Covalent modification of Cys-239 in β-tubulin by small molecules as a strategy to promote tubulin heterodimer degradation

Clinical microtubule-targeting drugs are functionally divided into microtubule-destabilizing and microtubule-stabilizing agents. Drugs from both classes achieve microtubule inhibition by binding different sites on tubulin and inhibiting or promoting polymerization with no concomitant effects on the protein levels of tubulin heterodimers. Here, we have identified a series of small molecules with diverse structures potentially representing a third class of novel tubulin inhibitors that promote degradation by covalent binding to Cys-239 of β-tubulin. The small molecules highlighted in this study include T0070907 (a peroxisome proliferator-activated receptor γ inhibitor), T007-1 (a T0070907 derivative), T138067, N,N′-ethylene-bis(iodoacetamide) (EBI), and allyl isothiocyanate (AITC). Label-free quantitative proteomic analysis revealed that T007-1 promotes tubulin degradation with high selectivity. Mass spectrometry findings showed covalent binding of both T0070907 and T007-01 to Cys-239 of β-tubulin. Furthermore, T007-1 exerted a degradative effect on tubulin isoforms possessing Cys-239 (β2, β4, and β5(β)) but not those containing Ser-239 (β3, β6) or mutant β-tubulin with a C239S substitution. Three small molecules (T138067, EBI, and AITC) also reported to bind covalently to Cys-239 of β-tubulin similarly induced tubulin degradation. Our results strongly suggest that covalent modification of Cys-239 of β-tubulin by small molecules could serve as a novel strategy to promote tubulin heterodimer degradation. We propose that these small molecules represent a third novel class of tubulin inhibitor agents that exert their effects through degradation activity.

tion. In view of the compelling research findings, microtubules present highly attractive targets for anticancer drug design (1,2). Microtubule inhibitors are functionally classified into two categories: microtubule-stabilizing and microtubule-destabilizing agents (3). The microtubule-stabilizing agents mainly bind to the paclitaxel site or the laulimalide site of tubulin (4,5), whereas the microtubule-destabilizing agents bind to the colchicine, vinblastine, maytansine, or pironetin sites (6). These agents achieve microtubule inhibition by binding different sites on tubulin and promoting or inhibiting polymerization with no concomitant effects on the protein levels of tubulin heterodimers (2). However, the commonly used clinical microtubule-targeting agents paclitaxel and vinblastine are susceptible to drug resistance (7,8), emphasizing the urgent medical need to identify novel effective microtubule inhibitors with diverse mechanisms of action.
Targeting protein degradation is a known efficacious therapeutic strategy. A few examples are degradation of promyelocytic leukemia-retinoic acid receptor ␣ oncoprotein induced by arsenic trioxide or retinoic acid and that of estrogen receptors induced by selective estrogen receptor modulators (9 -11). These earlier findings support the potential of tubulin degradation as a feasible anticancer approach. A number of small molecules have been identified that promote degradation of tubulin, such as T0070907, thymoquinone, isothiocyanates, and withaferin A (12)(13)(14)(15)(16)(17)(18), but their underlying mechanisms remain elusive at present.
In this study, we have identified five small molecules, T138067, N,NЈ-ethylene-bis(iodoacetamide) (EBI), 3 allyl isothiocyanate (AITC), T0070907, and T007-1, a T0070907 derivative, with diverse structures that promote tubulin degradation via covalent binding to Cys-239 of ␤-tubulin. Label-free quantitative proteomic analysis revealed that T007-1 promotes tubulin degradation with high selectivity. Mass spectrometry findings demonstrated that both T007-1 and T0070907 bind covalently to Cys-239 of ␤-tubulin. Furthermore, T007-1 exerted a degradative effect on ␤2-, ␤4-, and ␤5(␤)-tubulin iso-forms, which possess a cysteine residue at position 239, but not ␤3and ␤6-tubulin isoforms with serine at this position or mutant ␤-tubulin with a C239S substitution. The collective results suggest that this covalent binding activity of T007-1 accounts for its tubulin degradation effect. Notably, three small molecules (T138067, EBI, and AITC) reported to bind covalently to Cys-239 of ␤-tubulin also induced tubulin degradation while exerting no effect on mutant ␤-tubulin with C239S. All five small molecules identified in this study with totally different structures effectively promoted tubulin degradation by binding covalently to Cys-239 of ␤-tubulin. Based on the collective results, we propose that these small-molecule tubulindegradation agents exert their effects via covalent modification of Cys-239 of ␤-tubulin and constitute a third novel class of tubulin inhibitors.

