Attractive noncovalent interactions in asymmetriccatalysis: Links between enzymes and smallmolecule catalystsRobert R. Knowles and Eric N. Jacobsen1

Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138

Edited by David W. C. MacMillan, Princeton University, Princeton, NJ, and accepted by the Editorial Board September 5, 2010 (received for review May 8, 2010)

Catalysis by neutral, organic, small molecules capable of binding and activating substrates solely via noncovalent interactions—particularlyH-bonding—has emerged as an important approach in organocatalysis. The mechanisms by which such small molecule catalysts inducehigh enantioselectivity may be quite different from those used by catalysts that rely on covalent interactions with substrates. Attractivenoncovalent interactions are weaker, less distance dependent, less directional, and more affected by entropy than covalent interactions.However, the conformational constraint required for high stereoinduction may be achieved, in principle, if multiple noncovalent attractiveinteractions are operating in concert. This perspective will outline some recent efforts to elucidate the cooperative mechanisms respon-sible for stereoinduction in highly enantioselective reactions promoted by noncovalent catalysts.

enantioselectivity | H bonding | organocatalysis | transition state stabilization

Attractive noncovalent interactionsplay a central role in pharma-ceutical design, supramolecularchemistry, molecular biology,

sensing applications, materials, crystal en-gineering, and a host of other fields inthe chemical sciences (1–8). As a result ofthis broad importance, intensive researchefforts have been directed toward eluci-dating and quantifying these interactions,which include hydrogen bonding, elec-trostatic effects, π–π, cation–π, hydropho-bic, and Van der Waals forces (9–14).These studies have focused primarily onmolecular recognition phenomena, shed-ding light on the thermodynamic stabili-zation of bound complexes. Attractivenoncovalent interactions also play a keyrole in catalysis by lowering the kineticbarriers to reactions through transitionstate stabilization (15, 16). In fact, theseforces are responsible for many of the re-markable rate accelerations and stereo-selectivities characteristic of enzymaticcatalysis, wherein cooperative noncovalentinteractions with specific active site resi-dues can stabilize the electrostatic char-acter of a bound transition structurecomplex (Fig. 1) (17).Although enzymes are understood to

achieve selectivity through transition statestabilization, with the rate of the dominantpathway preferentially accelerated relativeto competing reactions, small moleculechiral catalysts are generally proposed tofollow a fundamentally different principlefor achieving enantioselectivity. The vastmajority of stereochemical models forsmall molecule catalysts invoke stericinteractions as a rationalization for ener-getic differentiation of the pathways lead-ing to enantiomeric products (e.g., Fig. 2)(18–21). In this conceptual framework, itis generally understood that transition

state assemblies leading to the undesiredconfiguration of the product are destabi-lized by repulsive steric interactions withsubstituents on the catalyst. Consequently,only the transition states that avoid thesesteric interactions are kinetically viable.The question of whether selectivity isachieved primarily through stabilizing ordestabilizing interactions represents a fun-damental difference in the way macromo-lecular and small molecule catalysts arethought to operate.Steric interactions—as modeled by the

Lennard-Jones potential—are stronglydistance dependent, such that the desta-bilizing effect can be alleviated by a verysmall reorganization (Table 1) (see ref. 10,pp 109–121). As such, steric destabiliza-tion is most effective as a defining elementof stereocontrol in strongly bound andconformationally restricted complexes,where any structural reordering away fromequilibrium incurs a substantial energeticpenalty. This is generally the case in re-actions mediated by organometallic com-plexes, Lewis acids, secondary amines, andother important classes of small moleculecatalysts, in which transition structuresare associated to the catalyst through well-defined covalent interactions (Fig. 2).By contrast, noncovalent interactions

are not only generally weaker but also lessdirectional and less distance dependentthan covalent and dative bonds (Table 1)(10, p 28). As such, in a complex wherea catalyst interacts with its substratethrough a single noncovalent interaction,many geometrical orientations of the sub-strate relative to the catalyst may be verysimilar in energy. Thus, significant struc-tural reorganizations can occur withoutenergetically destabilizing the binding in-teraction to any large extent. Such a lackof conformational order presents an obvi-

ous challenge to achieving high enantio-selectivity. This may be overcome, inprinciple, through implementation ofmultiple noncovalent attractive inter-actions operating in concert, because thiscan afford the conformational constraintrequired for high stereoinduction (22, 23).This cooperative model of binding isa defining feature of both biological andsynthetic receptors, and also underliesenzymatic catalysis.Several well-known asymmetric catalytic

