Kwan Lecture 34: Enamine and Iminium Organocatalysis
December 2, 2011 Enamine and Iminium Organocatalysis Scope of Lecture Eugene E. Kwan Key Questions proline aldol reactions enamine and iminium organocatalysis N H CO 2 H detection of enamines 1+rate law analysis List-Houk model and alternatives enantioselective reduction of iminium ions enantioselective Diels-Alder reactions cascade organocatalysis SOMO and photoredox catalysis Helpful References 1. "The Advent and Development of Organocatalysis." Macmillan D.W.C. Nature 2008, 455, 304-307. 2. "Theory of Asymmetric Organocatalysis..." Houk, K.N. et al. Acc. Chem. Res. 2004, 37, 558-569. 3. "Enamine Catalysis Is a Powerful Strategy..." List, B. Acc. Chem. Res. 2004, 37, 548-557. 4. "Iminium Catalysis." Pihko, P.M. et al. Chem. Rev. 2007, 107, 5416-5470. 5. "Organocatalytic Cascade Reactions..." Enders, D. et al. Nature Chemistry 2010, 2, 167-178. I thank Dr. Rob Knowles and Dr. Jaclyn Henderson for some helpful discussions and material for the preparation of this lecture. H O Br CO 2 Et CO 2 Et 83% yield, 95% ee H O CO 2 Et CO 2 Et visible light (fluorescent bulb) organocatalyst (20 mol%) Ru(bpy) 3 Cl 2 (0.5 mol%) 2,6-lutidine, DMF, 23 C + H O iPr NHPMP N H CO 2 Et PMP H O Me + Me N H COOH Me Me Me O O Me H + Me O Me OH N H CO 2 H why a turnover? O Me H O Me O Cl Cl Cl Cl Cl Cl O Me O H Me Cl EtOAc, 50 C 20 mol% catalyst + 86% yield 14:1 dr, 99% ee (1) Proline vs. Mannich Reactions (2) Organocascade Reactions (3) Organo-SOMO/Photoredox Catalysis mechanism? mechanism? Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Predicting the Future Society: What technologies will be important in the future? Organic Chemists: What new developments are going to propel the field forwards? People have been trying to predict the future of technology for quite a while. Unfortunately, they are very bad at it. In 1977, Ken Olson, CEO of Digital Equipment Corporation, a maker of large mainframe computers, famously said this at a convention of the World Future society: "There is no reason for any individual to have a computer in his home." ("Bill Gates: The Path to the Future." Gaitlin, J . Perennial Currents, 1999. ISBN: 0-38080-625-8, pg 39). But why are people so bad at it? The writer Vernor Vinge proposes that the creation of superhuman intelligence will lead to a breakdown in our ability to predict the future in what he calls a "singularity." In a more restrictive usage, technological singularities are breakthroughs beyond which technological progress becomes so fast that it makes any predictions of the future become impossible. Moore's Law is the classic example: Wikipedia Somewhat more controversial is the idea of accelerating change. Acccording the futurist Ray Kurzweil, paradigm shifts are occurring exponentially more frequently, and will soon lead to "change so rapid and profound it represents a rupture in the fabric of human history": Now, before superintelligent AIs become our inscrutable overlords, or the fabric of history ruptures, we will need to do some organic chemistry. In 1990, Seebach considered the future of organic chemistry ("Organic Synthesis--Where Now?" ACIEE 1990 29 1320) and wrote: "The primary center of attention for all synthetic methods will continue to shift towards catalytic and enantioselective variants..." This has certainly been true. However, he also felt that: "The discovery of truly new reactions is likely to be limited to the realm of transition metal organic chemistry, which will almost certainly...[deliver] miracle reagents..." This foreshadowed things like olefin metathesis, but did not really predict the explosion of interest in organocatalysis. Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Predicting the Future As Macmillan points out, mentions of the word "organocatalysis" have truly exploded in the literature (ISI Web of Knowledge, ref 1): Macmillan writes (ref 1): "Why was organocatalysis so long overlooked as an area of research?...One perspective worth considering is that it is impossible to overlook a field that does not exist yet... researchers cannot work on a problem that has not been identified." This may be a bit harsh, but there is a ring of truth to it. In Seebach's defense, hardly any of the reactivity that will be discussed in the first part of this lecture (proline aldol and Mannich reactions, reductions of o,|-unsaturated iminium ions, enantioselective Diels-Alder reactions) are actually new. Rather, they are "simply" enantioselective variants of old reactions. However, in the second part of the lecture, I will show you some genuinely new reactivity in the context of SOMO catalysis. The seminal work in this area came simultaneously from Hajos and Parrish (JOC 1974 39 1615) at Hoffmann-La Roche, and Eder, Sauer, and Weichert at Schering AG (ACIEE 1971 10 496). They reported that proline catalyzed this cyclization: Et O O O 3 mol% L-proline (Eder and co-workers used harsher conditions involving perchloric acid at elevated temperatures, but accomplished the same transformation.) Other amino acids, methylated proline, and pipecolic acid are either ineffective catalysts or give low enantioselectivity. As usual, these intramolecular reactions preceded the inter- molecular versions by quite a bit. In 2000, List and Barbas showed that intermolecular proline-catalyzed reactions can work (JACS 2000 122 2395): Me