Report

A Predictably Selective Aliphatic C–H Oxidation Reaction for Complex Molecule Synthesis

See allHide authors and affiliations

Science  02 Nov 2007:
Vol. 318, Issue 5851, pp. 783-787
DOI: 10.1126/science.1148597

Abstract

Realizing the extraordinary potential of unactivated sp3 C–H bond oxidation in organic synthesis requires the discovery of catalysts that are both highly reactive and predictably selective. We report an iron (Fe)–based small molecule catalyst that uses hydrogen peroxide (H2O2) to oxidize a broad range of substrates. Predictable selectivity is achieved solely on the basis of the electronic and steric properties of the C–H bonds, without the need for directing groups. Additionally, carboxylate directing groups may be used to furnish five-membered ring lactone products. We demonstrate that these three modes of selectivity enable the predictable oxidation of complex natural products and their derivatives at specific C–H bonds with preparatively useful yields. This type of general and predictable reactivity stands to enable aliphatic C–H oxidation as a method for streamlining complex molecule synthesis.

The 20th century witnessed tremendous advances in synthetic methods and strategies that have enabled small molecule targets of extraordinary complexity and biological importance to be synthesized in the laboratory (1). An important remaining challenge is to achieve syntheses with heightened levels of efficiency. Because many biologically relevant small molecules are oxidized hydrocarbons, reactions that incorporate oxidized functionality selectively into organic frameworks are of particular interest in this regard. Three general reaction classes have been developed for this purpose: functional group interconversions, C–C bond–forming reactions of preoxidized fragments, and olefin oxidations. With these reactions, modern synthetic planning often centers around the use and maintenance of preexisting oxidized functionality. A powerful new class of reactions is emerging that introduce oxidized functionality directly into aliphatic (sp3) C–H bonds. Oxidation reactions for isolated, unactivated sp3 C–H bonds capable of operating with predictable selectivities on complex substrates hold special promise for streamlining syntheses. Such reactions would provide a general way to install oxidized functionalities at a late stage, thereby reducing unproductive chemical manipulations associated with carrying them through a sequence (2, 3).

Despite important advances in the discovery of catalytic methods for aliphatic C–H bond hydroxylations, aminations, and alkylations (46), selective reactivity with complex substrates has only been demonstrated for activated C–H bonds (i.e., adjacent to a heteroatom or π system) (711) or via the use of substrate directing groups (1214). High-yielding oxidations of isolated, unactivated sp3 C–H bonds are rare, and predictable reactivity has only been shown with simple hydrocarbon substrates (10, 1517). The paradoxical challenge in solving this problem lies in discovering a catalyst that is both highly reactive and predictably selective for oxidizing these inert and ubiquitous C–H bonds. Moreover, to be useful in complex molecule synthesis, this reactivity and selectivity must be general for a broad range of densely functionalized substrates. Nature's design principles for creating such catalysts involve the use of elaborate protein binding pockets that inherently limit substrate generality. A different strategy was suggested to us by seminal work on site-selective olefin oxidations using bulky, electrophilic metal catalysts (18, 19). With these reagents, mono-oxidation of polyenes occurs predictably at the most electron-rich, least sterically hindered double bond. Moreover, polar functionality proximal to the olefin can direct oxidation, overriding electronic and steric effects. We hypothesized that site-selective oxidations of unactivated sp3 C–H bonds could similarly be predictably controlled if a suitably reactive metal catalyst could be discovered that is capable of discriminating the subtle electronic and steric differences between C–H bonds in complex molecules (Fig. 1). We herein report an electrophilic iron catalyst, 4, with a bulky ligand framework that uses H2O2, an inexpensive, environmentally friendly oxidant to effect highly selective oxidations of unactivated sp3 C–H bonds over a broad range of substrates. We demonstrate that the site of oxidation with 4 can be predicted in complex organic substrates on the basis of the electronic and steric environment of the C–H bond. Additionally, when carboxylate functionality is present, it can direct oxidations toward five-membered ring lactone formation.

Fig. 1.