processes involving Lewis acids and tran-sition metals have invoked attractive non-covalent interactions as elements ofstereocontrol, including the Sharplessdihydroxylation, the Noyori transfer hy-drogenation, and Corey’s oxazaborolidine-catalyzed cycloadditions (24–26). How-ever, over the past decade, remarkablysimple organocatalytic systems have beenidentified that operate exclusively throughnoncovalent interactions and yet inducehigh levels of enantioselectivity in syn-thetically important transformations. Elu-cidation of the attractive forces acting inconcert in such systems presents a signifi-cant challenge to modern mechanisticchemistry but one that seems well justifiedgiven the design principles and generalinsights into noncovalent interactions thatcould emerge. Here we discuss four chiralhydrogen-bond donor catalyst systems de-veloped in our laboratories that have be-gun to provide a wealth of information inthat regard.

Author contributions: R.R.K and E.N.J. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. D.W.C.M. is a guesteditor invited by the Editorial Board.1To whom correspondence should be addressed. E-mail:jacobsen@chemistry.harvard.edu.

20678–20685 | PNAS | November 30, 2010 | vol. 107 | no. 48 www.pnas.org/cgi/doi/10.1073/pnas.1006402107

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

DiscussionClaisen Rearrangement. Rate enhancementof the Claisen rearrangement by hydrogen-bond donors is known to occur in enzy-matic systems (Fig. 1) and has also beendemonstrated with protic solvents andelectron-deficient diaryl ureas (Fig. 3) (27–29). The observed accelerations are mostpronounced with allyl vinyl ethers bearingelectron-withdrawing and/or electron-donating substituents, an effect attribut-able to the notion that such substratesundergo rearrangement through polartransition states with substantial “eno-late”/“allyl cation” character. Electrostaticstabilization of the transition state can beimparted through H-bond interactions,which disperse the augmented anioniccharacter of the partially broken C–Obond (30–32).

With this precedent in mind, we soughtto develop a chiral H-bond donor catalystsystem for enantioselective Claisen rear-rangements (33). Substoichiometricquantities of diols, phosphoric acids, orthioureas were not found to induce mea-surable rate accelerations, but cationicdiphenyl guanidinium ion displayed sig-nificant catalytic activity in rearrange-ments of a diverse range of electronicallyactivated allyl vinyl ethers. Extensive de-velopment and optimization studies led tothe identification of chiral C2-symmetricguanidinium ion (3) as an effectiveand general catalyst for highly enantio-

selective Claisen rearrangements of ester-substituted allyl vinyl ethers (Fig. 4) (33).Structural and computational investi-

gation into the origins of enantioselec-tivity indicates that several attractivenoncovalent interactions may be actingin concert to serve as the basis for ster-eoinduction in these transformations. Inthe lowest-energy conformation of thecatalyst, both pyrrole rings of 3 are en-gaged in cation–π interactions with thecationic NH2 of the guanidinium ion,splaying the pendant phenyl substituentsof the pyrrole into an orientation thatcreates a well-defined box-like space sur-rounding the H-bond donor functionality(Fig. 4). In this conformation, the catalystis found to bind to substrate 1 througha dual hydrogen bonding interaction toboth the ether and ester oxygens. Thistwo-point binding interaction serves toorder the geometry of the complexation,and the presence of the ester substituentalso increases the degree of chargeseparation in the rearrangementtransition state.As noted above, this increased polari-

zation introduces cationic character onthe allyl fragment of the transition state aswell. Computational modeling indicatesthat this cationic character is engaged andstabilized by the catalyst phenyl ringsthrough a cation–π interaction in only oneof the two competing diastereomerictransition states (Fig. 4). This intriguing

Fig. 1. Noncovalent interactions between a tran-sition state analog and active site residues ofchorismate mutase. These attractive interactionsstabilize the electrostatic character of the pericy-clic transition state converting chorismate to pre-phenate, resulting in rate accelerations of up to 106.

Fig. 2. Stereochemical models for synthetically important enantioselective transformations whereinstereoselectivity is rationalized through steric destabilization of minor pathways.

Table 1. Distance dependencies of noncovalent interactions

Noncovalent interaction Energy dependence on distance

Charge–charge 1/r

Charge–dipole 1/r2

Dipole–dipole 1/r3

Charge-induced dipole 1/r4

Dipole-induced dipole 1/r5

Dispersion 1/r6

H-bond Complicated ∼1/r2

Steric repulsion 1/r12

Dependencies for entries 2–5 are only valid at values of r several times greater than the lengths of theinteracting dipoles. At or near the Van der Waals distances operating in the catalytic reactions discussedin the text, these interactions become largely electrostatic, displaying a ∼1/r dependence.