OH O Et OH O O O 71%, 99% ee Me O O O 3 mol% L-proline 52%, 74% ee O O NO 2 30 mol% L-proline + NO 2 O OH 68%, 76% ee The mechanism of these reactions is proposed to be analogous to how type I aldolases work (Bachrach, Section 5.3): Lys NH 2 + O Lys N -H 2 O Lys NH + R H O Lys N H O R +H 2 O Lys NH 2 + O OH R Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis Kinetics of Intermolecular Proline Aldol Reactions "Kinetic and Mechanisitic Studies of Proline-Mediated Direct Aldol Reactions." Blackmond, D.G. et al. Bioorg. & Med. Chem. Lett. 2009, 19, 3934-2937. "The Elusive Enamine Intermediate in Proline-Catalyzed Aldol Reactions: NMR Detection..." Gschwind, R.M. et al. ACIE 2010, 49, 4997-5003. Despite some recent controversy (see below), there is now little doubt that enamines are the nucleophilic species in these reactions and that C-C bond formation is rate-limiting. Gschwind and co-workers have looked at the self-aldolization of propionaldehyde under synthetically relevant conditions (20 mol% L-proline in d 6 -DMSO at 300 K) with real-time NMR: Me O H + Me O H OH O H + O H In agreement with previous studies, two diastereomeric aldol addition products are formed, along with some condensation product. Over the course of the reaction: When the catalyst loading is raised to 100%, the total amount of the intermediates rises from 8% to 25-30%, allowing for accurate quantitation. As others have noted before, oxazolidinones are detectable. But now, the enamine is detectable: (1) The enamine is E-configured and s-trans, regardless of [enamine]. Incidentally, this is the same conformation that Houk predicts is reactive (see discussion below). N Me CO 2 H E s-trans (2) NMR exchange spectroscopy ("EXSY") is a technique that allows the rate of interconversion between equilibrating species to be measured, so long as the exchange rate is suitable (for a more precise definition, you will have to wait until Chem 117 next semester). If two peaks have a "crosspeak" then there is exchange. The volume of the crosspeak is related to the rate of exchange. Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis Kinetics of Intermolecular Proline Aldol Reactions Here is a schematic for the exchange equilibria: Regardless of how enamine is formed, this is the proposed kinetic scheme for intermolecular proline aldol reactions (to maintain consistency with previous lectures, we're using a new set of letters to denote the various chemical species now): O Me H N Me CO 2 H N H CO 2 H + N Me CO 2 A I E N Me O O N Me O O Oa Ob (3) The iminium ion is spectroscopically invisible. Ob is more stable than Oa. (4) The dotted arrows represent crosspeaks. Note that exchange is not found between A and E. (5) The authors suggest that means that enamine only comes from oxazolidinones Oa and Ob, perhaps through an E2 mechanism. To bolster their claim, they show that: rate (Oa to A) rate (Oa to E) rate (Ob to A) rate (Ob to E) is different than the argument being that the partitioning to A and E should not depend on the starting point if a common intermediate I exists. , ? C CK K (CK)' A CKA P K 1 K 2 k 3 K I Q: Does this fit the observed kinetic data? To answer this question, we need to draw a 1+rate law for this. This is more complicated than the 1+rate laws we have considered. For example, step 1 produces water as well as CK and it reasonable to think that additional water would shift this equilibrium left towards starting materials. Assuming product release is irreversible, Assuming that we can ignore any catalysis (perhaps by water) of oxazolidinone formation, one can see that K 1 and K I will be merged into some smaller apparent equilibrium constant K 1 '. 1 2 3 2 T 1 1 I 1 2 2 2 2 2 [H O][K][A][C] [K] [K] [K] [A] [H O] 1 [H O] [H O] [H O] K K k v K K K K K = | | + + + | \ . N H CO 2 H N CO 2 H N O O H O R Me Me O H 2 O N Me CO 2 OH R H 2 O O Me OH R Me Me Me Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Kinetics of Intermolecular Proline Aldol Reactions 1 2 3 2 T ' ' 2 1 1 2 [H O][K][A][C] [H O] [K] [K] [A] K K k v K K K = + + if addition=rds, rate is approximately propotional to Similarly, [H 2 O] no longer appears in the numerator, since water will not speed up the reaction if aldol addition is rate-limiting. Additionally, the K 2 term is small, so it does not appear in the denominator. (1) does correspond to the power law above. To see this for [H 2 O], hold other concentrations constant (x=[H 2 O]): 0.7 0.6 0.9 2 T [H O] [K] [A] [C] v k
= Note that for a multi-step reaction, the exponents do not imply the molecularity of any particular step. Suppose the enamine formation is rate-limiting. In that case, the reaction becomes elementary and the rate law is (the pre-equilibrium assumption for step 1 is no longer valid since it's now the "last" step): T ' 2 1 [K][A][C] [H O] [K] K + T ' 2 1 [K][A][C] 1 [H O] [K] K x c o + + This is a bit less than negative first order in water, depending on how big c is. Finally, let y =[K], and we get: (1) T ' 2 1 [K][A][C] [H O] [K] y K d y o + + before the rate-determining step, slowing down the reaction. Increasing the amount of ketone increases the amount of this intermediate. However, because both water and ketone are incorporated before CK goes to product in a multi-step process, their kinetic orders have magnitudes less than one. Further Evidence Until a few years ago, the idea that these