Thomas M.
Beale
a,
Rebecca M.
Myers
*a,
James W.
Shearman
a,
D. Stephen
Charnock-Jones
bc,
James D.
Brenton
d,
Fanni V.
Gergely
d and
Steven V.
Ley
a
aDepartment of Chemistry, University of Cambridge, Lensfield Road, Cambridge, United Kingdom CB2 1EW. E-mail: rmm32@cam.ac.uk.; Fax: +44 1223 336442
bDepartment of Obstetrics and Gynaecology, University of Cambridge, The Rosie Hospital, Cambridge, United Kingdom CB2 0SW
cNational Institute for Health Research, Cambridge Comprehensive Biomedical Research Centre, Cambridge, UK
dCancer Research UK Cambridge Research Institute, Li Ka Shing Centre, Robinson Way, Cambridge, United Kingdom, CB2 0RE
First published on 23rd August 2010
The generally accepted view is that the 3,4,5-trimethoxy-substituted aromatic A-ring of combretastatin A-4 (CA-4) and its analogues should be conserved in order to maintain biological activity through enforcing an active molecular conformation. Contrary to this, we have found that substituting the larger meta-methoxy groups of CA-4 with smaller halogen atoms results in compounds that are equipotent or more potent than CA-4 itself in vitro.
![]() | ||
Fig. 1 |
Antitubulin agents change the extent of tubulin polymer assembly or disassembly through binding to one of three currently well-defined binding sites.3 These changes include MT-stabilisation (polymerisation) elicited by drug binding through the so-called paclitaxel binding site on β-tubulin, and MT-destabilisation (depolymerisation) in response to drug binding at either the Vinca or the COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcolchicine binding site. A more complex, concentration-dependent binding mode leads to suppression of MT dynamics.4 Such is the unabated interest in these agents there are at least 28 tubulin targeting agents in clinical development.5
The small natural product combretastatin A-4 (CA-4) (5)6 is highly effective at binding to the colchicine site on β-tubulin.7 At micromolar concentrations, despite its simple structure, 5 inhibits MT polymerisation, whilst at nanomolar concentrations it stabilises MT dynamics, arresting the cell cycle in metaphase and triggering apoptosis.8 Furthermore, unlike COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound2 and the Vinca alkaloid drugs, 5 is not a substrate for the multi-drug resistance efflux pump, P-glycoprotein.9 CA-4 (5) also has antivascular activity in vivo at doses less than 10% of the COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundMTD.10 However, an additional mode of action related to the endothelial effect has been suggested which could account for this activity.11
Antivascular therapy is an attractive strategy for the treatment of some cancers. Tumour vasculature is abnormal—the branching pattern is not hierarchical—leading to avascular hypoxic voids and vessels that are leaky and tortuous. These abnormalities are due to peturbation of the normal balance of pro- and antiangiogenic regulators. A consequence of this is that local capillary sprouting is enhanced. This process of neovascularisation is generally uncommon in healthy adult tissues. Strategies for targeting tumour vasculature involve antiangiogenic agents and vascular disrupting agents (VDAs).12 The antiangiogenic agents (such as those targeting the VEGF system) normalise abnormal tumour blood vessels which blocks further endothelial proliferation, reduces leakiness and improves perfusion allowing more effective treatment with traditional cytotoxic drugs.13,14 Despite success in some cancers, on-going clinical trials have found competing production of other proangiogenic factors by some tumour types to be a problem. In contrast, direct targeting of endothelial cells with VDAs causes the rapid collapse of abnormal blood vessels. The remaining viable rim of tumour cells at the periphery responds well to traditional chemotherapy when given before VDA treatment, cutting off the supply of oxygen and nutrients and leading to the rapid induction of tumour necrosis.15
The appeal of 5 is its simple stilbenic structure coupled with its remarkable anticancer activity, especially when these attributes are compared to highly complex structures such as COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound1 or COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound2. Although these latter two drugs target tubulin binding in dividing cancer cells, it is the fact that 5 has its most pronounced effect on the tumour endothelium that sets it apart from the other tubulin targeting agents. Analogues of 5 with improved or additional activity against proliferating endothelial cells giving faster blood vessel shut down at lower minimum concentrations are attractive targets.