T007-1 down-regulates tubulin protein with high selectivity
T0070907, a peroxisome proliferator-activated receptor ␥ (PPAR␥) inhibitor (Fig. 1A), promotes proteasome-dependent tubulin degradation and exerts anticancer effects on colorectal carcinoma cells (13,19). However, the mechanism underlying T0070907-mediated tubulin degradation remains to be established. To uncover the associated mechanisms and obtain anticancer lead compounds with improved activity, a series of T0070907 derivatives were synthesized. T007-1 (Fig. 1A, supporting methods), one of the T0070907 derivatives, showed better anticancer activity than the parent compound against the human cervical adenocarcinoma cell line, HeLa, and human colon colorectal carcinoma cell line, Hct116 (Fig. 1B). Labelfree quantitative proteomic analysis of T007-1-treated HeLa cells showed that T007-1 induced suppression of the protein levels of ␣and ␤-tubulin isoforms with high specificity. The top four most down-regulated proteins among the 1114 identified proteins were ␣and ␤-tubulin isoforms (␤, ␤4B, ␤8, and ␣1; Fig. 1C, supporting Data Set 1). Tubulin down-regulation was confirmed via immunofluorescence staining of T007-1treated HeLa cells. As shown in Fig. 1D, HeLa cells treated with T007-1 showed little tubulin staining (green staining with anti-␣-tubulin antibody). Immunoblot analysis revealed that T007-1 C, label-free quantitative proteomic analysis of total proteins from HeLa cells treated with 3 M T007-1 for 6 h. The graph shows fold-changes of 1114 proteins between T007-1 and vehicle treatment groups versus p value (t test; triplicate analysis). D, immunofluorescence of HeLa cells treated with or without 3 M T007-1 for 16 h (green, ␣-tubulin; blue, nucleus). E, Western blots of ␣-tubulin and ␤-tubulin in HeLa and Hct116 cells treated with the indicated concentrations of T007-1 for 16 h using GAPDH as a loading control. The lower graph depicts protein levels of ␣-tubulin and ␤-tubulin standardized to GAPDH levels. Quantitative data are presented as mean Ϯ S.D. of three independent experiments. ␣-Tub, ␣-tubulin; ␤-Tub, ␤-tubulin.
promotes ␣-tubulin and ␤-tubulin degradation in HeLa and Hct116 cells in a dose-dependent manner (Fig. 1E). Our results indicate that T007-1 mediates down-regulation of tubulin with high selectivity.

T007-1 promotes proteasome-dependent degradation of tubulin heterodimers
To determine whether T007-1 affects ␣-tubulin and ␤-tubulin mRNA expression, a quantitative PCR assay was performed. As shown in Figs. 2, A and B, T007-1 induced time-dependent down-regulation of ␣-tubulin and ␤-tubulin proteins in both HeLa and Hct116 cells but exerted no inhibitory effects on the corresponding mRNA levels. In fact, ␣-tubulin and ␤-tubulin mRNA levels were increased following reduction of the corresponding protein levels in both cell lines treated with T007-1. Our results suggest that ␣-tubulin and ␤-tubulin down-regulation by T007-1 is a post-transcriptional event and gene expression patterns are negatively regulated in response to the levels of the corresponding proteins. As T0070907 is reported to promote ␣-tubulin and ␤-tubulin degradation via a proteasome-dependent pathway, we examined whether T007-1 induces degradation via a similar mechanism. Pretreatment with the proteasome inhibitor, MG132, completely blocked ␣-tubulin and ␤-tubulin degradation by T007-1 (Fig. 2C), confirming similar proteasome-dependent activity to the parent compound.

