Experimental determination of the absolute enantioselectivity of an antibody-catalyzed Diels-Alder reaction and theoretical explorations of the origins of stereoselectivity

Article abstract of DOI:10.1021/ja020879d

The exo and endo Diels-Alder adducts of p-methoxycarbonylbenzyl trans-1,3-butadiene-1-carbamate and N,N-dimethylacrylamide have been synthesized, and the absolute configurations of resolved enantiomers have been determined. On the basis of this information, the absolute enantioselectivities of the Diels-Alder reaction catalyzed by antibodies 13G5 and 4D5 as well as other catalytic antibodies elicited in the same immunizations have been established. The effects of different arrangements of catalytic residues on the structure and energetics of the possible Diels-Alder transition states were modeled quantum mechanically at the B3LYP/6-311++G**//B3LYP/6-31+G** level of theory. Flexible docking of these enantiomeric transition states in the antibody active site followed by molecular dynamics on the resulting complexes provided a prediction of the transition -state binding modes and an explanation of the origin of the observed enantioselectivity of antibody 13G5.

Full text of DOI:10.1021/ja020879d

Published on Web 02/11/2003
Experimental Determination of the Absolute Enantioselectivity
of an Antibody-Catalyzed Diels-Alder Reaction and
Theoretical Explorations of the Origins of Stereoselectivity
Carina E. Cannizzaro, Jon A. Ashley, K. D. Janda, and K. N. Houk*
Contribution from the Department of Chemistry and Biochemistry, UniVersity of California,
Los Angeles, California 90095-1569, and Department of Chemistry and the Skaggs Institute for
Chemical Biology, The Scripps Research Institute, La Jolla, California 92037
Received June 24, 2002 ; E-mail: houk@chem.ucla.edu
Abstract: The exo and endo Diels-Alder adducts of p-methoxycarbonylbenzyl trans-1,3-butadiene-1-
carbamate and N,N-dimethylacrylamide have been synthesized, and the absolute configurations of resolved
enantiomers have been determined. On the basis of this information, the absolute enantioselectivities of
the Diels-Alder reaction catalyzed by antibodies 13G5 and 4D5 as well as other catalytic antibodies elicited
in the same immunizations have been established. The effects of different arrangements of catalytic residues
on the structure and energetics of the possible Diels-Alder transition states were modeled quantum
mechanically at the B3LYP/6-311++G**//B3LYP/6-31+G** level of theory. Flexible docking of these
enantiomeric transition states in the antibody active site followed by molecular dynamics on the resulting
complexes provided a prediction of the transition-state binding modes and an explanation of the origin of
the observed enantioselectivity of antibody 13G5.
and synthetic RNA fragments.8 There are proposals that enzyme-
Introduction
catalyzed Diels-Alder reactions may occur during biosynthesis
The Diels-Alder reaction is one of the most powerful
carbon-carbon bond-forming processes in organic chemistry.
For many synthetic applications, the value of the Diels-Alder
reaction resides in the degree of stereocontrol that can be
exercised. Extensive efforts have been focused on controlling
the diastereo- and enantioselectivity of the Diels-Alder reac-
tion.1 Some chiral Lewis acids effectively catalyze the [4 + 2]
cycloaddition reactions of various dienes and dienophiles,1 while
simple organic amine catalysts have been developed recently
to catalyze Diels-Alder reactions of different dienes with R,â-
unsaturated aldehydes.2 Several systems have also been reported
that accelerate the Diels-Alder reaction by noncovalent ca-
talysis, that is, by presenting a host cavity complementary to
the transition state. These include micelles,3 bovine serum
albumin,4 Baker’s yeast,5 cyclodextrins,6 Rebek’s tennis ball,7
of several secondary metabolites.9-12
Antibodies have been successful in catalyzing several types
of Diels-Alder reaction.13 Antibodies have attracted attention
because catalysis is not limited to metabolically important
reactions14 and might be developed for a broad range of organic
reactions.
(8) (a) Morris, K. N.; Tarasow, T. M.; Julin, C. M.; Simons, S. L.; Hilvert, D.;
Gold, L. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 13028-13032. (b) Seelig,
B.; Jaschke, A. Chem. Biol. 1999, 6, 167-176. (c) Tarasow, T. M.; Eaton,
B. E. J. Am. Chem. Soc. 2000, 122, 1015-1021.
(9) Ichihara, A.; Oikawa, H. In ComprehensiVe Natural Products Chemistry;
Sankawa, U., Ed.; Barton, D., Nakanishi, K., Met-Cohn, O., Series Eds.;
Polyketides and Other Secondary Metabolites, Vol. 1; Elsevier: New York,
1999; pp 367-408.
(10) (a) Pohnert, G. ChemBioChem 2001, 2, 873-875. (b) Auclair, K.;
Sutherland, A.; Kennedy, J.; Witter, D. J.; Van den Heever, J. P.;
Hutchinson, C. R. Vederas, J. C. J. Am. Chem. Soc. 2000, 122, 11519-
11520. (c) Hutchinson, C. R.; Kennedy, J.; Park, C.; Kendrew, S.; Auclair,
K.; Vederas, J. Antonie Van Leewenhoek 2000, 78, 287-295.
(1) Recent reviews of enantioselective Lewis acid-catalyzed Diels-Alder
reactions: (a) Togni, A.; Venanzi, L. M.; Angew. Chem., Int. Ed. Engl.
1994, 33, 497-526. (b) Maruoka, K.; Yamamoto, H. In Catalytic
Asymmetric Synthesis; Ojima, I., Ed.; VCH: New York, 1993; Chapter 9.
(c) Pindur U.; Lutz, G.; Otto, C. Chem. ReV. 1993, 93, 741-761. (d) Deloux,
L.; Srebnik, M. Chem. ReV. 1993, 93, 763-784. (e) Kagan, H. B.; Riant,
O. Chem. ReV. 1992, 92, 1007-1019.
(11) (a) Oikawa, H.; Katayama, K.; Suzuki, Y.; Ichihara, A. J. Chem. Soc.,
Chem. Commun. 1995, 1321-1322. (b) Oikawa, H.; Kobayashi, T.;
Katayama, K.; Suzuki, Y.; Ichihara, A. J. Org. Chem. 1998, 63, 8748-
8756. (c) Katayama, K.; Kobayashi, T.; Oikawa, H.; Honma, M.; Ichihara,
A. Biochim. Biophys. Acta 1998, 1384, 387-395. (d) Oikawa, H.; Suzuki,
Y.; Katayama, K.; Naya, A.; Sakano, C.; Ichihara, A. J. Chem. Soc., Perkin
Trans. 1 1999, 1225-1232.
(2) Ahrendt, K. A.; Borths, C. J.; MacMillan, W. C. J. Am. Chem. Soc. 2000,
122, 4243-4244.
(12) (a) Watanabe, K.; Mie, T.; Ichihara, A.; Oikawa, H.; Honma, M. J. Biol.
Chem. 2000, 78, 287-295. (b) For a recent review on Diels-Alderases:
Pohnert, G. ChemBioChem. 2001, 2, 873-875.
(3) Braun, R.; Schuster, F.; Sauer, J. Tetrahedron Lett. 1986, 27, 1285-1288.
(4) Colonna, S.; Manfredi, A.; Annunziata, R. Tetrahedron Lett. 1988, 29,
3347-3350.
(5) Rao, K. R.; Srinivasan, T, N.; Bhanumathi, N. Tetrahedron Lett. 1990, 31,
5959-5960.
(6) (a) Hudlicky, T.; Butora, G.; Fearnley, S. P.; Gum, A. G.; Persichini, P. J.,
III; Stabile, M. R.; Merola, J. S. J. Chem. Soc., Perkin Trans. 1 1995, 19,
2393-2398. (b) Alvira, E.; Cativiela, C.; Garcia, J. I.; Mayoral, J. A.
Tetrahedron Lett. 1995, 36, 2129-2132. (c) Sternbach, D. D.; Rossana,
D. M. J. Am. Chem. Soc. 1982, 104, 5853-5854.
(7) Kang, J.; Santamaria, J.; Hilmersson, G.; Rebek, J., Jr. J. Am. Chem. Soc.
1998 120, 7389-7390.
(13) (a) Hilvert, D.; Hill, K. W.; Nared, K. D.; Auditor, M. M. J. Am. Chem.
Soc. 1989, 111, 1, 9261-9262. (b) Braisted, A. C.; Schultz, P. G. J. Am.
Chem. Soc. 1990, 112, 7430. (c) Suckling, C. J.; Tedford, M. C.; Bence,
L. M.; Irvine, J. I.; Stimson, W. H. J. Chem. Soc., Perkin Trans. I 1993,
1925. (d) Gouverneur, V. E.; Houk, K. N.; de Pascual-Teresa, B.; Beno,
B.; Janda, K. D.; Lerner, R. A. Science 1993, 262, 204-208. (e)
Yli-Kauhaluoma, J. T.; Ashley, J. A.; Lo, C.; Tucker, L.; Wolfe, M. M.;
Janda, K. D. J. Am. Chem. Soc. 1995, 117, 7041-7047. (f) Meekel, A. P.;
Resmini, M.; Pandit, U. K. J. Chem. Soc., Chem. Commun. 1995, 571-
572. (g) Resmini, M.; Meekel, A. P.; Pandit, U. K. Pure Appl. Chem.,
1996, 68, 2025-2028.
