Correlation of hydrogen-bonding propensity and anticancer profile of tetrazole-tethered combretastatin analogues

Abstract

A series of 1,5-disubstituted tetrazole-tethered combretastatin analogues with extended hydrogen-bond donors at the ortho-positions of the aryl A and B rings were developed and evaluated for their antitubulin and antiproliferative activity. We wanted to test whether intramolecular hydrogen-bonding used as a conformational locking element in these analogues would improve their activity. The correlation of crystal structures with the antitubulin and antiproliferative profiles of the modified analogues suggested that hydrogen-bond-mediated conformational control of the A ring is deleterious to the bioactivity. In contrast, although there was no clear evidence that intramolecular hydrogen bonding to the B ring enhanced activity, we found that increased substitution on the B ring had a positive effect on antitubulin and antiproliferative activity. Among the various analogues synthesized, compounds 5d and 5e, having hydrogen-bonding donor groups at the ortho and meta-positions on the 4-methoxy phenyl B ring, are strong inhibitors of tubulin polymerization and antiproliferative agents having IC₅₀ value in micromolar concentrations.

Combretastatins (CAs) A-1¹ 1 and A-4² 2 (Chart 1) are natural products isolated from the African bush willow Combretum caffrum and are well known antimitotic agents causing vascular disruption. ³ CAs bind to the colchicine site of αβ-tubulin.⁴ Combretastatin A-1 phosphate (OXi4503; CA1P) and combretastatin A-4 phosphate (Fosbretabulin; CA4P) are well-known vascular disrupting agents currently in human clinical trials for the treatment of cancer.^5–7 CA1P demonstrated higher vascular disrupting and anti-tumor activity when compared with CA4P.⁸ The increase in efficacy of CA1P as a vascular disrupting agent is attributed to the presence of an additional hydroxyl substituent, which may result in the formation of a highly reactive ortho-quinone-like moiety through oxidative metabolism.^9,10

Chart 1.

Structures of CA-1 1, CA-4 2 and their known heteroaromatic analogues 3.

SAR studies of CAs analogues have suggested that there are three crucial structural components essential for activity, that is, the trimethoxy group on the A ring, cisoid configuration at the olefinic double bond and the methoxy group at the para-position on the B-ring.^11–13 The cis double bond in CAs can be converted into the more stable trans configuration in vivo. This would result in a drastic loss of activity¹⁴ and, therefore, several hetero-aromatic rings 3 (Chart 1), such as thiophene,¹⁵ furan,¹⁵ pyrazole,¹⁶ imidazole, ^16,17 thiazole,¹⁶ isoxazole,^18,19 1,2,3-thiadiazole,²⁰ isomeric triazoles, ^{15,16,21–23} and tetrazole,^16,24,25 have been evaluated to arrest the conformation in the cisoid form and for effects on cytotoxicity and on the tubulin target. So far, the restriction in conformation achieved has been essentially quasi-cisoid. It should be noted that in the absence of a conformational restriction, both the A and B rings of CA can rotate about the sp² bond that connects them.

In the present work, we attempted to arrest the cis-conformation effectively by introducing hydrogen-bond donor groups, such as a hydroxyl moiety, at ortho-positions on the A and B rings that might form stable six-membered hydrogen-bonds with acceptor atoms at the 2 or 4 or both 2 and 4-position of the central hetero- aromatic ring.

We have designed, synthesized and evaluated a series of 1,5-disubstituted tetrazole analogues with extended hydrogen-bond donors at the ortho-positions as potentially cis-restricted CA analogues. It is well-known in the case of triazole-based combretastatin analogues that when ring A is connected to a triazole nitrogen, the resultant analogue would show better antitubulin activity than analogues having a nitrogen connected to the B ring.²¹ We therefore initially designed analogues with a tetrazole nitrogen connected to the A ring rather than the B ring. However, we found that a hydroxyl group at the ortho-position of the B ring also enhanced activity. We therefore also introduced a hydroxyl or an amino group at the meta-position of the B ring. There are several modifications known with 2,3-dihydroxy groups on the B-ring as analogues of CA1,^23,26–28 but our modification of having 2-hydroxy and 3-amino groups on the B ring (5e, Chart 2) has not been previously reported.