Comparison between established modes of site-selective olefin oxidations and proposed modes for site-selective C–H bond oxidations with bulky, electrophilic metal catalysts. BG indicates bulky group; DG, directing group. (I) In asymmetric dihydroxylations (AD) catalyzed by electrophilic OsO4 with bulky quinuclidine ligands, dihydroxylation of polyenes occurs preferentially at the most electron-rich double bond (highlighted in yellow). (II) In AD, the most sterically accessible olefin site (yellow) is dihydroxylated preferentially. (III) Olefinic alcohols are epoxidized site-selectively with Mo(CO)6, VO(acac)2, or Ti(DET)(O-i-Pr)2. acac, acetylacetonate; DET, diethyl tartrate.

Several nonheme iron complexes have shown promising, stereospecific hydroxylation reactivities with unactivated sp3 C–H bonds in simple hydrocarbon substrates (2023). The application of these systems to preparative C–H oxidation chemistry has been prevented by the requirement for large excesses of substrate relative to oxidant, low catalyst turnover numbers, and poor selectivities for product formation. Nevertheless, iron(mep) complexes [mep is N,N'-dimethyl-N,N'-bis(2-pyridylmethyl)-ethane, 1,2-diamine] appeared promising for preparative C–H oxidations with complex substrates because they operate via an electrophilic metal oxidant (22, 23), have a bulky ligand framework amenable to modification, and have been used for preparative epoxidations of olefins containing functionality (24).

The attempted C-H oxidation of pivalate 1 with electrophilic [Fe(II)(mep)(CH3CN)2] (SbF6)2 complex, 3, under preparative conditions (substrate as the limiting reagent) resulted in only low conversion of starting material (12%) and modest selectivity for formation of tertiary hydroxylated product, 2 (56%, Fig. 2A, entry 1). Previous studies have shown a positive correlation between flexibility of the mep ligand and the lability of its iron complexes under oxidative conditions (25). Because unselective oxidations with nonheme iron complexes are often attributed to Fenton-type chemistry upon catalyst decomposition (25), we hypothesized that increasing the rigidity of the mep ligand may lead to improved selectivities. Exchanging the ethylene bridge with a cyclohexane ring had no effect. However, incorporating the methylamines into rigidifying pyrrolidine rings, which furnishes crystallographically characterized complex 4 (26), showed a notable improvement in selectivity (92%), translating into a doubling of the yield of 2 (14% yield, entry 2). The addition of acetic acid (AcOH), previously demonstrated to have a beneficial effect on epoxidations with 3 (24), increased the catalytic activities of both 3 and 4 without substantially changing their selectivities (entries 3 and 4, respectively). This resulted in notable improvement in yields with catalyst 4 (entry 4). Whereas increasing initial catalyst loadings, equivalents of AcOH, or equivalents of H2O2 (alone or in combination) gave no further improvements in yield, collective addition of all three components in a portionwise manner furnished preparatively useful amounts of hydroxylated product. Specifically, we found that three consecutive additions of catalyst 4 (5 mol%), AcOH (0.5 equivalent), and H2O2 (1.2 equivalents) over a period of 30 min afforded diastereomerically pure hydroxylated product 2 in 51% isolated yield (entry 5).

Fig. 2.

(A) Development of a preparatively useful aliphatic C–H oxidation reaction. Products resulting from unselective and overoxidation were observed in trace amounts by 1H nuclear magnetic resonance (NMR) and gas chromatograph analysis of the crude reaction mixture. (B) Structure of [Fe(S,S-PDP)(CH3CN)2](SbF6)2 catalyst (4) based on x-ray crystallographic analysis (anions are omitted for clarity). PDP indicates 2-({(S)-2-[(S)-1-(pyridin-2-ylmethyl)pyrrolidin-2-yl]pyrrolidin-1-yl}methyl)pyridine.

A preliminary investigation of the substrate scope highlights the selective, electrophilic nature of the oxidant generated with 4 and H2O2 (Fig. 3). In all cases examined, hydroxylation occurred preferentially at the most electron-rich tertiary (3°) C–H bond, despite the fact that secondary (2°) C–H bonds have a significant statistical advantage (entries 1 to 9). Although the dicationic iron catalyst 4 is Lewis acidic, a remarkable range of moderately Lewis basic groups were well tolerated. For example, cyclic ethers, esters, carbonates, and electron-deficient amides were compatible with this C–H oxidation reaction (Fig. 3). In all cases examined in which the 3° C–H bond is part of a stereogenic center, hydroxylation occurred with complete retention of stereochemistry (entries 6 and 7). When coupled to asymmetric alkylation methods for constructing stereogenic 3° alkyl centers, this reaction enables a very simple approach for accessing optically pure tertiary alcohols. In substrates in which no 3° C–H bonds were available, oxidation occurred at the methylene hydrogens to afford ketone product via the intermediacy of a 2° alcohol (entry 10). The site selectivities and stereochemical outcome of oxidations with 4 are consistent with a concerted mechanism mediated by an electrophilic oxidant (27).

