A focused library synthesis and cytotoxicity of quinones derived from the natural product bolinaquinone

Bolinaquinone is a natural product that is a structurally complex, cytotoxic sesquiterpene quinone. A scaffold simplification and focused library approach using a microwave-assisted Suzuki coupling gave 32 bolinaquinone analogues with good-to-excellent cytotoxicity profiles. Mono-arylbenzoquinones, Library A, were preferentially toxic towards BE2-C (neuroblastoma) cells with growth inhibition (GI50) values of 4–12 µM; only the 3,4-dimethoxyphenyl 23 and 3-biphenyl 28 variants were broad-spectrum active—HT29 (colon carcinoma), U87 and SJ-G2 (glioblastoma), MCF-7 (breast carcinoma), A2780 (ovarian carcinoma), H460 (lung carcinoma), A431 (skin carcinoma), Du145 (prostate carcinoma), BE2-C (neuroblastoma), MIA (pancreatic carcinoma) and SMA (spontaneous murine astrocytoma). Library B with a second aryl moiety exhibited broad-spectrum cytotoxicity with MCF-7 cells’ GI50 values of 5.6 ± 0.7 and 5.1 ± 0.5 µM for 2,5-dimethoxy-3-(naphthalene-1-yl)-6-(naphthalene-3-yl) 33 and 2,5-dimethoxy-3-(biaryl-2-yl)-6-(naphthalene-3-yl) 36, respectively. Similar potencies were also noted with 2,5-dimethoxy-3,6-diphenyl 30 against A2780 (GI50 = 5.9 ± 0.0 µM) and with 2,5-dimethoxy-3-(biaryl-3-yl)-6-(naphthalene-3-yl) 37 against HT29 (GI50 = 5.4 ± 0.4 µM), while the 3,4-dimethoxy mono-aryl analogue 23 exhibited good levels of activity against A2780 (GI50 = 3.8 ± 0.75 µM), the neuroblastoma cell line BE2-C (GI50 = 3 ± 0.35 µM) and SMA (GI50 = 3.9 ± 0.54 µM). Introduction of the amino-substituted Library C gave 2-(naphthalen-1-yl)-5-(naphthalen-3-yl)-3,6-bis(propylamino) 43, with excellent activity against HT29 (0.08 ± 0.0 µM), MCF-7 (0.17 ± 0.1 µM), A2780 (0.14 ± 0.1 µM), A431 (0.11 ± 0.0 µM), Du145 (0.16 ± 0.1 µM), BE2-C (0.08 ± 0.0 µM) and MIA (0.1 ± 0.0 µM).

hydrophobic decalin moiety of 1. The lack of a polar moiety associated with the pendant parent decalin of 1 and analogues 2-7 supported the introduction of simple aromatic moieties only. We were cognisant of the potentially detrimental effect on compound physico-chemical properties on the introduction of multiple lipophilic groups, but viewed that this may also improve cellular uptake. Thus Library A sought to explore the effect of decalin core replacement with less complex hydrophobic moieties such as simple aromatic rings (Type A analogues, figure 2). This route allowed the use of well-established Suzuki coupling approaches [33,34]. The introduction of a methoxy moiety offered the possibility of synthesis simplification and direct access to Library B through further scaffold arylation (Type B analogues, figure 2); in doing so, the hydroxyl moiety of 1 was modified to a methoxy moiety (Type A analogues, rsos.royalsocietypublishing.org R. Soc. open 2 ), Pd(dppf)Cl 2 , K 2 CO 3 , dioxane, µW, 120°C, 20 min, or arylboronic acid (R-B(OH) 2 ), Pd(dppf)Cl 2 , K 2 CO 3 , toluene, reflux, 72 h. figure 2). Finally, in the development of dynamin inhibitors we had shown that the introduction of amino moieties increased inhibition and analogue solubility, and viewed that this approach would potentially enhance cytotoxicity and ameliorate the poor physico-chemical characteristics of the hydrophobic analogues planned in Libraries A and B. Thus amine displacement of the methoxy moieties afforded Library C (Type C analogues, figure 2).