9
10.1021/ja020879d CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 2489-2506
2489

A R T I C L E S
Cannizzaro et al.
Scheme 1. Thermal Diels-Alder Reaction
Electron-rich dienes, such as alkoxy- or amino-substituted
dienes are especially valuable in synthesis,15 and Cr(III)-salen
complexes16 have been developed for the enantioselective
catalysis of the Diels-Alder reaction of such dienes.
A reaction of this type that has been catalyzed by antibodies
involves the cycloaddition of amino-substituted diene 1 and N,N-
dimethyl acrylamide (Scheme 1). Regioselectivity is high, and
only four ortho isomers shown in the scheme are observed
experimentally. To obtain only one adduct out of the four
possible isomers, the reaction must be both diastereoselective
(endo-exo) and enantioselective. This level of selection rep-
resents a great challenge to any particular catalyst.
In 1993, antibodies 22C8 and 7D4 were elicited against
racemic, bis-cyclooctene transition-state (TS) analogues, shown
in Figure 1a.13d Whereas the thermal reaction is only slightly
endo selective (Scheme 1), the antibodies catalyze enantiose-
lectively the formation of one enantiomer of the exo (trans) and
one enantiomer of the endo (cis) adducts, respectively.
Two years later, antibodies 13G5 and 4D5 were raised to
flexible ferrocene haptens, shown in Figure 1b.13e Indirect
comparison of the chiral HPLC retention times of the products
obtained with both sets of antibodies revealed that 13G5 and
4D5 catalyzed the formation of the same exo and endo adducts,
respectively. No studies of the absolute configuration of these
products were undertaken. Computational modeling that in-
volved locating gas-phase transition states followed by docking
into the crystal structure of 13G5 with Autodock V2.4 showed
that the four possible transition states can be accommodated in
the active site. Only the exo transition states maintained the
hydrogen-bonding motif observed in the crystal structure of an
inhibitor bound to exo-Diels-Alderase 13G5.17 It was suggested
that the exo-(3R,4R) transition structure is the most highly
stabilized but that definitive evidence could only be obtained
by determination of the absolute configuration of the product
obtained from 13G5 catalysis.17
Figure 1. Racemic haptens for the Diels-Alder reaction.
kcat/kuncat ) 6.9 M) requires a good understanding of the
interactions responsible for catalysis in the active site. To
program antibodies to catalyze the formation of each of the four
possible ortho adducts, a thorough understanding of the factors
determining the enantioselectivity is necessary.
We have now synthesized and established the absolute
configuration of the products obtained by antibody catalysis with
13G5, 4D5, and 15 other catalytic antibodies elicited for the
reaction in Scheme 1. The enantioselectivity of several of these
antibodies has been established experimentally, and some
considerations are made as to the origins of stereoselectivity
occurring by these antibodies, elicited to achiral or racemic
haptens. The reaction has been studied quantum mechanically
at a higher level of theory than previously, and the effects of
different arrangements of catalytic residues on the structure and
energetics of the possible transition states have been determined.
Flexible docking of these transition states in the antibody active
site, followed by molecular dynamical studies of the resulting
complexes, led to an explanation of the observed enantioselec-
tivity of antibody 13G5.
To improve the modest catalytic acceleration by these
antibodies (for example 13G5 has: kcat ) 1.2 × 10-3 min-1
;
(14) (a) Wentworth, P., Jr.; Janda, K. D. Cell Biochem. Biophys. 2001, 35, 63-
87. (b) Tremblay, M. R.; Dickerson, T. J.; Janda, K. D. AdV. Synth. Catal.
2001, 343, 577-585 and references therein. (c) Schultz, P. G.; Lerner, R.
A. Science 1995, 269, 1835. (d) Stewart, J. D.; Benkovic, S. J. Nature
1995, 375, 388-391. (e) Keinan, E.; Lerner, R. A. Isr. J. Chem. 1996, 36,
113-119. (f) Morrison, J. D., Ed. Chiral Catalysis; Asymmetric Synthesis,
Vol. 5; Academic Press: New York, 1985.
(15) Kozmin, S. A.; Rawal, V. H. J. Am. Chem. Soc. 1998, 120, 13523-13524.
(16) Huanh, Y.; Iwama, T.; Rawal, V. H. J. Am. Chem. Soc. 2000, 122, 7843-
7844.
Preparation of the Enantiomerically Pure Diels-Alder
Adducts and Determination of Their Absolute Stereochem-
istries. Synthetic Methodology. Diene 2 undergoes cycload-
dition with N,N-dimethylacrylamide at 110 °C in refluxing
toluene to afford a mixture of the stereoisomeric endo (cis) and
(17) Heine, A.; Stura, E. A.; Yli-Kauhaluoma, J. T.; Gao, C.; Deng, Q.; Beno,
B. R.; Houk, K. N.; Janda, K. D. and Wilson, I. A. Science 1998, 279,
1934-1940.
9
2490 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
Scheme 2. Diene Synthesis
A R T I C L E S
Scheme 3. Chiral Auxiliary Synthesis
Scheme 4. Dienophile Synthesis
exo (trans) adducts in a ratio of 66:34.18 When the same reaction
is carried out with the water-soluble diene 1 in phosphate-
buffered saline (PBS), an 85:15 mixture of the endo and exo
adducts is obtained. Both reactions are completely regioselective,
giving only the ortho regioisomers (Scheme 1).
improved methodology by Mathre et al.24 (Scheme 4) involves
nucleophilic attack of (S)-(-)-4-tert-butyl-2-oxazolidinone lithium
salt on the previously formed acryloyl anhydride.
The asymmetric Diels-Alder cycloaddition was carried out
first between equimolar quantities of diene 2 and dienophile
(S)-10 under the Lewis acid-catalyzed conditions (AlEt2Cl).23,25
Unfortunately, the carbamate moiety was reactive under these
conditions, and no Diels-Alder adduct formation was detected.
The reaction was carried out successfully without catalyst in
refluxing toluene to give endo-11 and exo-11 as diastereomeri-
cally pure adducts in a ratio of 67:33 (Scheme 5).
Compound endo-11 was recrystallized from ethyl ether:
hexanes (1:1 v/v), and the absolute configuration of the newly
created stereocenters was determined by X-ray diffraction
analysis. The X-ray structure is shown in Figure 2. From the
known S configuration of the chiral auxiliary, it is determined
that compound endo-11 has the (3R,4S) configuration. Note the
preferred s-cis conformation around the O-C bond of the
carbamate group (dihedral OdC-O-C ) 7.7°). The dihedral
angles around the O-C bond of the carbamate ester portion
are +27.2° for the gauche hydrogen, -93.4° for the methyl
benzoate group, and +146.5° for the hydrogen in an anti
arrangement.
The Evans regioselective hydrogen peroxide-mediated hy-
drolysis of oxazolidinone-type chiral auxiliaries26 was used for
the conversion of endo-11 into endo-12. Removal of the chiral
auxiliary without epimerization or carbamate hydrolysis was
performed at a reasonable rate at room temperature using 4 equiv
We expected, therefore, that the thermal cycloaddition
between diene 2 and dienophile (S)-10 would give direct access
to a separable mixture of the diastereomeric Diels-Alder
adducts, one endo and one exo, bearing a third chiral center of
known stereochemistry, a requirement for the unequivocal
assignment of the absolute configuration of the adducts by X-ray
crystallography in the absence of an anomalous dispersion atom.
The selection of the chiral auxiliary was based on the observa-
tion that crotonates with tert-butyl oxazolidinones as chiral
auxiliaries show the highest degree of diastereoselection in the
Et2AlCl-catalyzed Diels-Alder cycloaddition with isoprene;
diastereoselectivity is beyond the limits of detection (>100 ds).19
Dienes 1 and 2 were synthesized according to a modification
of Overman’s procedure20 (Scheme 2) described in full in the
Supporting Information.
The chiral auxiliary (S)-8 (Scheme 3) was obtained by
reduction of commercially available (S)-tert-leucine, (S)-6, with
borane generated in situ 21 to give the intermediate alcohol (S)-
7, followed by cyclization with phosgene.22
The chiral dienophile (S)-10 was first synthesized by addition
of the (S)-(-)-4-tert-butyl-2-oxazolidinone bromomagnesium
salt to 3-bromopropionyl chloride followed by dehydrohaloge-
nation of the resultant 3-bromopropionyl imide with triethyl-
amine, according to the strategy developed by Evans.23 An
(18) Overman, L. E.; Taylor, G. F.; Petty, C. B.; Jessup, P. J. J. Org. Chem.