Fig. 3.

Evaluation of functional group compatibility and substrate scope in 4-catalyzed oxidations of unactivated sp3 C–H bonds with H2O2. “Standard conditions” entail dropwise addition of a solution of H2O2 over ca. 45 to 75 s at room temperature [(50 weight % (wt %), 1.2 equivalents (equiv.) in CH3CN at 0.13 M] to a solution of 4 [5 mole % (mol %)], AcOH (0.5 equiv.), and substrate in CH3CN (0.67 M). After 10 min, a second portion of 4 (5 mol%) and AcOH (0.5 equiv.) in CH3CN (0.05 M) is added, followed by dropwise addition of H2O2 (50 wt %, 1.2 equiv.) in CH3CN (0.13 M); a third addition is then done in the same manner for a total of 15 mol % 4, 1.5 equiv. AcOH, and 3.6 equiv. of H2O2. Products resulting from unselective and overoxidation were observed in trace amounts by 1H-NMR analysis of the crude reaction mixture.

The mass balance of these reactions [average of circa (ca.) 51% mono-oxygenated product with ca. 29% recovered starting material (rsm)] indicates that substantial levels of indiscriminate oxidation are not incurred (Fig. 3). With a highly oxidized L-leucinol derivative, hydroxylation occurred exclusively at the 3° C–H bond (entry 8). Although greater than three iterations of 4, H2O2, and AcOH fail to increase product yields, recycling of isolated starting material provides an effective strategy for obtaining high yields with valuable substrates. For example, the L-leucinol derivative was recycled five times to obtain a 90% isolated yield of pure (–)-12 (entry 8).

Complex small molecules often contain multiple 3° C–H centers. We sought to investigate whether the site selectivity of oxidation with electrophilic catalyst 4 is sensitive to the electronic environment of the 3° C–H bond (Fig. 4A). A series of dihydrocitronellol derivatives were evaluated with electron withdrawing groups (EWGs) in α or β positions to one of the two 3° C–H centers (Fig. 4B). In substrates with no electronic bias, equimolar mixtures of hydroxylated products at both centers were formed (entry 1). In all other cases evaluated, hydroxylation with 4 and H2O2 occurred preferentially at the 3° C–H bond remote from the EWGs (entries 2 to 8). β-Acetate or halogen functionalities gave modest but useful site selectivities (entries 2 to 4), and α-electron withdrawing functionalities resulted in excellent selectivities for remote hydroxylation (entries 5 and 6). Site selectivities of >99:1 were observed when strongly electron-withdrawing carbonyls were incorporated in the α position relative to one of the 3° C–H bonds (entries 7 and 8). These results demonstrate that C–H oxidations with 4 are subject to electronic deactivation with EWGs in the α or β positions.

Fig. 4.

(A) Reactivity trends for oxidations catalyzed by 4 based on the electronics of the C–H bond. (B) Substrate electronic effects on site selectivity in hydroxylations of multiple 3° C–H bonds with 4. Only small amounts of diol byproducts were observed. (C) DFT-calculated three-dimensional structure of the lowest potential energy conformer of (–)-23 with corresponding calculated electrostatic atomic charges (eV) of the 3° C–H bonds of interest. (D) Selective hydroxylation of (–)-23 at C-1 with 4 based on steric effects. For standard conditions see, Fig. 3. Aliphatic C–H bonds that are oxidized to form product are indicated in red.

