An antibody-catalyzed allylic sulfoxide-sulfenate rearrangement

Article abstract of DOI:10.1021/jo991299a

Antibodies SZ-cis-39C11 and SZ-trans-28F8, which were elicited in response to N-aryl-3-methoxy-phenyl proline derivatives, catalyze the [2,3]- sigmatropic rearrangement of allylic sulfoxides to sulfenates. Reduction of the sulfenates with dithiothreitol in situ yields allylic alcohols as the final product. The antibodies achieve rate accelerations in the range 10²- 10³ over background and exhibit distinctive hapten-dependent substrate specificity and enantio- and diastereoselectivity. Of particular note is the effective chirality transfer from the sulfoxide center to the product alcohol in the SZ-cis-39C11-catalyzed conversion of (Z)-2-(4-methoxyphenyl)-but-2-en- 1-yl 4-nitrophenyl sulfoxide. These properties can be contrasted with those of bovine serum albumin (BSA) which accelerates the same reactions to a comparable extent but does not discriminate between substrate isomers. Partitioning of substrate from aqueous solution into the less polar environment of the protein pocket can account for much of the observed rate enhancement, whereas specific conformational constraints programmed by the haptens must orient the flexible substrate within the induced antibody- combining sites so as to favor certain reaction pathways over others. These studies thus expand the scope of antibody catalysis to an important new class of pericyclic reactions and illustrate how medium effects can be exploited together with conformational constraint to control reactivity and selectivity.

Full text of DOI:10.1021/jo991299a

8334
J . Org. Chem. 1999, 64, 8334-8341
An An tibod y-Ca ta lyzed Allylic Su lfoxid e-Su lfen a te
Rea r r a n gem en t
Zhaohui S. Zhou,† Alexander Flohr,† and Donald Hilvert*
Departments of Chemistry and Molecular Biology, The Skaggs Institute of Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La J olla, California 92037,
and Laboratory for Organic Chemistry, Swiss Federal Institute of Technology (ETH),
Universita¨tstrasse 16, CH-8092 Zurich, Switzerland
Received August 17, 1999
Antibodies SZ-cis-39C11 and SZ-trans-28F8, which were elicited in response to N-aryl-3-methoxy-
phenyl proline derivatives, catalyze the [2,3]-sigmatropic rearrangement of allylic sulfoxides to
sulfenates. Reduction of the sulfenates with dithiothreitol in situ yields allylic alcohols as the final
product. The antibodies achieve rate accelerations in the range 102-103 over background and exhibit
distinctive hapten-dependent substrate specificity and enantio- and diastereoselectivity. Of
particular note is the effective chirality transfer from the sulfoxide center to the product alcohol in
the SZ-cis-39C11-catalyzed conversion of (Z)-2-(4-methoxyphenyl)-but-2-en-1-yl 4-nitrophenyl sul-
foxide. These properties can be contrasted with those of bovine serum albumin (BSA) which
accelerates the same reactions to a comparable extent but does not discriminate between substrate
isomers. Partitioning of substrate from aqueous solution into the less polar environment of the
protein pocket can account for much of the observed rate enhancement, whereas specific
conformational constraints programmed by the haptens must orient the flexible substrate within
the induced antibody-combining sites so as to favor certain reaction pathways over others. These
studies thus expand the scope of antibody catalysis to an important new class of pericyclic reactions
and illustrate how medium effects can be exploited together with conformational constraint to control
reactivity and selectivity.
Enzymes that catalyze pericyclic reactions are rare in
binding energy to enhance rates and control selectivities.
Antibody catalysts have also been described for [2,3]-
sigmatropic eliminations of N-oxides9 and selenoxides.10
These reactions proceed via five- rather than six-
membered pericyclic transition states which are sub-
stantially less polar than their respective ground states.
Substituted tetrahydrofuran and pyrrolidine rings were
used to mimic these transition states, and preliminary
analysis9,10 suggests that the induced antibodies exploit
medium effects in addition to conformational constraint
to achieve rate accelerations of ∼103 over background.
In the case of the selenoxide elimination, these factors
are sufficient to overcome unfavorable eclipsing interac-
tions in the transition state and allow formation of a cis-
olefin (Scheme 1).10
We have begun to investigate various substrate modi-
fications as a means of mapping the stereoelectronic
properties of representative catalysts for the selenoxide
elimination. In the course of this work, we have found
that these antibodies also promote another [2,3]-sigma-
tropic process, the allylic sulfoxide-sulfenate (Evans-
Mislow) rearrangement (Scheme 2). Because sulfoxide
moieties are easily introduced and sulfenate esters
readily reduced, the Evans-Mislow rearrangement is
widely used for the synthesis of chiral allylic alcohols.11-13
nature1 but can be engineered for mechanistic and
synthetic studies via catalytic antibody technology. Ex-
amples of antibody-catalyzed Diels-Alder cycloadditions2
and Claisen3 and Cope4 rearrangements are notable in
this regard. Ongoing structural and mechanistic studies
on several of these catalysts5-8 are providing valuable
information on the ways in which proteins can exploit
* To whom correspondence should be addressed at the Swiss Federal
Institute of Technology. E-mail: hilvert@org.chem.ethz.ch.
† These authors contributed equally to this work.
(1) Pindur, U.; Schneider, G. H. Chem. Soc. Rev. 1994, 409-415.
(2) (a) Hilvert, D.; Hill, K. W.; Nared, K. D.; Auditor, M. T. M. J .
Am. Chem. Soc. 1989, 111, 9261-9262. (b) Braisted, A. C.; Schultz, P.
G. J . Am. Chem. Soc. 1990, 112, 7430-7431. (c) Gouverneur, V. E.;
Houk, K. N.; Pascal-Teresa, B.; Beno, K.; J anda, K. D.; Lerner, R. A.
Science 1993, 262, 204-208. (d) Meekel, A. A. P.; Resmini, M.; Pandit,
U. K. J . Chem. Soc., Chem. Commun. 1995, 571-572. (e) Yli-
Kauhaluoma, J . T.; Ashley, J . A.; Lo, C.-H.; Tucker, L.; Wolfe, M. M.;
J anda, K. D. J . Am. Chem. Soc. 1995, 117, 7041-7047. (f) Pitt, A. R.;
Stimson, W. H.; Suckling, C. J .; Marrero-Tellado, J . J .; Vazzana, C.
Isr. J . Chem. 1996, 36, 171-175.
(3) (a) Hilvert, D.; Carpenter, S. H.; Nared, K. D.; Auditor, M. T.
M. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 4953-4955. (b) Hilvert, D.;
Nared, K. D. J . Am. Chem. Soc. 1988, 110, 5593-5594. (c) J ackson,
D. Y.; J acobs, J . W.; Sugasawara, R.; Reich, S. H.; Bartlett, P. A.;
Schultz, P. G. J . Am. Chem. Soc. 1988, 110, 4841-4842.
(4) Braisted, A. C.; Schultz, P. G. J . Am. Chem. Soc. 1994, 116,
2211-2212.
(5) Haynes, M. R.; Stura, E. A.; Hilvert, D.; Wilson, I. A. Science
1994, 263, 646-652.