T007-1 induces G 2 /M cell cycle arrest and apoptosis in cancer cells
Because conventional tubulin inhibitors, such as paclitaxel and colchicine, inhibit spindle formation and induce G 2 /M phase cell cycle arrest in cancer cells (2,21), we further examined the effects of T007-1 on cell cycle arrest. As depicted in Fig.  3A, T007-1 promoted cell cycle arrest at the G 2 /M phase in a dose-dependent manner in HeLa and Hct116 cells. Tubulin inhibitors are reported to promote apoptosis in cancer cells after induction of G 2 /M phase cell cycle arrest (22). Accordingly, we investigated the effects of T007-1 on two apoptosis marker proteins, cleaved caspase-3 and PARP. T007-1 promoted time-dependent cleavage of caspase-3 and PARP, clearly indicative of apoptosis induction (Fig. 3B). T007-1-induced cleavage of caspase-3 and PARP was inhibited by the pancaspase inhibitor, Z-VAD-fmk, further suggesting that the apoptotic pathway is caspase-dependent. Based on these findings, we conclude that T007-1 induces G 2 /M phase cell cycle arrest and apoptosis in cancer cells in a similar manner to conventional tubulin inhibitors.

T007-1 binds covalently to Cys-239 of ␤-tubulin
As T007-1 promotes tubulin degradation at the post-transcriptional stage, we examined whether the agent directly binds tubulin with the aid of a microscale thermophoresis assay. As

A novel strategy for tubulin heterodimer degradation
shown in Fig. 4A, both colchicine and T007-1 interacted with tubulin with K D values of 4.96 Ϯ 1.93 and 0.39 Ϯ 0.11 M, respectively. We further conducted an in vitro tubulin polymerization assay to determine the effects of T007-1 on tubulin assembly by measuring the increase in absorbance of tubulin at 340 nm and 37°C using colchicine and paclitaxel as comparative agents. Paclitaxel clearly promoted, whereas colchicine inhibited tubulin polymerization (Fig. 4B). Analogous to colchicine, T007-1 inhibited tubulin polymerization in a dose-dependent manner. Earlier studies have shown that T0070907 binds covalently to Cys-313 of PPAR␥, resulting in inhibition of its activity (14). Because T007-1 is a T0070907 derivative, we speculated that this compound may similarly form covalent interactions with tubulin. T007-1 was incubated with purified tubulin heterodimers, digested with trypsin, and subjected to LC electrospray ionization tandem MS (LC-ESI MS/MS) for analysis. As expected, T007-1 (molecular mass of 474.5 Da) bound the 217 LTTPTYGDLNHLVSATMSGVTTCLR 241 peptide of ␤-tubulin (leading to mass shift from 2652 to 3089 Da with a 437-Da mass change in the modified tryptic peptide because of loss of one molecule of HCl (36.5) to allow for the formation of the S-C bond). This finding indicates that the fluorophenyl of T007-1 is the key functional group in covalent binding to tubulin. Fragmentation of this peptide revealed that T007-1 attaches to Cys-239 in ␤-tubulin (Fig. 4C). Using the same method, we showed that T0070907 similarly binds covalently to Cys-239 of ␤-tubulin (Fig. 4D). Because Cys-239 is adjacent to the colchicine site, we additionally conducted a competition assay. The two colchicine-binding site inhibitors (colchicine and plinabulin) suppressed tubulin degradation induced by T007-1 (Fig. 4E), suggesting that binding of T007-1 to tubulin underlies its degradation activity.

