|Home | About | Journals | Submit | Contact Us | Français|
Intramolecular 1,8-diketone aldol reactions were studied as a tool for the construction of the 7-membered rings of hypocrellin and shiraiachrome. Conditions were identified to obtain the relative stereochemistries present in the two natural products with excellent diastereoselectivity. In addition, a nine-membered ring congener, which has yet to be observed in nature, formed with high selectivity when a hindered amine was used in conjunction with silazide bases.
Hypocrellin (1) and shiraiachrome (2) are members of the perylenequinone class of natural products that also includes the elsinochromes, the calphostins, and cercosporin (Scheme 1).1 The novel helically chiral structures of 1 and 2 are each comprised of an oxidized pentacyclic core as well as a 7-membered carbocyclic ring. Our interest in hypocrellin, isolated from hypocrella bambusae,2 largely stems from its potential use as a photoactivated therapeutic agent.3 Existing as a 4:1 mixture of rapidly interconverting atropisomers (Eq 1),4 hypocrellin displays potent light-induced activity against bladder, brain, prostate and leukemia cell lines’5 as well as antiviral6 and immunotherapeutic properties.7
Scheme 1 illustrates our retrosynthetic analysis of hypocrellin and shiraiachrome A with respect to the ancillary rings. From a common precursor (3), the 7-membered ring will be generated via an intramolecular aldol cyclization using the axial configuration as a stereochemical relay. While there is precedent for highly diastereoselective aldol reactions involving ketone electrophiles,8 there are no instances involving acyclic 1,8-diketone substrates. In this report, we describe model studies to evaluate the feasibility and stereochemical outcome of such an aldol reaction.
To the best of our knowledge, the type of 1,8-diketone aldol reaction that we proposed to use had not been employed previously outside of bridged or macrocyclic architectures.9 In fact, aldol cyclizations of 1,8-dicarbonyls to form 7-membered rings are quite rare. A linear 1,8-dicarbonyl substrate (Eq 2) that does yield the 7-membered ring aldol product proceeds in low yield (35%) even with a more reactive aldehyde electrophile.10 In order to probe this type of ring-forming process further, we subjected 1,8-diketone 411 to LHMDS and were encouraged by the formation of the 7-membered aldol product 5, albeit in low yield (Eq 3). We postulated that the insertion of two further sp2 centers into the tether between the ketones (i.e., 3) would constrain the reacting groups in closer proximity and facilitate such a process.
We selected readily accessible racemic 10 as a model system to study the key transformation (Scheme 2). Importantly, the tether contains four sp2 centers permitting an evaluation of conformational rigidity on the process. Commercially available phenol 6 was protected as the benzoate and then iodinated to provide 7. Ullman coupling followed by deprotection afforded the racemic axially chiral biphenol, which was allylated to yield 8. Subsequent tandem Claisen-Cope rearrangement under thermal conditions provided 9.12 Further benzylation and regioselective Wacker oxidation afforded the desired 1,8-diketone 10.
With 10 in hand, conditions were explored to achieve the key intramolecular aldol transformation. On the basis of transition structure calculations, the Z-enolate (11Z) was expected to provide the syn aldol product with the relative configuration of hypocrellin, 1 (Scheme 3, 12a).13
No reaction was observed under Lewis acid aldol conditions (entries 1–2, Table 1). LDA did provide the product, but as a complex mixture (entry 3). On the other hand, silazide bases14 yielded only two diastereomers in all trials. The nature of the enolate cation affected the stereoselectivity with lithium enolates proving superior (entries 4–6). Warmer temperatures (entry 7) resulted in lower selectivity, but attempts to achieve a thermodynamic product ratio by further warming resulted in complex mixtures. Addition of HMPA gave further improvement (entries 8–9), but the optimal selectivity was obtained with Li(SiPhMe2)2 14b furnishing 12 as a 18:1 diastereomeric ratio and 100% conversion (entry 10).