By varying the bromination conditions, preferential access to the mono-or dibrominated products was possible. This simplified product isolation. For 3-bromo-2,5-dimethoxy-1,4-benzoquinone (10), the quinone was dissolved in DMF and heated to 100°C followed by the addition of 1 equivalent of NBS in a single portion. The dibrominated product, 3,6-dibromo-2,5-dimethoxy-1,4-benzoquinone (11), was accessed through the addition of 2 equivalents of NBS in DMF at 60°C. In both approaches, the desired product was isolated by flash chromatography. Subsequent application of a microwave-assisted Suzuki coupling approach with 10 gave facile access to Library A analogues 12-29 [33]. With the exception of sterically encumbered and aliphatic boronic acids, e.g. anthracene and cyclohexyl, microwave irradiation at 120°C for 20 min gave these analogues in 60-75% yield (scheme 1 and table 1). Within this library those reactions that displayed poor conversion to the desired analogue (14, 18, 19, 22 and 23) were also accessed by reflux approaches as extended microwave irradiation led to increasing levels of the undesired desmethoxy analogues. In general, these batch reactions were lower yielding when compared with the microwave conditions. Moreover, we noted that (in our hands) the Suzuki coupling failed to provide useful quantities of the target aryl-coupled quinones with the 2-hydroxy-, 3-hydroxy-and 4-methoxyphenyl boronic acids. The 2-tolyl boronic acid analogue failed to react to completion and gave rise to a complex, inseparable mixture; as such, this analogue was not pursued further.
Library B leveraged our ability to simultaneously conduct a Suzuki coupling with either two equivalents of an arylboronic acid or sequentially with one equivalent of two different arylboronic acids (scheme 2). Symmetrical 3,6-diaryl-dimethoxybenzoquinones were synthesized on treatment of 11 with two equivalents of an arylboronic acid and microwave irradiation at 120°C for 20 min as previously described, which afforded 30 and 31. Asymmetric analogues 32-37 were accessed by treatment of 11 with one equivalent of arylboronic acid under microwave conditions, and once TLC analysis had confirmed consumption of the starting materials, with ensuing addition of an equivalent of a second arylboronic -

4.3
(Continued.) Table 1.  acid and microwave irradiation at 120°C for 20 min as previously described. Analogues 30-37 were obtained in 65-70% yields after flash chromatography (scheme 2 and table 2). In some instances these Suzuki couplings failed, e.g. with the 1-naphthyl, 2-naphthyl and 4-biphenyl boronic acids as one of the substituents, which was most probably a consequence of steric hindrance. The asymmetric analogues were purified only in the last step to omit time-consuming purification of intermediates.
In the construction of Library C, a total of six analogues from Library A and Library B were treated with either n-propyl amine or N,N-dimethylpropane-1,3-diamine in methanol at room temperature to afford excellent yields (85%-95%) of the diaminoquinones (38-43) (scheme 3 and table 3).

Results and discussion
Our investigation of the bolinaquinone SAR commenced with an application of a scaffold simplification and focused library approach that saw the synthesis of mono-arylated dimethoxybenzoquinones (Library A). Of this initial library, analogues 14, 15, 18, 19, 21, 22, 24, 26 and 28 were deemed not sufficiently active at the initial screening concentration of 25 µM to proceed to GI 50 determination (table 1; electronic supplementary material). Of the other Library A members, phenyl 12 and 4-toluoyl 13 were essentially equipotent across the cell line panel with GI 50 values ranging from 10 ± 1.1 µM, (12, BE2-C) to greater than 50 µM (12, H460; and 13, HT29, U87, H460, A431 and MIA). However, 13 showed no activity (defined here as a GI 50 value greater than 50 µM) against HT29, U87, A431 and MIA cell lines, and was 50 µM potent against the normal cell line, MCF10A. The 3-toluoyl 14 analogue was inactive as was 4-butyl 15, which suggested that the position and nature of the simple aliphatic substituent affected the potency of these analogues with a 3-alkyl moiety and a larger 4-alkyl moiety not tolerated. Polar substituents on the introduced phenyl moiety such as 4-OH 16, 4-CH 2 OH 17 and 2,4-di-OCH 3 20 showed moderate activity against BE2-C with all GI 50 values approximately 11 µM, and were equipotent with 12 and 13 with similar activity profiles across the cell lines examined. The mono-OCH 3 analogues 18 and 19 were inactive. Of the di-OCH 3 analogues 20-23 examined, only 3,4-di-OCH 3 20 was active. Notably, of analogues 20-23, analogue 20 was the only active 2-substituted di-OCH 3 analogue, albeit modestly (except against BE2-C cells), which suggests that a 2-substituent may be detrimental to activity. The most active of the compounds with single aromatic moieties on the quinone scaffold was found to be 3,4-di-OMe 23, which was also the first broad-spectrum cytotoxic analogue in this series displaying GI 50 values of 3-15 µM across the cancer cell lines examined.