reactions go through enamine intermediates was itself controversial. Even the stoichiometry of the transition state was unclear. Here were some of the leading proposals for the Hajos-Parrish reaction: HO Me OH N H CO 2 Houk model Hajos model N O Absorbing K 1 and K I gives this expression: Experimentally, it has been determined that the rate can be fitted to an approximate power law expression: C CK / (CK)' CKA if enamine formation=rds, rate law is Since the reaction is about first order in aldehyde, this can be discounted. (Neither water nor aldehyde take the reaction forwards to product, so they don't appear in the numerator.) What if aldol addition is rate-limiting? O O H N O O H H O O H crystal surface O Swaminathan model N Me N CO 2 O H H R Agami model The Hajos and Houk models have the same stoichiometry, so we cannot distinguish between them with kinetics. These reactions work in compeletely homogeneous media, so the Swaminathan model is unlikely. Q: How can the number of prolines in the TS be determined? O O 1 T [K][C] k ee (Product) ee (Product) Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Further Evidence The classic experiment is to examine the relationship between catalyst enantiopurity and the product enantiopurity. This is called a nonlinear effects experiment. Despite some initial reports by Agami that these were involved in the Hajos-Parrish reaction, very careful studies showed that there is no such effect (Houk/List JACS 2003 125 16): If the homochiral catalyst-catalyst complexes are less reactive than heterochiral (meso) complexes, then the line would be curved upwards (asymmetric amplification). Similarly, ee is unchanged with dilution (more dilute =aggregation less likely): From microscopic reversibility arguments, the fact that the retro- aldol reaction is first-order in proline means that the forward aldol reaction is also first-order. This also has the advantage of definitively being related to the C-C bond forming step: Finally, performing the reaction in 18 O-labeled water gives incorporation of the label at the ketone. This is required by the mechanism, which generates an iminium ion which must be hydrolyzed by solvent (List PNAS 2004 101 5839): Me O Me O O 25 mol% (S)-proline 3 vol% H 2 18 O DMSO under Ar four days Me OH N O 2 C Me OH 18 O This excludes the Hajos model, which does not proceed via an iminium ion. (Initial reports were to the contrary, but careful experiments give incorporation.) 40% 50% 10% + + Me 18 O Me N HO 2 C HO Me OH N H CO 2 O O O O Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis The Houk-List Model for Stereoselectivity All of this evidence points to a one-proline enamine mechanism. Houk considered a variety of intermolecular reactions, both selective and unselective, and List measured their enantio- selectivities (JACS 2003 125 2475). One considered reaction: Me Me O O Me H + proline Me O Me OH A variety of parameters were looked at: N HO 2 C N CO 2 H syn or anti enamine R O H Re or Si face of aldehyde enamine-aldehyde rotamers As it turns out, unless the carboxylic acid and the aldehyde are engaged in hydrogen bonding, the incipient alkoxide is not effectively stabilized, and therefore the TS is very high in energy. The computations (B3LYP/6-31G*) led to this "metal-free" Zimmerman-Traxler model: O H N Me O O H Me O H N H O O Me Me favored disfavored (+1.0 kcal) Thus, anti-enamines are preferred, as is an equatorial aldehyde. Here are two equivalent renderings of the favored TS: This is the lowest energy TS leading to the minor product: Selectivity is predicted to be lower with cyclohexanone, since this makes both the equatorial and axial aldehyde conformers equally bad: O H N Me O O H 0.0 kcal/mol +0.5 kcal/mol O H N H O O Me attraction between negatively charged oxygen and positively charged N-C-H hydrogen atom 0 1 2 3 4 0 2 4 6 8 p r e d i c t e d
s e l e c t i v i t y
( k c a l / m o l ) observed selectivity (kcal/mol) y =1.7 x R =0.93 Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis The Houk-List Model for Stereoselectivity Of course, the computations are useless if they don't agree with reality. In general, the correlation is good, although the predicted selectivities are a bit too high: The Hajos-Parrish reaction seems to work along similar lines (Houk ACIE 2004 43 5766). The favored transition state is: Houk also examined the uncatalyzed and Hajos pathways: O Me O O H uncatalyzed TS (+10.2 kcal) Me OH N H CO 2 OH carbinolamine intermediate (+12.4) Since the carbinolamine leading up to the Hajos TS is already higher in energy than the uncatalyzed TS, Houk concludes the Hajos model is incorrect. If you don't believe any of this, the results for the closely related proline-catalyzed Mannich reactions are very compelling. Let's apply the same model to this reaction (Houk OL 2003 5 1249): Me Me O O H + S-proline Me O iPr NHPMP OMe NH 2 + Me Me E imines are more stable than Z imines, so this gives this TS: N H N H O O R Me PMP Indeed, this gives the correct facial selectivity and explains why the sense of induction is turned over from the aldol reaction. Note that for donor aldehydes, this favors syn-Mannich products: R N H N H O O R PMP R H O CO 2 Et NHPMP R So how can one obtain anti-Mannich products? N H CO 2 Et PMP H O R + 90%, 93% ee Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis Mannich Reactions A different