The syntheses of CA-4 analogues and structure activity relationships for the binding of COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcolchicine (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound4) to tubulin were reviewed recently.16 Linking these published structures, almost without exception, is a trimethoxyaryl ring (referred to as the A-ring) which is widely considered to be essential for biological activity. Superficially, the rationale for this appears to revolve around the preponderance of trimethoxyaryl motifs in nature, perhaps best illustrated by the vast family of plant COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundlignan natural products.17 Its presence in other natural product-derived anticancer drugs e.g. the topoisomerase II inhibitor COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundetoposide and other COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcolchicine binding site antitubulin agents e.g. steganacin and COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundpodophyllotoxin, have also been cited as compelling evidence. Indeed, it has been suggested that the trimethoxyaryl unit may make an additive contribution to the strength of binding to tubulin, serving as an anchor that maintains the B-ring portion of the molecule in the optimum orientation within the binding locus.18 Thus far, and possibly as a consequence of this, the number of A-ring substitution studies has been limited, with most effort having focussed on analogues with alternative bridging groups and B-rings.19 In one study, all three methoxy groups of 5 were replaced by methyl groups, which resulted in an approximate 10-fold loss in cytotoxicity (EC50 = 0.001 μM and EC50 = 0.02 μM in K562 cells respectively).18 Removal of a single meta-methyl group resulted in a further drop in activity to 0.07 μM. Removal of all three methyl groups (to an unsubstituted aryl ring) resulted in a much larger drop in activity to 0.5 μM. This pattern of activity has also been shown for the removal of methoxy groups from COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcolchicine analogues, with the most significant contribution to this loss arising from the 4-methoxy substituent.20 It has, therefore, been suggested that the 4-methoxy group is essential for binding to tubulin with the meta-methoxy groups required to enforce an active confirmation.21 It has also been reported that replacement of a single meta-methoxy group with a halogen atom (F, Cl, Br) resulted in little difference in human cancer cell line growth inhibition and antitubulin activity in comparison to 5,21 which seems to reinforce this hypothesis. However, the preferred conformation of these halogenated compounds in the crystal phase or when bound to tubulin is not known. As a semi-empirical study has suggested, the energy barrier to torsional rotation of 5 is negligible.22 Thus, it could be considered that the meta-halogen atoms are not merely enforcing an active conformation, but rather providing an electronic interaction with the protein arising from halogen bonding.
Although a familiar concept in the area of supramolecular chemistry,23 the importance of halogen bonding in biological systems has remained uninvestigated until relatively recently. The first extensive treatment of this phenomenon involved targeted data processing of the Protein Data Bank (PDB, July 2004 release). This resulted in the discovery of 113 distinct halogen-protein interactions (defined by distances less than the sum of the Van der Waal's radii of the atoms involved and a bonding angle of greater than 120°).24 A second study using the PDB (December 2008 release) found 248 halogen-protein interactions using the same definition but with a 140° bonding angle threshold.25 Following ab initio modelling of electrostatic potentials in different systems, it was postulated that this interaction arises between electronegative portions of protein backbone or amino acid side-chain and the electropositive cap found at the end of the halogen atoms along the C–X axis for X = Cl, Br, I. It was found that aromatic carbon-halogen bonds produce a larger electropositive cap and that the strength of interaction followed the series F≪Cl<Br<I, mirroring their trend of polarisability.25 Most commonly used molecular modelling platforms are presently inadequate for predicting this type of interaction.24,25
We were, therefore, interested in the activity of a series of 3,5-dihalogenated analogues of 5, maintaining the required 4-methoxy unit, to investigate whether two halogen atoms would maintain or possibly improve the activity of the parent compound. The 3,5-dibromo-4-methoxy motif is found in many marine natural products,26 including the ceratamines (6a and 6b, Fig. 2).27a The ceratamines have been reported to have an antimitotic effect through the stabilisation of MTs,27b and the syntheses and biological activity of these and a number of other dibromotyrosine-derived natural products are currently under investigation by our group.28 Since both 5 and 6 interact with tubulin, it was envisaged that compounds with structural features from both families may exhibit interesting or enhanced antimitotic and, potentially, antivascular activity. Furthermore, in the light of recent reports into the effects of halogen bonding on ligand-biomolecule interaction,24,25,29 we felt that halogenated A-ring analogues of 5 warranted further investigation.