We next investigated whether site selectivities of oxidation with bulky catalyst 4 are sensitive to the steric environment of the 3° C–H bond. We chose to examine (–)-acetoxy-p-menthane, 23 (Fig. 4D). Energy minimization calculations were performed on (–)-23 with use of density functional theory (DFT) followed by calculation of the electrostatic atomic partial charges. In the lowest potential energy conformer of (–)-23, the two 3° C–H bonds in the γ position to the acetate group (C-1 and C-8) are the least positive and have very similar atomic charges, suggesting a high similarity in their electron densities (Fig. 4C). Thus, only on the basis of the electronic factors, equivalent levels of oxidation would be predicted at these sites. However, we observed a strong preference for oxidation at the C-1 site, most likely because this site is less sterically hindered (Fig. 4D). The gem-dimethyl group of the isopropyl unit in the energy-minimized structure is oriented away from the acetate moiety to relieve unfavorable steric interaction. This conformation places the 3° C–H bond of C-8 proximal to the acetate group, making it sterically less accessible to the oxidant than the C-1 bond (Fig. 4C). These results demonstrate that, in molecules where C–H bonds of similar electron densities are present, sterics can provide a second handle for selectivity.

The interplay between electronic and steric factors in determining the selectivities of C–H oxidations with 4 was further illustrated in a study of methyl esters (Fig. 5A). Hexanoate (+)-26 was hydroxylated by 4 and H2O2 predominately at the 3° C–H site to afford, after an in situ lactonization, (+)-27 as the major product with methyl ketone (–)-28 as the minor product (Fig. 5A). This outcome was predicted on the basis of electronic effects. Increasing the steric bulk around this site by introducing a second methyl substituent in substrate (+)-29 reverses the selectivity and results in formation of methyl ketone (+)-31 as the major product. The 2° C–H bond oxidation by 4 also occurred at the most electron-rich, least sterically hindered site. This experiment shows that steric effects can override electronic effects in site selectivities of oxidation with 4 and suggests that oxidation at 2° C–H sites may operate with selectivities similar to those outlined above for 3° C–H sites.

Fig. 5.

(A) Three modes of selective aliphatic C–H bond oxidation catalyzed by 4. Aliphatic C–H bonds that are oxidized to form major product are indicated in red. (I) Oxidation occurs preferentially at the most electron-rich 3° C–H bond followed by in situ lactonization. Unoxidized (+)-26 was recovered in 23% yield from the reaction. (II) Oxidation occurs at the least sterically hindered, most electron-rich methylene site. Unoxidized (+)-29 was recovered in 16% yield from the reaction. (III) Oxidation is directed to the sterically hindered 3° C–H site by the free carboxylic acid. (B) Predictably selective aliphatic C–H bond oxidations with 4 of natural products and their derivatives. (I) Selective oxidation of 34 with small molecule catalyst 4 and with cultures of C. echinulata occurs at the most electron-rich and least sterically hindered 3°C–H bond to furnish (+)-35. (II) Structure of (-)-36, determined by x-ray analysis. When (–)-36 was exposed to standard reaction conditions, 92% of the starting material was recovered because of electronic deactivation of the core and steric deactivation of the isopropyl 3° C–H bond. (III) Carboxylate-directed lactonization of tetrahydrogibberellic acid analog (+)-37 via C-H oxidation to form lactone (+)-38 in 52% isolated yield (recycled once). The structures of (+)-37 and (+)-38 were determined by x-ray crystallographic analysis and are shown below. For substrates with carboxylic acid directing groups [i.e., (+)-32 and (+)-37], AcOH additive was omitted. For acid-sensitive substrates [i.e., (+)-34], AcOH additive was lowered to 10 mol % per addition.

On the basis of the known role of carboxylates as ligands for nonheme iron complexes and the beneficial role of acetic acid on the catalytic activity of 4, we postulated that a carboxylate group on the substrate could be used to direct the site of C–H oxidation (28). In support of this hypothesis, hexanoic acid (+)-32 furnished only the five-membered ring lactone (+)-30 in 70% isolated yield, whereas oxidation of the analogous methyl ester (+)-29 gave methyl ketone (+)-31 as the major product (Fig. 5A). The terminal carboxylic acid moiety in (+)-32 overrides the previously noted steric effects and directs hydroxylation to the hindered 3° site. Although a unique aspect of aliphatic C–H oxidations catalyzed by 4 is that they do not require a directing group for high selectivities, the ability to use this effect provides a third and powerful handle for selectivity. Predictable reactivity in response to all three such modes of selectivity has proven elusive in prior metalcatalyzed aliphatic C–H oxidations.