catalyst turns out to be highly anti-selective: H O iPr NHPMP 70%, 94:6 anti:syn >99% ee N H CO 2 Et PMP H O Me + Me N H COOH Me This catalyst was designed by computations, which predicted this transition state (Houk/Barbas JACS 2006 128 1040): N R N PMP EtO 2 C Me H O O Deletion of the methyl group reduces selectivity by about 1 kcal. Oxazolidinone Intermediates? Despite the success of the List-Houk mechanistic framework, Seebach and Eschemenmoser have proposed an alternate pathway that involves oxazolidinones (Helv Chim Acta 2007 90 425). Here is the List-Houk mechanism: N H CO 2 H N CO 2 H N O O H O R Me Me O H 2 O N Me CO 2 OH R H 2 O O Me OH R In that manifold, oxazolidinones are thought to be unproductive and "parasitic." In the Seebach-Eschenmoser scheme, the oxazolidinone can open to an enamine carboxylate, which is proposed to be the nucleophilic species: N H CO 2 H N CO 2 N O O H O R O H 2 O H 2 O O OH R H B: BH N O O HO R Me Me Me Me B (1) The enamine carboxylate might arise from an E2 elimination as depicted, or perhaps by deprotonation of the iminium ion formed from proline and acetone. (2) The carboxylate is proposed to assist the nucleophilicity of the enamine (similar to halolactonization): N H Me O O E + (3) From various X-ray structures and calculations, it's been suggested that the concave/convex shape of these structures might give rise to high stereoselectivity. N H N H O O R Me PMP R Chem 106 E. Kwan Lecture 31: Enamine and Iminium Organocatalysis Oxazolidinone Intermediates? (4) It's been suggested that the 9-membered hydrogen bonds required by the List-Houk model are implausible. However, the usual proscription against such rings is based on the transannular interactions in medium-sized cycloalkanes. Gellmann has shown that these H-bonds are, in fact, quite plausible (JACS 1990, 112, 8630; CSD: PAGTIA): N O O O H (5) However, in DMSO, a lot of these hydrogen bonds do get disrupted (DMSO is a very good H-bond acceptor). However, it seems plausible that in a transition state, H-bonding to an internal donor might be feasible due to entropic effects. Noncovalent interactions are also generally stronger in transition states, since they're more polarized. (6) The Seebach-Eschenmoser proposal does not take advantage of intramolecular general acid catalysis, and so an alkoxide is formed. Although DMSO is very polar, it is not a hydrogen bond donor, so the alkoxide is expected to be quite unstable. From the ground state side, the enamine carboxylate is not expected to be very well stabilized either. Sunoj has looked at the relative energetics of both mechanisms computationally using a variety of methods (ACIE 2010 49 6373). List-Houk (correct enantio- and diastereoselectivity) Seebach-Eschenmoser (correct ee, incorrect dr) predicted ee B3LYP/ 6-31+G** MP2/ 6-31+G** M05-2X/ 6-31+G** predicted dr experimental ee: 99%; dr: 4:1 anti:syn >99 >99 98.5 5.4:1 1.2:1 1:1.7 predicted ee predicted dr 95 >99 >99 1:3.9 1:4.6 1:2 So both models get the enantioface right, but the predictions of diastereoselectivity are better with the List-Houk enamine model. More compelling are the substantially higher barriers for the Seebach oxazolidinone pathway (by about 11 kcal/mol): Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Enamine vs. Iminium Activation Modes O H N Me O O H Me Q: Why is proline a catalyst for aldol reactions? i.e., Why does it lower the activation barrier? (1) general acid catalysis: COOH stabilizes forming alkoxide (2) enamine activation - most isolated carbonyl groups in ketones and aldehydes have very low enol content; the catalyst increases the amount of "enol" available - actually, it's an enamine, not an enol; on the Mayr scale, enamines are much more reactive N O OEt OEt OEt OEt - enols are not available on the scale, but this shows that you need approximately two oxygens to equal one nitrogen! OTMS 5.41 (s=0.91) 9.81 (s=0.81) 12.06 (0.80) 3.92 (s=0.90) 4.23 (s=1.00) If enamines are more nucleophilic, it's reasonable to expect that iminiums ions are more electrophilic. This is similar to how Lewis acids or hydrogen bond donors activate carbonyl groups: O activation O N N S H R R H O ML n N R R increasing covalency These are all LUMO-lowering strategies. While organocatalytic activation may not be as powerful in increasing reactivity as Lewis acid catalysis, it can be an advantage in getting catalyst turnover. Iminium ions are so easily hydrolyzed by water, they are generally considered to be in equilibrium with their carbonyl counterparts. MacMillan has taken advantage of this for Diels- Alder reactions (JACS 2000 122 4243): OAc + O H 20 mol% cat MeOH/H 2 O rt OAc CHO 11:1 endo:exo 72%, 85% ee N N H O Me Bn Me Me cat This also works with o,|-unsaturated ketones (JACS 2002 124 2458), but requires a different catalyst: Me + O 20 mol% cat 20 mol% HClO 4 EtOH, 30 C Me COEt 11:1 endo:exo 72%, 85% ee N N H O Me Bn cat Me Me Note that, in contrast to proline aldol reactions, these occur in protic media, where there may be some hydrophobic acceleration. Exactly how selectivity is obtained here is still up for debate (Houk Acc Chem Res 2004 37 558). Instead, I will discuss what's known about models for organocatalytic transfer hydrogenation (reviews: MacMillan Acc Chem Res 2007 40 1327; Adolfsson ACIE 2005 44 3340). Me O Me N H CO 2 Et EtO 2 C Me Me Hantzsch ester These reactions were developed