Chart 2.

Rationale for the design of conformationally restricted CA analogues (present work).

We followed the reported imidoyl chloride procedure for the construction of the tetrazoles.²⁹ 2,3,4-trimethoxy-benzaldehyde 6a was demethylated to afford 6b,³⁰ which was further dibenzylated by using BnBr in DMF to furnish di-benzyloxy-aldehyde 6c (Scheme 1). The aldehyde 6c was oxidized to the acid 8c by following a literature procedure.³¹ In another set of experiments, 2,3,4-trimethoxy-benzaldehyde 6a was nitrated to give nitrobenzaldehyde 6d.³² Aldehyde 6d was then converted to phenol 6e by the Dakin reaction.³³ Phenol 6e was O-benzylated by standard conditions to give benzyloxy derivative 6f, which was then reduced to its corresponding amine with stannous chloride to afford aniline 9a.

Scheme 1.

Reagents and conditions: (a) BCl₃, DCM, 24 h; (b) BnBr, K₂CO₃, DMF, 6 h; (c) NaH₂PO₄, NaClO₂, 30% H₂O₂, 6 h; (d) HNO₃, AcOH, 1 h; (e) mCPBA, DCM 6 h; (f) NaOH, H₂O, 4 h; (g) SnCl₂:2H₂O, AcOEt, 60 °C. 8 h.

β-Resorcylic acid 7a (Scheme 2) was dimethylated to give phenol 7b,³⁴ followed by benzyloxy protection to furnish ester 7c. Methyl ester 7c was hydrolyzed to give the acid 8a. Phenol 7b was nitrated to afford nitro phenol 7d in 25% yield,³⁵ then phenol 7d was benzyloxy-protected to afford ester 7e. Ester 7e was hydrolyzed to furnish the free acid 8d.

Scheme 2.

Reagents and conditions: (a) MeI, K₂CO₃, acetone, 6 h reflux; (b) BnBr, K₂CO₃, DMF, 6 h; (c) NaOH, THF, H₂O, 12 h; (d) Ac₂O, AcOH, HNO₃, 30 min, 0 °C to rt.

Amides 10a–f (Scheme 3) were synthesized from respective carboxylic acids 8a–d and anilines by the acid chloride method. Amides 10a–f were converted to imidoyl chloride by a modified literature protocol²⁹ and then reacted with sodium azide to form respective benzyloxy protected tetrazoles 11a–e. Tetrazoles 11a–e were subjected to hydrogenolysis to afford the target compounds 5a–e. The known analogue 5f (Chart 2) also synthesized following the reported protocol²⁵ for crystallographic studies.

Scheme 3.

Reagents and conditions: (a) (COCl)₂, cat. DMF, DCM, 2 h; (b) Et₃N, DCM, 6 h; (c) (COCl)₂, pyridine, DCM, 4 h; (d) NaN₃, DMF, 12 h, 60 °C; (e) 10% Pd/C, H₂, 60 psi, MeOH, 8 h.

The crystal structure of 5a (Fig. 1) clearly shows that rings A and B are not co-planar. Instead of the expected six-membered hydrogen- bonding between the phenolic OH on the trimethoxyphenyl A ring and a second nitrogen in the tetrazole ring, there was a fivemembered hydrogen bond formed within ring A. In the crystals of 5a and 5c, conformational restriction was observed by formation of a six-membered hydrogen bond between the phenolic OH of the para-methoxyphenyl ring and the fourth nitrogen in the tetrazole ring as shown in Figure 1. The crystal structures of 5a, 5c, and 5f showed that the aromatic rings having a trimethoxyphenyl structure and a para-methoxy structure were out of plane with each other. Crystallographic data for the structures 5a, 5b, and 5c have been deposited at the Cambridge Crystallographic Data Centre with the deposition numbers CCDC 910097 (5a), CCDC 910098 (5c), CCDC 910099 (5f).

