|Home | About | Journals | Submit | Contact Us | Français|
Discovery of efficient catalysts is one of the most compelling objectives of modern chemistry. Chiral catalysts are in particularly high demand, as they facilitate synthesis of enantiomerically enriched small molecules that are critical to developments in medicine, biology and materials science1. Especially noteworthy are catalysts that promote—with otherwise inaccessible efficiency and selectivity levels—reactions demonstrated to be of great utility in chemical synthesis. Here we report a class of chiral catalysts that initiate alkene metathesis1 with very high efficiency and enantioselectivity. Such attributes arise from structural fluxionality of the chiral catalysts and the central role that enhanced electronic factors have in the catalytic cycle. The new catalysts have a stereogenic metal centre and carry only monodentate ligands; the molybdenum-based complexes are prepared stereoselectively by a ligand exchange process involving an enantiomerically pure aryloxide, a class of ligands scarcely used in enantioselective catalysis2,3. We demonstrate the application of the new catalysts in an enantioselective synthesis of the Aspidosperma alkaloid, quebrachamine, through an alkene metathesis reaction that cannot be promoted by any of the previously reported chiral catalysts.
A chiral catalyst may promote a transformation with exceptionally high degrees of efficiency and enantioselectivity if the modes with which it associates with substrate molecules are sterically—as well as electronically—distinct4,5. Design of a chiral metal-based catalyst, where stereoelectronic interactions affect the energetics of the catalytic cycle, can lead to consideration of complexes that bear a stereogenic metal centre; stereoselective preparation and preservation of the stereochemical identity of the catalyst then become crucial issues. In the limited number of chiral stereogenic-at-metal catalysts prepared, stereoselective synthesis is addressed through multidentate ligands6–8. As stereogenic-at-metal complexes can undergo stereo-mutation9, polydentate ligation ensures minimal erosion of stereochemical integrity. The rigidity of bi- or polydentate ligation may be detrimental, however, if catalyst structural fluxionality gives rise to enhanced activity and/or enantioselectivity. Chiral complexes that bear a stereogenic metal centre and only monodentate ligands, therefore, offer attractive opportunities in enantioselective catalysis10,11. Such options are, nonetheless, almost entirely unexplored; it is particularly challenging to design and synthesize enantiomerically pure stereogenic-at-metal catalysts that are devoid of polydentate ligands, do not readily stereomutate and can thus serve as effective chiral catalysts.
The growing list of alkene metathesis transformations that cannot be promoted by the existing metal complexes has underlined the need for more effective classes of catalysts1,12,13. One instance that points to such limitations is a key step in an enantioselective synthesis14 of the adrenergic blocker quebrachamine15 (Fig. 1): the conversion of achiral 1 to the strained chiral tetracycle 2 by reaction of sterically hindered alkenes. The results illustrated in Fig. 1, regarding molybdenum alkylidene 3 (refs 12, 16) and ruthenium carbene 4a (ref. 17), although being somewhat inefficient, represent the optimal among available achiral catalysts. The faster initiating 4b (ref. 18) proceeds only to 48% conversion, most likely as a result of lower stability of the carbene intermediates. In relation to an enantioselective quebrachamine synthesis, the existing chiral catalysts, represented by 5–7 (ref. 12), are entirely ineffective in promoting the formation of 2 (≤5% conversion with up to 50 mol% loading after up to 48 h at 22–80 °C). With 16 mol% 8 (ref. 19) at 80 °C, there is approximately 50% conversion, but only rac-2 is generated; related monodentate N-heterocyclic chiral ruthenium carbenes20 are equally ineffective.
As the first step towards identifying a new set of chiral catalysts, we considered reasons for the higher activity of alkylidene 3 in comparison with 5–7, a variance that exists despite the similar electron-withdrawing ability of the oxygen-based ligands. We surmised that such differences may originate from the structural rigidity of the diolates and the resulting higher energy transition states and intermediates in the catalytic cycle. The O–Mo–O angle in tetrahedral 5a ((O–Mo–O), ~127° (ref. 21)), compared to that in a related square pyramidal tungstacyclobutane ((O–W–O), ~99° (ref. 22)), supports the hypothesis that structural adjustments within the catalytic cycle may be better accommodated in a less rigid metal complex.