Knowles and Jacobsen PNAS | November 30, 2010 | vol. 107 | no. 48 | 20679

SPECIA

LFEATURE:PERSP

ECTIV

ED

ownl

oade

d by

gue

st o

n A

ugus

t 23,

202

0

rationale for stereoinduction was testedexperimentally through the preparationand evaluation of aryl-substituted deri-vatives of 3. Dimethylamino-substitutedcatalyst 5 induced increased enantiose-lectivity relative to the parent catalyst,whereas the corresponding fluorophenylderivative 4 proved less enantioselective(Table 2). This lends direct support to thehypothesis that differential transition statestabilization responsible for enantiose-lectivity is achieved through intermole-cular cation–π interactions (34, 35).This chiral guanidium-catalyzed Claisen

rearrangement reaction highlights the po-tential of spatially resolved transition statecharge recognition to serve not only asa basis for rate acceleration but also, to theextent that the recognition stabilizes onediastereomeric transition state preferen-tially, as a design principle for stereo-induction in noncovalent catalysis.

Cationic Polycyclizations. Enantioselectivityin the catalytic Claisen rearrangementdescribed in the previous section relies onselective catalyst stabilization of transitionstates bearing a relatively small degreeof charge separation. In principle, co-operative noncovalent interactions may beaccentuated in reactions involving fullyionic intermediates and transition states,and in the past few years a variety of urea-and thiourea-catalyzed enantioselectivetransformations of cationic electrophileshas been reported (36–39). The successof these processes is predicated on theability of the urea or thiourea catalystto bind the anion associated with thepositively charged electrophile throughhydrogen-bond interactions (Fig. 5). Innonpolar media where ion pairing is par-ticularly strong, this binding ensures thatthe chiral catalyst remains in close prox-imity to the cationic electrophile duringthe enantioselectivity-determining step ofthe catalytic cycle.For the catalyst to achieve the high de-

gree of transition state organization nec-essary for high enantioinduction, thepresence of secondary binding elementscapable of directly engaging and stabilizingthe cationic character of electrophilic

Table 2. Effect of catalyst arene substituents on the enantioselectivity of Claisenrearrangement of 1 is consistent with the proposed transition state cation–π interaction

MeO

O

O Me

20 mol% catalyst

30 o C, 72 h, hexanes MeO

O

O

Me

N N H

NH 2 +

N H N

BArF -

cat al yst

R R

2 1

Cat al yst Substituent (R) % Experiment al ee Cal c. ΔΔΕ‡. (kcal / mol )

. 2 5 7 H 3 64

. 1 2 6 F 4 73 e M N 5 2 . 3 0 9 45

Fig. 3. Examples of rate acceleration in Claisen rearrangements of polarized allyl vinyl ethers facilitatedby hydrogen bonding interactions.

Fig. 4. Diastereomeric transition structures for the enantioselectve Claisen rearrangements of ester-substiuted allyl vinyl ether 1 catalyzed by guanidinium 3 calculated using density functional theory(B3LYP 6-31G). Hydrogen bonding interactions are indicated in black and the stereodifferentiatingcation–π interaction is highlighted in red.

20680 | www.pnas.org/cgi/doi/10.1073/pnas.1006402107 Knowles and Jacobsen

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

species has proven critical. This is illus-trated in a particularly striking way inthe aryl-terminated bicyclization ofhydroxylactam derivatives such as 6,catalyzed by thiourea derivatives 8–11(Fig. 6) (40).The design of this system was inspired by

recent advances in the understanding ofbiosynthetic polyene cyclizations. In theseenzymatic transformations, olefins in apolyunsaturated substrate undergo se-quential additions to pendant carbocations

to generate structurally and stereochemi-cally complex polycyclic products. Mecha-nistic and structural studies of relevantenzymes such as oxidosqualene cyclasehave provided strong evidence that thecationic intermediates and transition statesin these cyclizations are stabilized by cati-on–π interactions with aromatic residueswithin the enzyme active site (41–44).In line with this biosynthetic proposal,

we explored the construction and appli-cation of thiourea catalysts bearing spe-

cifically positioned aryl substituents thatcould engage in analogous cation–πinteractions to direct the stereochemicalcourse of a cationic polycyclization. In themodel bicyclization reaction of hydrox-ylactam derivative 6 (Fig. 6), it was dis-covered that both the reactivity and degreeof asymmetric induction observed in thesetransformations was strongly correlatedwith the expanse of the arene withina common catalyst framework, with largerarenes proving more effective. Given thecationic nature of the reaction and factthat larger polycyclic aromatic hydro-carbons bind cations more strongly thantheir smaller analogs, this trend suggestedthat stabilizing cation–π interactions wereprofoundly influencing the degree ofasymmetric induction (45).Several experimental results and obser-