as a synthetic analog to natural hydride reductions which occur with NADH or FADH 2 . |,|-Disubstituted aldehydes can be reduced effectively (MacMillan JACS 2005 127 32, List ACIE 2005 44 108) with Hantzsch esters as the hydride source: H H + NR 2 * R Ar NR 2 * R Ar O H 20 mol% cat Hantzsch ester PhMe, 30 C 96% conv., 75% ee N N H O Me tBu cat Me Ph TFA O H Me Ph Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Organocatalytic Hydride Reductions The convergence of E and Z starting materials to the same products suggests a Curtin-Hammett scenario where a dienamine intermediate can rotate to isomerize the olefins: Bn An anomalous case (see below) is this tert-butyl catalyst: O H 20 mol% cat Hantzsch ester CHCl 3 , 30 C 91%, 93% ee N N H O Me tBu cat Me Ph TFA O H Me Ph Under slightly different conditions, the reaction improves: O H 20 mol% cat Hantzsch ester dioxane, 13 C 77%, 90% ee N N H O Me tBu cat Me Ph TCA O H Me Ph Bn advantages: bench-stable reagent; a good alternative to hydrogenation (usually needs optimization) or copper-based hydrides (unreliable); E and Z isomers converge to the same product disadvantages: the Hantzsch pyridine is a stoichiometric byproduct, which must be removed by chromatography (acid- base extraction doesn't work); the Hantzsch pyridine is quite a large reagent for delivering just one hydrogen An initial result: N CO 2 Et EtO 2 C Me Me byproduct Hantzsch pyridine N N O Me H Ar Me N N O Me H Ar N N O Me H Ar Me fast slow O H Me Ar O H Me Ar Stereochemical Model N N Ph Me O Me H 3 C CH 3 CH 3 (1) In the left-hand rotamer, the bulky tert-butyl group ensures that the benzyl group rotates over the |-face of the electrophile, making the hydride reagent come from the bottom. (2) Both the attractive tt/cation-t interactions and repulsive steric interactions (as the carbons pyramidalize) are expected to become more important in the transition state. What controls the stereoselectivity? Having established the favored iminium ion geometry, we must now consider two iminium ion rotamers: N N Me Ph O Me H 3 C CH 3 CH 3 better than tt or cation-t interaction? preferred attack N N H 2 O Me Me Me N N O Me Me Me Ph H The ground state conformational preferences of these systems have been studied (Tomkinson et al. OL 2009 11 133). The X-ray structures shown below are similar to the solution phase structures, as shown by NMR studies. Here is an X-ray structure of a free imidazolidinone catalyst (CSD: POPRES): Bn Bn Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Organocatalytic Hydride Reductions Calculations show the penalty for rotating the benzyl group in the ground state of the iminium ion is about 1 kcal/mol--very small. I doubt these ground state pictures tell us much about the transition state; the fact that the benzyl group doesn't seem to be covering the |-face very effectively in the X-ray structure is meaningless. The anomaly is that the catalyst which only has a tert-butyl group is still very selective: This is the X-ray structure of the corresponding iminium ion (CSD: DOSKIG01). Note the rotation in the benzyl group: N N Ph Me O Me H 3 C CH 3 CH 3 still preferred (?) I don't have a good explanation for this. Enones also work in this chemistry, but with a slightly different system (MacMillan JACS 2006 128 12662; Houk OL 2009 11 4298): O Bu 20 mol% cat Et 2 O, 0 C N H CO 2 tBu tBuO 2 C Me Me H H + O Bu 82%, 90% ee The product can be explained by this model, but it might not be right (see the Houk paper): N N Bu O Me preferred attack O Me Q: Why can't we make taxol in ten steps? Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Organocatalytic Cascades "Strategies to Bypass the Taxol Problem..." Walji, A.M.; MacMillan, D.W.C. Synlett 2007, 10, 1477-1489. Taxol is a potent anticancer agent used in the treatment of lung, ovarian, breast, neck, and other cancers, with sales of $1.6 billion US in 2000. Until 1993, taxol had to be extracted from the bark of the pacific yew tree and prepared via semi-synthesis. Now, BMS uses cultured Taxus cells in a fermentation process. Taxol is extracted and purified directly from the broth. Analysts estimate that the world will need 1 040 kg of taxol/year by 2012. Despite our best efforts, total synthesis is clearly not up to the job: Each of these groups used cutting-edge chemical methodology and very clever synthetic logic, but none of their routes is even remotely amenable to a large-scale process route. This problem is certainly not unique to taxol. Here's what I think the biggest problem is: the complexity of molecules scales exponentially with size, but the complexity introduced by a synthetic sequence scales even less than linearly with its length. MacMillan points out that the yield of a synthetic sequence also drops exponentially with its length. A somewhat unsatisfactory, literal explanation is that the total yield is just the product of the yields of every step. But why don't reactions give 100% yield? (1) The reactions are intrinsically "dirty" and give side products which account for the remainder of the mass balance. (2) Purification is "lossy." J ust how lossy is purification? Hudlicky has recently done some simple experiments (Synlett 2010 18 2701) which show that, in general, one cannot expect more than a ca. 94% yield for any given reaction which involves workup and chromatography. AcO Me HO OBz O OH O OAc O O OH BzHN H