Structure Optimization of Gatastatin for the Development of γ-Tubulin-Specific Inhibitor

Abstract

Gatastatin (O⁷-benzyl glaziovianin A) is a γ-tubulin-specific inhibitor that is used to investigate γ-tubulin function in cells. We have previously reported that the unsubstituted phenyl ring of the O⁷-benzyl group in gatastatin is important for γ-tubulin inhibition. To obtain further structural information regarding γ-tubulin inhibition, we synthesized several gatastatin derivatives containing a fixed O⁷-benzyl moiety. Modifications of the B-ring resulted in drastic decrease in cytotoxicity, abnormal spindle formation activity, and inhibition of microtubule (MT) nucleation. In contrast, various O⁶-alkylated gatastatin derivatives showed potent cytotoxicity, induced abnormal spindle formation, and inhibited MT nucleation. We had previously reported that O⁶-benzyl glaziovianin A is a potent α/β-tubulin inhibitor; thus, these new results suggest that the O⁶-position restricts affinity for α/β- and γ-tubulin. Considering that an O⁷-benzyl group increases specificity for γ-tubulin, more potent and specific γ-tubulin inhibitors can be generated through O⁶-modifications of gatastatin.

Microtubules (MTs) are dynamic polymers consisting of α- and β-tubulin heterodimers that are involved in cell motility, vesicle trafficking, and chromosome segregation.^1,2 In cells, polymerization of MTs starts from the MT nucleator γ-tubulin ring complex (γTuRC).³⁻⁵ γ-Tubulin is an essential component of γTuRC; however, the molecular mechanisms of MT nucleation and spatiotemporal regulation of γTuRC still remain to be revealed. One approach for revealing such mechanisms is the application of γ-tubulin-specific inhibitors.⁶ In addition, γ-tubulin is an attractive target molecule for both basic biological studies and antitumor medicines, because γ-tubulin is activated during mitosis^7,8 and is overexpressed in certain types of glioblastomas and nonsmall cell lung cancers.^9,10 Using this approach, we screened γ-tubulin-specific inhibitors and identified gatastatin (Figure 1, 1),¹¹ which is a derivative of the MT dynamics inhibitor glaziovianin A (AG1) (Figure 1, 2).^12,13 Gatastatin (1) has a higher affinity for γ-tubulin than toward α- and β-tubulin heterodimers and inhibits the progression of anaphase.¹¹ The cytotoxicity of gatastatin (1) is relatively weak compared to that of conventional MT inhibitors, for example paclitaxel and vinblastine, although γ-tubulin is essential for cell viability. It is therefore important to identify more potent γ-tubulin inhibitors.

Figure 1.

Structures of gatastatin (1) and glaziovianin A (2).

Previously, we focused on the O⁷-benzyl group in gatastatin (1) in a structure–activity relationship study and systematically optimized the aromatic ring at the O⁷-benzyl group in accordance with an operational Topliss tree.^14,15 Some derivatives showed stronger cytotoxicity in HeLa cells than gatastatin (1), but the cytotoxicity of these derivatives was independent of their inhibition of α/β- and γ-tubulin, suggesting strong recognition of the O⁷-benzyl group by γ-tubulin. We also revealed that B-ring-modified AG1 does not inhibit α/β-tubulin polymerization activity¹³ and that benzylation of the O⁶-position of AG1 increases cytotoxicity in HeLa cells.¹⁶ We thus assume that modifications of the B-ring portion and O⁶-position increase specificity for γ-tubulin (Figure 2). Consequently, we prepared B-ring-modified gatastatin derivatives (3–5), O⁶-modified gatastatin derivatives (6–10), and O⁶-modified 3′,4′-acetonide gatastatin derivatives (11–15) (Figure 3).^17,18

Figure 2.

Summary of results from our previous structure–activity relationship study of gatastatin (1) and AG1 (2).

Figure 3.

Structures of B-ring-modified gatastatin derivatives (3–5), O⁶-modified gatastatin derivatives (6–10), and O⁶-modified 3′,4′-acetonide gatastatin derivatives (11–15).

We first evaluated HeLa cell growth inhibition by gatastatin derivatives 3–15 (Table 1). All B-ring-modified derivatives (3–5) showed drastically reduced cytotoxicity, with IC₅₀ values over 100 μM. Only derivative 4 showed weak cytotoxicity at 100 μM. In contrast, O⁶-modified gatastatin derivatives (7–10) showed more potent cytotoxicity than gatastatin (1). This potent cytotoxicity was abolished by increasing the steric hindrance of C3′ and C4’ in the B-ring portion (derivatives 13–15). This result is similar to results of the structure–cytotoxicity relationship study of AG1.^19,20

Table 1. Cytotoxicity of Gatastatin Derivatives in HeLa Cells^a.