vations offer support for this hypothesis.The “strength” of an attractive non-covalent interaction is generally reflectedin the enthalpic contribution to the freeenergy of association (46). As such, if thelarger arenes were indeed engaging instronger or more extensive cation–π in-teractions in the dominant transition statestructure, this should be manifested in thedifferences in the magnitude of the dif-ferential enthalpy term between the moreand less enantioselective catalysts. Con-sistent with this reasoning, an Eyringanalysis of enantioselectivity for catalysts9, 10, and 11 revealed that the degree ofasymmetric induction in these reactionswas enthalpically controlled and that theextent of the differential enthalpy in-creased dramatically with the increasingsize of catalyst arene (Fig. 7). A corre-sponding and opposing increase in thedifferential entropy terms across the serieswas also observed, consistent with thegreater degree of differential transitionstate ordering expected as a consequenceof a stronger noncovalent interaction.Moreover, it was found that the degree ofenantioselectivity observed under a stan-dard set of conditions for catalysts 8–11correlated with the quadropole momentand polarizability of the arene found ineach of these catalysts (Fig. 7). Becausethese two molecular properties dictate thestrength of the electrostatic and dispersioncomponents of the cation–π interaction,these correlations offer further support forthe view that stabilizing cation–π inter-actions are a principal determinant ofenantioselectivity in these transformations(35, 47, 48).The examples provided above underline

the potential importance of cation–π in-teractions in differential stabilization ofcompeting diastereomeric transitionstates. Given these observations, it wouldbe reasonable to expect that other typesof stabilizing electrostatic interactionscould play a role in controlling the

Fig. 5. Generalized reaction scheme for anion-binding thiourea catalysis.

Fig. 6. Effects of catalyst aromatic group on the efficiency and enantioselectivity of the polycyclizationof hydroxylactam 6.

Knowles and Jacobsen PNAS | November 30, 2010 | vol. 107 | no. 48 | 20681

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

stereochemical outcome of organo-catalytic reactions of cationic inter-mediates. Indeed, as illustrated in thefollowing two examples, cation stabiliza-tion by weakly basic substituents in thio-urea and urea catalysts has been found toplay a pivotal role in the mechanisms ofsynthetically valuable enantioselectivereactions.

Strecker Reaction. In the course of exploringsimple chiral thiourea derivatives as po-

tential catalysts for the hydrocyanation ofimines, we observed a remarkable effect ofthe structure of the tertiary amide group onenantioselectivity (Fig. 8) (49). Thiourea16 proved to be a practical and broadlyapplicable catalyst for the Strecker re-action using either trimethylsilyl cyanideor potassium cyanide as the cyanide source(50). An extensive kinetic and computa-tional analysis of the mechanism of thistransformation revealed that differentialcation stabilization afforded by the amide

carbonyl of catalyst 16 plays a prominentrole in directing the stereochemicalcourse of these transformations (Fig. 8).Hammett studies, catalyst structure/ac-

tivity relationships, and computationalinvestigations paint a consistent mecha-nistic picture wherein the addition of HCNto imines mediated by 16 proceeds througha catalyst-bound cyanide/iminium ionpair (Fig. 9). In this pathway, which isqualitatively similar to the proposedmechanism for Strecker reactions carriedout in protic solvents, a thiourea-boundHCN (or HNC) first transfers its acidicproton to the imine substrate (51). Theresulting contact ion pair then undergoesrearrangement to separate the chargedspecies, which is accomplished by trans-ferring the hydrogen bonding interactionof the protioiminium ion N-H from thebound cyanide to the carbonyl of the cat-alyst amide. This charge-separated pairthen collapses to form the α-aminonitrileproduct (Fig. 9).Charge separation is energetically costly

in nonpolar media, and the ability of thepolar functionality of the catalyst to facil-itate this event is a plausible rationalefor catalysis. Yet in addition to being ratelimiting, this rearrangement is also enan-tioselectivity determining in that theresulting isomeric ion pairs collapse toproduct stereospecifically. Transition statestructures for the rearrangement stepleading to each enantiomer of the productwere identified computationally for eightstructurally distinct amido(thio)urea cata-lysts. In every case, the calculated enan-tioselectivities were found to correlate wellwith the values obtained experimentally,providing strong support for the validity of

Fig. 7. (A) Differential activation parameters for the competing diastereomeric pathways in the polycyclization of hydroxylactam 6 catalyzed by 9, 10, and 11.These values were derived from an Eyring analysis of enantioselectivity over a temperature range of 70 °C. (B and C) Linear correlations between ln(er) and thepolarizability and quadropole moment of the catalyst aromatic group obtained for the reaction of 6 with catalysts 8–11 under the reaction conditions de-scribed in Fig. 6.