mass recovery filtration aqueous extraction fraction collection separation 1001.7 mg 978.6 mg 985.0 mg 980.1 mg This is bad news if you want to use a sequence of 37 steps. The top synthesis, Wender's, already has an average yield of 85%, suggesting that there is nothing wrong with the reactions themselves; they're simply not generating enough complexity per step. O O OH Nicolaou (1994) 51 steps Me Wender (1997) 37 steps 0.4% overall yield H O Me Kuwajima (1998) 47 steps NH 2 O HO OH Mukiyama (1999) 38 steps O O Me 47 steps Danishefsky (1995) Me Me Me Me O Holton (1994) 41 steps (1000 mg scale) Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Organocatalytic Cascades MacMillan proposes that we circumvent the defects of "stop-and- go" synthesis by combining several steps into one step. For example, consider this sequence (JACS 2005 127 15053): This represents the merging of two catalytic cycles. First, the catalyst forms an iminium ion with the aldehyde (A), which undergoes a Friedel-Crafts alkylation. Then, the product tautomerizes into a nucleophilic enamine (B), which is chlorinated. (1) This reduces the number of purifications required, which is good. However, simply performing reactions in a cascade, rather than stop-and-go sense does not by itself improve the intrinsic yield of a reaction. (2) Cascade reactions are not new. In fact, we've already seen quite a few such reactions in the course already. However, most of them are diastereoselective in nature. For example, this cascade reaction was discussed in Lecture 10 on page 14: Me Ph Me Ph Me HO oxy-Cope/ carbonyl-ene O Me H O Me O Cl Cl Cl Cl Cl Cl O Me O H Me Cl EtOAc, 50 C N N Me O N Bn tBu 20 mol% catalyst Me N N Me O N Bn tBu Me O Me A B Cl O Me + Without catalyst, one is at the mercy of the complex energy landscape that is intrinsic to the reaction; there is no guarantee that the desired reaction lies on the minimum energy path. Nor is it guaranteed that there is only one minimum energy path, such that the reaction will be "clean" and have no other side products. So using a tunable catalyst is a real advantage in that sense. (3) One complaint about this chemistry is that it makes strings of stereocenters efficiently, but is relatively ineffective at building structural complexity. Is this true? You decide. This area has been reviewed recently: "Organocatalytic Cascade Reactions..." Enders, D. et al. Nature Chemistry 2010, 2, 167-178. Here is another example from the Hayashi group for the synthesis of oseltamivir (ACIE 2009 48 1304): 86% yield 14:1 dr, 99% ee RO O H tBuO 2 C NO 2 + N H Ph Ph OTMS Michael addition Cs 2 CO 3 P CO 2 Et O EtO EtO conjugate addition/ HWE olefination NO 2 tBuO 2 C RO CO 2 Et 5 mol% 5:1 dr undesired diastereomer The first step uses a silylated Hayashi-J orgenson diarylprolinol catalysts. Unfortunately, the second step delivers the incorrect stereocenter at the nitro group. (All of this occurs in one pot.) O H NO 2 tBuO 2 C RO Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Organocatalytic Cascades NO 2 tBuO 2 C RO CO 2 Et O Me H O Me OTMS Me O O O Me H O OH Me H Me O H O Me O H O Me O O O H O Me O N HO 2 C Me type I/type II cross-metathesis (Grubbs II) Me O N N O Me tBu Ph + H selective for iminium catalysis three sequential one-pot operations Friedel-Crafts alkylation (imidizolidinone) ketone aldol (proline) selective for enamine catalysis 64%, 95% ee, 5:1 dr Fortunately, a hetero-conjugate addition/epimerization strategy, followed by some other manipulations delivers Tamiflu in 57% overall yield from the nitroalkene, which is pretty good. Overall, there are nine reactions, three separate one-pot operations, and one column chromatography purification. TolSH 70% NO 2 tBuO 2 C RO CO 2 Et STol = tBuO 2 C O 2 N RO CO 2 Et STol 1. TFA 2. (COCl) 2 , cat.DMF 3. NaN 3 NO 2 RO CO 2 Et STol O N 3 1. AcOH/Ac 2 O 2. Zn/TMSCl 3. NH 3 ; K 2 CO 3 one-pot one-pot no purification purified by flash chromatography NH 2 AcHN RO CO 2 Et oseltamivir (82% from A) A (1) The second one-pot procedure sequentially cleaves the tert-butyl ester, converts the liberated carboxylic acid into the acyl chloride, and displaces the chloride with azide. (2) The third one-pot procedure first converts the acyl azide into the amine via a Curtius rearrangement. The advantage of these conditions is that the reaction occurs at ambient temperatures. The amine is trapped by acetic anhydride. Zinc then reduces the nitro group. Finally, ammonia is added to chelate the zinc, and potassium carbonate initiates a retro-conjugate addition to generate the olefin. Obviously, this is a very nice synthesis, but the fact that the conjugate addition is not catalyst controlled means that its dr is at the mercy of the various stereocenters present: O H NO 2 tBuO 2 C RO P CO 2 Et O EtO EtO + NO 2 tBuO 2 C RO CO 2 Et 5:1 dr undesired diastereomer A more sophisticated "cycle-specific" strategy is to use more than one catalyst at the same time (MacMillan ACIE 2009 48 4349). This requires the catalysts be orthogonal: Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Stereoselectivity in Cascades A common feature in these cascade reactions is that they give very high overall enantioselectivities, even if the component reactions are not very enantioselective. How is this possible? Consider a two-step cascade in which the first step produces an x : 1-x enantiomeric ratio