Covalent binding to Cys-239 accounts for tubulin degradation induced by T007-1
We further investigated whether T007-1-induced tubulin degradation is mediated through binding to Cys-239. Data obtained from label-free quantitative proteomic analysis showed no degradation activity of T007-1 on ␤6-tubulin in which cysteine was substituted with serine at position 239 (Fig. A novel strategy for tubulin heterodimer degradation 5A). Immunoblot analysis consistently revealed that T007-1 specifically promotes degradation of ␤2-, ␤4-, and ␤5-tubulin isoforms containing cysteine at position 239, whereas exerting no effect on ␤3and ␤6-tubulin isoforms, which possess serine at this position (Fig. 5B). These findings gave rise to the speculation that covalent modification of Cys-239 by T007-1 contributes to its tubulin degradation activity. To validate this theory, a FLAG tag was fused to the C termini of WT and C239S ␤-tubulin genes, which were subsequently cloned into MSCV-IRES-GFP expression vectors. FLAG-wt or FLAG-C239S ␤-tubulin and GFP were co-expressed in HeLa cells. T007-1 clearly promoted FLAG-wt ␤-tubulin degradation while exerting no effect on FLAG-C239S ␤-tubulin (Fig. 5C), confirming that Cys-239 is necessary for tubulin degradation induced by T007-1. As determined from LC-ESI MS/MS, fluorophenyl of T007-1 was the key functional group in covalent binding to Cys-239. We further synthesized a T007-1 derivative, 180422 (Fig. 5D, the synthesis method is described in supporting methods), lacking fluorophenyl that could not bind covalently to Cys-239 of ␤-tubulin. As expected, 180422 displayed loss of anti-proliferative and tubulin degradation activities (Fig. 5, E and F). Our collective data suggest that covalent modification of Cys-239 is a prerequisite for tubulin degradation by T007-1.

Covalent modification of Cys-239 in ␤-tubulin by small molecules presents a novel tubulin degradation strategy
Next, we investigated whether tubulin degradation is a common biochemical consequence of covalent modification of Cys-239 in ␤-tubulin by small molecules. To this end, small molecules reported to form covalent interactions with Cys-239 of ␤-tubulin, such as T138067 and EBI (Fig. 6A), were selected for analysis (23,24). Similar to T0070907 and T007-1, T138067 and EBI promoted tubulin degradation in a dose-dependent manner (Fig. 6B). Pretreatment with colchicine inhibited T138067and EBI-induced tubulin degradation (Fig. 6C). Furthermore, T138067 and EBI specifically promoted degradation of FLAG-wt ␤-tubulin and not that of FLAG-C239S ␤-tubulin (Fig. 6D). These results support the theory that covalent modification of Cys-239 by T138067 and EBI plays a role in tubulin degradation by these two compounds.

Discussion
Tubulin inhibitors are widely used as chemotherapeutic agents for cancer therapy (25,26), such as the microtubulestabilizing agents paclitaxel and docetaxel and destabilizing agents vinblastine, vincristine, vinorelbine, and eribulin (27)(28)(29)(30). These drugs inhibit or promote tubulin polymerization to suppress tubulin dynamic instability, inhibit spindle formation during mitosis, and subsequently induce G 2 /M phase cell cycle arrest and apoptosis in cancer cells (20,27), whereas exerting no effects on tubulin protein levels.
In this study, we have identified small molecules that promote tubulin degradation by covalently binding Cys-239 of ␤-tubulin. Although previous reports have shown that certain small molecules, for instance, T0070907, thymoquinone, isothiocyanates, and withaferin A, promote tubulin degradation (12,14,15,17,18), the underlying mechanisms are currently unclear, such as whether these agents bind tubulin and the specific binding sites. Our study revealed that these agents might also promote tubulin degradation by covalently binding to Cys-239 of tubulin and, because they can all react with thiol groups in proteins (31,32). In addition, covalent modifiers of Cys-239 of tubulin appeared to promote tubulin degradation with high selectivity. Using label-free quantitative proteomic analysis of HeLa cells treated with T007-1, we observed a significant down-regulation of only four among the 1114 identified proteins (protein ratios lower than 2/3, compared with controls) (Fig. 1C), which were all tubulin isoforms. This high specificity implies that Cys-239 presents an ideal binding site for drug design based on tubulin degradation activity.
Fluorophenyl of T007-1 is the key functional group mediating covalent bond formation. In the current study, compound 180422 without fluorophenyl showed total loss of anti-proliferative activity, implying that T007-1 acts through covalent binding and not noncovalent binding behavior because anti-proliferation activity is retained upon conventional noncovalent inhibitor binding at the colchicine site (21). This finding indicates that the fluorophenyl of T007-1 is important for both binding and degradation activities.
Cysteine residues play an important role in protein structural stability, and oxidation of even a single cysteine could cause protein unfolding and aggregation (20,35). The key residue mediated by tubulin degradation agents is Cys-239. The underlying mechanism is proposed as covalent modification of Cys-239