1978, 43 2164.
(19) Evans, D. A.; Chapman, K. T.; Hung, D. T.; Kawaguchi, A. T. Angew.
Chem., Int. Ed. Engl. 1987, 26, 1184-1186.
(20) Jessup, P. J.; Petty, C. B.; Roos, J.; Overman, L. E. Organic Syntheses;
Wiley: New York, 1988; Collect. Vol. VI, p 95.
(24) Mathre, D. J.; Ho, G.-J. J. Org. Chem. 1995, 60, 2271-2273.
(25) Gawley, R. E.; Aube´ J. Principles of Asymmetric Synthesis; Pergamon:
Oxford, 1996; p 263.
(26) Evans, D. A.; Britton, T. C.; Ellman, J. A. Tetrahedron Lett. 1987, 28,
6141-6144.
(21) Abiko, A.; Masamune, S. Tetrahedron Lett. 1992, 33, 5517-5518.
(22) Newman, M. S.; Kutner, A. J. Am. Chem. Soc. 1951, 73, 4199-4204.
(23) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc. 1988, 110,
1238-1256.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2491

A R T I C L E S
Cannizzaro et al.
Scheme 5. Thermal Asymmetric Diels-Alder Reaction
of hydrogen peroxide and 2 equiv of LiOH. The desired acid,
endo-12, was obtained in 95% yield, based upon 81% conver-
sion.
Palladium-catalyzed hydrogenation of the double bond fol-
lowed by hydrogenolysis of the carbamate moiety27-29 of acid
endo-12 gave â-amino acid endo-15. The absolute stereochem-
istry of this â-amino acid derivative, and therefore of endo-
12, was determined by comparing its optical rotation [R]20
+2.0 (c ) 0.20 in methanol) with literature values of [R]25
)
)
D
D
+1.7 (c ) 0.60 in water) for the corresponding hydrochloride.30
This experiment proved that no significant epimerization had
occurred during removal of the chiral auxiliary (optical rotation
of the 3R,4R epimer: [R]25 ) -45.4 (c ) 1.04 in water) for
D
the corresponding hydrochloride,30 and [R]16D ) -65.5 (c ) 2
in water) for the neutral compound33) and also provided an
alternative method for the determination of the absolute con-
figuration of the exo adduct.
Figure 2. X-ray crystal structure of endo-11 adduct.
carbamate group axial and the dimethylamide equatorial is most
stable and should be present to the extent of about 90% at
equilibrium at room temperature due to A1,3 strain involving
the carbamate. These computations explain why epimerization
does not readily give the trans isomer. Efforts were then directed
toward the isolation of the exo-11 adduct obtained along with
endo-11 adduct from the thermal Diels-Alder reaction mixture.
Four consecutive column chromatographies were necessary to
isolate approximately 0.7 g of pure exo-11 adduct by NMR.
Attempts to recrystallize this glassy material were unsuccessful.
An analytical sample of this material was further purified by
HPLC, but crystallization remained elusive. To establish the
absolute stereochemistry of the pure exo-11 adduct and obtain
a sample of the final exo product, we followed the same
synthetic route developed for the endo-11 adduct. Transforma-
tion of exo-11 into exo-12 was carried out successfully, and
the acid was isolated as an amorphous white solid (Scheme 8).
Amidation of acid endo-12 was carried out successfully by
activation with carbonyl diimidazole followed by addition of
dimethylamine as reported by Jung et al.31 Amide endo-13 was
obtained in 82% crude yield. Final hydrolysis of the methyl
ester gave the endo-14 final target in 98% yield with the 3R,4S
configuration (Scheme 6).
We attempted to prepare the exo-(3R,4R) adduct by epimer-
ization of the enantiomerically pure endo-(3R,4S) adduct with
4 equiv of potassium tert-butoxide in tert-butyl alcohol at room
temperature. However, 1H 500 MHz NMR analysis of each spot
showed no signs of epimerization. Starting material, p-hy-
droxymethylbenzoic acid, and decomposed material were found.
We explored the equilibrium between the exo-(3R,4R) and
endo-(3R,4S) adducts with semiempirical AM1 calculations
summarized in Scheme 7. The endo (cis) conformer with the
(27) Jackson, A. E.; Johnstone, R. A. W. Synthesis 1976, 685-687.
(28) Meienhofer, J.; Kuromizu, K. Tetrahedron Lett. 1974, 37, 3259-3262.
(29) ElAmin, B.; Anantharamaiah, G. M.; Royer, G. P.; Means, G. E. J. Org.
Chem. 1979, 44, 3442-3444.
Treatment of the acid exo-12 with the different optically active
amines, (+)-(2S,3R)-4-(dimethylamino)-3-methyl-1,2-diphenyl-
2-butanol or (R)-methylbenzylamine, gave needle-shaped crys-
tals that were too small to produce a useful diffraction pattern
in either case. Since crystallization failed, we turned to chemical
correlation. The â-amino acid, exo-15, was obtained from a pure
(30) Davies, S. G.; Ichihara, O.; Lenoir, I.; Walters, I. A. S. J. Chem. Soc.,
Perkin Trans. 1 1994, 1411-1415.
(31) Jung, M. E.; Da´mico, D. C. J. Am. Chem. Soc. 1995, 117, 7379-7388.
(32) Nohira, H.; Ehara, K.; Miyashita, A. Bull. Chem. Soc. Jpn. 1970, 43, 2230-
2233.
(33) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (b) Becke,
A. D. J. Chem. Phys. 1993, 98, 5648.
9
2492 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Scheme 6. Final Steps toward the Enantiomerically Pure endo-14 Adduct
sample of exo-12 by catalytic hydrogenation of the double bond
followed by hydrogenolysis of the carbamate moiety.27-29 The
absolute stereochemistry of the â-amino acid exo-15, and
therefore of exo-12, was determined by comparing its optical
rotation [R]20D ) +63.0 (c ) 0.25 in methanol) with literature
values of [R]17D ) +66.5 (c ) 2 in water)32 for the same amino
Immunization with the CPY hapten (Figure 1b) had originally
produced six exo-Diels-Alderases 13G5, 4D3, 7G8, 2A3, 6D7,
and 14D9, in order of decreasing effectiveness of catalysis
relative to the uncatalyzed reaction, and one endo-Diels-
Alderase, antibody 2A3.13e The immune system responded to
the CPQ hapten (Figure 1b) to generate seven endo-Diels-
Alderases 4D5, 8G2, 12F10, 15C1, 9B4, 1F4, and 12D2,
presented again in order of decreasing efficacy of catalysis
relative to the uncatalyzed reaction, along with one exo-Diels-
Alderase, antibody 7D10.13e
acid, and [R]25 ) +47.4 (c ) 1.14 in water) for the
D
corresponding hydrochloride.30 After amidation of exo-12 and
hydrolysis of the methyl ester of exo-13, the enantiomerically
pure exo-(3S,4S)-14 adduct was obtained as an amorphous white
solid.
Determination of the Absolute Stereoselectivity of the
Diels-Alderase Catalytic Antibodies. The order in which the
four isomeric Diels-Alder adducts appear in the chiral HPLC
trace was first identified by comparisons with our authentic exo
and endo adducts. The Diels-Alder reaction was then performed
in the presence of catalytic amounts of different antibodies, and
the mixture of stereoisomeric cycloadducts was analyzed by
chiral HPLC (see Supporting Information for details).
Comparisons of HPLC traces of the enantiomerically pure
products and the products of the antibody-catalyzed reactions,
showed that antibody 13G5, the most efficient of all exo-Diels-
Alderases, catalyzes the formation of the exo-(3S,4S) adduct.
All of the other exo-Diels-Alderases catalyzed the formation
of the same exo-(3S,4S) adduct, regardless of their relative rates
of catalysis. Even exo-Diels-Alderase 7D10, elicited to a
different hapten (CPQ), also catalyzes the formation of the exo-
(3S,4S) adduct. These results are summarized in Table 1.
Results
We also found that all endo- Diels-Alderases 4D5, 8G2,
12F10, 15C1, 9B4, 1F4, and 12D2 (from hapten CPQ) and
2A3 (from hapten CPY) catalyzed the formation of the endo-
(3R,4S) adduct.
Earlier results4e showed that exo-Diels-Alderase 22C8,
elicited from the bicyclooctene hapten, BCO-exo, catalyzes the
formation of the same exo cycloadduct as 13G5, while endo-
Diels-Alderase 7D4 from BCO-endo catalyzes the formation
of the same endo adduct as 4D5. Despite important differences
in the structures of the haptens used to generate them, the
racemic nature of the BCOs and achiral nature of the CPY and
Figure 3a shows the chiral HPLC traces of a standard mixture
of the four stereoisomers made by combining equal amounts
of racemic endo and exo adducts. Figure 3b is a trace of the
enantiomerically pure sample of exo-(3S,4S)-14, and Figure 3c
is a trace of the enantiomerically pure sample of endo-(3R,4S)-
14.