(6) Haynes, M. R.; Lenz, M.; Taussig, M. J .; Wilson, I. A.; Hilvert,
D. Isr. J . Chem. 1996, 36, 151-159.
(7) (a) Ulrich, H. D.; Mundorff, E.; Santarsiero, B. D.; Driggers, E.
M.; Stevens, R. C.; Schultz, P. G. Nature 1997, 389, 271-275. (b)
Driggers, E. M.; Cho, H. S.; Liu, C. W.; Katzka, C. P.; Braisted, A. C.;
Ulrich, H. D.; Wemmer, D. E.; Schultz, P. G. J . Am. Chem. Soc. 1998,
120, 1945-1958.
(8) (a) Romesberg, F. E.; Spiller, B.; Schultz, P. G.; Stevens, R. C.
Science 1998, 279, 1929-1933. (b) Heine, A.; Stura, E. A.; Yli-
Kauhaluoma, J . T.; Gao, C.; Deng, Q.; Beno, B. R.; Houk, K. N.; J anda,
K. D.; Wilson, I. A. Science 1998, 279, 1934-1940.
(9) Yoon, S. S.; Oei, Y.; Sweet, E.; Schultz, P. G. J . Am. Chem. Soc.
1996, 118, 11686-11687.
(10) Zhou, Z. S.; J iang, N.; Hilvert, D. J . Am. Chem. Soc. 1997, 119,
3623-3624.
(11) Evans, D. A.; Andrews, G. C. Acc. Chem. Res. 1974, 7, 147-
155.
(12) Hoffmann, R. W. Angew. Chem., Int. Ed. Engl. 1979, 18, 563-
572.
(13) Braverman, D., In The Chemistry of Sulphones and Sulphox-
ides; Patai, S., Rappoport, Z., Stirling, C. J . M., Eds.; J ohn Wiley:
London, 1988; Chapter 14.
10.1021/jo991299a CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/13/1999

Allylic Sulfoxide-Sulfenate Rearrangement
J . Org. Chem., Vol. 64, No. 22, 1999 8335
3.3H), 3.81 (s, 0.23H), 3.88 (d, 0.23H, 12.6 Hz), 3.99 (d, 1H,
12.8 Hz), 4.26 (d, 1H, 12.8 Hz), 5.71 (q, 0.23H, 7.0 Hz), 6.07
(q, 1H, 7.1 Hz), 6.81 (d, 2H, 8.9 Hz), 6.85 (d, 0.45H, 8.9 Hz),
7.10 (d, 0.45H, 8.9 Hz), 7.19 (d, 2H, 8.9 Hz), 7.71 (dd, 2.45H,
8.9 Hz), 8.25 (d, 2H, 8.9 Hz), 8.34 (d, 0.45H, 8.9 Hz). 13C NMR
(CDCl3, 125 MHz): δ 14.7, 15.4, 55.3, 59.3, 68.4, 77.7, 113.9,
114.0, 123.7, 123.9, 125.3, 125.5, 127.2, 128.7, 129.8, 130.2,
131.7, 133.0, 149.4, 151.0, 159.1. DEPT 135: CH-CH3, δ 14.7,
15.4, 55.3, 77.7, 113.9, 123.7, 123.9, 125.3, 125.5, 127.1, 129.8,
130.1, 131.6; CH2, δ 59.3, 68.4. UVmax (water): 252 nm (ꢀ )
16 100 M-1 cm-1). LRMS (FAB+, NBA): m/z 332 (M + H+).
HRMS: calcd for C17H18NO4S 332.0957, found 332.0951.
The product is a 4:1 mixture of the Z and E isomers as
shown by NOE difference measurements (CDCl3, 300 MHz).
Irradiation of dCH-Me results in large positive NOE's to C-2′,
C-5′ of the neighboring aromatic ring and to dCH-Me. Irradia-
tion of dCH-Me gives small positive NOEs to dCH-Me and to
CH2SO.
Sch em e 1
Sch em e 2
(E)-2-(4-Meth oxyp h en yl)bu t-2-en -1-ol (5c). Stille cou-
pling of (E)-2-(tri-n-butylstannyl)-2-buten-1-ol19 (12) and
4-iodoanisole using triphenylarsine, dipalladium tris(diben-
zylideneacetone), and copper iodide as the catalyst20 yielded
1
compound 5c as a colorless oil (35% yield). H NMR (CDCl3,
In contrast to selenoxides, which undergo rapid racem-
ization via achiral hydrates,14,15 sulfoxides are configu-
rationally stable, making possible an investigation of
antibody-controlled chirality transfer.
400 MHz): δ 1.64 (d, 3H, 6 Hz), 3.81 (s, 3H), 4.30 (d, 2H, 6
Hz), 5.78 (q, 1H, 6 Hz), 6.90 (dt, 2H, 9 Hz, 3 Hz), 7.16 (dt, 2H,
11 Hz, 3 Hz). 13C NMR (CDCl3, 125 MHz): δ 14.3, 55.1, 113.6,
123.0, 129.7, 133.5, 140.2, 158.5. UVmax (H2O): 200 nm, 240
nm. LRMS (FAB+, NBA/NaI): m/z 178 (M+). HRMS: calcd for
C
11H14O2 178.0990, found 178.0994.
3-(4-Meth oxyp h en yl)bu t-3-en -1-yl) 4-Nitr op h en yl Su l-
Exp er im en ta l Section
foxid e (3c). Compound 5c was converted to 3c in 40% yield
according to the protocol described for 3a . Neat sulfoxide 3c
rearranges within minutes to 3b via a [1,3]-shift. Because of
this rapid rearrangement, flash-chromatography fractions
containing product were diluted with acetonitrile and concen-
trated at 0 °C three times, taking care not to evaporate the
sample to dryness. The concentration of 3c in the final
Gen er a l Meth od s. 1H and 13C NMR spectra were recorded
on Bruker AC 250 and AMX 400 and 500 instruments. Mass
spectra were obtained on a VG ZAB-2VSe double-focusing
high-resolution mass spectrometer. HPLC analyses were
performed on Merck LiChrosorb RP18 10µ, LiChrospher 100
CH-18/2 (5µ), Rainin Microsorb-MV C18 300 Å and 86-203
(C-18) stationary phases using various acetonitrile in water
(0.05% TFA) gradients at elution rates of 1.0 mL min-1. For
all kinetic runs, a guard column filled with Whatman Pelli-
cular ODS (C-18) was added upstream of the analytical
column. Melting points are uncorrected.
1
acetonitrile solution was determined by integration of the H
NMR signals against an internal standard. HPLC and NMR
1
indicate a 1:1 mixture of diastereomers. H NMR (MeCN-d3,
400 MHz): δ 1.39 (d, 1.5H, 7 Hz), 1.49 (d, 1.5H, 7 Hz), 3.99 (q,
0.5H, 7 Hz), 4.35 (q, 0.5H, 7 Hz), 5.02 (s, 0.5H), 5.32 (s, 0.5H),
5.43 (s, 0.5H), 5.60 (s, 0.5H), 6.75 (dt, 1H, 9 Hz, 3 Hz), 6.86
(dt, 1H, 9 Hz, 3 Hz), 7.14 (dt, 1H, 9 Hz, 3 Hz), 7.28 (dt, 1H, 9
Hz, 3 Hz), 7.56 (dt, 1H, 9 Hz, 3 Hz), 7.80 (dt, 1H, 9 Hz, 3 Hz),
8.11 (dt, 1H, 9 Hz, 3 Hz), 8.26 (dt, 1H, 9 Hz, 3 Hz). 13C NMR
(MeCN-d3, 125 MHz): δ 12.4, 13.9, 55.9, 63.5, 65.5, 114.5,
114.8, 116.9, 117.1, 124.1, 124.8, 127.1, 127.2, 128.6, 128.7,
133.5, 142.4, 145.0, 149.9, 151.8, 160.6. UVmax (MeCN): 256
nm (ꢀ 12 300 M-1 cm-1). LRMS (FAB+, NBA/NaI): m/z 332
(M + H+). HRMS: calcd for C17H17NO4S 332.0957, found
332.0949.