![]() | ||
Fig. 2 |
![]() | ||
Fig. 3 |
COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound3,4,5-Trimethoxybenzaldehyde (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound7) and COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound3,5-difluoro-4-methoxybenzaldehyde (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound8) are commercially available and aldehydes 9, COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound10 and COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound11 were readily synthesised (Scheme 1). After some experimentation, we found that treatment of COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound4-hydroxybenzaldehyde (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound12) with COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundsulfuryl chloride under microwave irradiation gave the best ratio of the desired dichlorinated (13) to monochlorinated product (7:
1). Without purification, methylation of 13 gave 9 in 69% yield over two steps. COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound3,5-Dibromo-4-methoxybenzaldehyde (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound10) was obtained in 90% yield via methylation of COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundphenol COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound14. Iodination of COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound12 was achieved using COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundpotassium iodide, COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundsodium chloride and COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundsodium periodate which gave 15 in excellent yield (97%).30 Subsequent methylation produced COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound3,5-diiodo-4-methoxybenzaldehyde (COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound11) in 90% yield.
![]() | ||
Scheme 1 Reagents and conditions: (a) for 13, SO2Cl2, CH2Cl2, μW, 80 °C, 10 h; (b) for 15, KI, NaCl, NaIO4, AcOH/H2O (9![]() ![]() Read more about this on ChemSpider Download mol file of compoundacetone, 60 °C, 9 69% over 2 steps, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compound10 91%, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compound11 90%. |
COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundn-Butyllithium was used to generate the corresponding ylide from COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundphosphonium salt 16,31 required for the Wittig olefinations. This was followed by addition of the requisite aldehyde as a solution in COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundTHF (Table 1).32 As expected, the Wittig reactions gave 17–21 as approximately 1:
1 mixtures of the (E)- and (Z)-isomers which were either separated at this stage or after TBDMS deprotection. A similar ratio of alkene isomers of 17 was obtained when COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundpotassium hexamethyldisilazide (KHMDS) was used as base to form the ylide at −78 °C.
Compound | X | (E):(Z)b | Isolated (E):(Z):mixture/% | Yield/%c |
---|---|---|---|---|
a Reagents and conditions: (a) n-BuLi, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundTHF, −20 °C to rt, 10 min; (b) aldehyde, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundTHF, rt, 30 min. b Ratio determined by 1H NMR spectroscopy. c Total isolated yield. d Isomers separated after deprotection. |
||||
17 | OMe | 53![]() ![]() |
40![]() ![]() ![]() ![]() |
77 |
18 | F | 53![]() ![]() |
36![]() ![]() ![]() ![]() |
84 |
19 | Cl | 57![]() ![]() |
90 | |
20 | Br | 60![]() ![]() |
96 | |
21 | I | 58![]() ![]() |
97 |
Interestingly, inverting the sense of Wittig coupling; i.e. reaction of the ylide derived from 22 using KHMDS as the base with TBDMS-protected COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundisovanillin, increased the ratio of isomers towards desired (Z)–17 (E:
Z ratio 22
:
78) when compared with the Pettit procedure above (E
:
Z ratio 53
:
47) without a reduction in overall yield (Scheme 2). This represents an improvement to a similar reported procedure.33 In contrast, when applied to the synthesis of 20 there was no improvement in the ratio of isomers, presumably due to electronic effects operating in the initial coupling event.
![]() | ||
Scheme 2 Reagents and conditions: (a) KHMDS, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundTHF, −78 °C; (b) TBDMS-isovanillin, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundTHF, −78 °C to rt, 77%. |
Deprotection of the desired (Z)-isomers of 17 and 18 with TBAF in COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundTHF proceeded smoothly to give 5 and 23 in 73% and 92% yield respectively. Separation of the (E)- and (Z)-isomers for 24, 25 and 26 was performed following deprotection (Table 2). In deuterated COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundchloroform (Z)-26 isomerised to afford exclusively the (E)-isomer over the course of a few days; this was not observed for 5 or 23, 24 or 25.