Compound	IC50 (μM)	Growth (%) at 100 μM
Gatastatin (1)	10.5 ± 2.5	N.T.b
3	>100	100.9 ± 11.4
4	>100	77.3 ± 13.3
5	>100	100.4 ± 7.1
6	>100	99.1 ± 3.1
7	5.8 ± 0.4	N.T.
8	9.9 ± 0.9	N.T.
9	0.8 ± 0.2	N.T.
10	1.1 ± 0.2	N.T.
11	>100	100.8 ± 7.9
12	>13c	N.T.
13	>100	107.4 ± 7.9
14	>100	102.5 ± 7.7
15	>100	101.9 ± 5.6

^aHeLa cells were treated with various concentrations of each derivative for 48 h. Cell growth was determined using a WST-8 cell counting kit (Dojindo), and the resulting IC₅₀ values were calculated.

^bNot tested.

^cDerivative 12 precipitated at concentrations above 13 μM.

Next, we investigated the effects of the derivatives on MT polymerization in vitro (Figure 4) and in cells (Figure 5). For in vitro MT polymerization, a polymerization inducer, such as 1 M glutamate, is required for efficient MT polymerization. In these experiments, we used 0.8 M glutamate because the high polymerization activity in the case of 1 M glutamate prevents examination of the inhibitory effects of derivatives (Figure 4A). Under these conditions, the percentages of polymerized MTs with vehicle control and colchicine, a MT polymerization inhibitor, were 84.8 ± 9.9% and 43.0 ± 18.0%, respectively. MT polymerization in the presence of 30 μM gatastatin derivative ranged from 69–92%, with no significant differences compared to the vehicle control.

Figure 4.

Effects of gatastatin derivatives on MT polymerization in vitro. Porcine brain tubulin (1 mg/mL) was incubated with 10 μM colchicine, 10 μM paclitaxel, and 30 μM gatastatin derivative in the presence of either (A) 0.8 M or (B) 0.2 M glutamate for 30 min at 37 °C. Polymerized and unpolymerized tubulin were separated by ultracentrifugation (200,000g, 10 min, 25 °C). Samples were then separated using SDS-PAGE using a 10% gel and the amount of tubulin quantified by ImageJ. Values represent mean ± SD of three independent experiments.

Figure 5.

None of the tested gatastatin derivatives affected the MT networks in interphase cells. Cells were treated with 100 nM colchicine (Col) and paclitaxel (Tax) for 6 h. Cells were treated with 30 μM gatastatin (1) and gatastatin derivatives (3–15) for 24 h. Scale bar = 25 μm.

We also examined MT polymerization in the presence of 0.2 M glutamate (Figure 4B), in which MT polymerization is not stimulated without addition of an assembly inducer like paclitaxel. Polymerized MT levels (%) in the presence of vehicle control and paclitaxel were 36.3 ± 8.1% and 91.6 ± 5.9%, respectively, while the addition of 30 μM gatastatin derivative led to polymerization levels of 31–41%, with no significant difference from with vehicle control. Taken together, these results suggest that all derivatives have minimal effects on MT polymerization.

This conclusion is supported by MT network structures observed in interphase HeLa cells (Figure 5). MT networks in cells treated with 100 nM colchicine or paclitaxel were depolymerized or bundled, respectively. In contrast, networks treated with 30 μM of a derivative were identical to those in DMSO-treated cells. Together with the results of in vitro polymerization assays, we conclude that all derivatives have little impact on α/β-tubulin heterodimers.

We next investigated the effects of derivatives on spindle morphology in mitotic cells, as γ-tubulin activity is required for bipolar spindle formation during mitosis. As shown in Figure 6A, most DMSO-treated cells show normal bipolar spindles; however, gatastatin (1) induces misaligned chromosomes and multipole spindles, as previously reported.¹¹ At 30 μM, most B-ring-modified derivatives (3, 5, 11–15) did not induce abnormal spindles. To our surprise, 3′,4′-dimethoxy gatastatin (4) induced misaligned chromosomes (65 ± 2.3%) despite showing only weak cytotoxicity (Table 1, Figures 6A and 6B). These results suggest that while the B-ring portion is important for γ-tubulin interaction, derivative 4 stills interacts with γ-tubulin. Furthermore, these results indicate that derivative 4 can prevent proper spindle formation but is not enough to constitutively activate the spindle checkpoint. Alternatively, derivative 4 may be relatively less stable in cells, thus decreasing its inhibitory activity. Further investigation is required to resolve this controversy.

Figure 6.

Gatastatin derivatives (4, 7–10) induce abnormal spindle formation. (A) Effects of gatastatin derivatives on spindle formation in cells. Exponentially growing HeLa cells are treated with 30 μM of each for 24 h and the resulting spindle morphology is classified. Values represent mean ± SD of three independent experiments. Greater than 100 cells are examined for each experiment. (B) Spindle morphology of cells treated with 30 μM gatastatin (1) or a gatastatin derivative (3–15) for 24 h. Abnormal bipolar spindles with misaligned chromosomes are observed in cells treated with gatastatin (1) and derivatives 4 and 7–10. Red arrows indicate misaligned chromosomes. Scale bar = 10 μm. (C) Compound 9, but not 10, dose-dependently induces multipolar spindle formation.