The diastereomers were separated by preparative HPLC on
LiChrospher 100CH-18/2 (5 µm) using water/acetonitrile
gradients (without TFA). In acetonitrile at 5 °C, 3c does not
rearrange to 3b over the course of several weeks, but within
days, the separated diastereomers reequilibrate to the 1:1
mixture.
cis- a n d tr a n s-1-(4-Nitr op h en yl)-3-(4-m eth oxyp h en yl)-
p r olin e (cis-8 a n d tr a n s-8). A mixture of 3-(4-methoxyphe-
nyl)proline hydrochloride 721 (244 mg, 0.95 mmol), 1-fluoro-
4-nitrobenzene (0.53 mL, 5.0 mmol), and potassium carbonate
(848 mg, 8.0 mmol) in 3.5 mL of dry DMSO was heated at 120
°C for 23 h. The mixture was then poured into 50 mL of
saturated aqueous sodium carbonate solution and stirred at
room temperature for 1 h to hydrolyze the 4-nitrophenyl ester.
The orange-red solution was washed with ethyl acetate,
adjusted to pH 1.5 with concentrated aqueous HCl and then
extracted with diethyl ether. The bright yellow organic extract
2-(4-Meth oxyp h en yl)p r op -2-en -1-yl 4-Nitr op h en yl Su l-
foxid e (3a ). n-Butyllithium (2.0 M in hexane, 1.0 mL, 2.0
mmol) and 4-nitrobenzenesulfenyl chloride (398 mg, 2.1 mmol)
were added sequentially to a solution of 2-(4-methoxyphenyl)-
prop-2-en-1-ol16,17 (5a ) (328 mg, 2.0 mmol) in dry THF at 0
°C. After being stirred for 30 min, the reaction was diluted
with ethyl acetate, washed with aqueous phosphate buffer (400
mM, pH 6.8) and brine, dried with MgSO4, and concentrated
in vacuo. Flash chromatography on silica gel with hexane/ethyl
acetate 2:1 afforded 55% of the yellow solid 3a . Mp: 99.9-
100.7 °C. TLC: Rf 0.25 (hexane/ethyl acetate 2:1). 1H NMR
(CDCl3, 500 MHz): δ 3.80 (s, 3H), 3.91 (d, 1H, 12.8 Hz), 4.16
(d, 1H, 12.8 Hz), 5.02 (s, 1H), 5.49 (s, 1H), 6.84 (d, 2H, 8.9
Hz), 7.29 (d, 2H, 8.9 Hz), 7.72 (d, 2H, 8.9 Hz), 8.27 (d, 2H, 8.9
Hz). 13C NMR (CDCl3, 125 MHz): δ 55.2, 64.1, 113.9, 118.7,
123.6, 125.4, 127.1, 127.6, 130.5, 135.8, 149.3, 150.8, 159.6.
DEPT 135: CH-CH3, δ 113.9, 123.6, 125.4, 127.1; CH2, δ 64.1,
118.7. UVmax (water): 258.3 nm (ꢀ ) 10 430 M-1 cm-1). LRMS
(FAB+, NBA/NaI): m/z 318 (M + H+). HRMS: calcd for
C
16H15N4O 318.0800, found 318.0807.
(Z)-2-(4-Meth oxyph en yl)bu t-2-en -1-yl 4-Nitr oph en yl Su l-
foxid e (3b). This compound was obtained from 3-(4-methoxy-
phenyl)but-3-en-2-ol18 (5b) following the procedure described
for the preparation of 3a . Yellow solid (61%). Mp: 101 °C.
TLC: Rf 0.25 (hexane/ethyl acetate 3:2). 1H NMR (CDCl3, 500
MHz): δ 1.58 (d, 3H, 7.1 Hz), 1.69 (d, 0.7H, 7.0 Hz), 3.79 (ms,
(14) Davis, F. A.; Reddy, R. T. J . Org. Chem. 1992, 57, 2599-2606.
(15) Shimizu, T.; Yoshida, M.; Kobayashi, M. Bull. Chem. Soc. J pn.
1987, 60, 1555-1557.
(18) Cabri, W.; Candiani, I.; Bedeschi, A.; Santi, R. J . Org. Chem.
1992, 57, 3558-3563.
(16) Gassman, P. G.; Harrington, C. K. J . Org. Chem. 1984, 49,
2258-2273.
(19) Ensley, H. E.; Buescher, R. R. J . Org. Chem. 1982, 47, 404-
(17) Mickelson, T. J .; Koviach, J . L.; Forsyth, C. J . J . Org. Chem.
1996, 61, 9617-9620.
408.
(20) Flohr, A. Tetrahedron Lett. 1998, 39, 5177-5180.

8336 J . Org. Chem., Vol. 64, No. 22, 1999
Zhou et al.
was washed with brine, dried with MgSO4, and concentrated
in vacuo. Flash chromatography on silica gel (hexane/ethyl
acetate/HOAc 1:1:0.002) afforded cis-8 (47 mg, 15%) as a thick
yellow oil. Subsequent elution with more polar solvent (hexane/
ethyl acetate/HOAc 1:1:0.003) yielded trans-8 (114 mg, 35%)
as a thick yellow oil.
1H, J ) 8.5 Hz), 6.51 (d, 2H, J ) 8.9 Hz), 6.86 (d, 2H, J ) 8.7
Hz), 7.22 (d, 2H, J ) 8.7 Hz), 7.33 (d, 2H, J ) 8.9 Hz), 7.97
(m, 1H). 13C NMR (CDCl3, 125 MHz): δ -1.7, 17.0, 27.9, 29.6,
47.3, 47.8, 55.2, 62.9, 66.4, 111.7, 113.7, 122.3, 126.4, 129.0,
129.1, 144.4, 158.9, 162.9, 172.5. LRMS (FAB+, NBA) m/z 532/
524 (M+, C25H33BrN2O4Si calcd 532/534).
cis-8. Rf: 0.35 (hexane/ethyl acetate/HOAc 1:1:0.002). 1H
NMR (CDCl3, 500 MHz): δ 2.36 (m, 1H), 2.78 (m, 1H), 3.41
(m, 1H), 3.78 (s, 3H), 3.84 (m, 3H), 4.51 (d, 1H, J ) 8.5 Hz),
6.54 (d, 2H, J ) 9.2 Hz), 6.84 (d, 2H, J ) 8.7 Hz), 7.21 (d, 2H,
J ) 8.7 Hz), 8.14 (d, 2H, J ) 9.2 Hz). 13C NMR (CDCl3, 125
MHz): δ 47.3, 48.1, 55.2, 65.8, 110.8, 114.0, 126.3, 127.4, 128.9,
138.1, 150.8, 159.1, 174.9. LRMS (FAB+, NBA/NaI) m/z 343
(M + H+). HRMS: calcd for C18H18N2O5 343.1294, found
343.1290.