Figure 1.

Crystal structures of 5a, 5c and 5f with polar hydrogen showing hydrogen bonds.

The 1,5-disubstituted tetrazole analogues of Combretastatin 5a–e were evaluated for their inhibition of the growth of four different human cancer cell lines, that is, human cervix carcinoma (HeLa), human non-small-cell lung carcinoma (A549 and H1299), and breast adenocarcinoma (MCF-7). A comparison was made with the data obtained earlier for compound 5f²⁵ (Table 1). Compounds 5a and 5b, which has an ortho-hydroxyl group in the A-ring, were least active against the cancer cell lines tested with IC₅₀ >45 μM. Compound 5c bearing ortho-hydroxyl group in the B-ring was shown to enhance antiproliferative activity, in comparison with 5a and 5b. Although activity further enhanced with increased substitution in the B-ring, as in compounds 5d and 5e, lesser antiproliferative activity was observed when compared to 5f²⁵ in the common cell lines tested, that is, HeLa, A549 and MCF-7.

Table 1.

IC₅₀ ^a values (expressed in μM) of compounds 2, 5a–f

Compound	HeLa	A549	MCF-7	H1299
5a	45 ± 4	50 ± 0.8	46 ± 1.2	49 ± 2.1
5b	92 ± 0.3	>100	72 ± 1.4	40 ± 1.8
5c	2.4 ± 0.09	6.4 ± 0.43	8.2 ± 0.09	7.9 ± 0.8
5d	0.95 ± 0.001	6.7 ± 0.29	6.6 ± 0.04	8.8 ± 0.7
5e	0.9 ± 0.0016	0.52 ± 0.0009	2.9 ± 0.07	4.0 ± 0.4
5f	0.0026 ± 0.0b	0.222 ± 0.059b	0.0385 ± 0.0058b	n.d.b
2	0.004 ± 0.0012	0.28 ± 0.05	0.098 ± 0.003	0.0062 ± 0.002

^aIC₅₀ = compound concentration required to inhibit tumor cell proliferation by 50%. Data are expressed as the mean ± SE from the dose–response curves of at least three experiments.

^bValues taken from the literature for comparison.²⁵ n.d. indicates not determined.

The new tetrazole compounds 5a–e were also examined for potential inhibition of tubulin assembly, inhibition of [³H]colchicine binding to tubulin. A comparison was made with the data obtained earlier for compound 5f²⁵ (Table 2).

Table 2.

Inhibition of tubulin polymerization and colchicine binding by compounds 5a–f

Compound	Tubulin assemblya IC50 ± SD (μM)	Colchicine bindingb %inhibition ± SD
5a	>20
5b	>20
5c	2.7 ± 0.2	52 ± 0.6
5d	2.5 ± 0.1	62 ± 0.4
5e	2.2 ± 0.1	69 ± 0.3
5f	2.5 ± 0.3c	54 ± 0.3c
2	1.1 ± 0.1	99 ± 0.06

^aInhibition of tubulin polymerization (tubulin was at 10 μM).

^bInhibition of [³H]colchicine binding (tubulin, colchicine, and tested compounds were at 1, 5, and 5 μM, respectively).

^cValues taken from the literature for comparison.²⁵

Compounds 5a and 5b, both of which had an ortho-hydroxyl group in the A ring, were inactive as inhibitors of tubulin polymerization. Compound 5c, analogous to 5f except for an ortho-hydroxyl group in the ring B shown in the crystal structure to hydrogen bond to N-4 of the tetrazole ring, was moderately less active than 5d and 5e in both tubulin assays. Based on historical data, in fact, 5c had activity indistinguishable from 5f, so that the orthohydroxyl moiety neither enhanced nor reduced activity in this class of agent. Activity was, however, increased by adding a meta-substituent in the B ring, between the other two substituents, with the amino group of 5e increasing antitubulin activity somewhat more than the hydroxyl group of 5d.

