|Home | About | Journals | Submit | Contact Us | Français|
Starting from Dane’s diene and methylcyclopentenedione, (+)-estrone is synthesized along the Quinkert−Dane route in 24% total yield. The key step is an enantioselective Diels−Alder reaction promoted by an amidinium catalyst as efficiently as by a traditional Ti-TADDOLate Lewis acid.
Estrone 7, due to its relatively simple structure and its considerable pharmaceutical importance, has become―and still is―a most popular target compound for the development of novel synthetic methodologies.1,2 This applies in particular to the Diels−Alder reaction:(3) numerous milestone achievements of constitutional and stereochemical control are closely associated with synthetic approaches toward estrone and related estrogens. Beginning with Dane’s unsuccessful attempt to synthesize rac-3 from diene 1 and diketone 2,(4) Valenta’s variant(5) of Johnson’s synthesis(6)―both starting from diene 1(7)―is among the early applications of Lewis acid catalyzed Diels−Alder reactions(8) in natural product synthesis. With the advent of effective chiral Lewis acids, Dane’s concept regained attention. A TADDOL (α,α,α′,α′-tetraaryl-1,3-dioxolane-4,5-dimethanol) based titanium Lewis acid(9) allowed Quinkert for the first time to implement Dane’s original concept with high levels of constitutional control and enantioselection.(10) More recently Corey has demonstrated the utility of his oxazaborolidinium catalysts by synthesizing (+)-estrone from diene 1.7b,7c
Although the field of enantioselective catalysis was dominated in the past by metal containing catalysts, the past decade has seen an important number of highly selective reactions catalyzed by hydrogen bond donors such as chiral ureas and thioureas.(11) The acceleration of Diels−Alder reactions by hydrogen bonds was reported even prior to Lewis acid catalysis.(12) However, when compared to traditional strong Lewis acids, hydrogen bond donors are mostly considered to be rather weak electrophiles that are inferior at least in terms of rates. In contrast to this view, we will show below that a metal free catalyst forming up to three hydrogen bonds with diketone 2 promotes the Diels−Alder reaction with diene 1 as efficiently as the best titanium TADDOLates used earlier in this step, thus leading to a new organocatalytic variant of the Quinkert−Dane synthesis of (+)-estrone (Scheme 1).
In previous studies we demonstrated the hydrogen bond-mediated complexation of diketone 2 and lipophilic amidinium ions, resulting in a significant acceleration of the cycloaddition with Dane’s diene 1.(13) Axially chiral amidine 8a(14) was intended to form three hydrogen bonds with dienophile 2.(15) However, reaction rates suggested that the resorcinol moiety of 8a does not participate in substrate binding and activation. A crystal structure of a related compound later revealed a considerable distortion of the molecular framework displacing the hydroxy groups of 8a thus preventing hydrogen bond formation with diketone 2 (Figure (Figure1).1). In spite of this drawback, amidinium salt 8a, like a Lewis acid catalyst, shifted the constitutional selectivity of the cycloaddition in favor of isomer 5 and induced up to 26% ee (43% ee when the methyl group of 2 was replaced by ethyl). The predominant configuration of product 5 is explained by an endo attack of the diene to the Si,Si face of dienophile 2 when bound to the chiral amidine.(15)
Having identified the reason for the moderate success of compound 8a, we recently embarked on the synthesis of an optimized amidinium catalyst 9a (Scheme 2). By changing the substitution pattern of the central naphthalene unit, the hydroxy groups are now closer to the amidinium ion and steric shielding around the substrate binding site is improved. Starting from 1-hydroxynaphthalene, resorcinol, and 8-iodo-1-naphthonitrile 14, a sequence of bromination and cross coupling reactions afforded precursor molecule 15 as a racemic mixture of two conformers, slowly interconverting by rotation around the naphthalene−naphthalene axis (see the Supporting Information). Attachment of (S)-alaninol (→16), base induced stereoselective cyclization(14) (→17), and removal of protective groups led to amidine 9a as a single isomer, accessible in gram amounts. A weakly coordinating counterion is crucial for the effective binding of neutral substrates and also increases catalyst solubility in nonpolar solvents at low temperatures.(13)
A crystal structure of amidinium salt 9b (counterion: formate) reveals how hydrogen bond accepting guest molecules may bind to the cationic cleft (Figure (Figure2).2). The formate oxygens sit on top of the amidinium NH with N···O distances of 2.80 and 2.82 Å, respectively. One of the resorcinol OH groups points toward a formate oxygen (O···O: 3.91 Å). Both groups are well placed to form a third host−guest hydrogen bond in solution, which is prevented in the crystal by a competing OH···O bond from a neighbor molecule (O···O: 2.66 Å, see the Supporting Information).