trans-10. Pale yellow oil (54%). TLC: Rf 0.3 (hexane/ethyl
acetate 2:1). H NMR (CDCl3, 500 MHz): δ 0.01 (s, 9H), 0.96
1
(m, 2H), 2.06 (m, 1H), 2.55 (m, 1H), 3.55 (m, 1H), 3.64 (m,
1H), 3.78 (s, 3H), 4.01 (s, 2H), 4.25 to 4.14 (m, 2H), 4.36 (d,
1H, J ) 3.5 Hz), 6.55 (d, 2H, J ) 8.9 Hz), 6.83 (d, 2H, J ) 8.7
Hz), 7.06 (d, 2H, J ) 8.7 Hz), 7.35 (d, 2H, J ) 8.9 Hz), 7.97
(m, 1H). 13C NMR (CDCl3, 125 MHz): δ -1.5, 17.4, 29.6, 32.3,
47.5, 48.5, 55.3, 63.5, 67.7, 112.2, 114.1, 122.3, 126.5, 127.7,
134.9, 144.2, 158.5, 163.0, 173.8. LRMS (FAB+, NBA): m/z 532/
524 (M+, C25H33BrN2O4Si calcd 532/534).
trans-8. Rf: 0.25 (hexane/ethyl acetate/HOAc 1:1:0.002). 1H
NMR (CDCl3, 500 MHz): δ 2.16 (m, 1H), 2.58 (m, 1H), 3.72
(m, 4H), 3.78 (s, 3H), 4.45 (d, 1H, J ) 3.4 Hz), 6.56 (d, 2H, J
) 9.2 Hz), 6.86 (d, 2H, J ) 8.7 Hz), 7.07 (d, 2H, J ) 8.7 Hz),
8.15 (d, 2H, J ) 9.3 Hz), 13C NMR (CDCl3, 125 MHz): δ, 20.6,
32.0, 47.9, 48.4, 55.3, 67.1, 111.3, 114.3, 126.3, 127.6, 133.5,
138.3, 150.7, 158.8, 177.3. DEPT 135: (CH-CH3): δ 48.4, 55.3,
67.1, 111.3, 114.4, 126.3, 127.5; (CH2): δ 32.0, 47.9. HRMS
(FAB, NBA/NaI) m/z 343.1290 (M + H+ C18H18N2O5 requires
343.1294).
cis- a n d tr a n s-1-(4-Nitr op h en yl)-3-(4-m eth oxyp h en yl)-
p r olin e Tr im eth ylsilyleth yl Ester (cis-9 a n d tr a n s-9).
Gen er a l P r oced u r e. 1,3-Dicyclohexylcarbodiimide (DCC)
(0.28 mmol) was added to a solution of the desired isomer of 8
(0.23 mmol), 2-trimethylsilylethanol (0.28 mmol), and 4-(dim-
ethylamino)pyridine (DMAP, 0.05 mmol) in 5.0 mL dry aceto-
nitrile at 0 °C. After being stirred at 4 °C for 48 h, the mixture
was filtered. The filtrate was diluted with 50 mL of ethyl
acetate, washed with 0.1 M aqueous HCl, aqueous NaHCO3,
and brine, dried with MgSO4, and concentrated in vacuo. Flash
chromatography on silica gel (hexane/ethyl acetate 5:1) af-
forded the desired ester.
cis-9. Pale yellow solid (50%). TLC: Rf 0.30 (hexane/ethyl
acetate 5:1). 1H NMR (CDCl3, 500 MHz): δ -0.07 (s, 9H), 0.47
(m, 1H), 0.61 (m, 1H), 2.33 (m, 1H), 2.83 (m, 1H), 3.62 (m,
1H), 3.72 (m, 1H), 3.90 to 3.78 (ms, 6H), 4.47 (d, 1H, J ) 8.5
Hz), 6.48 (d, 2H, J ) 9.0 Hz), 6.88 (d, 2H, J ) 8.7 Hz), 7.22 (d,
2H, J ) 8.7 Hz), 8.12 (d, 2H, J ) 9.3 Hz). 13C NMR (CDCl3,
125 MHz): δ -1.7, 17.0, 27.5, 47.2, 48.1, 55.2, 63.4, 66.2, 110.7,
113.8, 126.2, 127.9, 129.0, 137.7, 150.9, 159.1, 171.1.
trans-9: Brown solid (82%). Mp: 95.5-96.0 °C. TLC: Rf 0.30
(hexane/ethyl acetate 5:1). 1H NMR (CDCl3, 500 MHz): δ 0.02
(s, 9H), 0.97 (m, 2H), 2.15 (m, 1H), 2.57 (m, 1H), 3.68 (m, 4H),
3.79 (s, 3H), 4.24 (m, 2H), 4.39 (d, 1H, J ) 3.7 Hz), 6.53 (d,
2H, J ) 9.2 Hz), 6.86 (d, 2H, J ) 8.7 Hz), 7.07 (d, 2H, J ) 8.7
Hz), 8.15 (d, 2H, J ) 9.4 Hz). 13C NMR (CDCl3, 125 MHz): δ
-1.6, 17.5, 32.0, 47.9, 48.4, 55.3, 64.2, 67.7, 111.2, 114.3, 126.2,
127.6, 133.7, 138.0, 150.9, 158.7, 172.3. LRMS: (FAB+, NBA/
NaI): m/z 454 (M+). HRMS: calcd for C25H34N2O4Si 454.2288,
found 454.2274.
cis- a n d tr a n s-1-(4-[2-Br om oa cetyla m in o]p h en yl)-3-(4-
m eth oxyp h en yl)p r olin e Tr im eth ylsilyleth yl Ester (10).
Gen er a l P r oced u r e. A mixture of the desired isomer of 9
(26 mg, 0.06 mmol) and palladium on activated carbon (5%, 7
mg) in 3.0 mL of ethanol was stirred under hydrogen (1 atm)
at room temperature overnight. The reaction mixture was
filtered through Celite, and the filtrate was concentrated in
vacuo to give a colorless oil. The oil was dissolved in 5.0 mL of
dry CH2Cl2, and triethylamine (41 µL, 0.3 mmol) was added.
The mixture was then cooled to 0 °C before bromoacetyl
bromide (16 µL, 0.18 mmol) was added. After being stirred for
an additional 10 min, the reaction mixture was diluted with
30 mL of ethyl acetate, washed with 0.05 M aqueous HCl, 5%
aqueous NaHCO3, and brine, dried with MgSO4, and concen-
trated in vacuo. Flash chromatography on silica gel afforded
the desired amide.