Within Library A the introduction of a larger aromatic moiety resulted in a modest increase in cytotoxicity with 1-naphthyl 25 showing preferential activity towards the BE2-C cell line, with a GI 50           value of 8 ± 0.5 µM; however, 2-naphthyl 24 was inactive. In a similar manner, the benzothiophene 26 and 3-biphenyl 28 analogues did not proceed to GI 50 determination. However, the introduction of a 2and especially a 4-biphenyl moiety with 29 and 27, respectively, resulted in increased cytotoxicity with the latter returning broad-spectrum activity from 3.9 to 19 µM (table 1), but also a significant increase in toxicity towards the MCF10A cell line with 27 returning a GI 50 of 3.9 µM. This in turn suggested that the orientation of the aromatic or hydrophobic moiety relative to the core quinone moiety is critical to the retention of cytotoxicity. The two most active compounds in this library, 4-biphenyl 27 and 3,4-di-OMe 23, both contain 4-disposed moieties. The introduction of a 2-substituent results in the aromatic moieties twisting out of plane relative to the quinone moiety, e.g. 18 and 20-22, which has an adverse impact on the observed cytotoxicity (not shown). Library A demonstrated that the introduction of large aromatic moieties resulted in good levels of broad-spectrum cytotoxicity with 27; as such, with Library B we sought to explore the effect of increasing the aromatic content of these analogues on compound cytotoxicity. Interestingly, the introduction of a second aryl moiety with Library B gave rise to a different activity profile to that observed with Library A. The parent diphenyl (30) analogue showed moderate-to-good broad-spectrum cytotoxicity, e.g. A2780 GI 50 = 5.9 ± 0.0 µM and SMA GI 50 = 6.2 ± 0.3 µM, except with the H460, A431 and Du145 cell lines (table 2; GI 50 > 50 µM). The introduction of an alkyl chain with 31 was detrimental to activity (table 2). However, this did not appear to be a consequence of a steric clash as the naphthyl substituted 32-37, even in the presence of the 2-and 3-biphenyl moieties, returned good-to-excellent cytotoxicity across the cell lines examined. In these instances, the GI 50 values ranged from 5.1 ± 0.5 µM (36; MCF-7) to 35 ± 6.9 µM (34; H460). The bis-naphthyl (33) saw an overall reduction in cytotoxicity, except with the ovarian and breast cell lines MCF-7 and A2780, with GI 50 values of 5.6 ± 0.7 and 6.9 ± 1.2 µM, respectively. The replacement of the decalin moiety of 1 with a phenyl moiety (12) resulted in a decrease in cLogP (5.10 to 1.85) and polar surface area (PSA, 48.97 to 41.65 Å 2 ); a 4-biphenyl (27) moiety resulted in cLogP of 3.65 and a PSA of 41.90 Å 2 (see the electronic supplementary material). The introduction of a second aromatic moiety was highly detrimental to cLogP with 37 displaying a cLogP of 6.71. However, as we suspected from our earlier dynamin studies, the introduction of amino moieties improved the cLogP values of the resultant analogues such as 41 (cLogP 5.21), but the presence of the amino moieties failed to overcome the effect of two naphthyl moieties with 43 (cLogP 7.91). Despite this significant deterioration in cLogP, we noted no effect of compound precipitation during our cytotoxicity studies, with all analogues soluble in DMSO at 40 mM and maintaining solubility in the MTT assay.