Determination of the relative configuration proved difficult. With no vicinal protons, coupling constants could not be used. Analysis of the lowest energy conformational families obtained with MM2*15 indicated a lack of characteristic nOe’s to distinguish amongst all the four diastereomers without pure samples of each. Ultimately, debenzylation to 13a and co-crystallization with methyl phenyl sulfoxide afforded suitable crystals (Figure 1). Pleasingly, the relative stereochemistry of 12a matches that of the major atropisomer of hypocrellin.16
Buoyed by this success, we turned again to transition structure calculations which indicated that the E-enolate (11E) should provide the anti aldol product 12c (Scheme 3)13 corresponding to the relative stereochemistry of shiraiachrome A (2). Following a report by Collum, we utilized hindered bases with LiHMDS to initiate formation of the E-enolate.17 These conditions in this system, however, produced a 9-membered ring product (12f) (Table 1, entries 11–13). Presumably, the very large bases create hindered LiHMDS adducts that cause enolization to the less hindered position (11A, see below) even though it is substantially less acidic. In line with this hypothesis, progressively larger bases (Ph3N>i-Pr2EtN>Et3N) provided greater amounts of 12f. With Ph3N, 12f accounted for 90% of the product (entry 13).
Thus, attention turned to the tetramethylpiperide (TMP) bases18 to generate the E-enolate and the shiraiachrome diastereomer. While LiTMP alone provided the 9-membered aldol product (Table 1, entry 14), LiTMP•HBr provided 12d, which had been observed as the minor diastereomer in the Li(SiPhMe2)2 reaction (entry 10), as the major product (entry 15).
To determine the structure of 12d, several experiments were undertaken on the debenzylated versions of 12a–d (13a–d) (Figure 2). First, a comparison of the crystal structure of 13a to the MM2* calculated structure showed remarkable agreement,13 indicating that calculations could be used to estimate distances and dihedral angles. Second, 13d was established as an anti diastereomer from the chemical shifts of the C15–H. In 13a (Figure 2), the C15-H (3.40 ppm) is trans to the C14-OH in 13a (Figure 2). The observation of a higher chemical shift of 3.60 ppm for the C15-H of the new compound implied a closer proximity of the C15-H to the C14-OH (i.e., a cis relationship) narrowing the field to 13c or 13d. Third, nOe experiments were performed on 13a and on the debenzylated form of the E-enolate product (Figure 2). For 13a nOe’s between H11 and H18 and H15 and H18 were observed, consistent with the distances from the crystal and calculated structures. For the E-enolate product, lack of an H15–H18 nOe indicates a trans array confirming an anti aldol adduct (13c or 13d). The strong nOe between H11 and H18, however, is only consistent with 13d, implying that 12d is the major product with LiTMP•HBr (Table 1, entry 15). This result was not in accord with our expectations that 12c would predominate (see TS calculations in Scheme 3).
Overall, the results indicate that the biaryl stereochemistry controls the facial approach to the ketone electrophile thereby establishing the C14 stereocenter (see Scheme 3). On the other hand, the enolization method controls the relative configuration between C14 and C15 (syn aldol for 12a and anti aldol for 12d).
These results support the use of 3 (Scheme 1) with Li(SiPhMe2)2 to produce hypocrellin (1), a tactic which has proved successful.16 On the other hand, the results also indicate that LiTMP•HBr would not produce shiraiachrome A (2) from 3 raising the interesting question of how one might access 2. We propose that ent-3 would provide the enantiomeric array of that illustrated in 12d (Scheme 3) yielding atrop-2 which would only equilibrate with the major form 24 after formation of the 7-membered ring that lowers the atropisomerization energy barrier.16
With the above results in hand for 12a and 12d, the stereochemistry of the 9-membered ring product was expected to arise from a similar facial attack on the carbonyl controlled by the biaryl axis (i.e., 12e, Scheme 5). However, experiments show no H11–H18 nOe, pointing to 12f as the product. This result may be explained by the biaryl stereochemistry controlling the approach to the carbonyl in lower energy conformational isomer 11A’ where steric interactions between H11 and C18 are minimized relative to 11A. It is unlikely that this result is due to formation of a mixture of 12e and 12f followed by a thermodynamic equilibration, due to the high atropisomeric stability of these compounds. 1b
The high yields observed in the 1,8-diketone aldol reaction of model system 10 were unexpected in light of past precedent which indicates that a bridged scaffold is necessary to effect an efficient aldol reaction between ketone centers in a 1,8-relationship. Furthermore, the transannular aldol reaction is expected to induce significant dihedral strain since the biaryl angle becomes compressed going from 10 (88° calc) to 12a/12d (55° X-ray). Since the biaryl angle in 3 (31° calc, Scheme 1) positions the reacting groups in even closer proximity, formation of 1 or 2 (23–24° X-ray) does not require much movement and was expected to occur even more readily. Thus, these results strongly supported the proposed aldol pathway that was subsequently employed to form the natural products hypocrellin (1) and shiraiachrome (2) following the plan outlined in Scheme 1.16
In conclusion, we have developed an efficient method for installing the seven-membered ring of hypocrellin (1) or shiraiachrome A (2) stereoselectively via intramolecular 1,8-diketone aldol cyclizations. Furthermore, use of hindered bases (Ph3N) in conjunction with LiHMDS furnished a nine-membered aldol product via the acetate enolate, which provides an avenue to interesting nonnatural perylenequinones for further study.