Cou p lin g of Ha p ten s cis-6 a n d tr a n s-6 to Ca r r ier
P r otein s. The haptens were deprotected immediately prior
to coupling. Approximately 5 mg of cis-10 or trans-10 were
dissolved in 500 µL of neat TFA. After 1 h at room tempera-
ture, the reaction was evaporated to dryness, and crude 6 was
redissolved in 200 µL of DMF. The carrier proteins bovine
serum albumin (BSA) and thyroglobulin (TG) were modified
with 2-iminothiolane (Traut’s reagent) in deoxygenated phos-
phate-buffered saline solution (PBS, 10 mM sodium phosphate,
100 mM NaCl, pH 8.0, containing 1.0 mM EDTA) to convert
surface amine residues to thiols. The modified proteins were
purified by gel filtration on a Sephadex G-50 column eluted
with PBS and then immediately reacted with the DMF
solutions of hapten trans-6 or cis-6 under an atmosphere of
argon at room temperature for 2 h. The ratio of haptens to
thiols in the reaction mixtures ranged from 10:1 to 35:1.
The carrier protein-hapten conjugates were purified by gel
filtration on a G-50 Sephadex column eluted with PBS. Protein
concentration was determined by the BCA method,22 and
thiol concentration was determined by titration with 5,5′-
dithiobis(2-nitrobenzoic acid).23 The epitope density was 8.5
(BSA) and 9.8 (TG) for cis-6 and 8.3 (BSA) and 26.0 (TG) for
trans-6.
P r ep a r a tion a n d P u r ifica tion of Mon oclon a l An tibod -
ies. Mice (129 GIX+ strain) were immunized with the thyro-
globulin conjugates of cis-6 and trans-6. Serum titer was
checked with the corresponding BSA conjugates by enzyme-
linked immunosorbent assay (ELISA).24 Hybridomas were
obtained by fusion of spleen cells with SP2/0+ myeloma cells
according to standard protocols.25 High affinity anti-cis-6 and
anti-trans-6 antibodies were identified by screening tissue
culture supernatants by ELISA. The corresponding hybrido-
mas were subcloned twice and propagated as mouse ascites.
Monoclonal antibodies were obtained from ascites fluid by
ammonium sulfate precipitation, followed by chromatography
on DEAE-Sephacel using NaCl gradients (0 to 500 mM) in 10
mM Tris, pH 8.0. Antibodies were further purified by FPLC
on Protein G Sepharose followed by Mono Q HR 10/10.
Antibody concentrations were determined by absorbance at
280 nm (l mg mL-1 ) 1.37 OD, assuming a molecular weight
of 160 000 for IgG).
Kin etic Assa ys. Assays were performed in 100 mM Tris-
HCl and 100 mM NaCl (pH 8) in thermostated quartz cuvettes
at 20 °C. The substrate sulfoxide (3) was incubated with or
without antibody, and upon addition of DTT, the production
of 4-nitrothiophenol was monitored either spectrophotometri-
cally at 410 nm or by reversed-phase HPLC with p-methoxy-
acetophenone as an internal standard. Initial rates were
determined from the linear portion of the time course (typically
5-10% conversion) and were unaffected by the order of
(21) Chung, J . Y. L.; Wasicak, J . T.; Arnold, W. A.; May, C. S.;
Nadzan, A. M.; Holladay, M. W. J . Org. Chem. 1990, 55, 270-275.
(22) Walker, J . M. Methods Mol. Biol. 1994, 32, 5-8.
(23) Riddles, P. W.; Blakeley, R. L.; Zerner, B. Methods Enzymol.
1983, 91, 49-60.
(24) Clark, B. R.; Engvall, E. Enzyme-Immunoassay; CRC Press:
Boca Raton, Florida, 1980.
(25) Harlow, E.; Lane, D. Antibodies: A Laboratory Manual; Cold
Spring Harbor Lab: New York, 1988.
cis-10. Pale yellow oil (34%). TLC: Rf 0.3 (hexane/ethyl
acetate 3:1). 1H NMR (CDCl3, 500 MHz): δ -0.07 (s, 9H), 0.51
(m, 1H), 0.60 (m, 1H), 2.27 (m, 1H), 2.77 (m, 1H), 3.51 (m,
1H), 3.80 to 3.64 (m, 7H), 3.87 (m, 1H), 4.01 (s, 2H), 4.36 (d,

Allylic Sulfoxide-Sulfenate Rearrangement
J . Org. Chem., Vol. 64, No. 22, 1999 8337
addition of the reagents. p-Methoxyacetophenone does not act
as an inhibitor or activator of the antibodies, nor is it converted
under the assay conditions. To account for substrate depletion
during assays, the data were fit to the equation vo ) [(kcat/2)
× [(E + S + Km) - [(E + S + Km)2 - 4ES]1/2]], where vo is the
initial rate, kcat and Km are the steady-state kinetic parameters,
and E and S are the total concentrations of antibody binding
site and substrate, respectively.26 All reaction products were
identified by HPLC analysis by co-injection of authentic
samples under at least two different separation conditions. The
stoichiometry of the reaction was similarly determined by
monitoring the total time course of the reaction spectropho-
tometrically or by HPLC.
Sch em e 3
In h ibition Stu d ies. Inhibition studies were performed
under standard assay conditions in the presence of increasing
concentrations of the hapten analogues cis-8 and trans-8. The
ratio of inhibitor to antibody binding site was varied between
0.25 and 4.0. Control experiments with inhibitor in the absence
of antibodies showed no affect of the course of reaction.
Deter m in a tion of En a n tioselectivity. Substrate 3b (470
µM) was incubated at 21 °C with antibody (15 µM) and DTT
(645 µM) in 640 µL of buffer solution (100 mM Tris-HCl, 100
mM NaCl, pH 8) for 80 min. The reaction was quenched by
addition of 50 mM iodoacetic acid sodium salt. After being
stirred for an additional 30 min, the aqueous solution was
extracted with ethyl acetate. Solvent was then removed with
a stream of nitrogen and the crude product redissolved in
hexanes/2-propanol (80:20). The enantiomers of the product
alcohol and the unreacted substrate sulfoxide were then
separated by HPLC on a Chiralcel OD-H column eluted
isocratically at 0.5 mL min-1 with hexanes/2-propanol (85/15).
The product enantiomers (5b) eluted with retention times of
11.0 and 11.9 min and the substrate enantiomers (3b) at 62.4
and 65.6 min, respectively.
Sch em e 4
Resu lts
Syn th esis. The pyrrolidine derivative 6 was originally
designed to mimic the five-membered transition state for
the [2,3]-sigmatropic elimination of selenoxides illus-
trated in Scheme 1.10 Antibodies elicited in response to
this compound were expected to provide a low dielectric
environment capable of constraining the flexible sub-
strate in a reactive conformation and desolvating the
polar selenoxide moiety. Like the selenoxide elimination,
the Evans-Mislow rearrangement of allylic sulfoxides to
sulfenate esters proceeds via a five-membered transition
state that is less polar then the starting sulfoxide.11,27,28
Although the precise steric demands of the two reactions
differ, it seemed conceivable that antibodies generated
against 6 might also promote the reaction shown in
Scheme 2.
Hapten 6 was synthesized as shown in Scheme 3. 3-(4-
Methoxy)proline 7 was prepared by the methods of
Chung et al.21 and Cox et al.29 and alkylated at nitrogen
with 1-fluoro-4-nitrobenzene to give 8 as a 1:2 mixture
of diastereomers. The diastereomers were separated by
silica gel chromatography and elaborated separately.