Fig. 8. Effect of tertiary amide structure on enantioselectivity in the thiourea-catalyzed hydrocyanationof imine 12.

20682 | www.pnas.org/cgi/doi/10.1073/pnas.1006402107 Knowles and Jacobsen

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

the computational analysis and the accu-racy of the proposed mechanism (Fig. 10).In an effort to elucidate the basis for

asymmetric induction, we undertook ananalysis of the structural and geometricfeatures of the noncovalent interactionsassociating these competing rearrange-

ment transition states with the catalysts.The degree of stabilization of the cyanidenucleophile was found not to correlatewith the observed levels of enantiose-lectivity, because the sums of the bondlengths between the bound cyanide andthe two (thio)urea protons are essentially

equal in the transition states leading toboth the (R) and (S) enantiomers forall of the catalysts examined (d1 + d2,Fig. 10). However, the relative stabiliza-tion of the iminium cation in the re-arrangement transition states was foundto correlate strongly with the calculatedand observed enantioselectivities. Spe-cifically, in the most enantioselectivecatalysts, the lengths of the stabilizinghydrogen bonds to the N-H of theprotioiminium ion from the amide car-bonyl and the cyanide anion (d3 + d4,Fig. 10) are shorter, and consequentlymore stabilizing, in the lower energytransition state assembly than are theanalogous hydrogen bonds in the minorpathway. The steric demand of the cata-lyst amino acid and amide substituentsprevents the minor assembly from ac-cessing an optimal hydrogen bondingarrangement. This represents a well-characterized example of the manner inwhich stabilizing and destabilizing non-covalent interactions can act in concertto energetically differentiate qualitativelysimilar catalyst–ion pair complexes asmeans of achieving high enantiose-lectivity.

Povarov Reaction. Whereas the Streckerreaction proceeds through a transient,catalyst-bound iminium/cyanide pair thatrapidly collapses to form product, it standsto reason that iminium ions coupled withless nucleophilic anions may persist ascatalyst-bound intermediates if the ener-getics of the complexation are sufficientlyfavorable. This manner of strong groundstate stabilization was found to be a keymechanistic aspect of enantioselectivePovarov reactions between N-aryliminesand electron-rich olefins cocatalyzed bysulfonic acids and sulfinamide urea 21(Fig. 11) (52). These reactions proceedthrough highly reactive iminium sulfonateintermediates, which participate readily in[4+2] cycloadditions with electron-richdienophiles, such as dihydrofuran, in theabsence of any additional catalyst (Fig.11). However, these ion pairs also formstrong noncovalent complexes with sulfi-namide urea 21, wherein the sulfonateanion is hydrogen bonded in a bidentatefashion to the urea, and the iminium

Fig. 10. Correlation of transition structure bond lengths with enantioselectivity for Strecker reactions ofimine 12. Plots of the sum of the cyanide-(thio)urea H-bond lengths (d1 + d2, Left) and cyanide N-iminiumH + amide O-iminium H bond lengths (d3 + d4, Right) in B3LYP 6-31G(d) transition structures for eightstructurally distinct H-bond donor catalysts. This analysis points to differential iminium ion stabilizationthrough hydrogen bonding interactions as a basis for enantioselectivity.

Fig. 9. Proposed mechanism for the thiourea-catalyzed Strecker reaction.

Fig. 11. Representative asymmetric Povarov reaction catalyzed by urea 21. Illustrated below are thehydrogen bonding interactions that lead to the strong binding observed between the 21 and the imi-nium sulfonate intermediate.

Knowles and Jacobsen PNAS | November 30, 2010 | vol. 107 | no. 48 | 20683

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

formyl and N-H protons are simulta-neously engaged in H-bonding inter-actions with the sulfonate and sulfinamideoxygens (Fig. 11) (53).These complexed iminium ions also

undergo cycloaddition, but at rates severaltimes slower than those observed for thecorresponding unbound ions pairs. Thiskinetic disparity arises because catalyst21 stabilizes the iminium sulfonate com-plex more strongly than it stabilizes to thecycloaddition transition state, leading torate deceleration in the catalytic pathwayrelative to the background reaction(Fig. 12). Attenuation of the reactivity of