A novel strategy for tubulin heterodimer degradation
resulting in tubulin unfolding and subsequent proteasome-dependent degradation of unfolded tubulin. However, this is only a theoretical assumption and requires further investigation.
Overall, our findings suggest that covalent modification of Cys-239 of ␤-tubulin by small molecules presents an effective strategy to promote tubulin heterodimer degradation. We propose that these small-molecule tubulin degradation agents represent a third novel class of tubulin inhibitors.

Cell lines and cultures
The human cervical adenocarcinoma cell line, HeLa, and human colon colorectal carcinoma cell line, Hct116, were obtained from American Type Culture Collection and cultured with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin. The temperature was set at 37°C with an atmosphere of 5% CO 2 . All cell lines were authenticated by short tandem repeat testing and free of mycoplasma.

Cell viability detection
Cells cultured in 96-well plates were treated with different concentrations of compounds for 72 h. Next, 20 l of MTT (mg/ml) was added to each well and cultured for another 4 h. The supernatant was removed and 150 l of DMSO added to each well. Plates were gently shaken for 10 min, and absorbance at 570 nm was measured using a microplate reader (Biotek, USA).

Label-free quantitative proteomic analysis
Cells were incubated with or without T007-1 for 6 h, washed with phosphate-buffered saline (PBS), and lysed with radioimmunoprecipitation assay (RIPA) buffer (containing 2 mM phenylmethylsulfonyl fluoride and proteinase inhibitor mixture) for 30 min at 4°C. Samples were centrifuged at 13,000 rpm for 30 min at 4°C. The supernatant fractions were collected and subjected to a BCA Protein Assay for determination of protein concentrations. Three biological replicates were examined for each experiment.
Trypsin digestion-Protein (300 g) was denatured at 100°C for 5 min. Samples were cooled to room temperature and further denatured using urea buffer (8 M urea, 150 mM Tris-HCl, pH 8.0). Subsequently, cysteine residues of protein were blocked with iodoacetamide (100 l of 50 mM) in urea buffer for 30 min at room temperature in the dark. Protein was digested with 4 g of trypsin (Promega, Madison, WI) at 37°C for 16 -18 h, the resulting peptides were collected and their concentrations were determined at OD 280 .
LC-ESI MS/MS-Each sample (1 g) was loaded onto a Thermo Scientific EASY column for separation using a segmented gradient on an Easy-nLC nanoflow HPLC system (Thermo Fisher Scientific). The eluent was further analyzed with an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific) in a positive ion mode. Mass spectra were acquired over a range of 350 -2000 m/z. The maximum ion injection time was set at 50 ms for the survey scan and 150 ms for MS/MS scans for the 16 most intense signals in the acquired mass spectra.
Analysis of differentially abundant proteins-The Andromeda peptide search engine was employed to identify proteins. Peptide and protein false discovery rates were estimated with the Self database, and the maximum protein and peptidespectrum match false discovery rates set to 0.01. Cysteine carbamidomethylation was set as a fixed modification and methionine oxidation as a variable modification. We calculated label-free quantification using MaxQuant. This value was obtained by dividing protein intensity by the number of theoretically observable tryptic peptides between 5 and 30 amino acids and was highly correlated with protein abundance.