Under the conditions used to separate the four isomers, the
first peak corresponds to the exo-(3S,4S) cycloadduct (retention
time ) 12.15 min), and the fourth peak corresponds to the endo-
(3R,4S) adduct (retention time ) 21.77 min).
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2493

A R T I C L E S
Cannizzaro et al.
Scheme 7. AM1 Calculations on the Base-Catalyzed Equilibrium between endo-14 and exo-14 Diels-Alder Adducts
Scheme 8. Final Steps toward the Enantiomerically Pure exo-14 Adduct
CPQ haptens, all of the antibodies catalyze the formation of
one exo and one endo adduct. Having established the absolute
stereochemistries of the products from one set of antibodies,
we can conclude that antibody 22C8 catalyzes the formation
of the exo- (3S,4S) adduct and 7D4 catalyzes the formation of
the endo-(3R,4S) adduct.
9
2494 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
an increased peak at 21.73 min with respect to the peak for the
enantiomer at 17.08 min, while the exo mixture remains racemic.
On the basis of the experimentally determined amounts of
enantiomeric adducts in the antibody-catalyzed reactions, we
calculated an % ee ) 98 ( 2% ee for exo Diels-Alderase 13G5
and 90 ( 2% ee for endo Diels-Alderase 4D5, after subtracting
the amount of racemic adducts produced in the background
reaction at the same time. The small, random fluctuations of
the enantiomeric ratios observed on every single measurement
are within experimental error, eliminating the possibility of
product inhibition. A steady decrease of enantiomeric ratio with
time would have been expected under product inhibition
conditions. These enantiomeric excesses indicate that there are
differences in free energies of activation of at least 2.7 and 1.8
kcal/mol, respectively, for formation of the major and minor
enantiomers by 13G5 and 4D5, respectively, at 37 °C.
Computational Explanations of the Origins of Enantiose-
lective Catalysis. Methodology. Quantum mechanical calcula-
tions summarized in Figure 8 assessed the effect of hydrogen-
bonding residues on the potential energy surface of the exo
Diels-Alder reaction between methyl N-butadienyl carbamate
(diene 15) used as a model for diene 1, and N,N-dimethylacryl-
amide shown in Figure 6.
All stationary points were optimized and characterized by
frequency analysis at the B3LYP/6-31+G** level. The reported
energies correspond to single-point calculations at the B3LYP33/
6-311++G** level on these B3LYP/6-31+G** geometries. The
reported energies include zero-point energy corrections scaled
by 0.9804.34
Figure 3. Determination of the absolute configuration of the antibody-
catalyzed adducts.
Calculations on the transition states of the Diels-Alder
reaction between diene 1 and N,N-dimethylacrylamide were then
carried out to obtain a more realistic structure for the docking
procedure (Figure 9). All stationary points for this larger system
were fully optimized at the B3LYP/6-31G* level with zero-
point energy corrections scaled by 0.9804.33 Calculations were
performed with Gaussian 98.35
Quantitative Determination of the Enantioselectivity of
the endo Diels-Alderase 4D5 and the exo Diels-Alderase
13G5. Five buffered solutions (pH ) 7.4, PBS) of 13G5 and
three buffered solutions of 4D5 were prepared. The solutions
each contained 20 µM of the corresponding antibody, 8 mM of
the diene, and 4 mM of the dienophile.
Docking of the enantiomeric exo transition states within the
X-ray structure of exo-Diels-Alderase 13G5 was performed
with Autodock V3.036 that allows a flexible ligand to interact
with a rigid macromolecule.
The antibody-catalyzed reactions were analyzed once a day,
following the HPLC protocol described in the Supporting
Information. The rate of formation of exo adduct produced in
the reaction catalyzed by 13G5 (R2 ) 0.97) is constant (Figure
4a). Similar results were obtained for the reaction catalyzed by
4D5 (R2 ) 0.99, Figure 4b). The five background solutions were
also analyzed and gave a good linearity (R2 ) 0.96 for the exo
adducts in Figure 4a and R2 ) 0.99 for the endo adducts in
Figure 4b). It is clear from the plotted data in Figure 4 that
neither exo-Diels-Alderase 13G5 catalyzed the formation of
any of the endo adducts, nor did endo-Diels-Alderase 4D5
catalyze the formation of any of the exo adducts.
All molecular dynamics simulations were performed with the
AMBER 5.0 suite of programs37 and the parm94 parameter set.38
After flexible docking, the lowest-energy structure from the
lowest-energy cluster of each competing enantiomeric exo
(34) Wong, M. W. Chem. Phys. Lett. 1996, 256, 391-399.
(35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian,
Inc.: Pittsburgh, PA, 1998.
Figure 5 shows a background trace A after 139.37 h (about
6 days) of reaction. The peaks at 12.05 and 15.01 min
correspond to equal amounts of exo-(S,S) and exo-(R,R) adducts,
respectively, and the peaks at 17.02 and 21.69 min correspond
to equal amounts of endo-(3S,4R) and endo-(3R,4S) cycload-
ducts, respectively. Traces B and C correspond to 13G5- and
4D5-catalyzed reactions that were stopped and analyzed after
approximately the same reaction time (165.77 and 167.07 h,
respectively). In trace B the peak at 12.04 min is noticeably
enlarged with respect to the peak for the enantiomer at 15.00
min, while the endo mixture remains racemic. Trace C shows
(36) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W. E.;
Belew, R. K.; Olson, A. J. J. Comput. Chem. 1998, 19, 1639-1662.
(37) Pearlman, D. A.; Case, D. A.; Caldwell, J. W.; Ross, W. S.; Cheathman,
T. E., III; DeBolt, S.; Ferguson, D.; Seibel, G.; Kollman, P. Comput. Phys.
Commun. 1995 91, 1-41.
(38) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M., Jr.;
Ferguson D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P.
J. Am. Chem. Soc. 1995, 117, 5179-5197.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2495

A R T I C L E S
Cannizzaro et al.
Table 1. Enantioselectivity of the Entire Pool of Catalytic Antibodies Elicited along with 13G5 and 4D5; Filled Circles Represent the Main
Stereoisomer Produced by Each Antibody
Figure 4. (a) 13G5 production of exo and endo adducts. (b) 4D5 production of exo and endo adducts.
transition state was selected as the initial system for the
molecular dynamics simulations.
Quantum Mechanical Theozymes. Three key residues in
the active site of antibody 13G5 appeared to be responsible for
the observed catalysis, as determined by X-ray crystallography
studies of the exo Diels-Alderase inhibitor complex at 1.95 Å
resolution (Figure 7).16 It was previously postulated that Tyr-
L36 activates N,N-dimethylacrylamide for nucleophilic attack
9
2496 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Figure 5. Chiral HPLC traces of Diels-Alder adducts produced by 13G5 and 4D5.
0.5 kcal/mol with respect to the uncatalyzed reaction (Figure
8a) due to a preferential stabilization of the hydrogen-bonded
reactant diene (-4.3 kcal/mol) over the transition state (-3.8
kcal/mol). This result is in agreement with the fact that electron-
withdrawing groups increase the energy of activation of a
normal39 electron demand Diels-Alder reaction between an
electron rich diene and electron-deficient dienophile.40
A water molecule hydrogen-bonded to the dienophile car-
bonyl decreases the energy of activation of the exo transition
state (Figure 8c) by 1.1 kcal/mol with respect to the uncatalyzed
reaction, due to preferential stabilization of the transition state
(-6.5 kcal/mol) over the hydrogen-bonded reactant dienophile
(-5.4 kcal/mol).
The strongest accelerating effect occurs when a formate anion
is hydrogen-bonded to the carbamate NH group (Figure 8d).
The energy of activation of the exo transition state is decreased
by 7.4 kcal/mol due to a preferential stabilization of the
transition state (-29.0 kcal/mol) over the hydrogen-bonded
reactant diene (-21.6 kcal/mol).
The asparagine residue probably does not catalyze the
reaction, but rather deactivates the diene, although it may play
a role during the initial molecular recognition event between
the diene and antibody 13G5. The tyrosine and aspartate residues
do activate diene and dienophile toward the exo Diels-Alder
reaction, through hydrogen bonding.
Calculations on the exo transition state catalyzed simulta-
neously by molecules of water and formate show that there is
a synergistic stabilizing effect of 1.9 kcal/mol on the energy of
activation. In this case, the energy of activation (Figure 8e)
decreases by 10.4 kcal/mol with respect to the uncatalyzed
reaction, due to a preferential stabilization of the transition state
Figure 6. Dienes and dienophile used in the calculations.
by hydrogen bonding to the carbonyl group.16 Asparagine-L91
and aspartic acid-H50 hydrogen bond to the carboxylate side
chain of the inhibitor that substitutes for the carbamate group
of the diene 1 substrate. As stated earlier, the four possible ortho
transition states, endo and exo, can be accommodated in the
active site of 13G5, but only the exo transition states maintained
the hydrogen-bonding motif that accounts for inhibitor recogni-
tion.16 In this exo conformation, tyrosine-L36 still hydrogen
bonds to the carbonyl of N,N-dimethylacrylamide, while aspar-
tate-H50 forms a hydrogen bond with the carbamate NH, and
asparagine-L91, with the carbamate carbonyl.