The above considerations imply that a chiral molybdenum catalyst, bearing monodentate ligands—either two that are chiral (identical enantiomers, non-stereogenic-at-Mo; see Fig. 2) or one achiral and one chiral ligand (stereogenic-at-Mo)—would be more active. Recent theoretical studies suggest that high-oxidation-state complexes containing two electronically distinct ligands should be particularly effective promoters of alkene metathesis4,5 (Fig. 2). According to theoretical explorations, an acceptor ligand (A in I, Fig. 2) ensures sufficient metal Lewis acidity, necessary for effective binding of the Lewis basic alkene. Efficient alkene coordination, however, requires the presence of a sterically accessible ligation site, made available through alteration of the structure of the initial tetrahedral complex I (Fig. 2). It has been proposed that the donor group (D) causes I to distort dissymmetrically; ligand D preferentially interacts with the most available metal orbital such that a trigonal prismatic complex bearing an open ligation site is rendered energetically more accessible. The donor ligand thus occupies an apical site in II (the alkylidene, acceptor and imido ligands constitute the basal plane of the trigonal prism), coordinated opposite only to a weakly bound alkene (trans effect). The resulting complex, III, leads to trigonal bipyramidal IV, which undergoes facile cycloreversion (Fig. 2) in which the metallacyclobutane carbons, constituting the alkene being released, are positioned trans to D, affording V. The aforementioned electronic effects thus facilitate formation of complex V as well. Such a scenario suggests acceleration at two critical stages of the catalytic cycle: substrate–catalyst association and metal-lacyclobutane decomposition. The above hypothesis finds support in the efficient reaction shown in Fig. 2: 1 mol% rac-9 (ref. 23) promotes formation of rac-2 within 1 h. Such a level of activity is in stark contrast to 3 (Fig. 1), a complex with two hexafluoro-t-butoxides.
Stereoselective synthesis of stereogenic-at-Mo complexes became our next objective. One approach would involve diastereoselective mono-protonation of bis-pyrrolides 10a and 10b (Fig. 3)24 with 1 equiv. of a chiral enantiomerically pure alcohol. Preliminary experiments (T. Pilyugina, A.H.H. and R.R.S., unpublished work), involving chiral diols (for example 5–7, Fig. 1), indicated that pyrrole molecules, released upon alcohol exchange, are not deleterious to catalyst activity. These observations suggested that mono-aryloxides might be prepared and used in situ. To this end, we favoured binaphthol-derived alcohols, because this class of ligand possesses several important attributes: (1) ease and low cost of synthesis, (2) facility of modification and (3) appropriate electron-withdrawing ability.
We subjected bis-pyrrolides 10a and 10b to mono-protected diols derived from binaphthol. Treatment of 10a with 1 equiv. of enantiomerically pure 11 affords 12a in 19:1 d.r.; with 10b, 12b is generated as a 7:1 mixture. Under identical conditions, 13a and 13b are obtained with similar selectivity. The identity of the major diastereomer of 13b was established by X-ray crystallography (Supplementary Information page SI18); assignment of other isomers in Fig. 3 is by inference. The configurational stability of the above complexes is indicated by the absence of detectable variations in the diastereomeric ratios; for example, even after one month, a toluene solution of 13b remains a 7:1 mixture of the same diastereomers.
Attempts to prepare the mono-aryloxides with the parent ligand lacking the bromides led to a mixture containing bis-aryloxides as well as the unreacted bis-pyrrolides. Synthesis of such complexes can be accomplished with 1 equiv. of the related octahydrobinaphthol, but the process is less stereoselective (d.r., 1–2.5:1) and, unlike the bromides, excess alcohol causes bis-aryloxide formation. In reactions to form 13a and 13b, bis-aryloxides are not detected even under relatively forcing conditions (60 °C, 2 h). The bromine atoms are therefore required for efficient and stereoselective formation of 12 and 13 (minimal non-stereogenic-at-Mo bis-aryloxide). Delineation of principles that govern stereoselective ligand exchange reactions is a topic of ongoing investigations.