Compound | X | Isolated (E):(Z):mixture/% | Yield/%b |
---|---|---|---|
a Reagents and conditions: (a) TBAF, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundTHF, rt, 30 min. b Total isolated yield of single or multiple components. c Deprotection performed on purified (Z)-isomer. |
|||
5 | OMe | 73 | |
23 | F | 92 | |
24 | Cl | 47![]() ![]() ![]() ![]() |
100 |
25 | Br | 43![]() ![]() ![]() ![]() |
79 |
26 | I | 59![]() ![]() ![]() ![]() |
75 |
HUVEC Isolate | Time/h | Compound EC50/nM | ||||
---|---|---|---|---|---|---|
5 (CA-4) | 23 (F) | 24 (Cl) | 25 (Br) | 26 (I) | ||
a HUVECs were seeded at 7.5 × 103 cells per well and incubated for 5 h prior to drug addition. Real-time monitoring every 30 min for 45 h allowed extraction of EC50 data. | ||||||
1 | 24 | 1.9 | 193.3 | 3.8 | 2.2 | 0.9 |
45 | 1.8 | 224.4 | 2.3 | 2.2 | 1.0 | |
2 | 24 | 2.2 | 263.1 | 2.6 | 2.3 | 1.7 |
45 | 0.9 | 318.0 | 2.4 | 1.1 | 0.9 | |
3 | 24 | 2.3 | 448.6 | 11.0 | 3.0 | 2.6 |
45 | 3.0 | 563.6 | 15.8 | 3.2 | 1.9 | |
4 | 24 | 2.2 | 1088.0 | 28.3 | 4.5 | 2.5 |
45 | 3.0 | 1276.0 | 49.1 | 5.5 | 2.4 |
In this assay, CA-4 (5) was highly potent in the low nanomolar range with mean EC50 values across all four independent HUVEC isolates of 2.2 nM at both 24 h and 45 h (Table 3). Interestingly, 26 (I) proved to be marginally more potent than 5 with mean EC50 values of 1.9 nM at 24 h and 1.6 nM at 45 h. The activity of 26 (I) agrees with that previously reported (EC50 = 0.6 nM, HUVECs).3525 (Br) was slightly less potent than 5 with mean EC50 values across all isolates of 3.0 nM at both 24 and 45 h with very little variance across the HUVEC isolates. The pattern became disrupted with 24 (Cl) which retained good potency against isolates 1 and 2 (EC50 = 1.1–3.8 nM). However, a considerable drop was observed for isolates 3 and 4 (EC50 = 11.0–49.1 nM). The mean EC50 values for 24 (Cl) across all four isolates were 22.9 nM at 24 h and 17.4 nM at 45 h. The most striking loss of potency was observed for 23 (F) across all four HUVEC isolates (mean EC50 = 498.3 nM at 24 h and 595.9 nM at 45 h) with the most significant loss again seen with isolates 3 and 4. Therefore, the order of activity of dihalogenated CA-4 analogues against HUVECs can be described as F≪Cl<Br<OMe<I.
![]() | ||
Fig. 4 FACS analysis of SK-OV-3 (ovarian cancer) cells stained with propidium iodide after 24 h exposure to control, 5 and 24–26 at 5 nM. (a) Control, (b) 5, (c) 24, (d) 25, (e) 26. Data is representative of three independent experiments. |
Treatment with CA-4 (5) at a concentration of 5 nM increased the amount of cells in G2/M to 30.5% (Fig. 4b). Treatment with 24 (Cl) at 5 nM increased the percentage further to 32.2% (Fig. 4c) and levels of mitotic block increased significantly on treatment with 25 (Br) at 5 nM (51.6%, Fig. 4d), reaching a maximum of 83.6% for 26 (I) at the same concentration (Fig. 4e). Treatment with 23 (F) resulted in a large percentage of G2/M cells (73.7%), however, this required a concentration of 200 nM (see ESI,† Fig. S1). Therefore, the ability of these analogues to induce mitotic block was shown to increase in the order F≪Cl≈OMe<Br<I. This order fits with the proposed strength of halogen bonding24 and correlates with halogen Van der Waal's radii. For all of these compounds the population increase in G2/M is consistent with the hypothesis that they are acting as disruptants to the action of MTs in cell division.