To test the ability of these residues to preferentially bind the
exo transition state, a model system was studied. It consisted
of N,N-dimethylacrylamide, methyl N-butadienyl carbamate as
the diene, a formamide molecule to model the asparagine residue
(Asn-L91), a water molecule for the tyrosine residue (Tyr-L36),
and a formate molecule to mimick the aspartate residue (Asp-
H50). The local minima corresponding to the hydrogen-bonding
situations in the antibody were located. When formamide is
hydrogen-bonded to the carbamate carbonyl, the energy of
activation of the exo transition state (Figure 8b) increases by
(39) Sustmann, R. Tetrahedron Lett. 1971, 2721-2724.
(40) Sustmann, R.; Shubert, R. Angew. Chem., Int. Ed. Engl. 1972, 11, 840.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2497

A R T I C L E S
Cannizzaro et al.
Figure 7. Exo Diels-Alderase 13G5 inhibitor complex at 1.95 resolution.
Figure 8. Relative energies of activation (Ea) in kcal/mol of the exo Diels-Alder reaction catalyzed by different arrangements of hydrogen-bonding groups.
Higher bars correspond to the energies of the transition states shown (a-f), and lower bars represent the corresponding substrates, hydrogen-bonded to
formamide, water or formate when required.
(-37.4 kcal/mol) over the corresponding hydrogen-bonded
reactants (-27.0 kcal/mol). These two catalytic hydrogen-
bonding interactions each cause asynchronicity and charge
separation in the transition state. When both are present, the
effects are cooperative, and the maximum charge separation is
favored. The orthogonal orientation of Tyr-L36 and Asp-H50
within the active site of 13G5 permits maximum delocalization
of charge in the exo transition state. Inclusion of the formamide
molecule at this point increases the energy of activation by 1.0
kcal/mol (Figure 8f), showing that the deactivating effect of
the asparagine residue is enhanced in the presence of the
catalytic residues.
The B3LYP/6-31+G** optimized geometries are substan-
tially different from the HF/3-21G geometries of simplified
models reported previously,12d although the qualitative conclu-
sions about activation energies are the same.
The effects of both catalytic residues on the conformation
and relative energies of the endo and exo transition states
between diene 1 and N,N-dimethylacrylamide were explored
at the B3LYP/6-31G* level of theory.
9
2498 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Figure 9. Effect of the catalytic groups on the diastereoselectivity of the Diels-Alder reaction.
The endo transition state (Figure 9A) is 0.7 kcal/mol more
stable than the exo TS in the gas phase (Figure 9B), in agreement
with the experimentally observed diastereoisomeric ratio of
methyl ester derivatives obtained under kinetic control in
refluxing toluene (66:34, 0.5 kcal/mol). The diastereoselectivity
increases when the reaction is carried out in water at 37 °C
(85:15, 1.0 kcal/mol).
close contact with the inhibitor in the active site and is an exo
Diels-Alderase.
Docking Studies. Docking of the lowest-energy exo-(R,R)
transition state, shown in Figure 9E, and its mirror image exo-
(S,S) were carried out.
HF/6-31G charges were fitted to the electrostatic potential
of these quantum mechanically obtained transition states,
according to the Merz-Singh-Kollman scheme. The transition
states were manually predocked into 13G5 using the Insight
II/Discover software package.42 The values of the charges
induced on the water molecule by the reacting dienophile were
used to manually modify the charges in the side-chain atoms
of the Tyr-L36 that the water mimicks. Likewise, charges
developed on the formate molecule by the reacting diene were
translated onto the Asp-H50 residue. These changes were
performed under the assumption that the two hydrogen bonds,
one to Asp-H50 and the second to Tyr-L36, are crucial for
enantioselection and catalysis.
When the formate anion hydrogen-bonds to the carbamate
NH group and the water hydrogen-bonds to the dienophile
carbonyl, the exo TS (Figure 9D) becomes 1.1 kcal/mol more
stable than the endo TS (Figure 9C).
As noted earlier, the catalytic hydrogen-bonding interactions
cause asynchronicity and charge separation in the transition state.
The charge of the carbonyl oxygen of the dienophile changes
from -0.58 (Figure 9, A or B) to -0.63 (Figure 9, C or D)
after hydrogen-bonding to a water molecule. The charge on the
formate oxygen hydrogen-bonded to the carbamate HN changes
from -0.60 to -0.57 upon hydrogen-bonding. The switch in
steroselectivity observed upon hydrogen-bond formation to these
two catalytic groups could be a consequence of the increased
negative charge on these two oxygens and their mutual
electrostatic repulsion in the endo TS (Figure 9C). In the exo
transition states ( Figure 9, D and E) this electrostatic repulsion
is minimized because the two oxygens are separated from each
other. Antibody 13G5 has aspartate and tyrosine residues in
Autodock V3.0 utilizes a genetic algorithm technique to
sample potential binding modes. This Lamarckian genetic
algorithm was used to search the conformational space of the
TS within the active site. It calculates the interaction energy
(41) Vargas, R.; Garza, J.; Dixon, D. A.; Hay, B. P. J. Am. Chem. Soc. 2000,
122, 4750-4755.
(42) InsightII/Discover; Biosym/Molecular Simulations Inc.: San Diego, CA.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2499

A R T I C L E S
Cannizzaro et al.
Figure 10. Inhibitor bound to exo Diels-Alderase 13G5 (left) and the exo-(S,S) transition state, (Figure 9E), middle. They are superimposed on the right.
Figure 11. Rigid transition-state docking. The exo-(S,S)-TS is on the left, and the exo-(R,R)-TS is on the right.
utilizing grid-based affinity potentials to describe the protein.
Interaction energies produced by Autodock V3.0 have been
parametrized to reproduce experimental free energies of binding
from aqueous solution.35
The second strategy involved docking the flexible transition
states into the rigid active site of antibody 13G5, followed by
molecular dynamics relaxation of the hydrated antibody around
the bound and frozen transition state. This latter approach was
designed to study the effect of the conformational flexibility of
the diene in the recognition event during the docking simulation
and to address the role of induced-fit on binding.
Docking the frozen transition states within the rigid frame-
work of antibody 13G5 resulted in only one conformational
cluster for the exo-(S,S) transition state and two clusters for the
exo-(R,R). The mean docked energy of the exo-(S,S) cluster is
-12.5 kcal/mol. The mean docked energy of the lowest-energy
exo-(R,R) cluster containing 118 out of 120 total runs is -11.1
kcal/mol. The corresponding docked structures are shown in
Figure 11. When no conformational changes are allowed in
either the TS or the antibody, the experimentally observed exo-
(S,S) TS fits perfectly in the active site of Ab 13G5 in a
productive orientation that preserves the three hydrogen bonds
responsible for inhibitor recognition. A closer look at the active
Two different docking simulations were performed with both
enantiomeric exo transition states. The first strategy involved
docking the frozen transition states, exo-(R,R) and exo-(S,S),
within the rigid framework of antibody 13G5. Figure 10 shows
the structure of the inhibitor bound in the crystal structure of
13G5 on the left, and the computed exo-(S,S) transition state
(enantiomer of figure 9E) with the water and formate molecules
removed in the middle. The superimposed structures on the right
of Figure 10 show that the three hydrogen-bonding groups,
namely the carbonyls of the diene and dienophile and the NH
group of the diene carbamate, occupy almost identical regions
of space. This observation strongly suggests that the experi-
mentally catalyzed exo-(S,S) transition state fits nicely in the
active site, while the enantiomeric exo-(R,R) transition state does
not.
9
2500 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Figure 12. Flexible transition-state docking. The exo-(S,S)-TS is on the left, and the exo-(R,R)-TS is on the right.
site reveals that both oxygens of the Asp-50 carboxylate are
hydrogen-bonded to the carbamate NH with distances of 2.31
and 2.27 Å. Asn-91 is hydrogen-bonded to the carbamate
carbonyl (3.01 Å), and Tyr-36 forms a hydrogen bond to the
dienophile carbonyl (2.09 Å). The electron-poor benzoate group
of the diene is π-stacked between the two electron-rich aromatic
residues of Tryptophan-H33 (Trp-H33) and Tyr-H102 with inter-
ring distances between 3 and 4 Å. The backbone NH of Tyr-
H102 constitutes the fourth point of recognition by weakly
hydrogen-bonding the benzoate oxygen (2.86 Å).
been calculated for the rotation of a carboxylate out of the plane
of a phenyl ring.50
Rotation around the OCO- -CH2R bond of esters necessitates
rotational barriers only from 0.6 to 1.4 kcal/mol,51 and rotational
barriers of the Ph- -CH2X bond require 1-1.6 kcal/mol.52 During
the flexible docking procedure, only the dihedral angles around
the C-O and the O-Ph bonds of the diene were allowed to
change to avoid the appearance of unrealistic conformations.
