25 mg/kg, and 50 mg/kg groups were 30.57%, 50.51%, and 65.65%, respectively.

Subsequently, the key molecular events SKLB-23bb triggered in the xenografts were evaluated. Mice bearing hct116 xenografts were given a single oral 50 mg/kg dose of SKLB-23bb. Analysis of the protein levels of the tumor samples indicated that SKLB-23bb induced selective acetylation of α-tubulin, phosphorylation of H₃ and activation of PARP (Fig. 7D; Supplementary Fig. S7). Taken together, our data inferred that the molecular modulations that resulted from SKLB-23bb treatment could be retained in experimental animal systems, proposing that SKLB-23bb could be effective in the treatment of various tumor types.

In previous study, we report an orally bioavailable, HDAC6-selective inhibitor SKLB-23bb with a broad antitumor spectrum. In this work, interests focused on why SKLB-23bb, compared with ACY1215, an agent in clinical trials, exhibited superior antitumor efficiency, especially toward tumor models that tended to be robust against HDAC6 inhibition. Through the research procedure, it was discovered that SKLB-23bb additionally targets tubulin.

Altered tubulin acetylation status could influence cell behaviors, such as cytoskeleton organization, intracellular cargo transporting, and cell migration (41–44)). Whether HDAC6 could impact the basic tubulin assembly process via modulation of acetylation of the α-tubulin subunit is still unclear (45, 46). Thus, it would be essential to clarify whether SKLB-23bb directly targets tubulin or if its impact on the microtubule system was caused indirectly by the agent's inhibition of HDAC6. In our experimental procedures, we found that SKLB-23bb could affect microtubule modality independently of the HDAC6 status, and the HDAC6 inhibitor ACY1215 failed to impact the cellular microtubule arrangement and cell-cycle distribution (Fig. 4A; Supplementary Figs. S3 and S4). The opinion that another HDAC inhibitor, TSA, could not influence microtubule polymerization or depolymerization (47) also suggested that, despite their epigenetic modifying function, HDAC inhibitors could not disrupt the basic microtubule assembly disassembly progress. Furthermore, the evidence that SKLB-23bb simultaneously disrupted cellular microtubules and induced α-tubulin acetylation (Supplementary Fig. S3) demonstrated that SKLB-23bb was a dual targeting agent.

Considering the “dose–effect” relationship, it could be discovered that the cellular concentration required for SKLB-23bb to effectively inhibit HDAC6, impact the microtubule system and subsequently block the cell cycle at G₂–M phase were all lower than 100 nmol/L. The cytotoxic IC₅₀ values of SKLB-23bb against most tumor cell lines were also under 100 nmol/L (Table 1). Taken together, these clues suggested that for both the targets of HDAC6 and tubulin, SKLB-23bb has good “on-target effect (48)”.

Microtubule-targeting agents were well reported to trigger the instinct apoptotic pathway (32, 49), with familiar progress including disruption of BCL2 family proteins, mitochondrial membrane potential loss, caspase-9 activation, and caspase-3 activation (50). The HDAC6 inhibitors ACY1215 and WT161 were reported to trigger the unfolded protein response, leading to caspase family activation (21, 26). Hence, the signaling pathways SKLB-23bb impacted and the effect of SKLB-23bb to induce apoptosis were in accordance with the two cellular targets.

The final evaluation of the antitumor activity of SKLB-23bb utilizing xenografts representing various tumor types indicated that SKLB-23bb was active in vivo. In HBL-1 model, ACY1215 was evaluated for comparison in the same dose, frequency and routine as SKLB-23bb. Rationally, in clinical trials, ACY1215 was administrated orally (10). However, the activity of ACY1215 as a single agent was limited and inferior to SKLB-23bb, which is consistent with the data from B-cell lymphoma Ramos xenografts assessed in our previous work (28). The published data describing the efficiency of ACY1215 against solid tumor models suggested that it was not active when administrated alone (51). On the basis of this information, ACY1215 was no longer engaged in the studies of solid tumor xenograft models.