a highly reactive intermediate by a catalystfinds precedent in enzymology and hasbeen termed negative catalysis (54). Nor-mally, such a situation would precludeachieving high enantioselectivity becausethe starting material would be consumedin the racemic pathway preferentially.However, the total iminium ion concen-tration never exceeds that of the catalyst21 because of the use of catalytic levelsof Brønsted acid, and the equilibrium ofassociation for the iminium sulfonate andcatalyst 21 so strongly favors the ternarycomplex 22 that the concentration of freeiminium sulfonate in solution is vanish-

ingly small. Thus, despite proceeding ata rate substantially slower than thecompeting racemic pathway, the conver-sion of the substrate to product is chan-neled entirely through the asymmetricreaction pathway involving the catalyst-bound ion-pair.Further kinetic and computational

investigations were undertaken to eluci-date the basis for stereoselectivity in thesereactions. These studies indicate that theN-H•••Osulfinamide and C-H•••Osulfonatehydrogen bonding interactions observedin the ground state complex are main-tained in the transition state structuresand that the cycloaddition step is anasynchronous but concerted process.Enantioselectivity arises through theagency of a stabilizing π–π interactionbetween the (bis)trifluoromethylphenylgroup of the catalyst and aniline arene ofthe substrate that is observed in the tran-sition state leading to the major enantio-mer of the product, but which is notpresent in the competing diastereomerictransition state (Fig. 13).This reaction provides another demon-

stration of how multiple, weak noncovalentinteractions can operate cooperativelyin a small molecule catalyst to stabilizea highly reactive intermediate in a syn-thetically useful context. More broadly, theuse of a strong protic acid in conjunctionwith a second ion-binding catalyst offersa unique and potentially general approachto inducing asymmetry in reactions thatproceed through similar specific acidmechanisms.

ConclusionsThe four systems described in this per-spective illustrate some of the differentways that small-molecule catalysts canpromote enantioselective reactions solelythrough the agency of noncovalent inter-actions. The relatively small size of theorganic catalysts and the availability ofimproved methods for modeling electro-static forces render these reactions wellsuited to theoretical characterization.Direct correlation of computed andexperimentally determined structure/selectivity relationships offers strong vali-dation for the accuracy of a given mecha-nistic hypothesis. Full elucidation of thebasis for stereoinduction remains ex-tremely challenging but is a necessarystep in the ultimate quest for rationalcatalyst design. As such, the combinationof experimental and computational meth-ods to elucidate catalytic mechanismsand elements of stereocontrol will likelycontinue to increase in importanceas an enabling aspect of research inorganocatalysis.On a fundamental level, these systems

demonstrate how high enantioselectivityis tied directly to the capacity of the

Fig. 13. Calculated transition states structures for the Povarov reaction catalyzed by 21. A stabilizing,transition state π–π interaction is proposed as a basis for enantioselectivity and is highlighted in thestructure leading to the (R) enantiomer of product.

Fig. 12. Plots of initial rate and enantiomeric excess in the Povarov reaction versus [21] at three dif-ferent concentrations of triflic acid. This graph was reproduced from reference 54.

20684 | www.pnas.org/cgi/doi/10.1073/pnas.1006402107 Knowles and Jacobsen

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

catalysts to induce differential stabiliza-tion of charge distributions in competingdiastereomeric transition structures. Thiscapacity arises from the engagement ofcooperative mechanisms by the multi-functional catalysts, wherein multiple,spatially resolved, noncovalent interac-tions operate collectively to stabilize thecharge-separated character of one di-

astereomeric transition state selectively.Although this manner of electrostaticcomplementarity has long been recog-nized as a central feature of the macro-molecular structures of enzymes, theexamples provided here show how rela-tively simple, low-molecular-weight or-ganic catalysts can readily accommodateall of the functionality necessary to do so

as well. There is likely great future op-portunity in applying transition statestabilization strategies in organocata-lyst design.

ACKNOWLEDGMENTS. We thank Stephan Zuendand Christopher Uyeda for important discussions.This work was supported by National Institutes ofHealth Grants GM-43214 and P50 GM-69721 andfellowship support (to R.R.K.).

1. Bissantz C, Kuhn B, Stahl M (2010) A medicinalchemist’s guide to molecular interactions. J Med Chem53:5061–5084.

2. Bohm H-J, Schneider G (2003) Protein-LigandInteractions—From Molecular Recognition to DrugDesign (Wiley-VCH, Weinheim).

3. Lehn J-M (1995) Supramolecular Chemistry. Conceptsand Perspectives (Wiley VCH, Weinheim).

4. Schneider HJ (2009) Binding mechanisms in supramole-cular catalysis. Angew Chem Int Ed 48:3924–3977.

5. Osada Y, De Rossi DE (2000) Polymers, Sensors,Actuators (Springer, Berlin).

6. Hatada K (1997) Macromolecular Design of PolymericMaterials (Marcel-Dekker, New York).

7. MoultonB,ZaworotkoMJ (2001) Frommolecules tocrystalengineering: Supramolecular isomerism and polymor-phism in network solids. Chem Rev 101:1629–1658.