![]() | ||
Fig. 5 Confocal images of HeLa (cervical cancer) cells. (a) Treatment with 25. Accumulations of MTs encircling some nuclei were seen at 100 nM (arrow, panel 3) along with asters 10 nM (asterisk, panel 4). (b) Treatment with CA-4 (5). Staining: anti-tubulin, Dm1a (green), DNA, DAPI (blue). Representative images shown, scale bar 10 μm. |
![]() | ||
Fig. 6 Redistribution of MTs around the nuclear envelope of HeLa cells treated with 25 at 100 nM, scale bar 5 μm. |
Treatment with 5 produced similar phenotypic effects at 10-fold lower concentrations i.e. MT depolymerisation in interphase cells was observed at 100 nM (Fig. 5b, panel 2) and 10 nM (Fig. 5b, panel 3) with interphase MT networks intact at 1 nM (Fig. 5b, panel 4). Consistent with a mitotic block phenotype, a large number of mitotic cells were observed at 1 nM, however, circular accumulations of MTs around nuclei were not seen. These data suggest that 25 induces mitotic block at 10 nM as cells had a similar appearance to those treated with 5 at 1 nM. Photomicrographs were also obtained for SK-OV-3 and SK-OV-3TR cell lines that reflect the observations in the HeLa cells and their relative EC50 values in the MTS assay (ESI,† Fig. S2 and S3).
On treatment with a 10 nM concentration of 25 this effect on mitotic cells was still evident. The cell in late G2 at t = 0 min (Fig. 7, panel 1) successfully entered prophase at t = 11 min, whereupon the nuclear envelope broke down and MT spindles began to form (Fig. 7, panel 2). At t = 21 min bipolar MT spindles could be seen in the mitotic cell (Fig. 7, panel 3). This cell would normally progress through anaphase but treatment with 25 resulted in collapse of the mitotic spindle and the cell failed to exit mitosis after t = 31 min (Fig. 7, panels 4–6). Interphase cells were unaffected by drug treatment until t = 51 min, where the cell denoted by an asterisk began to bleb, whether this is a direct response to 25 is unclear (Fig. 7, panel 6). These responses were not due to photobleaching or phototoxicity as control cells progressed normally through mitosis under the same conditions (Fig. 8).
![]() | ||
Fig. 7 GFP-tubulin HeLa (cervical cancer) cells after treatment with 25 at 10 nM. The asterisk denotes blebbing interphase cell at t = 51 min. Images taken at 10 min intervals immediately after drug addition. Scale bar, 12 μm. |
![]() | ||
Fig. 8 GFP-tubulin HeLa (cervical cancer) cells after treatment with control. Scale bar, 12 μm. |
The data from the FACS analysis and confocal microscopy demonstrates that 25 induces significant G2/M arrest through an interaction with tubulin. At high concentrations (>100 nM) 25 has a depolymerising effect on MTs. At concentrations closer to its EC50 value (1–10 nM) 25 has a larger effect on mitotic over interphase MTs. It is thought that at these concentrations 25 is influencing MT dynamics and in doing so induces mitotic block in dividing cells, whilst interphase cells are unaffected.
In summary, we have shown that replacement of two meta-methoxy groups of CA-4 (5) with F, Cl, Br or I atoms results in conservation of the antitubulin and antimitotic effects of the parent compound. Difluorination, however, results in a significant drop in potency over those substituted with Cl, Br or I. A similar finding was reported for trifluorinated derivatives,18,37 whereas mono-fluorination gave no significant reduction in efficacy.21 Dichlorination resulted in a moderate loss of activity, which is significant since mono-chlorination was reported21 to conserve the activity of 5. We have demonstrated that dibromination or diiodination results in compounds (25 and 26) that have equal or superior potency to 5 in both cell viability and cell cycle analysis assays. This seems to undermine the hypothesis that the steric bulk of the meta-groups results in a more favourable binding geometry, given that bromine (1.85 Å)24 and iodine (1.98 Å)24 have smaller van der Waal's radii than a methoxy (2.60 Å)38 group. This raises the possibility that an increased ability to interact with protein through a halogen bonding mechanism could be responsible for the activity of the dibrominated and diiodinated analogues, as these atoms are those expected to have a halogen bond with the greatest strength. This challenges the established structure-activity relationship formulated for analogues of 5, in which the 3,4,5-trimethoxyaryl group is considered optimal for tubulin binding, and other meta-substitutions merely maintain the same conformation.
Footnote |
† Electronic supplementary information (ESI) available: Experimental procedures for the synthesis of 5 and 23–26 with full spectral data, along with procedures for HUVEC isolation, biological assays and microscopy methods. See DOI: 10.1039/c0md00095g |
This journal is © The Royal Society of Chemistry 2010 |