In contrast to B-ring-modified derivatives, all O⁶-modified derivatives except O⁶-tetrahydropyran gatastatin (6) induced abnormal spindles at 30 μM (Figure 6A). We then chose to focus on derivatives 9 and 10, since these derivatives show potent cytotoxicity (Table 1). As shown in Figure 6C, both derivatives induced abnormal spindles at concentrations as low as 1 μM. O⁶-Propargyl gatastatin (9) showed dramatic dose-dependent-induction of multipolar spindle formation (Figure 6C), suggesting that chromosome misalignment leads to enough cytotoxicity and multipolar spindle formation may reflect the degree of γ-tubulin inhibition. In contrast, O⁶-benzyl gatastatin (10) induced formation of abnormal spindles containing misaligned chromosomes in a dose-independent manner. These results suggest that derivative 10 may be efficiently excluded from cells compared to derivative 9 and that the concentration of derivative 10 inside cells remains constant, although further investigation of this possibility is required. Taken together, these results suggest that modification of the O⁶-position potentiates interaction with γ-tubulin.

We have previously reported that 30 μM gatastatin (1) inhibits γ-tubulin-dependent MT nucleation of centrosomes in human cells without affecting γ-tubulin localization.¹¹ We therefore evaluated the inhibition of MT nucleation in mitotic cells by gatastatin derivatives. Gatastatin (1) significantly inhibits MT nucleation from centrosomes in mitotic cells at 30 μM but not at 10 μM (Figure 7A). B-ring-modified derivatives 3 and 5 and O⁶-tetrahydropyranyl gatastatin (6) show no inhibitory activity, even at 100 μM (Supporting Information Figure 1). 3′,4′-Dimethoxy gatastatin (4) inhibits MT nucleation at 30 μM but not at 10 μM, indicating that the inhibitory activity of 4 is similar to that of gatastatin (1) (Supporting Information Figure 2). In contrast, O⁶-modified gatastatin derivatives 7–10 inhibited the MT nucleation at 30 μM (Supporting Information Figure 3A). Among them, derivative 9 showed the most potent activity and inhibited MT nucleation at 0.3 μM (Figure 7B). Furthermore, we confirmed derivative 9 inhibited GTP-binding on γ-tubulin in vitro more potently than gatastatin (Supporting Information Figure 4). Taken together with the results from in vitro MT polymerization (Figure 4) and MT network in interphase cells (Figure 5), these suggest that derivatives 7–10 are potent inhibitors of γ-tubulin-dependent MT nucleation of centrosomes.

Figure 7.

Derivative 9 inhibits MT nucleation from mitotic centrioles more potently than gatastatin. The MT areas nucleated from centrosomes (n = 20–25) were measured. (A) Gatastatin (1) inhibits MT nucleation at 30 μM. The MT polymerization inhibitor colchicine (Col) is the positive control. (B) Derivative 9 inhibited MT nucleation at concentrations above 0.3 μM. We have used GraphPad Prism 6 (GraphPad Software, San Diego, CA) to perform ANOVA. ****, p < 0.0001; **, p < 0.01; ns, no significance.

In this study, we investigated the structure–activity relationships of gatastatin (1). B-ring modification drastically decreases cytotoxicity of gatastatin, although 3′,4′-dimethoxy gatastatin (4) still possesses weak cytotoxicity (Table 1) and induces abnormal spindle structure (Figures 6A and 6B). In contrast, some modifications of the O⁶-position in gatastatin (1) produce derivatives that greatly increase cytotoxicity (Table 1), induce abnormal spindle formation (Figures 6A and 6B), and inhibit γ-tubulin-dependent nucleation (Supporting Information Figure 3). In particular, our results suggest that O⁶-propargyl gatastatin (derivative 9) is a more potent γ-tubulin-specific inhibitor than gatastatin (1); therefore, we named this compound as gatastatin G2.²¹ This compound would be a useful tool for investigating γ-tubulin function in cells.

Glossary

Abbreviations

AG1

glaziovianin A

MT

microtubule

γTuRC

γ-tubulin ring complex

THP

tetrahydropyran

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.9b00526.

• Experimental details for synthesis and analysis of compounds 3–15; conditions for biological assays (PDF)

Author Present Address

^∇ I.H.: Graduate School of Integrated Basic Sciences, Nihon University 3-25-40 Sakurajosui, Setagaya-ku, Tokyo 156-8550, Japan

Author Contributions

^‡ K.S. and H.E. contributed equally.

The authors declare no competing financial interest.