8. Aakeroy CB, Seddon KR (1993) The hydrogen bond andcrystal engineering. Chem Soc Rev 22:397–407.

9. Jeffrey GA (1997) An Introduction to HydrogenBonding (Oxford Univ Press, New York).

10. Israelachvili J (1991) Intermolecular and Surface Forces(Academic Press, London), 2nd Ed.

11. Meyer EA, Castellano RK, Diederich F (2003)Interactions with aromatic rings in chemical biologicalrecognition. Angew Chem Int Ed Engl 42:1210–1250.

12. Ma JC, Dougherty DA (1997) The cation-π interaction.Chem Rev 97:1303–1324.

13. Tanford C (1980) The Hydrophobic Effect (Wiley, NewYork), 2nd Ed.

14. Stone AJ (1996) The Theory of Intermolecular Forces(Oxford Univ Press, New York).

15. Pauling L (1946) Molecular architecture and biologicalreactions. Chem Eng News 24:1377.

16. Jencks WP (1989) Cataylsis in Chemistry and Enzymology(McGraw Hill, New York).

17. Warshel A, et al. (2006) Electrostatic basis for enzymecatalysis. Chem Rev 106:3210–3235.

18. Trost BM, Machacek MR, Aponick A (2006) Predictingthe stereochemistry of diphenylphosphino benzoicacid (DPPBA)-based palladium-catalyzed asymmetricallylic alkylation reactions: a working model. Acc ChemRes 39:747–760.

19. Austin JF, MacMillan DW (2002) Enantioselectiveorganocatalytic indole alkylations. Design of a newand highly effective chiral amine for iminium catalysis.J Am Chem Soc 124:1172–1173.

20. Corey EJ, Ishihara K (1999) Highly enantioselectivecatalytic Diels-Alder addition promoted by a chiral bis(oxazoline)-magnesium complex. Tetrahedron Lett 45:6807–6810.

21. Noyori R, Tokunaga M, Kitamura M (1995) Stereoselec-tive organic synthesis via dynamic kinetic resolution. BullChem Soc Jpn 68:36–56.

22. Sculimbrene BR, Morgan AJ, Miller SJ (2002)Enantiodivergence in small-molecule catalysis ofasymmetric phosphorylation: Concise total synthesesof the enantiomeric D-myo-inositol-1-phosphate and

D-myo-inositol-3-phosphate. J Am Chem Soc 124:11653–11656.

23. Copeland GT, Miller SJ (2001) Selection of enantio-selective acyl transfer catalysts from a pooled peptidelibrary through a fluorescence-based activity assay:An approach to kinetic resolution of secondaryalcohols of broad structural scope. J Am Chem Soc123:6496–6502.

24. Kolb HC, Andersson PG, Sharpless KB (1994) Toward anunderstanding of the high enantioselectivity in theosmium-catalyzed asymmetric dihydroxylation (AD). 1.Kinetics. J Am Chem Soc 116:1278–1291.

25. Yamakawa M, Yamada I, Noyori R (2001) CH/πattraction: The origin of enantioselectivity in transferhydrogenation of aromatic carbonyl compoundcatalyzed by chiral η6-arene-ruthenium(II) complexes.Angew Chem Int Ed 40:2818–2821.

26. Corey EJ, Rohde JJ (1997) The application of the formylC-H—O hydrogen bond postulate to the understandingof enantioselective reactions involving chiral boronLewis acids. Tetrahedron Lett 38:37–40.

27. Cload ST, Liu DR, Pastor RM, Schultz PG (1996) Mutagenesisstudy of active site residues in chorismate mutase fromBacillus subtilis. J Am Chem Soc 118:1787–1788.

28. Curran DP, Kuo LH (1995) Acceleration of a dipolarClaisen rearrangement by hydrogen bonding to asoluble diaryl urea. Tetrahedron Lett 36:6647–6650.

29. Coates RM, Rogers BD, Hobbs SJ, Peck DR, Curran DP(1987) Synthesis and Claisen rearrangement ofalkoxyallyl enol ethers. Evidence for a dipolartransition state. J Am Chem Soc 109:1160–1170.

30. Severance DL, Jorgensen WL (1992) Effects ofhydration on the Claisen rearrangement of allyl vinylether from computer simulations. J Am Chem Soc 114:10966–10968.