The catalytic potential of amidinium salt 9a was characterized by determining the cycloaddition kinetics of Dane’s diene 1 (45 mM) and diketone 2 (30 mM) in CH2Cl2 at 4−5 °C. Reaction rates are calculated from the reduction of diene concentrations as detected by HPLC. When a minor correction term is introduced for the dimerization of 1,(16) the reaction of 1 and 2 obeys a second order rate law. Compared to the uncatalyzed reaction (an upper limit for k2 of 4 × 10−8 mM−1 s−1 can be estimated) 1 equiv of 9a gives a roughly 550-fold acceleration (k2 = 2.2 × 10−5 mM−1 s−1), 4-fold above the rates seen with our previous catalyst 8a (k2 = 5.3 × 10−6 mM−1 s−1). Thus, reaction kinetics supports the view of three hydrogen bonds involved in the catalyst−substrate complex. Further analysis is complicated by the fact that of the primary cycloadducts 3 and 4 only the latter tautomerizes readily to the keto enol stage.(13) Since we have observed quite different tautomerization rates for 3 and ent-3 induced by chiral amidinium catalysts, a proper analysis of constitutional ratios and enantioselectivities requires complete tautomerization to the keto enols 5 and 6. This can be enforced by addition of water prior to HPLC analysis. On the basis of HPLC, catalyst 9a under these conditions gives 92% total yield of cycloadducts, a ratio of 5 and 6 of 4.7:1 and 60% ee favoring the non-natural enantiomer ent-5. Interestingly, amidinium salt 9a directs the cycloaddition to the opposite face (Re,Re) of the diketone as catalyst 8a. Lowering the reaction temperature to −30 °C increased the ee to 76% and the ratio of 5 and 6 to 7.5:1. Reduction of the catalyst load to 10 mol % just caused a minor loss of selectivity (73% ee). The absolute configuration of ent-5, established previously by chemical correlation,(15) could be confirmed by converting the compound into its TBDPS ether. Recrystallization led to an enantiomerically pure sample that allowed the determination of the stereochemistry by the anomalous X-ray dispersion effect of silicon (via Flack parameters, see the Supporting Information).
A simplistic approach to explain the antipodal influence of catalysts 8a and 9a, both exhibiting identical absolute configurations, is shown in Figure Figure3.3. While the OH groups of catalyst 8a seem not to form hydrogen bonds with substrate 2 their steric influence will favor orientation b over a. Shielding of the backface by the naphthalene moiety should then direct diene 1 to the Si,Si face of 2, leading to the observed natural configuration of product 5. In catalyst 9a, the OH group comes closer to the diketone. Thus by formation of bifurcated H-bonds the substrate may be tilted relative to the catalyst (bottom: c, d). Steric clashes with the amino ether and naphthalene moieties are reduced in orientation c, now exposing the Re,Re face of the diketone to Dane’s diene. The Curtin−Hammet principle is respected insofar as stronger H-bonds will stabilize a complex while increasing at the same time its reactivity. It should be kept in mind, however, that additional attractive or repulsive interactions between catalyst and diene or the counterion, which are entirely neglected in Figure Figure3,3, may be decisive for the total transition state energies.