We then turned our attention to probing the ability of mono-aryloxides to perform as catalysts for enantioselective ring-closing metathesis (RCM). The transformation in Table 1 (14 →15) served as the model process; reactions were performed with complexes prepared and used in situ. With 1 mol% 12a (d.r., 19:1; entry 1, Table 1), RCM proceeds to 50% conversion, affording (R)-15 in 18% e.e. With 12b (d.r., 7:1; entry 2, Table 1), complete conversion is achieved and (S)-15 is obtained in 50% e.e. The above trend is again observed in reactions involving 13a and 13b (entries 3 and 4, Table 1), but with a wider selectivity gap: with 1 mol% 13a, RCM proceeds to 54% conversion to afford (R)-15 in 13% e.e., whereas in the presence of 13b there is >98% conversion and (S)-15 is isolated in 93% e.e. With 13b, the RCM can be performed in a fume hood at 22 °C with 1 mol% loading (30 min, 96% conversion, 86% yield, 92% e.e.). These results illustrate the significant potential of the new catalysts in practical procedures for enantioselective synthesis. Despite providing slightly lower selectivity, 13b is more effective than the corresponding optimal molybdenum diolate (5 mol% 5a: 20 min, 95% conversion, 98% e.e. (ref. 25)).
The reduced reactivity and selectivity in reactions of complexes that are more stereochemically pure (that is, 12a and 13a) raises the question of whether the minor diastereomers are more active. To investigate this, a stereochemically pure sample of 13b (d.r., >25:1) was used to initiate enantioselective RCM of 14 (1 mol%, 22 °C, 30 min): (S)-15 was isolated in 93% e.e. and 94% yield, results identical to those obtained with the 7:1 mixture (entry 4, Table 1). Moreover, by monitoring the reaction progress spectroscopically (400-MHz 1H NMR) in the presence of 13b, generated in situ, we established that >98% of the major isomer is consumed, presumably through initiation with the alkene substrate, whereas the minor diastereomer remains largely intact (>95%). Related experiments indicate that in the case of binaphtholate 12b, both isomers are initiated with nearly equal facility, a finding that might explain why the product is obtained with lower selectivity with this complex.
A variety of enantioselective RCM reactions illustrate the special utility of the new catalysts (Table 2). None of the previously reported diolates promotes RCM of secondary allylamine 16 (entry 1, Table 2)26; by sharp contrast, with 2.5 mol% 13b, there is 94% conversion to piperidine 17 within 1 h, furnishing the desired product in 89% yield and 67% e.e. Diolate 7 (see Fig. 1) and 13c (see Fig. 4 for structure; d.r., 5:1) promote formation of 19 (entry 2, Table 2) with high enantioselectivity (98% and 91% e.e., respectively), but RCM with the mono-aryloxide is significantly more efficient: 3 mol% dichloro-substituted 13c furnishes 95% conversion to 19 within 1 h, whereas 48 h are required with 10 mol% diolate 7. Reaction of amine 20 (entry 3, Table 2), which requires 15 mol% 5b, proceeds to only 75% conversion after 24 h, yielding 21 with low selectivity (e.r., 65:35). By contrast, 1 mol% 13b is sufficient for >98% conversion within 1 h, affording 21 in >98% yield and 92% e.e. Similarly, enantioselective synthesis of azepine 23 (entry 4, Table 2) is more efficient (1 h versus 20 h) and substantially more selective (81% e.e. versus 40% e.e.) when 13c is used (versus diolate 5b). RCM of arylamine 24 (entry 5, Table 2) proceeds with high enantioselectivity when diolate 5a (ref. 25) or mono-aryloxide 13c is used; with 1 mol% 13c, however, there is >98% conversion in 1 h (versus 7 h with 2 mol% 5a). With enantioselective RCM of silyl ether 26 (entry 6, Table 2), diolate 5c initiates a slightly more selective ring closure (97:3 e.r. versus 94:6 e.r. with 13c); with 1 mol% 13c, conversion of 26 to 27 is complete in 1 h (versus 5 mol% and 12 h with 5c)27. Three points regarding the transformations in Table 2 merit mention. First, in certain cases, dichloro complex 13c affords similar, but higher, selectivity in comparison with dibromo complex 13b. Second, the molybdenum centre undergoes two inversions in the course of each catalytic cycle (Fig. 2). The high enantioselectivities observed might suggest that adventitious isomerization occurs at a minimum or not at all, as such isomerizations would furnish the alternative product enantiomers. Third, chiral ruthenium-based alkene metathesis catalysts developed so far only promote enantioselective RCM of trisubstituted alkenes with high selectivity (≥80% e.e. or ≥90:10 e.r.)20,28,29.