Using the 1SA0 crystal structure of α,β-tubulin as a template,³⁶ compounds 5c–f were docked into the colchicine binding site to provide a molecular basis for understanding their activity, and, concomitantly, to explain the lesser activity of compounds 5a and 5b. In the docking studies, compound 5e nicely overlaps with all structural features of colchicine in the binding site (Fig. 2A). Similarly, the trimethoxyphenyl moieties of compounds 5c, 5d and 5f occupied similar conformational space as the trimethoxyphenyl motif of colchicine, which is perhaps the structural basis for the observed bio-activity of these compounds.

Figure 2.

Predicted binding model from Schrodinger. (A) binding model of compound 5e overlaid with that of colchicine in the binding site on β-tubulin, which is shown in cyan ribbon with binding site amino acids rendered in cyan stick. The ligands are rendered in stick with oxygen, nitrogen, and hydrogen atoms colored red, blue, and white, respectively, and the carbon atoms are colored yellow and pink for compound 5e and for colchicine, respectively. (B) Corey–Pauling–Koltun (CPK) rendering of compound 5e docked in β-tubulin showing the tight molecular packing of the trimethoxy aromatic ring. (C) Superimposition of the crystal structure of compound 5a (grey carbon atoms) with potential binding modes (green and cyan carbon atoms) of compound 5a in the colchicine site. Red dashes indicate probable steric clash.

The tetrazole groups of 5c–f are bioisosteric with the B ring of colchicine. The 4-methoxy aromatic rings of compound 5c–f occupy common conformational space with the tropolone C ring of colchicine. Additionally, the binding models show that the conformations of compound 5c–f provide an excellent stereoelectronic fit to the colchicine binding site of tubulin, which is consistent with other potent colchicinoids (Fig. 2B).

The binding models indicate that compounds 5c–f are characterized by the same pharmacophoric features that previously had been used to describe colchicine binding site inhibitors.³⁷ On initial examination, there were potentially two stable conformations of the trimethoxyphenyl group of compounds 5a and 5b in the colchicine site. However, both conformations resulted in highly unfavorable, intramolecular steric clashes involving the hydroxyl group, as shown by red dashed lines (Fig. 2C). The unfavorable intramolecular strain generated by the 2-hydroxy substituent on the trimethoxyphenyl ring may be the factor responsible for the inactivity of compounds 5a and 5b with tubulin.

In conclusion, we synthesized five tetrazole analogues of CA-1 and CA-4 and evaluated their effects on tubulin polymerization, colchicine binding to tubulin and antiproliferative activity against human cancer cell lines. These analogues were designed to evaluate the idea that extended hydrogen bonds at the ortho-positions would introduce conformational restriction and lock the compounds into a cisoid form. The crystal structure of 5a revealed, however, that restriction of conformation of the trimethoxyphenyl ring was not achieved. Instead, a five-membered hydrogen-bond with the vicinal para-methoxy on the A ring occurred. Further, introduction of this hydroxyl group into the trimethoxyphenyl ring A resulted in the loss of antitubulin activity in both 5a and 5b. Introduction of polar groups, such as hydroxyl and amino groups, into the B ring were tolerated. We achieved the desired intramolecular hydrogen bond with the tetrazole ring with the ortho-hydroxyl of 5c, but there was no enhancement of antitubulin activity, although, antiproliferative activity has been reduced relative to compound 5f, lacking hydrogen bond, as demonstrated using crystallographic data. We were able to enhance the antitubulin as well as antiproliferative activity of 5c by incorporating either a hydroxyl (5d) or, especially, an amino (5e) moiety at the meta-position of ring B, between the ortho-hydroxyl and the para-methoxyl moieties.