To synthesize naturally configurated (+)-estrone 7, we prepared the amidinium salt ent-9c (TPPB salt, tetrakis(pentafluorophenyl)borate) from intermediate 15 and (R)-alaninol (Scheme 3). The preparative Diels−Alder reaction was then conducted in CH2Cl2 at −30 °C, using 15 mol % of catalyst ent-9c. After 2 d the reaction product―containing cycloadducts 3 and 5―was isomerized by addition of HCl. High yields of ketoenols 18 + ent-18 (91%) and only traces of constitutional isomers (resulting from cycloadduct 6) could be isolated. Reduction of the enol triflates by published methods10,17 led to the enones 19 + ent-19 (71%). At this stage an enantiomeric excess of 81% ee was determined. It could be raised to 99.1% ee by a single recrystallization from MeOH (83%).(18) Torgov’s diene 20(1b) was obtained from 19 in a deprotonation/protonation sequence under kinetic control (74%). This compound could be stereoselectively reduced by a known two-step procedure consisting of catalytic hydrogenation followed by treatment with acid and Et3SiH (→21, 82%).2f,10 Ether cleavage finally led to (+)-estrone 7 (74%; ee >99.9%). Thus, a total yield of 24% can be achieved from diketone 2 including all purification steps. Catalyst 9a also promoted the cycloaddition of 1 and the ethyl analogue of diketone 2. Replacing the methyl group of 2 by ethyl retarded the Diels−Alder reaction 3-fold and also lowered the enantiomeric excess to a still significant 58% ee (CH2Cl2, −30 °C, 20 mol % of 9a).
How does catalyst ent-9c compare to titanium TADDOLates? To achieve high levels of enantioselectivities, sterically demanding TADDOL derivatives are required. For the best case (9-phenanthryl, 2 equiv of of catalyst, 1.5 equiv of 1, 1 equiv of 2, −80 °C, 2 d) 64% yield of ketoenols 18 + ent-18 was reported with 93% ee.(10c) With use of substoichiometric amounts of Ti-TADDOLate (0.25 equiv, −80 °C) the ee dropped to 85%. However, after prolonged reaction time (7 d) the yield improved to 78%.(10c) The total yield of recrystallized, enantiomerically pure intermediate 19 attainable by this method equals fairly well the results obtained with catalyst ent-9c (0.15 equiv of ent-9c, 1.5 equiv of 1, 1 equiv of 2, −30 °C, 2 d, 91%, 81% ee). Thus, axially chiral amidinium ions can compete in terms of rates and stereoselectivity with a well-established class of traditional chiral Lewis acids. The synthesis of ent-9c, undeniably, is more complex than the preparation of TADDOLs. It is also true that Lewis acids of the type TiCl2(OR)2 are milder than TiCl4 or BF3. The remaining challenge therefore is to identify simple cationic hydrogen bond donors that may rival the reactivity and selectivity of the most potent chiral Lewis acids. Toward this aim, we have already found chiral bisamidinium salts(19) that combine good synthetic accessibility with rate effects distinctly beyond that of axially chiral amidinium salts.(19a)
Diketone 2 (225 mg, 2.04 mmol) and catalyst ent-9c (TPPB salt, 400 mg, 0.30 mmol, 0.15 equiv) were dissolved in CH2Cl2 (24 mL), cooled to −30 °C in a polypropylene tube,(20) and then treated with a precooled solution of diene 1 (570 mg, 3.06 mmol, 1.5 equiv) in CH2Cl2 (10 mL). After 2 days at −30 °C, the mixture was warmed overnight to rt and concd HCl (5 drops) was added. It was intensively stirred for 30 min, filtered over MgSO4, and evaporated to dryness. The residue was chromatographed on silica gel (n-hexane/EtOAc 4:1) to yield enol 18 + ent-18 as an orange foam (550 mg, 91%). It was not further purified by recrystallization to allow a proper ee determination after the subsequent step. 1H NMR (300 MHz, CDCl3) δ 7.07 (d, J = 8.7 Hz, 1 H), 6.71 (dd, J = 8.7, 2.6 Hz, 1 H), 6.70 (s, 1 H), 6.54 (d, J = 3.1 Hz, 1 H), 5.36 (s, 1 H, OH), 3.80 (s, 3 H, OCH3), 3.01 (br s, 1 H), 2.84−2.69 (m, 2 H), 2.47−2.36 (m, 2 H), 2.26−2.09 (m, 2 H), 1.94 (ddd, J = 13.1, 7.0, 4.5 Hz, 1 H), 1.64 (ddd, J = 13.1, 8.3, 4.5 Hz, 1 H), 1.24 (s, 3 H); 13C NMR (75 MHz, CDCl3) δ 208.4, 158.3, 150.0, 136.7, 129.7, 128.72, 128.67, 128.4, 123.4, 113.5, 111.0, 55.3, 49.3, 45.7, 32.0, 28.6, 27.4, 22.8, 22.4. Anal. Calcd for C19H20O3·0.3 EtOAc: C, 75.16; H, 6.99. Found: C, 75.28; H, 6.86.