Protection of the carboxylic acid as a trimethylsilylethyl
ester, hydrogenation of the nitro group, and acylation of
the resulting amine with bromoacetyl bromide yielded
protected haptens cis-10 and trans-10. Deprotection with
TFA afforded cis-6 and trans-6. The activated bromides
were coupled to bovine serum albumin (BSA) and thy-
roglobulin (TG) that had been previously derivatized with
2-iminothiolane.
All sulfoxide substrates were prepared from the cor-
responding sulfenate esters. Sulfenate esters 4a and 4b
were obtained by reacting allylic alcohols 5a 16,17 and 5b18
with n-butyllithium and 4-nitrobenzenesulfenyl chloride
in THF (Scheme 4). They rearranged spontaneously in
solution to afford 3a and 3b. Compound 3b was obtained
as a 4:1 mixture of diastereomers that were not separable
by HPLC. The major diastereomer was identified by
difference NOE-measurements to be the cis isomer (as
shown in Scheme 4).
Sulfoxide 3c was synthesized as shown in Scheme 5.
Stille coupling of (E)-2-(tri-n-butylstannyl)-2-buten-1-ol
(26) Segel, I. H. Enzyme Kinetics: Behavior and Analysis of Rapid
Equilibrium and Steady-State Enzyme Systems; Wiley: New York,
1975; pp 72-74.
(27) Bickart, P.; Carson, F. W.; J acobus, J .; Miller, E. G.; Mislow,
K. J . Am. Chem. Soc. 1968, 90, 4869-4876.
(29) Cox, D. A.; J ohnson, A. W.; Mauger, A. B. J . Chem. Soc. 1964,
5024-5029.
(28) Tang, R.; Mislow, K. J . Am. Chem. Soc. 1970, 92, 2100-2104.

8338 J . Org. Chem., Vol. 64, No. 22, 1999
Zhou et al.
to catalyze the selenoxide elimination reaction.10 None
of the other antibodies showed detectable activity. The
most active catalyst, SZ-cis-39C11, and SZ-trans-28F8
were investigated in greater detail.
Sch em e 5
The antibody-catalyzed reactions are competitively
inhibited by the corresponding transition state analogues.
As reported previously,10 SZ-cis-39C11 recognizes only
one enantiomer of the hapten analogue cis-8, as shown
by the 2:1 stoichiometry of inhibition (per binding site)
by racemic hapten. In contrast, SZ-trans-28F8 binds both
enantiomers of trans-8 equally well. The rate of the
catalyzed reaction is independent of the DTT concentra-
tions used for the assay (>0.3 mM), showing that the
rate-limiting step proceeds within the induced binding
pocket.
(12)19 with 4-iodoanisole20 yielded the Z-configured allylic
alcohol 5c. Treatment of the latter with n-butyllithium
and 4-nitrobenzenesulfenyl chloride afforded sulfenate
4c, which spontaneously rearranged to sulfoxide 3c.
Compound 3c could not be handled in the absence of
solvent because it rapidly undergoes a [1,3]-allyl shift in
the solid phase to give 3b as a 2:1 mixture of diastere-
omers. R,R-Disubstituted allylic sulfoxides undergo an
analogous [1,3]-allyl shift,11 presumably due to steric
crowding near the sulfur atom. Davis et al.14 have also
reported large rate accelerations for racemization of
selenoxides in the solid state which they attributed to
association of the polar Se-O bond moieties. In solution,
however, 3c is stable for weeks at 5 °C. HPLC and NMR
indicate that it is present as a 1:1 mixture of diastereo-
mers. Although the diastereomers could be separated by
reversed-phase HPLC, the individual isomers reequili-
brate to the 1:1 diastereomeric mixture within days.
Allylic 4-nitrobenzenesulfoxide 13 was used to evaluate
the importance of the methoxy-substituted aromatic ring.
This compound was synthesized as described by Tang
and Mislow.28
Steady-state kinetic parameters were determined from
plots of initial rates versus substrate concentration and
are presented in Table 1. The low Km values and the
sensitivity of the antibodies to changes in aromatic
substitution pattern (data not shown) suggest that the
aryl rings contribute significantly to binding. Compound
13, lacking the p-methoxy-substituted aromatic ring, is
not a substrate for either antibody, suggesting that the
aromatic rings help to preorganize the pericyclic system
for reaction. Rate enhancements over the corresponding
uncatalyzed reactions (kcat/kuncat) are in the range 102-
103 as is typical of other antibody-catalyzed sigmatropic
rearrangements.3,4,9,10 Catalytic proficiency [(kcat/Km)/kun
-
cat] provides a measure of the antibody’s affinity for the
rate-limiting transition state,30 and the values of 106-
107 M-1 obtained here are also typical for these systems.
The chirality of the sulfoxide substrates provides a
means of assessing the degree of stereocontrol achieved
by the antibodies. In the case of racemic 3a , the rear-
rangement is promoted with comparable efficiency but
different enantioselectivity by the two catalysts. Both
enantiomers are converted to product 5a in the presence
of DTT and antibody SZ-cis-39C11, albeit at somewhat
different rates, but only one is a substrate for SZ-trans-
28F8 (Figure 1A). The differences in selectivity are
magnified with methyl-containing substrates. Thus, com-
pound 3b, which has a γ-methyl group, is the poorest
substrate for SZ-trans-28F8, not being detectibly con-
verted to product. It is the best substrate for SZ-cis-
39C11, however, subject to a >800-fold acceleration over
background. SZ-cis-39C11 not only recognizes the chiral-
ity of the sulfoxide with 80:20 selectivity, but produces
alcohol 5b in an enantiomeric ratio of 70:30. The observed
enantiomeric excesses presumably represent lower limits
on the antibody’s true selectivity, given that the substrate
is a 4:1 mixture of cis and trans isomers. The uncatalyzed
background reaction and some racemization during
workup may have further eroded the measured enantio-
selectivities.
An tibod y Ca ta lysis. The TG conjugates of cis-6 and
trans-6 were used separately to immunize mice, and
monoclonal antibodies were prepared in the usual way.25
Screening of tissue culture supernatants by ELISA with
the corresponding BSA-hapten conjugates yielded 20
antibodies with high affinity for trans-6 and 28 antibodies
with high affinity for cis-6. Hybridomas were subcloned
twice and propagated in mouse ascites. Antibodies were
purified by ammonium sulfate precipitation and chro-
matography on DEAE-Sepharose, Protein G, and Mono
Q ion exchange columns.
Although the interconversion of allylic sulfoxides 3 and
sulfenate esters 4 is reversible, with the equilibrium
greatly favoring the sulfoxides in aqueous solution, the
sulfenate esters can be trapped with dithiotreitol (DTT)
in situ, shifting the equilibrium to the right and releasing
product alcohol and an easily detected aryl thiol. All 48
antibodies were consequently screened individually for
the ability to accelerate the formation of 4-nitrothiophe-
nol from sulfoxide 3a in the presence of DTT (Scheme
2). Product formation was monitored at 410 nm. Under
the assay conditions neither antibody precipitation
nor sulfoxide reduction was observed. Only three anti-
bodies showed significant rate enhancements over back-
ground: SZ-cis-39C11, SZ-trans-28F8, and SZ-cis-42F7.