A solution of the diol 9 (275 mg, 0.71 mmol) in DMF (20 mL) was cooled to 0 °C. NaH (60% in oil, 140 mg, 3.6 mmol) was added and stirred for 15 min. Benzyl bromide (0.43 mL, 3.56 mmol) and n-Bu4NI (26 mg, 0.071 mmol) were added at 25 °C and the reaction stirred for 1 h. The mixture was acidified with 1 M HCl, diluted with EtOAc, washed with H2O (6×) and brine, and dried (Na2SO4). The concentrate was chromatographed (40% EtOAc/hexanes) to afford 1-[4,6,2’,4’-tetramethoxy-5,3’dibenzyloxy-6’-(allyl)-biphenyl-2-yl]-propan-2-ene as a yellow resin (281 mg, 70%): 1H NMR (500 MHz, CDCl3) δ 7.51 (d, J = 6.9 Hz, 4H), 7.35 (t, J = 6.9 Hz, 4H), 7.29 (t, J = 7.2 Hz, 2H), 6.62 (s, 2H), 5.84-5.76 (m, 2H), 5.08 (d, J = 11.1 Hz, 2H), 5.05 (d, J = 11.1 Hz, 2H) 5.01-4.98 (m, 2H), 4.98-4.96 (m, 2H), 3.88 (s, 6H), 3.67 (s, 6H), 3.03-2.93 (m, 4H); 13C NMR (125 MHz, CDCl3) δ 153.2, 152.1, 139.2, 138.1, 137.3, 135.1, 128.6, 128.4, 128.1, 123.2, 116.1, 107.9, 75.2, 60.9, 56.1, 37.9; IR (film) 2930, 2856, 1594, 1455 cm−1; HRMS (ESI+) calcd for C36H38O6Na+ (MNa+) 589.2566, found 589.2571.
To a solution of PdCl2 (1.3 g, 7.1 mmol) and CuCl (3.5 g, 3.5 mmol) stirred in a DMF/H2O mixture (7:1 v:v, 50 mL) at room temperature for 30 min, 1-[4,6,2’,4’-tetramethoxy-5,3’dibenzyloxy-6’-(allyl)-biphenyl-2-yl]-propan-2-ene (1.0 g, 1.8 mmol) was added. After stirring under oxygen 18 h, the reaction mixture was acidified with 1 M HCl, diluted with EtOAc, washed with H2O (6X) and brine, and dried (Na2SO4). The corresponding concentrate was chromatographed (20–50% EtOAc/hexanes) to afford 10 (520 mg, 50%) as a yellow oil: 1H NMR (500 MHz, CDCl3) δ 7.42 (d, J = 7.4 Hz, 4H), 7.29 (t, J = 7.1 Hz, 4H), 7.23 (t, J = 6.6 Hz, 2H), 6.58 (s, 2H), 5.07 (s, 4H), 3.88 (s, 6H), 3.60 (s, 6H), 3.35 (d, J = 16.7 Hz, 2H), 3.30 (d, J = 16.7 Hz, 2H), 1.94 (s, 6H); 13C NMR (125 MHz, CDCl3) δ 206.6, 153.4, 152.0, 139.7, 138.0, 130.5, 128.6, 128.5, 128.1, 123.2, 109.5, 75.2, 60.8, 56.1, 47.9, 29.9; IR (film) 2937, 1710, 1594 1455 cm−1; HRMS (ESI+) calcd for C36H38O8Na+ (MNa+) 621.2464, found 621.2463.