Having established the effects of aryl substitutions on the cytotoxicity of modified bolinaquinone analogues, we next examined the displacement of the -OCH 3 moieties with simple amines to generate Library C. Of the six Library C analogues, only 41 and 43 returned sufficient activity at the initial 25 µM screening dose to proceed to GI 50 determination (table 3) (table 1) and 33 (table 2), respectively, highlights a greater than or equal to 20-fold potency enhancement on the introduction of the amine substituents, potentially suggesting the presence of additional binding domains that these substituents are capable of accessing. Of 41 and 43, the tri-substituted quinone moiety of 41 is capable of both DNA adduction and redox cycling, whereas the tetra-substituted quinone moiety of 43 is not.
While the exact cytotoxic mechanism was not determined, these data support 43 mediating its effects through a redox-cycling mechanism and validate our phenotypic approach to the identification of to note that an in vitro assay is a poor determinant of human toxicity, noting that compound toxicity can only really be measured in vivo [38]. It is accepted that animal models are required to predict potential clinical toxicity.
The lead was deemed to be structurally too complex to facilitate rapid focused library development while retaining the parent hydrophobic core. However, application of a scaffold simplification approach focusing on the hydrophobic core combined with a microwave-assisted Suzuki coupling (and, in some cases, reflux conditions) permitted the synthesis of 32 modified bolinaquinone analogues. These novel analogues spanned three focused libraries. Simple mono-arylbenzoquinones displayed preferential toxicity towards the BE2-C neuroblastoma cell line with GI 50 values of 3-12 µM, apart from the 3,4-dimethoxy (23) and 4-biphenyl (27), which returned broad-spectrum cytotoxicity with an average GI 50 of 10.2 µM. The introduction of a second aryl moiety with Library B failed to enhance BE2-C specificity, but resulted in enhanced broad-spectrum cytotoxic activity in general. Despite the transition towards broad-spectrum cytotoxicity, good levels of activity were apparent against the MCF- Of the analogues reported herein, the highest levels of cytotoxicity were observed in those with bulky aromatic moieties such as a biphenyl and naphthyl, especially in the presence of additional propyl amine moieties. These modified bolinaquinone analogues are promising leads in the search for new cytotoxic agents and we will report on further developments in due course.

In vitro growth inhibition assay
Cells in logarithmic growth were transferred to 96-well plates. Cytotoxicity was determined by plating cells in duplicate in 100 µl medium at a density of 2500-4000 cells per well. On day 0 (24 h after plating), when the cells were in logarithmic growth, 100 µl of medium with or without the test agent was added to each well. After 72 h, drug exposure growth inhibitory effects were evaluated using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide) assay and the absorbance read at 540 nm. Percentage growth inhibition was determined at a fixed drug concentration of 25 µM. A value of 100% is indicative of complete cell growth inhibition. Those analogues showing appreciable percentage growth inhibition underwent further dose-response analysis allowing for the calculation of a GI 50 value. This value is the drug concentration at which cell growth is 50% inhibited based on the difference between the optical density values on day 0 and those at the end of drug exposure [40,41].
A suspension of 9 (0.34 g, 2.02 mmol) in DMF (15 ml) was stirred for 10 min at 60°C and then NBS (0.79 g, 4.50 mmol) was added quickly in one portion to the solution. The reaction mixture was cooled to room temperature (25°C) and stirred for 8 h. Next, water (50 ml) was added and the mixture was extracted with EtOAc (2 × 30 ml), dried over MgSO 4 and the solvent evaporated in vacuo. The crude product was purified by flash chromatography (5% EtOAc-95% hexanes). The title compounds were obtained as orange and red crystalline solids, respectively, with identical spectral data to the previous report [31,32].
(holding time) with magnetic bar stirrer. After cooling with compressed air to 40°C, the reaction mixture was diluted with dioxane (10 ml) and filtered through the Celite ® . The reaction crude was dried in vacuo and subjected to column chromatography for purification (15% EtOAc-85% hexanes) to afford an orange-red solid.
(B) Where none of the desired product or large amounts of starting material was found after microwave irradiation, the reactions were investigated under reflux conditions. A mixture of 10, the requisite boronic acid (1.20 equiv), K 2 CO 3 (2.5 equiv), Pd(dppf)Cl 2 (0.2 equiv) and toluene (15 ml) was heated at reflux and monitored by TLC and LCMS until the starting material had been consumed. When the reaction was deemed complete, the reaction mixture was absorbed onto silica and purified by automated column chromatography (0-100% EtOAc in hexanes).