Two conformational clusters for each transition state were
obtained; the lowest in energy containing most of the structures
in both cases. The left side of Figure 12 shows the lowest-energy
conformer of the (S,S) transition state docked into Ab 13G5.
The mean docked energy of the lowest-energy cluster containing
118 out of 120 total runs, is -13.1 kcal/mol; this value also
corresponds to the energy of the lowest-energy conformer within
the cluster. The right side of Figure 12 shows the lowest-energy
conformer of the exo-(R,R) transition state docked into Ab 13G5.
The average docked energy of the lowest-energy cluster
containing 119 out of 120 total runs is -13.0 kcal/mol and also
corresponds to the energy of the lowest conformer within the
cluster. There is now no preference for the exo-(S,S) transition
state leading to the experimentally observed enantiomerically
On the contrary, the exo-(R,R) TS does not dock in a
productive orientation. In the lowest-energy complex shown in
Figure 11, the benzoate group of the Diels-Alder TS is placed
between the two aromatic residues, Trp-H33 and Tyr-H102, in
the same orientation as the exo-(S,S) TS. Nevertheless, the
transition state as a whole has to rotate to preserve the
conformation of the diene chain. This rotation places the
carbamate NH and the dienophile carbonyl perpendicular to the
polar residues Asp-H50 and Tyr-L36, preventing the formation
of catalytic hydrogen bonds. Only one hydrogen bond between
Asn-L91 and the carbamate carbonyl is formed.
These results suggest that the mature exo-Diels-Alderase
13G5 is capable of binding the experimentally observed exo-
(S,S) TS without undergoing important changes in its quaternary
structure. Essentially a lock-and-key mechanism43,44 explains
the enantioselectivity in this case.
(43) Wedemayer, G. J.; Patten, P. A.; Wang, L. H.; Schultz, P. G.; Stevens, R.
C. Science 1997, 276, 1665-1669.
(44) Amit, A. G.; Mariuzza, R. A.; Phillips, S. E. V.; Poljak, R. J. Science 1986,
233, 747.
(45) For reviews, see: (a) Stewart, W. E.; Siddall, T. H., III. Chem. ReV. 1970,
70, 517. (b) Oki, M. Applications of Dynamic NMR Spectroscopy to Organic
Chemistry; VCH: Deerfield Beach, 1985; Chapter 2.
(46) (a) Drakenberg, T.; Dahlqvist, K.-I.; Forsen, S. J. Phys. Chem. 1972, 76,
2178-2183. (b) Eberhardt, E. S.; Loh, S. N.; Hinck, A. P.; Raines, R. T.
J. Am. Chem. Soc. 1992, 114, 5437-5439.
Flexible docking studies were also carried out to assess the
role of transition-state flexibility on docking. Only rotations
about bonds with relatively low barriers need to be included in
such calculations (<4 kcal/mol). The barriers of rotation around
the N- -CO of amides45 are about 17-18 kcal/mol, increasing
up to 20 kcal/mol as the dielectric constant and hydrogen-bond
donating ability of the solvent increase,46 and decrease to 15-
17 kcal/mol47 in carbamates becoming insensitive to solvent
polarity. Rotational barriers around the O- -CO bond of car-
bamates are about 16 kcal/mol,48 decreasing to 10 kcal/mol49
in methyl esters. Rotational barriers of around 6 kcal/mol have
(47) Cox, C.; Lectka, T. J. Org. Chem. 1998, 63, 2426-2427.
(48) Van der Werf, S.; Olijnsma, T.; Engberts, J. B. F. N. Tetrahedron Lett.
1967, 689-693.
(49) (a) Review: Jone, G. I. L.; Owen, N. L. J. Mol. Struct. 1973, 18, 1. (b)
Grindley, T. B. Tetrahedron Lett. 1982, 23, 1757-1760.
(50) Nelson, M. R.; Borkman, R. F. J. Mol. Struct. (THEOCHEM) 1998, 432,
247-255.
(51) Karabatsos, G. J.; His, N.; Orzech, C. E., Jr. Tetrahedron Lett. 1966, 4639-
4643.
(52) (a) Abraham, R. J.; Bakke, J. M. Tetrahedron 1978, 34, 2947-2951. (b)
Mons, M.; Robertson, E. G.; Simons, J. P. J. Phys. Chem. A 2000, 104,
1430-1437.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2501

A R T I C L E S
Cannizzaro et al.
Figure 13. Conformational changes on the enantiomeric transition states upon flexible docking. The exo-(S,S)-TS is on the left, and the exo-(R,R)-TS is on
the right.
pure product. Both transition states are oriented within the active
site in a fashion very similar to that of the rigidly docked exo-
(S,S) transition state in Figure 11. The catalytic tyrosine residue
(Tyr-L36) is hydrogen-bonded to the dienophile carbonyl, with
a hydrogen-bond length of 1.86 Å in the exo-(S,S) TS and 1.73
Å in the exo-(R,R) TS. Both oxygen atoms of the catalytic
aspartate residue (Asp-H50) are hydrogen-bonded to the diene
NH. The bond lengths are 2.34 and 2.44 Å in the exo-(S,S) TS
and 2.08 and 2.41 Å in the exo-(R,R) TS. Asn-L91 forms a
hydrogen bond with the carbamate carbonyl that is slightly
longer in the exo-(S,S) TS (2.24 Å) than in the exo-(R,R) TS
(1.94 Å). In both cases, the fourth contact required for
enantiodifferentiation53 is obtained by a hydrogen bond between
the backbone NH of Tyr-H102 and the benzoate oxygen.
structure. Parts a and b of Figure 13 show the calculated
structures of the enantiomeric transition states as mirror images
of each other. Parts c and d of Figure 13 are the corresponding
structures after flexible docking. Both the rigid and flexible
docking studies show that the exo-(S,S) TS fits better in the
active site.
Molecular Dynamics Studies
After flexible docking, antibody 13G5 was allowed to relax
around the constrained docked structures of both transition
states. In each case, 1000 steps of energy minimization were
followed by a 280-ps molecular dynamics simulation of the
antibody around the frozen transition states. The exo-(S,S) and
exo-(R,R) structures in the protein were capped with a sphere
of explicit water molecules centered at the transition structure
with a radius of 20 Å. A Na+ counterion was added to neutralize
the carboxylate of the diene. The sodium ion was positioned
using the “addions” functionality of xleap, which defines an
electrostatic grid and places the ion at the most negative point
on that grid. In both cases, the ion was at least 22 Å distant
from the closest atom in the docked transition states. The
nonbonded cutoff for the van der Waals and electrostatic
interactions occurred at 14 Å. All the protein residues and the
water molecules were allowed to move (i.e., included in the
“belly” option of AMBER) while the transition states remained
frozen. A constant dielectric function was used that mimics the
presence of water around the system when explicit water
molecules are present.
The rigid and flexible docking of the exo-(S,S) TS gives rise
to the same binding mode with very close hydrogen-bonding
distances. Therefore, allowing conformational changes during
the docking simulation does not improve the binding of exo-
(S,S) TS. On the other hand, for the exo-(R,R) TS to fit in a
productive orientation within the active site, a conformational
change of the diene chain is necessary. To assess the energetic
cost of this conformational change, single-point energy evalu-
ations at the B3LYP/6-31G* level of theory were performed
on the exo-(R,R) TS after the rigid and the flexible docking.
The new conformation obtained after flexible docking is 1.6
kcal/mol higher in energy than the original transition-state
(53) Mesecar, A. D.; Koshland, D. E., Jr. Nature 2000, 403, 614-615.
9
2502 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Figure 14. Molecular dynamics relaxation of exo-Diels-Alderase 13G5 around the flexibly docked transition states. The exo-(S,S)-TS is on the left, and
the exo-(R,R)-TS is on the right.