The most notable demonstration of the attributes of molybdenum mono-aryloxides is in the context of enantioselective synthesis of quebrachamine (Fig. 1). In the presence of 1 mol% 13c, generated in situ, triene 1 is transformed entirely in 1 h to 2 in 84% yield and 96% e.e. (Fig. 4). The target alkaloid is subsequently obtained in high enantiomeric purity and yield (97%). The molybdenum-catalysed process in Fig. 4 thus constitutes the application of a highly effective catalytic enantioselective RCM (e.r., ≥95:5) to the total synthesis of a relatively complex natural product1,12.
The chiral complexes discovered in this study are examples of configurationally stable, enantiomerically pure, diastereomerically enriched, stereogenic-at-metal catalysts that bear only monodentate ligands. We have demonstrated that constitutionally fluxional stereogenic-at-metal complexes that effectively exploit stereoelectronic factors should be considered as viable and attractive options in future catalyst design.
The general procedure for in situ catalyst preparation and catalytic enantioselective alkene metathesis is as follows. An oven-dried (135 °C) 25-mL round-bottom flask, equipped with a magnetic stir bar, is charged with Mo bis-pyrrolide (0.01 equiv.) under N2 atmosphere (dry box). The flask is sealed with a septum, taped, and brought to a fume hood, in which all manipulations are performed. An 8-mL vial is charged with the alcohol (0.01 equiv.), which is subsequently dried by azeotropic distillation with C6H6. Toluene (0.02 M) is added to the alcohol and the resulting solution immediately transferred by syringe to the Mo bis-pyrrolide (flask pressurized on addition; that is, no outlet needle). The mixture is allowed to stir for the appropriate period of time at 22 °C. A 20-mL vial is charged with substrate (1 equiv.), which is dried by azeotropic distillation with C6H6. Toluene (final substrate concentration, 0.2 M) is added and the solution immediately transferred by syringe to the 25-mL flask (flask pressurized on addition); the mixture is allowed to stir for the required period of time. The reaction is quenched by addition of wet diethyl ether and concentrated in vacuo (percent conversion determined by 400-MHz 1H NMR analysis). Purification is performed by silica gel chromatography and enantiomeric purity of the product determined by gas or high-performance liquid chromatography analysis in comparison with authentic racemic material.
This research was supported by the US National Institutes of Health, Institute of General Medical Sciences (grant GM-59426 to A.H.H. and R.R.S.). We are grateful to B. C. Bailey and K. Wampler for assistance in obtaining the X-ray structure of the major diastereomer of molybdenum complex 13b. We thank A. R. Zhugralin for numerous discussions regarding the mechanistic aspects of these investigations and R. Singh for experimental suggestions. Mass spectrometry facilities at Boston College are supported by the US National Science Foundation (grant DBI-0619576).
Author Contributions S. J. Malcolmson and S. J. Meek were involved in the discovery and development of the new catalysts. E.S.S. designed and developed the synthesis route to racemic quebrachamine. A.H.H. and R.R.S. designed and directed the research program. A.H.H. wrote the manuscript.
Author Information X-ray crystallographic data have been deposited at the Cambridge Crystallographic Data Centre, UK; CCDC 703841 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre (www.ccdc.cam.ac.uk/data_request/cif). Reprints and permissions information is available at www.nature.com/reprints.