These are the same antibodies that were previously found
The diastereoselectivity of the antibodies was investi-
gated with compound 3c, which contains two chiral
centers. UV and HPLC assays show that all isomers react
at roughly comparable rates in the absence of catalyst
but only one pair of enantiomers is accepted by SZ-trans-
28F8 and SZ-cis-39C11 (Figure 1B). At low substrate
concentrations, the antibodies further discriminate be-
tween the two possible enantiomers of the reactive
diastereomer, processing one more than an order of
magnitude faster than its antipode. At high substrate
(30) Radzicka, A.; Wolfenden, R. Science 1995, 267, 90-93.

Allylic Sulfoxide-Sulfenate Rearrangement
J . Org. Chem., Vol. 64, No. 22, 1999 8339
Ta ble 1. Kin etic P a r a m eter s for th e An tibod y-Ca ta lyzed Eva n s-Mislow Rea r r a n gem en t of Allylic Su lfoxid es 3a
kcat/Km
(kcat/Km)/kuncat
(M-1
catalyst
substrate
kcat (min-1
)
Km (µM)
(M-1 min-1
)
kcat/kuncat
)
SZ-cis-39C11
3a (R1 ) H, R3 ) H)
3b (R1 ) H, R3 ) Me)
3c (R1 ) Me, R3 ) H)
3a (R1 ) H, R3 ) H)
3b (R1 ) H, R3 ) Me)
3c (R1 ) Me, R3 ) H)
3a (R1 ) H, R3 ) H)
3b (R1 ) H, R3 ) Me)
3c (R1 ) Me, R3 ) H)
0.0076
0.14
0.027
0.0040
ND
20
28
10
<1
ND
3.3
>700
470
230
380
5100
2700
>4000
ND
160
820
47
7.9 × 106
3.0 × 107
4.7 × 106
>8.3 × 107
4.7 × 107
1.7 × 105
2.8 × 105
7.4 × 105
SZ-trans-28F8
80
0.090
27 000
8.0
160
BSA
0.022
0.10
47
430
130
170
a
The reactions were assayed at 20 °C in 100mM Tris-HCl, 100 mM NaCl (pH 8) in the presence of DTT. ND: no activity detected.
Discu ssion
Mech a n ism . A reversible pericyclic mechanism has
been proposed for the interconversion of allylic sulfoxides
and sulfenate esters.11,27,28 The entropies of activation for
the forward and reverse reactions are in the range -0.7
to -10 eu,28 consistent with a conformationally restricted
five-membered cyclic transition state. Electron withdraw-
ing substituents greatly accelerate the rearrangement of
allylic aryl sulfoxides, as do nonpolar solvents.28 In
contrast, the reverse process starting from the sulfenate
ester is relatively insensitive to solvent change. These
results, together with recent calculations,32 suggest that
the transition state more closely resembles the less polar
sulfenate ester in its structure and solvation properties
than the polar sulfoxide.
F igu r e 1. (A) Total time course of the antibody-catalyzed and
uncatalyzed rearrangement of allylic sulfoxide 3a . The reac-
tions were performed at 20 °C in 100 mM Tris-HCl and 100
mM NaCl (pH 8) in the presence of 0.5 mM DTT ([3a ] ) 60
µM; [SZ-trans-28F8] ) 5 µM; [SZ-cis-39C11] ) 5 µM) and
monitored at 410 nm. After 800 min, SZ-cis-39C11 was added
to the SZ-trans-28F8 reaction to convert the remaining sub-
strate to product. (B) HPLC traces monitoring the SZ-cis-
39C11-catalyzed conversion of 3c ([3c] ) 14 µM; [SZ-trans-
28F8] ) 10 µM). The diastereomers of 3c were separated on
Rainin Microsorb-MV 86-203 (C-18, 4.6 mm i.d.) eluted with
acetonitrile/water (0.05% TFA) (20/80 ratio for 10 min, followed
by a linear gradient over 5 min to a 50/50 ratio that was then
maintained for 45 min). The two diastereomers eluted with
retention times of 23.6 and 24.0 min, respectively. Top, t ) 0
min; middle, t ) 15 min (25% conversion of substrate); bottom,
t ) 110 min (50% conversion of substrate). Acetophenone was
used as an internal standard.
Consistent with these conclusions, catalysis of the [2,3]-
sigmatropic rearrangement of allylic sulfoxides 3 by
antibodies SZ-trans-28F8 and SZ-cis-39C11 and BSA can
be largely ascribed to a simple medium effect. Partition-
ing of properly configured substrates from aqueous
solution into the relatively apolar environment of the
antibody or BSA active sites would shift the internal
equilibrium in favor of the sulfenate ester 4, which can
then be trapped by DTT either within the binding pocket
or upon dissociation into free solution. The rate accelera-
tions we observe are comparable to those seen previously
for the medium-sensitive selenoxide eliminations also
catalyzed by these proteins.10 The latter reaction is,
however, considerably more sensitive to solvent change
than the rearrangement of allylic sulfoxides as judged
by correlations of reaction rate with the solvent polarity
parameter ET. For example, the selenoxide elimination
is accelerated by a factor of 40 000 in going from aqueous
buffer (ET ) 65) to cyclohexane (ET ) 30),10 whereas the
rate of the allylic sulfoxide-sulfenate rearrangement is
enhanced only by a factor of about 100.28,33
The importance of a medium effect is supported by the
similarity in the rate accelerations (kcat/kun) achieved by
the antibodies and BSA for most of the reactions exam-
ined (Table 1). The conversion of 3b to 4b by SZ-cis-
39C11 appears to be an exception to this general state-
ment, as the rate enhancement in the presence of this
antibody is nearly an order of magnitude faster than the
comparable reaction with BSA. The γ-methyl substituent
in 3b, which lacks a counterpart in the hapten, presum-
ably makes serendipitous but favorable interactions
concentrations both enantiomers are converted to prod-
uct, suggesting a higher Km for the less reactive species
and thus poorer binding as the basis for discrimination.
The nonreacting diastereomer is a competitive inhibitor,
lowering the catalytic efficiency of the antibodies when
present. Accordingly, binding alone appears to be insuf-
ficient for catalysis. This conclusion is further strength-
ened by the observation that the large majority of anti-6
antibodies lack any detectable catalytic activity.
BSA Ca ta lysis. As seen for other medium-sensitive
transformations,10,31 the Evans-Mislow rearrangement is
also catalyzed by BSA, albeit with lower efficiencies (for
3a and 3b) and substantially higher Km values than with
the antibodies. These results suggest weaker and less
specific binding of substrate. Consistent with this inter-
pretation, BSA does not differentiate between the enan-
tiomers of 3a or 3b and product 5b is generated in
racemic form. BSA also converts both diastereomers of
3c to the corresponding allylic alcohols.
(32) J ones-Hertzog, D. K.; J orgensen, W. L. J . Am. Chem. Soc. 1995,
117, 9077-9078.
(33) Preliminary investigations of the uncatalyzed rearrangement
of compound 3b in a limited range of solvents shows that it and the
substituted allylic phenyl sulfoxides studied by Tang and Mislow28 are
comparably sensitive to solvent change.
(31) (a) Kikuchi, K.; Thorn, S. N.; Hilvert, D. J . Am. Chem. Soc.
1996, 118, 8184-8185. (b) Hollfelder, F.; Kirby, A. J .; Tawfik, D. S.
Nature 1996, 383, 60-63.

8340 J . Org. Chem., Vol. 64, No. 22, 1999
Zhou et al.