A solution of 10 (15 mg, 0.033 mmol) in THF (20 mL) was cooled to −78 °C under argon. Lithium tetramethyldiphenyl disilylamide (34 µL, 0.040 mmol, 1.2 M in THF) was added and the mixture stirred for 1 h. The mixture was quenched with saturated NH4Cl, extracted with EtOAc, washed with brine, dried (Na2SO4) and concentrated. Analysis of this material (500 MHz 1H-NMR) indicated 100% conversion and an 18:1 diastereomeric ratio of 12a:12d. The concentrate was chromatographed which separated the diastereomers (30% EtOAc/hexanes) to afford 12a (11 mg, 70%) as a yellow resin: 1H NMR (500 MHz, CDCl3) δ 7.50, (d, J = 7.4 Hz, 4H), 7.39-7.35 (m, 4H), 7.32-7.29 (m, 2H), 6.56 (s, 1H), 6.32 (s, 1H), 5.02-5.18 (m, 4H), 3.88 (s, 3H), 3.84 (s, 3H), 3.74 (s, 3H), 3.66 (s, 3H), 3.39 (s, 1H), 2.38 (d, J = 13.0 Hz, 1H), 2.32 (d, J = 13.0 Hz, 1H), 2.17 (s, 3H), 1.34 (s, 3H); 13C NMR (90 MHz, CDCl3) δ 213.6, 153.5, 153.4, 152.5, 152.3, 141.1, 140.2, 138.2, 138.1, 133.2, 130.1, 128.7, 128.6, 128.5, 128.4, 128.1, 128.0, 123.9, 121.8, 108.7, 105.9, 80.8, 75.4, 75.3, 61.2, 61.1, 60.3, 60.2, 56.3, 46.9, 32.4, 26.9; IR (film) 3497, 2937, 1698, 1598 cm−1; HRMS (ESI+) calcd for C36H38O8Na+ (MNa+) 621.2464, found 621.2449.
To a solution of 12a (single isomer, 69 mg, 0.12 mmol) in a MeOH/THF mixture (1:1 v:v, 10 mL) was added 10% Pd/C (130 mg, 1.2 mmol). A hydrogen balloon was added and the reaction mixture was stirred for 20 min. The mixture was filtered through silica (10% MeOH/CH2Cl2) to afford 13a as a colorless oil (51 mg, 100%): 1H NMR (500 MHz, CDCl3) ™1.33 (s, 3H), 2.13 (s, 3H), 2.30 (d, J = 13.1 Hz, 1H), 2.36 (d, J = 13.1 Hz, 1H), 3.41 (s, 1H), 3.52 (s, 3H), 3.67 (s, 3H), 3.90 (s, 3H), 3.94 (s, 3H), 4.52 (s, 1H), 5.63 (s, 1H), 5.66 (s, 1H), 6.33 (s, 1H), 6.56 (s, 1H); 13C NMR (125 MHz, CDCl3) ™26.5, 31.9, 46.4, 56.2, 56.3, 59.8, 60.3, 60.4, 80.7, 104.9, 107.7, 120.6, 122.9, 125.4, 128.3, 137.4, 138.2, 145.3, 145.5, 147.0, 147.1, 213.4; IR (thin film) 3439, 2937, 1695, 1610 cm−1; HRMS (ESI) calcd for C22H26O8Na (MNa+) 441.1525, found 441.1544.
To a solution of 13a (23 mg, 0.061 mmol) in a benzene/hexane mixture (1:1 v:v, 0.40 mL) was added racemic phenylmethyl sulfoxide (17 mg, 0.12 mmol). The mixture was stirred at room temperature for 12 h. The precipitate was recrystallized from CH2Cl2/hexane (1:1 v:v, 1 mL) to afford the co-crystal product. A crystal structure was obtained securing the relative configuration (see Supporting Information).