Western blotting
Cells were collected and lysed in RIPA buffer. Equal amounts of total protein samples were loaded onto an SDS-PAGE system for separation and electrophoretically transferred to polyvinylidene difluoride membranes. Polyvinylidene difluoride membranes were blocked in 5% skimmed milk for 1 h and incubated with primary antibody at 4°C overnight. Next, membranes were washed three times with PBS and Tween 20 (3 ϫ 10 min) before incubation with secondary antibody for 45 min and re-washed with PBS containing Tween 20 (3 ϫ 10 min) before detection of immunoreactivity using enhanced chemiluminescence reagents (Millipore).

Mass spectrometry
Purified tubulin (20 M) was incubated with a slight excess of 25 M T0070907 or 25 M T007-1 for 3 h at room temperature. Excess compounds were removed via an ultrafiltration method and samples were denatured via SDS-PAGE. Tubulin bands were excised for subsequent MS, which was conducted by rigorously following a published protocol (6).

Microscale thermophoresis assay
Binding of T007-1 and colchicine to tubulin was detected with a microscale thermophoresis assay utilizing a Monolith NT.115 instrument (NanoTemper Technologies). Purified tubulin was labeled using a Monolith protein labeling kit RED-NHS (NanoTemper Technologies). Different concentrations (100 M to 3 nM) of T007-1 or colchicine were incubated with labeled tubulin (40 nM) in assay buffer (80 mM PIPES, pH 6.9, 0.5 mM EGTA, 2 mM MgCl 2 , 1 mM GDP) for 10 min at 4°C. Samples were loaded into glass capillaries for detection. K D values were obtained using NanoTemper software.

Immunofluorescence staining
Cover glasses were placed in the bottoms of 6-well plates. Cells were seeded onto cover glasses and incubated for 24 h before treatment with T007-1 for 16 h. The culture medium was removed and washed with PBS for 2 min. Cells were incubated with 50% methanol and 50% acetone for 2 min, followed by primary antibody for 4 h at room temperature. Next, cells were washed with PBS (4 ϫ 5 min) and incubated with fluorescent secondary antibodies and DAPI for 45 min at room temperature, followed by re-washing with PBS (4 ϫ 5 min). Images were acquired using a fluorescence microscope (Olympus, Japan).

In vitro tubulin polymerization assay
Samples of purified tubulin (2 mg/ml) in protein expression medium buffer (80 mM PIPES, pH 6.9, 0.5 mM EGTA, 2 mM MgCl 2 , 1 mM GTP) with 15% glycerol were incubated with different compounds at 4°C for 1 min, transferred to pre-warmed (37°C) 96-well plates and optical densities at 340 nm were determined once per minute using a microplate reader (Biotek, USA) at 37°C.

Cell cycle analysis
Cells were seeded onto six-well plates and incubated for 24 h before treatment with different concentrations of T007-1 for 16 h. Treated cells were washed with PBS, collected, and fixed in 70% ethanol at 4°C for 24 h. Fixed cells were washed three times with PBS, stained with 50 g/ml of propidium iodide for 30 min, and analyzed using a flow cytometer (BD FACSCalibur).

Vector construction
Complete sequences of TUBB (␤-tubulin) with FLAG tag fused to the C terminus were synthesized by Genewie (Suzhou, China), containing BamHI and PacI restriction sites at all ends. The gene was cloned into MSCV-IRES-GFP expression vector co-expressing GFP and target proteins. TUBB containing the C239 mutation was constructed using a Q5 site-directed mutagenesis kit (New England Biolab E0554S).

Transfection experiments
HeLa cells were seeded onto 6-well plates and cultured for 24 h before transfection. Plasmid DNA (2.0 g) was dissolved in 300 l of Opti-MEM (Thermo) and incubated for 5 min. During this time, 7.5 l of Lipofectamine 2000 was added to 300 l of Opti-MEM and incubated for 5 min. Next, plasmid DNA and Lipofectamine 2000 were mixed and incubated for 20 min before addition to HeLa cells in 6-well plates for a 24-h incubation period. After addition of the appropriate compounds, cells were incubated for a further 16 h. Total protein was extracted and lysed with RIPA buffer before Western blot analysis using FLAG and GFP antibodies. GAPDH was employed as the loading control.