31. Sehgal A, Shao L, Gao J (1995) Transition structure andsubstituent effects on aqueous acceleration of the Claisenrearrangement. J Am Chem Soc 117:11337–11340.

32. Cramer CJ, Truhlar DG (1992) What causes aqueousacceleration of the Claisen rearrangement? J Am ChemSoc 114:8794–8799.

33. Uyeda C, Jacobsen EN (2008) Enantioselective Claisenrearrangements with a hydrogen-bond donor catalyst.J Am Chem Soc 130:9228–9229.

34. Ngola SM, Dougherty DA (1998) Concerning the effectsof aromatic ring fluorination on the cation-π interactionand other molecular recognition phenomenon in aqueousmedia. J Org Chem 63:4566–4567.

35. Mecozzi S, West AP, Dougherty DA (1996) Cation-πinteractions in simple aromatics: electrostatics providea predictive tool. J Am Chem Soc 118:2307–2308.

36. Taylor MS, Jacobsen EN (2004) Highly enantioselectivecatalytic acyl-pictet-spengler reactions. J Am Chem Soc126:10558–10559.

37. Raheem IT, Thiara PS, Peterson EA, Jacobsen EN (2007)Enantioselective Pictet-Spengler-type cyclizations ofhydroxylactams: H-bond donor catalysis by anionbinding. J Am Chem Soc 129:13404–13405.

38. Reisman SE, Doyle AG, Jacobsen EN (2008) Enantio-selective thiourea-catalyzed additions to oxocarbeniumions. J Am Chem Soc 130:7198–7199.

39. Zhang Z, Schreiner PR (2009) (Thio)urea organocatalysis—what can be learnt from anion recognition? Chem SocRev 38:1187–1198.

40. Knowles RR, Lin S, Jacobsen EN (2010) Enantioselectivethiourea-catalyzed cationic polycyclizations. J AmChem Soc 132:5030–5032.

41. Hoshino T, Sato T (2002) Squalene-hopene cyclase:Catalytic mechanism and substrate recognition. ChemCommun (Camb) (4):291–301.

42. Morikubo N, et al. (2006) Cation-π interaction in thepolyolefin cyclization cascade uncovered by incorporatingunnatural amino acids into the catalytic sites of squalenecyclase. J Am Chem Soc 128:13184–13194.

43. Wendt KU, Poralla K, Schulz GE (1997) Structureand function of a squalene cyclase. Science 277:1811–1815.

44. Thoma R, et al. (2004) Insight into steroid scaffoldformation from the structure of human oxidosqualenecyclase. Nature 432:118–122.

45. Vijay D, Sastry GN (2008) Exploring the sizedependence of cyclic and acyclic π-systems on cation-πbinding. Phys Chem Chem Phys 10:582–590.

46. Calderone CT, Williams DH (2001) An enthalpiccomponent in cooperativity: the relationship betweenenthalpy, entropy, and noncovalent structure in weakassociations. J Am Chem Soc 123:6262–6267.

47. Ngola SM, Dougherty DA (1996) Evidence for theimportance of polarizability in biomimetic catalysisinvolving cyclophane receptors. J Org Chem 61:4355–4360.

48. Cubero E, Luque FJ, Orozco M (1998) Is polarizationimportant in cation-π interactions? Proc Natl Acad SciUSA 95:5976–5980.

49. Zuend SJ, Jacobsen EN (2009) Mechanism of amido-thiourea catalyzed enantioselective imine hydrocyanation:transition state stabilization via multiple non-covalentinteractions. J Am Chem Soc 131:15358–15374.

50. Zuend SJ, Coughlin MP, Lalonde MP, Jacobsen EN (2009)Scaleable catalytic asymmetric Strecker syntheses ofunnatural alpha-amino acids. Nature 461:968–970.

51. Arnaud A, et al. (2000) Theoretical study of theaddition of hydrogen cyanide to methanimine in thegas phase and in aqueous solution. J Am Chem Soc122:324–330.

52. Xu H, Zuend SJ, Woll MG, Tao Y, Jacobsen EN (2010)Asymmetric cooperative catalysis of strong Brønstedacid-promoted reactions using chiral ureas. Science327:986–990.

53. Mayr H, Ofial AR, Wurthwein EU, Aust NC (1997) NMRspectroscopic evidence for the structure of iminium ionpairs. J Am Chem Soc 119:12727–12733.

54. Retey J (1990) Enzymic reaction selectivity bynegative catalysis or how do enzymes deal with highlyreactive interemediates? Angew Chem Int Ed Engl 29:355–361.

Knowles and Jacobsen PNAS | November 30, 2010 | vol. 107 | no. 48 | 20685

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020