As seen in the 2 models of hct116 and A2780s, 12.5 mg/kg or 6 mg/kg TIW oral dosing of SKLB-23bb exhibited significant antitumor effect, respectively. Notably, in the MCF-7 xenograft, the outcome of 12.5 mg/kg group was not satisfying. The probable explanation is that MCF-7 exhibited relatively low HDAC6 expression and enzyme activity (Fig. 2A; Supplementary Fig. S1A). Thus, the threshold required for SKLB-23bb to exhibit potential toward this model depended more on the agent's microtubule-targeting ability.

In this study, data suggested that ACY1215 was active only toward specific tumor types, typically, multiple myeloma and lymphoma models (Table 1). The probable explanation is that these tumor subtypes tend to be highly dependent on cellular mis-folded protein clearance, in which HDAC6 participated as a key factor in the proteasome-independent proteolysis mechanism. Thus, these tumors could be more susceptible to HDAC6 inhibition. Moreover, based on the results shown in this study as well as other ACY1215 reports, the possibility that the undesired class I HDAC inhibition contributed to the activity of this compound could not be excluded, which was supported by the fact that ACY1215 could elevate histone H₃ acetylation level at concentrations above 1 μmol/L (Fig. 2B), and, in correlation, the cytotoxic effects did not appear until such concentrations were reached. The pharmacodynamic results in a recent clinical trial might indicate the same opinion, as high doses of ACY1215 also increased histone acetylation in the tested samples (10).

In the effort to establish HDAC6 stably knockout cell lines, it was shown that the loss of HDAC6 was insufficient to impact cell growth in the hct116^−HDAC6− and A2780s^−HDAC6− cell lines (Supplementary Fig. S2), suggesting that HDAC6 is less critical in the aspect of tumor survival in the two cell lines. This observation may explain the reason why HDAC6-selective inhibitors ACY1215 and ACY241 are mainly evaluated in combination approaches.

In summary, our work determined the advantage of SKLB-23bb against ACY1215 in terms of the cancer therapeutic potential. According to our literature research, SKLB-23bb might be the first orally bioavailable HDAC6 and tubulin dual-targeting agent with tumor therapy potential. The certification of tubulin as an additional target of SKLB-23bb gifted this compound with broad antitumor spectrum and superior activity. In the future, it may be worthy to evaluate SKLB-23bb in various tumor types, especially solid tumor subtypes that HDAC inhibitors are rarely tested in clinical, thus breaking through the bottleneck restricting HDAC6-selective inhibitors and HDAC pan inhibitors. SKLB-23bb might also serve as an example and encourage the design and development of small molecules with multiple targets in cancer therapy (52).

No potential conflicts of interest were disclosed.

Conception and design: F. Wang, L. Zheng, C. Nie, Y. Wei, L. Chen

Development of methodology: L. Zheng, Y. Yi, Q. Qiu, Y. Hu, L. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Wang, L. Zheng, Y. Yi, X. Wang, W. Yan, P. Bai, J. Yang, D. Li, H. Pei, L. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Wang, L. Zheng, Y. Yi, H. Ye

Writing, review, and/or revision of the manuscript: F. Wang, L. Zheng, Y. Hu, L. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Wang, Z. Yang, Q. Qiu, T. Niu, Y. Hu, S. Yang

Study supervision: C. Nie, Y. Hu, S. Yang, L. Chen

This work was supported by the National Natural Science Foundation of China (U1402222, to L. Chen; 81703344, to Z. Yang), and Guangdong Innovative Research Team Program (grant 2011Y073, to L. Chen). The authors thank Ronghong Zhang, Xuefei He, Mengshi Hu, Linlin Xue, Xue Yuan for assistance in xenograft studies. The authors thank Minghai Tang for checking the purity of the synthesized compounds. The authors thank Yuxi Wang, Chengyong Wu and Hua Jiang for providing their valuable advices.

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.