To a stirring solution of 10 (90 mg, 0.15 mmol) in THF (16.0 mL) at −78 °C under Ar was added the LiTMP-LiBr complex (15.0 mL, 0.8 M LiTMP, 0.75 mmol). After stirring the mixture for 2 h at −78 °C, the mixture was quenched with satd NH4Cl, extracted with EtOAc, washed with brine, dried over anhydrous Na2SO4, and concentrated. The concentrate was chromatographed (2% MeOH in 1:1 Et2O/hexanes) yielding 12d (64 mg, 71 %) as pale yellow oil: 1H NMR (500 MHz, CDCl3) δ 7.53, (d, J = 7.2 Hz, 2H), 7.49 (d, J = 7.2 Hz, 2H), 7.38-7.34 (m, 4H), 7.32-7.29 (m, 2H), 6.65 (s, 1H), 6.44 (s, 1H), 5.16 (d, J = 11.0 Hz, 1H), 5.15 (d, J = 11.0 Hz, 1H), 5.07 (d, J = 10.1 Hz, 1H), 5.06 (d, J = 10.1 Hz, 1H), 3.88 (s, 3H), 3.82 (s, 3H), 3.72 (s, 3H), 3.66 (s, 3H), 3.58 (s, 1H), 2.81 (s, 1H), 2.44 (d, J = 14.0 Hz, 1H), 2.18 (d, J = 14.0 Hz, 1H), 2.13 (s, 3H), 1.39 (s, 3H); 13C NMR (90 MHz, CDCl3) δ 210.4, 153.7, 153.4, 152.5, 152.3, 140.9, 140.2, 138.1, 137.9, 133.1, 131.3, 128.6, 128.5, 128.4, 128.11, 128.05, 124.2, 121.7, 108.7, 106.0, 81.4, 75.4, 75.3, 63.7, 61.2, 61.1, 56.5, 56.2, 48.0, 31.4, 29.9, 24.0; IR (film) 3250 (br), 2943, 2881, 2835, 1715, 1599, 1460, 1413, 1104, 749 cm−1; HRMS (ESI+) calcd for C36H38O8Na+ (MNa+) 621.2462, found 621.2451.
To a stirring solution of 10 (90 mg, 0.15 mmol) and Ph3N (1.5 g, 6.0 mmol) in PhCH3 (20 mL) at −78 °C under Ar was added lithium hexamethyl disilylamide (5.0 mL, 1.0 M in PhCH3, 5.0 mmol). After stirring at −78 °C for 2 h, the mixture was quenched with satd NH4Cl, extracted with EtOAc, washed with brine, dried over anhydrous Na2SO4, and concentrated. The concentrate was chromatographed (2% MeOH in 1:1 Et2O:hexanes) yielding 12f (60 mg, 67 %) as pale yellow oil: 1H NMR (500 MHz, CDCl3) δ 7.47 (d, J = 7.4 Hz, 4H), 7.37-7.32 (m, 4H), 7.30-7.28 (m, 2H), 6.65 (s, 1H), 6.60 (s, 1H), 5.13 (d, J = 11.1 Hz, 1H), 5.09 (d, J = 11.1 Hz, 1H), 5.04 (d, J = 11.1 Hz, 1H), 5.01 (d, J = 11.1 Hz, 1H), 4.06 (s, 1H), 3.88 (s, 3H), 3.84 (s, 3H), 3.70 (s, 3H), 3.64 (s, 3H), 3.32 (d, J = 16.6 Hz, 1H), 3.16 (d, J = 16.6 Hz, 1H), 2.53 (d, J = 13.8 Hz, 1H), 2.50 (d, J = 13.8 Hz, 1H), 2.30 (d, J = 13.2 Hz, 1H), 1.91 (d, J = 13.2 Hz, 1H), 1.19 (s, 3H); 13C NMR (90 MHz, CDCl3) δ 211.0, 154.0, 153.8, 152.1, 151.41, 140.3, 140.0, 138.0, 137.7, 133.1, 131.7, 128.7, 128.5, 128.4, 128.2, 128.0, 124.7, 123.3, 110.4, 109.8, 77.6, 75.6, 75.0, 74.7, 61.0, 60.9, 56.23, 56.16, 52.7, 48.2, 46.0, 31.4; IR (film) 3247 (br), 2927, 2858, 1730, 1692, 1599,1460, 1406, 1097, 1027, 749 cm−1; HRMS (ESI+) calcd for C36H38O8Na+ (MNa+) 621.2462, found 621.2416.
This work was finanically supported by the NIH (CA-109164). Partial instrumentation support was provided by the NIH for MS (1S10RR023444) and NMR (1S10RR022442). Thanks to Eli Lilly (E.O.B), the Vagelos Scholars Program (J.L.), and the Defence Science & Technology Agency of Singapore (J.L.) for fellowships.
Supporting Information Available Additional experimental descriptions, NMR spectra, calculated structures, and crystallographic data. This material is available free of charge via the Internet at http://pubs.acs.org.