During the initial minimization performed to remove bad
contacts in the X-ray structure of 13G5, distance constraints
were introduced in the docked transition states among the carbon
and hydrogen atoms of the reacting parts of the diene and
dienophile, using the “NMR constraints” functionality of the
SANDER module. This procedure ensured that the core of the
transition state remained structurally as obtained quantum
mechanically, allowing changes of the bond lengths of only
(0.003 Å. Minimizations were run with a combination of 10
cycles of steepest descent, followed by 990 steps of the full
conjugate gradient method without any constraints on protein
bond lengths. After the initial minimization, heating and
equilibration of the system were carried out for a total of 80
ps. Equilibration was assessed by monitoring temperature and
total and potential energies versus time. A productive run of
200 ps was carried out, and 400 coordinates were saved for
analysis. The dynamics were run with constraints on all bonds
using the SHAKE54 algorithm, and a time step of 1 fs. The
average value of the temperature was maintained at the
experimental temperature of 310 K, by means of the Berendsen
coupling algorithm.55
Visual inspection of the trajectories obtained for both (S,S)
and (R,R) transition states revealed two important changes when
compared to the initial structures originated during the flexible
docking procedure. First, after the equilibration period, only one
water molecule entered the active site when the exo-(S,S) TS
was bound as a consequence of a tighter fitting between this
TS and antibody 13G5 (Figure 14). This water molecule is
strongly hydrogen-bonded to Asp-H50 (1.59 Å) and regularly
forms a second hydrogen bond with the alkoxide oxygen of
the carbamate. Asp-H50 moves closer to the carbamate NH
(1.75 Å), strengthening this important catalytic interaction. By
contrast, three water molecules entered the active site when exo-
(R,R) TS was bound. They formed three strong hydrogen bonds
with the catalytic Asp-H50 with bond lengths that vary from
1.59 to 1.64 Å. These interactions decrease the electron density
on the catalytic residue and keep it at 2.08 Å from the transition
state, weakening the catalytic hydrogen bond between Asp-H50
and the carbamate HN.
Second, the decelerating interaction between Asp-L91 and
the carbamate carbonyl is completely lost when the experimen-
tally observed exo-(S,S) TS is bound, but it remains when the
exo-(R,R) TS is bound. Catalytic residue Tyr-L36 reorients and
hydrogen bonds the dienophile carbonyl trans to the CdC with
shorter bond lengths (1.70 Å for exo-(S,S) TS and 1.64 Å for
the exo-(R,R)).
These observations suggest that when the diene is recognized
by Asp-H50 and Trp-H33 of the heavy chain, the si face is
exposed, and the incoming dienophile can approach the diene
from the back, presenting its si face to give the experimentally
observed (S,S) product. When the diene is recognized simulta-
neously by the heavy (Asp-H50 and Trp-H33) and the light
(Asn-L91) chains, the re face is exposed, and the dienophile
can only approach the diene from the front, presenting its re
face. This last arrangement is disfavored for three reasons: (1)
It requires a conformational change of the exo-(R,R) transition
state to fit the active site in a productive manner, with an
energetic cost of 1.6 kcal/mol, (2) the most important electro-
static interaction from the catalytic point of view between Asp-
H50 and the carbamate NH is weakened by solvation, and (3)
the unfavorable hydrogen bond between Asp-L91 and the
carbamate carbonyl, that disfavors binding of the transition state
relative to the ground state, is still present.
Discussion
Hapten CPY can rotate freely around its axis to give a mixture
of rotamers in solution. The functional arrangement in these
rotamers corresponds to a racemic mixture of endo and exo
Diels-Alder TS analogues. In principle, this hapten could be
bound in a suitable conformation mimicking any of these four
stereoisomeric transition states. Our experimental results show
(54) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1997,
23, 327-341.
(55) Berensen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.;
Haak, J. R. J. Chem. Phys. 1984, 81, 3684.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2503

A R T I C L E S
Cannizzaro et al.
two L residues, and this fact has recently been applied toward
the design and synthesis of a potent and selective protease-
resistant antibacterial agent variety of D-polypeptides.57
In 1969 Jencks introduced the idea58 that catalytic antibodies
elicited in response to transition-state mimics might catalyze
reactions by selective stabilization of the transition state more
than the corresponding reactants, in analogy to enzymes. Many
catalytic antibodies catalyze the hydrolysis of ester and amide
bonds, or the formation of ester and amide bonds where one of
the reactants is an amino acid of known stereochemistry, and
other related reactions such as transesterification and ester
aminolysis. Therefore, we could in principle determine whether
the immune system possesses an inherent bias toward one
chirality by analyzing the absolute stereochemistry of the
products formed by antibodies elicited by racemic haptens. If
the immune system were biased, all antibodies generated from
racemic haptens would be expected to catalyze the formation
of the same enantiomer of the product in a particular reaction.
Although no chiral centers are created during these processes,
antibodies elicited by racemic peptide-like haptens have been
shown to selectively react with only one enantiomer of the
substrate.59-61 The absolute stereochemistry of the products was
either (R)26,27 or (S).28 In some cases the enantioselective
discrimination was achieved by preferential binding of the
substrate with the same chirality observed in the product.26 In
other cases, one enantiomer of the product was formed
preferentially, despite the higher affinity of the antibody for other
enantiomer of the hapten.27
The observations suggest that, in principle, there is no innate
reason the immune system would select a particular stereo-
chemistry. Antibody 13G5 catalyses the Diels-Alder reaction
between a diene bearing a carbamate group and a dienophile
with a dimethylamide moiety. This highly enantioselective
reaction could be regarded as the formation of a â-dipeptide
where two new stereocenters are created. Therefore, antibodies
that catalyze the formation of the other two diastereoisomeric
transition states of the Diels-Alder reaction, (R,R) and (S,R),
should have been expected. The fact that no antibodies capable
of catalyzing the formation of the (R,R) adduct were detected
along with 13G5, despite the low concentration of hapten used
during the immunization,62 could be explained by the fact that
they all evolved from the same germ line or, alternatively, by
the fact that antibodies were selected for binding to the TS
analogue, TSA, and then for catalysis. The hapten used for
Figure 15. Postulated energy profile of the Diels-Alder reaction catalyzed
by 13G5.
that antibody 13G5 catalyzes the formation of the exo-(S,S)
adduct, and Table 1 indicates that all exo-Diels-Alderases
elicited with hapten CPY also catalyze the formation of the
same enantiomer. X-ray crystallography and computational
studies suggested that Tyr-L36 and Asp-H50 catalyze the
reaction by activating the dienophile and the diene, respectively.
Asn-L91, on the other hand, increases the energy of activation
of the reaction. Docking and molecular dynamics simulations
showed that this latter deactivating interaction is lost when
antibody 13G5 relaxes around the exo-(S,S) TS and remains
when antibody 13G5 relaxes around the exo-(R,R) TS. Further-
more, the tighter fitting of the exo-(S,S) TS impedes solvation
of the active site, maximizing the stabilizing interactions
between the exo-(S,S) TS and the catalytic Tyr-L36 and Asp-
H50 residues. It is reasonable to assume that the diene and
dienophile form a van der Waals complex within the antibody-
combining site in an orientation similar to that of the corre-
sponding transition state before reacting. Although antibody
13G5 might be capable of stabilizing the exo-(R,R) complex to
a greater extent than the enantiomeric exo-(S,S) complex, the
exo-(S,S) TS is preferentially stabilized over the exo-(R,R) TS,
giving rise to the (S,S) cycloadduct. (Figure 15)
This analysis correlates with an earlier observation on the
enantioselectivity of antibody 14D9. This antibody catalyzes
the transformation of achiral enol ethers into the corresponding
(S) ketals despite preferential binding of the (R) intermediate.56
The surprising consistency of the whole panel of antibodies
obtained from the ferrocene hapten CPY raises a number of
important questions about the origins of this stereoselectivity.
First, why are only catalysts that induce one type of chirality
in the product obtained from racemic haptens? One explanation
is that all evolved from a single germline. If this were true,
then all antibodies would be expected to have a high degree of
conserved residues and probably to present a very similar spatial
arrangement of catalytic groups. Clearly, it would be interesting
to obtain all sequence data to analyze this possibility in detail.
Alternatively, if the panel of catalytic antibodies came from
different germlines, it could be that they preferentially catalyze
the formation of the exo-(S,S) TS because the immune system
possesses an inherent bias toward one chirality due to the fact
that all antibodies are formed from L-amino acids. It is well-
known that proteases can only hydrolyze peptidic bonds between
(57) Gill, T. J., III; Kunz, H. W.; Papermaster, D. S. J. Biol. Chem. 1967, 242,
3308.
(58) Jenks, W. P. Catalysis in Chemistry and Immunology; McGraw-Hill: New
York, 1969; p 288.
(59) (a) Janda, K. D.; Benkovic, S. J.; Lerner, R. A. Science 1989, 244, 437-
440. (b) Nakatani, T.; Hiratake, J.; Shinzaki, A.; Umeshita, R.; Suzuki, T.;
Nishioka, T.; Nakajima, H.; Oda, J. Tetrahedron Lett. 1993, 34, 4945-
4948. (c) Tanaka, F.; Kinoshita, K.; Tanimura, R.; Fujii, I. J. Am. Chem.
Soc. 1996, 118, 2332-2339.
(60) (a) Cochran, A. G. J. Am. Chem. Soc. 1991, 113, 6670-6672. (b) Martin,
M. T.; Angeles, T. S.; Sugasawara, R.; Aman, N. I.; Napper, A. D.; Darsley,
M. J.; Sanchez, R. I.; Booth, P.; Titmas, R. C. J. Am. Chem. Soc. 1994,
116, 6508-6512.