The value of this aliphatic C–H oxidation reaction for late-stage synthesis rests on how predictive the electronic, steric, and carboxylate-directing modes of selectivity are in complex molecular settings. In order to evaluate this question, we examined the C–H oxidation of several natural products and their derivatives with 4. Antimalarial compound (+)-artemisinin 34 displays five 3° C–H bonds along its tetracyclic skeleton (Fig. 5B). In addition to the site-selectivity issue posed in this substrate, a chemo-selectivity challenge is present in the form of a sensitive endoperoxide moiety known to be prone to Fe(II)-mediated cleavage (29). On the basis of the selectivity rules outlined above, we predicted that the electron-rich and sterically unencumbered 3° C–H bond at C-10 would be oxidized preferentially. The remaining 3° C–H bonds are in an α and/or β position to electron-withdrawing ester and endoperoxide moieties. We were gratified to find that the selectivity rules for oxidations with 4 developed on relatively simple substrates could be extended to this complex natural product. (+)-10β-Hydroxyartemisinin, 35, was generated as the major product in 34% yield (41% rsm, Fig. 5B). By recycling this valuable starting material through the reaction twice, we obtained diastereomerically pure (+)-35 in 54% isolated yield. With the same protocol, catalyst 3 afforded (+)-35 in only 23% isolated yield. Interestingly, (+)-34 has previously been enzymatically transformed to (+)-35 in 47% yield with microbial cultures of Cunninghamella echinulata (29). Catalyst 4 gives higher yields than the enzymatic reaction with substantially shorter reaction times (three 30-min reactions versus 4 days) and a 10-fold higher volume throughput (0.033 M versus 0.0035 M). The ability of a simple, small molecule catalyst with broad substrate scope to achieve P-450-like tailoring enzyme selectivities is remarkable. We expect that 4 will find widespread use for oxidative modifications to the core structures of natural products and pharmaceuticals.

The high sensitivity of catalyst 4 to steric effects is further illustrated in the attempted oxidation of (–)-α-dihydropicrotoxinin, 36 (Fig. 5B). All of the C–H bonds on the highly oxygenated core are electronically deactivated toward oxidation. Thus on the basis of electronic factors alone, the 3° C–H bond of the exocyclic isopropyl moiety should undergo selective oxidation with 4/H2O2. However, treating (–)-36 under standard hydroxylation conditions resulted in 92% recovered starting material. Examination of an x-ray structure reveals that the isopropyl moiety, in order to avoid severe unfavorable steric interactions, is oriented with its gem-dimethyl group projecting away from the ring system (Fig. 5B). This conformation orients the isopropyl 3° C–H bond underneath the ring and renders it inaccessible to 4. The demonstrated stability of a densely functionalized natural product derivative to this oxidation reaction serves to underscore the remarkably mild nature of this method.

A powerful application of 4-catalyzed hydroxylations is to effect carboxylate-directed, diastereoselective lactonizations at 2° C–Hsites. We chose to evaluate this application with tetrahydrogibberellic acid analog, (+)-37 (Fig. 5B). The carboxylate moiety of (+)-37 may direct five-membered ring lactonizations by 4 to one of four C–H bonds. On the basis of x-ray crystallographic analysis of (+)-37, we predicted that hydroxylation would occur selectively at the 15α 2° C–H bond on the D-ring that is closest to the carboxylate moiety. We were gratified to find that hydroxylation with 4 and H2O2 furnished five-membered ring lactone (+)-38 as a single diastereomer in 52% isolated yield (recycled once). With the same protocol, catalyst 3 afforded (+)-38 in only 26% isolated yield. This large difference in yields with catalyst 4 versus 3 has proven general with all complex substrates examined. Importantly, oxidation of the corresponding methyl ester of 37 resulted in mostly recovered starting material and mixtures of undefined oxidation products, none of which is (+)-38.

Given the predictable reactivity and broad scope demonstrated in this study, we anticipate that this general aliphatic C–H oxidation reaction, and others like it, will fundamentally alter the ways in which complex molecules and pharmaceuticals are synthesized in the laboratory.

Supporting Online Material

www.sciencemag.org/cgi/content/full/318/5851/783/DC1)

Materials and Methods

Fig. S1

Tables S1 to S4

References

References and Notes

View Abstract

Navigate This Article