Enol 18 + ent-18 (550 mg, 1.85 mmol) was dissolved in dry CH2Cl2 (27 mL) and treated simultaneously at 0 °C with Tf2O (0.68 mL, 4.05 mmol, 2.2 equiv) and 2,6-lutidine (0.50 mL, 4.32 mmol, 2.3 equiv). After 1 h the solvent was evaporated and the residue purified on silica gel (n-hexane/EtOAc 4:1) to obtain an orange-brown oil that was dissolved in dry DMF (10 mL) along with PdCl2(dppf) (32 mg, 0.039 mmol, 0.02 equiv). After heating to 60 °C, Et3SiH (0.80 mL, 5.00 mmol, 2.7 equiv) was added and the solution was stirred for 20 min at this temperature. Then the reaction mixture was partitioned between Et2O and H2O, the aqueous phase was extracted with Et2O, and the combined organic phase was washed with sat. NaHCO3 and brine. After drying (MgSO4) the solvent was removed in vacuo and the residue purified on silica gel (n-hexane/EtOAc 4:1) yielding enone (19 + ent-19) as a slightly pink solid (370 mg, 71%), the ee of which was determined by chiral HPLC (DAICEL OJ, n-hexane/iPrOH 10 + 4, 0.8 mL·min−1, tmajor = 12.4 min, tminor = 21.7 min) to be 81%. Recrystallization from MeOH improved the ee to 99.1% (28 mL MeOH, reflux →4 °C; 370→307 mg, 83%). Recrystallized 19: mp 156−158 °C (lit.(7a) mp 151−153 °C, lit.(10c) mp 160 °C); 1H NMR (300 MHz, CDCl3) δ 7.64 (dd, J = 5.6, 2.6 Hz, 1 H), 7.09 (d, J = 7.8, 0.8 Hz, 1 H), 6.72 (dd, J = 7.8, 2.7 Hz, 1 H), 6.70 (s, 1 H), 6.12 (dd, J = 5.8, 2.2 Hz, 1 H), 3.80 (s, 3 H, OCH3), 3.15 (br s, 1 H), 2.90−2.70 (m, 2 H), 2.51−2.36 (m, 2 H), 2.30−2.21 (m, 1 H), 2.20−2.08 (m, 1 H), 1.92 (ddd, J = 13.0, 7.1, 4.6 Hz, 1 H), 1.60 (ddd, J = 13.0, 8.2, 4.6 Hz, 1 H), 1.20 (s, 3 H); 13C NMR (75 MHz, CDCl3) δ 214.4, 162.5, 158.3, 136.7, 130.6, 129.6, 128.7, 128.4, 123.4, 113.6, 111.0, 55.6, 55.3, 46.9, 31.6, 28.7, 27.5, 22.5, 22.3. Anal. Calcd for C19H20O2: C, 81.40; H, 7.19. Found: C, 81.34; H, 7.19.
Financial support by the DFG is gratefully acknowledged (SPP 1179, Organocatalysis)
Experimental details for the syntheses of amidinium salt ent-9c and of (+)-estrone 7, characterization data for all compounds, 1H and 13C NMR spectra, determination of ee values and of reaction kinetics, and crystal structure data for compounds ent-5 and 9b. This material is available free of charge via the Internet at http://pubs.acs.org.