of products, where x is the mole fraction of R configuration produced at stereocenter 1. Similarly, step 2 has a selectivity of y: starting material step 1 selectivity x x R 1 (1-x) S 1 step 2 selectivity y xy R 1 -R 2 x(1-y) R 1 -S 2 (1-x)y S 1 -R 2 (1-x)(1-y) S 1 -S 2 If this seems a bit abstract, maybe revisiting this example will help: O Me H O Me O Cl Cl Cl Cl Cl Cl O Me O H Me Cl EtOAc, 50 C 20 mol% catalyst + 86% yield 14:1 dr, 99% ee stereocenter 1 stereocenter 2 The Fridel-Crafts alkylation is step 1, and has some intrinsic enantioselectivity x. The chlorination is step 2, and has another intrinsic selectivity y. (1) Without a loss in generality, suppose that the R 1 -R 2 product is the desired one. How much of it is formed? A mole fraction of xy. (2) Where does the enantiomer of the desired product, S 1 -S 2 , come from? It comes from an incorrect first step, followed by an incorrect second step. Hence, a mole fraction of (1-x)(1-y) is formed. Since 1-x and 1-y are small fractions (i.e., less than 1), multiplying them together gives an even smaller number. (3) The enantiomeric ratio of the product is therefore xy divided by (1-x)(1-y). Division by a very small number between 0 and 1 increases the size of xy by a lot. This is why the enantiomeric ratio of the product is very large. (4) The enantiomeric ratio of the product can be big, even if the enantioselectivity of the component steps is not that great. For example, suppose steps 1 and 2 have ee's of 80%. That would be considered good, but great. In our language, this would mean x = y = 0.9. This gives an enantiomeric ratio of 81:1, or an ee of 97.6% (which, for one step, would be considered excellent). Thus, it is relatively easy to get almost enantiopure product: 0.5 0.6 0.7 0.8 0.9 1 2 4 8 16 32 64 128 256 e n a n t i o m e r i c
r a t i o
o f
p r o d u c t selectivity for second step (y) x=0.9 x=0.8 x=0.7 x=0.6 Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis SOMO Activation So far, we've seen a variety of activation modes: O R H R NR 2 enamine: more nucleophilic O R H R N iminium: more electrophilic O H-A O ML n O R H O R H R M = H (acid catalysis) M = metal (Lewis acid catalysis) more electrophilic hydrogen bonding more electrophilic O R H R OH R N RN NHC carbonyl anion more nucleophilic O R OR R O N or R O R N RN NHC carbonyl pseudohalide more electrophilic nucleophilic catalysis more electrophilic or Nuc: O R OR H O R B: base catalysis Q: What's in between an enamine and an iminium? "Enantioselective Catalysis Using SOMO Activation." MacMillan et al. Science 2007, 316, 582-585. MacMillan's clever idea is that enamines are very electron rich, and therefore more easily oxidized than amines or iminiums: H O TMS N H N O Me Ph tBu 20 mol% CAN (2 equiv.) NaHCO 3 , DME 24h, 20 C H O 75% yield, 94% ee N R R O H + N H -H 2 O H N R H +2 e - 9.8 eV 8.8 eV 7.2 eV N R H -1 e - N R H N R H (The ionization potentials are shown in bold; the lower the number, the easier it is to oxidize.) Oxidation of the enamine produces a radical cation, which is an ambident electrophile: enamine radical cation more reactive In the jargon of organocatalysis, this is "SOMO activation"-- the generation of a singly occupied molecular orbital which can undergo many subsequent reactions. One of the first examples was an aldehyde allylation, which is very useful synthetically: + CAN is ceric ammonium nitrate (NH 4 )Ce(NO 3 ) 6 , which is a strong oxidant (even stronger than Cl 2 !) which reacts to generate a chiral enamine radical cation. H R R H O N H N O Me Ph tBu 20 mol% CAN (2 equiv.), H 2 O DTBP, acetone, 24h, 20 C Ph OTMS 74% yield, 93% ee H O Ph O Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis SOMO Activation stereochemical model N N O Me H 3 C CH 3 CH 3 R o + (Historical Note: This use of stoichiometric CAN for similar reactions was previously reported by Narasaka (Chem Lett 1992 2099.) Aldehydes can be coupled with enolate equivalents, which produces umpolung 1,4-dioxygenated relationships (MacMillan JACS 2007 129 7004): (DTBP =2,6-di-tert-butylpyridine) nucleophiles approach from below cationt interaction between benzyl group and allyl radical cation? + If the nucleophile is allyltrimethylsilane, then this produces: N R N R TMS The regioselectivity can be explained by a radical version of the |-silicon effect; donation of the Si-C bonds into the radical is good. (However, the |-silicon effect for radicals is much less than for carbocations.) The radical gets oxidized to the carbocation (which is why two equivalents of CAN are required), and then base comes in and does a Peterson olefination. Si CH 3 CH 3 CH 3 |-silyl radical stabilized by o SiC to n* C donation N R Si CH 3 CH 3 CH 3 -1 e - :B CAN N R H 2 O O R H H H H H (catalyst substituents omitted) CN Me Me CN H H H H O CN Me Me CN O N N O Me Nap tBu Cu(OTf) 2 NaTFA/TFA iPrCN/DME 23 C CN CN Me Me [Ox] - H 2 O - 1e - N N O Me Nap tBu CN CN Me Me [Ox] - catalyst - H + , - 1e - 56%, 92% ee A dramatic demonstration of the power of this activation mode to produce complex architectures as well as stereodefined arrays is this cascade polyene cyclization (MacMillan JACS 2010 132 5027): CAN is ineffective here due to an inner-sphere nitrate transfer. It is interesting that both the yield and ee go up when the aryl group on the catalyst changes from phenyl to naphthyl. To me, this strongly suggests an enhanced cation-t interaction with a larger aromatic surface. Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Cascade Reactions In the previous examples, the radical produces by the coupling reaction gets oxidized to a carbocation, which is then eliminated. What happens if we intercept the carbocation with an internal nucleophile first? Good stuff (Macmillan JACS 2010 132 10015): Ph OMe + O H N H N O Me Ph tBu 20 mol% catTFA Fe(phen) 3 (SbF 6 ) 3 Na 2 HPO 4 , THF 10 C, 12 h OMe O H Ph First, the aldehyde is alkylated by styrene to give a stable benzylic radical. Then, oxidative radical-polar crossover gives a carbocation which is engaged in Friedel-Crafts alkylation: OMe O H Ph -1 e - OMe O H Ph alkylation, then elimination 76%, >20:1 dr 94% ee (1) The diastereoselectivity is rationalized based on a chair TS. (2) The yield includes 7% ortho-coupled product. (3) Because the selectivity of the second step is largely based on the conformational preferences of the substrate, the ee mainly reflects the enantioselectivity of the first step, while the dr mainly reflects the diastereoselectivity of the second step. (4) Thus, the inherent enrichment in ee that is commonplace in largely catalyst-controlled reactions is bypassed here. Photoredox Catalysis The use of stoichiometric oxidants is undesirable in terms of atom economy, the ability of the substrate to tolerate oxidizing conditions, purification, etc. The latest advance is to use a photosensitized metal-ligand complex as the redox shuttle. This is particularly convenient, as visible light can be used as the energy source (MacMillan Science 2008 322 77): H O Br CO 2 Et CO 2 Et 83% yield, 95% ee H O CO 2 Et CO 2 Et visible light (fluorescent bulb) organocatalyst (20 mol%) Ru(bpy) 3 Cl 2 (0.5 mol%) 2,6-lutidine, DMF, 23 C + N N O Me Me tBu CO 2 Et CO 2 Et . N N O Me Me tBu EtO 2 C EtO 2 C N N O Me Me tBu EtO 2 C EtO 2 C * Ru(bpy) 3 2+ Br CO 2 Et CO 2 Et Ru(bpy) 3 + Ru(bpy) 3 2+ light SET SET [O] -1 e - The blue cycle is started by sacrificing a few molecules of enamine. Thereafter, it is driven by oxidation of the product radical (A) to an iminium ion and reduction of the bromo malonate (B) to its corresponding malonate radical. Thus, every equivalent of product formed requires one turn of the blue cycle. B A Q: Where is chemistry going? Chem 106 E. Kwan Lecture 34: Enamine and Iminium Organocatalysis Looking Ahead chemical biology - find small molecule modulators for every cellular function (Schreiber) - sequence the genome of a single cell (Xie) - find ways to visualize cellular functions; understand how DNA controls cell funtions, cancer (Xie) physical chemistry - find an accurate, but cost-effective electronic structure method (Aspuru-Guzik) - predicting the boiling point of water (Kahne) - protein folding (Cohen) - determining the structure of a compound from a single molecule (protein, DNA, etc.) (Cohen) inorganic chemistry catalysis: - figure out what to do with carbon dioxide (Whitesides, Friend) - conversion of carbon dioxide to water and CO (for syngas) (Betley) - new catalysts for commodity chemicals and fuels (Friend) - metal-organic frameworks for catalysis; self-assembled monoliths that are catalysts but don't need to support life (Betley) - catalysts for reducing nitrogen to ammonia - recyclable plastics from renewable sources (Cohen) organic chemistry - de novo design of enzymes for new reactions (Xie) - rational drug design (Whitesides) - turning cellulose into food for humans (Cohen) - "green synthesis" (J acobsen) - cheaper and better catalysis with non-toxic first-row transition metals; catalytic oxidations with O 2 (J acobsen) - predicting non-covalent interactions and transition states (J acobsen) energy: - materials for energy transfer, storage (Aspuru-Guzik, Friend) - catalytic water oxidation for energy storage (Betley, Gordon) - cheap semiconductors (Gordon) - room temperatures superconductors (Cohen) Where is organic chemistry going in 30 years? (speculation) (1) Organic chemistry will not be "dead," and making small molecule modulators will only be but one area of research. (2) Fundamentals: We still won't be able to predict what happens when two molecules get mixed together, but we'll get much better at it. Strides in ultrafast spectroscopy will allows us to probe TSs directly. Modeling of solvent effects, ion-pairing, radicals, and dynamic effects, and electron correlation phenomena will attain a high level of accuracy. (2) New reactivity: Organometallic chemistry will give us much better redox methods, but recent progress in N-heterocyclic carbenes and SOMO activation suggests that new main group reactivity patterns will be discovered. (3) Methodology: Rational computer-aided catalyst design will become common. Each catalyst will have a binding pocket that is specifically effective for binding a particular transition state. Thus, catalyst efficiency will improve, but substrate scope will diminish. But that won't be seen as a problem. Catalysts today are mostly good at functionalizations, but future catalysts will be useful for building architecture, too. (4) Synthesis: People will design sophisticated mixtures of catalysts to do cycle-specific cascade sequences. Thus, it will be much more common for drugs to have complicated arrays of stereocenters or architectures. Here is some of what CCB faculty have to say (Nov. 2010):