(61) (a) Guo, J.; Huang, W.; Scanlan, T. S. J. Am. Chem. Soc. 1994, 116, 6062-
6069. (b) Lo, C.-H. L.; Wentworth, P.; Jung, K. W.; Yoon, J.; Ashley, J.
A.; Janda, K. D. J. Am. Chem. Soc. 1997, 119, 10251-10252.
(62) For both the CPY and CPQ immunizations, 100 µg (of a 5 mg/mL KLH
conjugate) per mouse per injection were used. Four mice were immunized
each with 100 µg CPY-KLH (or CPQ-KLH) in 200 µL RIBI adjuvant
intraperitoneally. Two weeks later the mice were boosted again with another
100 µg. The mice were bled 1 week later, and the titers were determined
by ELISA. After 1 month, a final injection with 50 µg of hapten-KLH in
100 µL PBS was performed intravenously in the tail vein. Three days later
the spleen was removed for fusion.
(56) Shabat, D.; Shulman, H.; Itzhaky, H.; Reymond, J.-L.; Keinan, E. J. Chem.
Soc., Chem. Commun.1998, 16, 1759-1760.
9
2504 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003

Antibody-Catalyzed Diels−Alder Reactions
A R T I C L E S
Figure 16. Sequences of the germ-line precursors, VMS9 and Vk1a, of the VH and VL regions of structurally related antibodies 39-A11, DB3, and TE33.
The first two sequences correspond to antibodies 13G5 and 4D5 elicited toward a racemic TSA of a Diels-Alder reaction, very similar to the hapten used
to generate the other Diels-Alderase 39-A11. Dashes indicate identical residues, and stars correspond to gaps in the sequences.
screening antibodies for binding to the TS analogue resembles
the van der Waals complex between the reactants more closely
than the transition state because the distance between the
cyclopentadienyl rings in the ferrocene hapten is longer than
the distances of the forming bonds in the transition state.
Screening for binding would select antibodies that bind pref-
erentially the rotamer of the hapten that mimics the (R,R) TS
analogue as depicted in the energetic profile shown in Figure
15. Considering that this (R,R) TS analogue would resemble a
dipeptide made from D-amino acids, this postulate is in
agreement with the experimental observation that D-polypeptides
are more antigenic.63 Despite binding preferentially the van der
(63) Schechter, I.; Sela, M. Biochemistry 1967, 6, 897.
9
J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003 2505

A R T I C L E S
Cannizzaro et al.
several comparisons between germline and mature catalytic
antibodies have been carried out. Particularly relevant to the
present study is the case of antibody 39-A11, that catalyzes a
Diels-Alder reaction very similar to the one shown in Scheme
1. Structural and functional studies of the germline precursor
suggest that 39-A11 and related antibodies (DB3, TE33, and
1E9) derive from a family of germline genes that have been
selected through evolution for the ability of the encoded proteins
to form a polyspecific combining site (Figure 16).69 Recent
findings about the evolution of catalytic antibody 28B4 are also
consistent with the notion that certain combinations of Vκ-Jκ
and VH give a polyspecific combining site which is to be tuned
by CDR H3 for optimal specificity.70 Analysis of hapten binding
and catalytic determinants in a family of catalytic antibodies
raised against a common TS analogue, which accelerate an oxy-
Cope rearrangement, also revealed close homologies among the
heavy chains of the catalytically active members, which derive
mainly from a single germline gene.71 The light chains could
be traced to several different but related germline genes.
Construction and characterization of hybrid antibodies indicated
a strong conservation of binding-site structure among the
catalytically active clones. Finally, comparison of X-ray struc-
tures of the binding sites of catalytic antibodies raised to bind
different phosphonates showed that, despite their different
sequences, they all exhibit a tetrahedral array of hydrogen-bond
donors complementary to the tetrahedral anion.72
Figure 17. Racemic haptens used for generating Diels-Alderases that
catalyze a similar transformation.
Waals complex between the reactants in the (R,R) configuration,
these antibodies would catalyze the formation of the (S,S)
product because of the particular distribution of hydrogen-bond
donor and acceptor groups in the active site. This interpretation
is in agreement with previous observations that not every
antibody that binds the TSA will be an effective catalyst.64
Screening for actual catalytic activity65 rather than for TSA
binding might help identify antibodies that enantioselectively
catalyze the formation of each of the four diastereoisomeric
Diels-Alder adducts.
The catalytic active site of Ab 13G5 consists of a deep
hydrophobic pocket made of the variable domains (VL and VH)
of the Fab. The polar residues that interact with the inhibitor
and the transition states are located as follows: Tyr-L36 in the
framework, at the border of the L1 loop, Asn-L91 in the L3
loop, Trp-H33 in the H1 loop, Asp-H50 in the H2 loop, and
Thr-102 in the H3 loop. The sequence and size of these
hypervariable regions or CDRs normally determine the specifi-
city of the antibody-antigen interactions. Our results suggest
that to further accelerate this reaction while maintaining the
observed enantioselectivity, Tyr-L36 could be mutated to Lys.
The perturbed pKa of the side chain of lysine, that drops from
10.54 to 6.3-6.7 in a hydrophobic cavity,66 would increase the
strength of the hydrogen bond with N,N-dimethylacrylamide,
further activating the dienophile toward the Diels-Alder
reaction.
In the absence of antigen, the immune system presents an
impressive diversity of antibodies (greater than 108 different
binding sites).67 Genetic and biochemical studies68 have ex-
plained the origin of this sequence diversity of antibodies that
derives from the combinatorial association of only 103-104 gene
fragments. Once the immune system is challenged by an antigen,
the process of affinity maturation provides further diversity by
somatic mutation of the bases encoding for the variable region,
leading to antibodies with increased affinity and specificity. In
an effort to understand the immunological evolution of antibody
catalysis, and ultimately, the structural basis for the transforma-
tion of sequence diverse antibodies into high affinity receptors,
Despite the fact that 13G5 and 4D5 derive from immunization
with the same mixture of racemic diastereoisomeric haptens,
their sequences show only 40% homology (Figure 16). More
surprising is the fact that neither of them evolves from the
germline of antibody 39-A11, despite the important structural
similarity of the haptens (Figure 17).
These observations suggest that each hapten (CPY and CPQ,
Figure 1b), and the hapten used to elicit 39-A11 (Figure 17)
are recognized by different germline antibodies during the
primary response. The different polar groups present in the
CDRs of antibodies 13G5, 4D5, and 39A11 suggest that through
selection and mutation, nature found different ways to catalyze
the same chemical transformation.
Acknowledgment. We are grateful to the National Institute
of General Medical Sciences, National Institutes of Health
(GM61402 to K.N.H. and GM43858 to K.D.J.), the National
Science Foundation (CHE-9616772 to K.N.H.), and The Skaggs
Institute for Chemical Biology (K.D.J.) for financial support of
this research.
Supporting Information Available: Experimental details
(PDF). X-ray crystallographic file in CIF format. This material
is available free of charge via the Internet at http://pubs.acs.org.
(64) Mader, M. M.; Bartlett, P. A. Chem. ReV. 1997, 97, 1281-1301.
(65) (a) Benedetti, F.; Berti, F.; Flego, M.; Resmini, M.; Bastiani, E. Anal.
Biochemistry 1998, 256, 67-73. (b) Shulman, H.; Eberhard, A.; Eberhard,
C.; Ulitzur, S.; Keinan, E. Bioorg. Med. Chem. Lett. 2000, 10, 2353-2356
and references therein.
(66) (a) Lee, J. K.; Houk, K. N. Science 1997, 276, 942-945. (b) Daopin, S.;
Anderson D. E.; Baase, W. A.; Dalhquist, F. W.; Matthews, B. W.
Biochemistry 1991, 30, 11521.
(67) Burton, D. R. Acc. Chem. Res. 1993, 26, 405-411.
(68) (a) Burnet, F. M. In The Clonal Selection Theory of Acquired Immunity;
Vanderbilt University Press: Nashville, TN, 1959; p 53. (b) Talmage, D.
W. Science 1959, 129, 1649. (c) Lederberg, J. Science 1959, 129, 1649.
(d) Tonegawa, S. Nature 1983, 302, 575. (e) French, D. L.; Laskov, R.;
Scharff, M. D. Science 1989, 244, 1152.
JA020879D
(69) Romesberg, F. E.; Spiller, B.; Schultz, P. G.; Stevens, R. C. Science 1998,
279, 1929-1933.
(70) Yin, J.; Mundorff, E. C.; Yang, P. L.; Wendt, K. U.; Hanway, D.; Stevens,
R. C.; Schultz, P. G. Biochemistry 2001, 40, 10764-10773.
(71) Ulrich, H. D.; Schultz, P. G. J. Mol. Biol. 1998, 275, 95-111.
(72) Tantillo, D. J.; Houk, K. N. Chem. Biol. 2001, 535-545.
9
2506 J. AM. CHEM. SOC. VOL. 125, NO. 9, 2003