Sch em e 6
Sch em e 8
Sch em e 7
Sch em e 9
substituent. In addition to differentiating between the
two substrate enantiomers, SZ-cis-39C11 apparently
stablizes one of these transition states selectively. In
contrast, the BSA-catalyzed reaction is nonselective,
suggesting a rather loose, nonspecific interaction with
substrate. At the other extreme, SZ-trans-28F8 does not
catalyze the rearrangement of 3b at all, presumably
because steric clashes between the γ-methyl group and
the protein pocket prevent the flexible substrate from
adopting a conformation suitable for reaction.
within the SZ-cis-39C11 binding pocket that contribute
to selective transition state stabilization. The active site
of SZ-trans-28F8 is less tolerant of substitution at the
γ-position of substrate since the rearrangement of 3b is
not detectably catalyzed by this antibody.
Selectivity. The low Km values and the fact that
compound 13 is not a substrate for SZ-trans-28F8 or SZ-
cis-39C11 indicate that the aryl rings contribute signifi-
cantly to productive binding. They are also clearly
important for imparting selectivity and catalytic profi-
ciency to the antibody-catalyzed reactions. Although the
rate accelerations for the antibody and BSA-catalyzed
reactions (kcat/kuncat) are comparable, the catalytic profi-
ciencies of the former are 50-500-fold larger. This
difference is manifest most dramatically in the contrast-
ing stereoselectivities of the antibody and BSA-catalyzed
conversions. All the antibody reactions, with the excep-
tion of the rearrangement of 3a promoted by SZ-cis-
39C11, are highly diastereo- and/or enantioselective.
BSA, on the other hand, discriminates poorly, if at all,
between the stereoisomers of substrates 3a -c. Interest-
ingly, SZ-trans-28F8 is considerably more selective than
might have been expected based on its ability to bind both
enantiomers of trans-8 and its lack of selectivity toward
chiral selenoxides;10 selectivity in the case of SZ-cis-
39C11 appears to depend critically on the presence of an
R- or γ-substituent.
A complete understanding of the origins of the observed
selectivities, and how they correlate to hapten recogni-
tion, will have to await detailed structural studies on
these catalysts. In principle, the [2,3]-sigmatropic rear-
rangement can occur either via an endo or an exo
transition state as shown in Scheme 6 for substrate 3a .
Endo transition states are generally favored due to their
completely staggered conformations,12,13,27 but the pres-
ence of the â-aryl moiety and the R- or γ-methyl groups
as in 3b and 3c complicates matters due to competing
1,2- and 1,3-pseudodiaxial interactions.
For 3c, we have found that the product of the catalyzed
and uncatalyzed reactions is exclusively trans-allylic
alcohol 5c. Since the different 3c diastereomers react at
similar rates in the uncatalyzed reaction (as judged by
HPLC), this product must be formed via both endo and
exo transition states of similar energy. Scheme 8 il-
lustrates this with two respresentative diastereomers
(note that the exo-(S,R) and endo-(S,S) conformers of 3c
would give rise to the cis-allylic alcohol). We assume that
the R-methyl substituent will adopt a pseudoequatorial
orientation in the accessible endo and exo transition
states to avoid a very unfavorable 1,2-diaxial interaction
with the aryl group on the sulfoxide. In the antibody
catalyzed reaction, only one pair of enantiomers (the
same for both antibodies) is recognized and converted to
product, again indicating effective discrimination be-
tween competing transition states. The fact that the
antibodies catalyze the rearrangement of both enanti-
omers of this isomer, albeit at significantly different
rates, allows the prediction to be made that the preferred
substrate has the (S,R) or (R,S) configuration. The two
enantiomeric endo transition states of this diastereomer
are more similar to one another, because of the pseudo
plane of symmetry bisecting the molecule, than the two
enantiomeric exo transition states of the other diastere-
omer (Scheme 8). If this hypothesis can be verified, it
would suggest that an endo transition state may also be
preferred in the antibody-catalyzed conversions of 3a and
3b.
Although SZ-cis-39C11 does not discriminate between
the enantiomers of 3a , the successful transfer of chirality
from sulfur to carbon in the rearrangement of sulfoxide
3b shows that this antibody can distinguish differently
shaped transition states. As depicted in Scheme 7, only
modest stereoselectivity is expected in the uncatalyzed
reaction of 3b because both substrate enantiomers can
yield the same product enantiomer via competing exo and
endo transition states, both of which have unfavorable
1,3-pseudodiaxial interactions between the aryl group on
the sulfoxide and either the â-aryl or the γ-methyl
Although the selenoxide elimination (Scheme 1) and
the allylic sulfoxide-sulfenate rearrangement (Scheme
2) share several common features, the corresponding
transition states are subtly different in their steric
demands as a consequence of the differing hybridization
state of the carbon â to the selenoxide/sulfoxide (sp3 in
the selenoxides, sp2 in the sulfoxides). For selenoxide 14
(Scheme 9), which is analogous to sulfoxide 3c, minimi-
zation of eclipsing interactions between the R-methyl and
â-aryl substituents favors formation of the trans-olefin
product. In the presence of SZ-cis-39C11, this intrinsic

Allylic Sulfoxide-Sulfenate Rearrangement
J . Org. Chem., Vol. 64, No. 22, 1999 8341
preference can be partially overcome to yield a 45:55
mixture of cis- and trans-olefins, as programmed by the
cis-hapten.10 Because only 50% of the substrate is con-
verted to product, this antibody appears to be extremely
sensitive to the stereochemical configuration at C-R (the
selenoxide moiety epimerizes rapidly under the experi-
mental conditions) in line with the results obtained here
for the allylic sulfoxide rearrangement of substrate 3c.
The fact that some trans product is also formed suggests
that alternative productive binding modes for substrate
are accessible, however. Like antibody DB3, which
recognizes a series of conformationally distinct steroids
with high affinity, SZ-cis-39C11 may have a bifurcated
binding pocket. In contrast, SZ-trans-28F8 catalyzes the
elimination of both epimers of 14 and yields only trans
product.10 The greater selectivity observed for the allylic
sulfoxide-sulfenate rearrangement suggests that sulfox-
ides and selenoxides may bind in quite different ways to
the SZ-trans-28F8 antibody.
Com p a r ison w ith Oth er An tibod ies. The X-ray
structures of antibodies 1F75 and AZ-28,7 which catalyze
the Claisen rearrangement of chorismate and the oxy-
Cope rearrangement of a 2,5-diaryl-3-hydroxyhexadiene,
respectively, have been solved. These structures show
how hydrogen bonding, electrostatic, and van der Waals
interactions can be exploited within an antibody-com-
bining site to promote sigmatropic rearrangements.
Antibody AZ-28 was elicited with a 2,5-diarylcyclohex-
anol derivative, and as we observe for the antibodies
reported here, the aryl groups of the ligand appear to be
key epitopes for immune recognition and catalysis.4,7 The
hapten binds to the immunoglobulin in a deep cylindrical
pocket. Its 5-phenyl substituent sits at the bottom of the
pocket tightly packed by the side chains of several
aromatic and hydrophobic residues. The cyclohexyl ring
adopts the expected chair conformation with the hydroxyl
and aryl moieties in equatorial orientations. The hydroxyl
group is hydrogen bonded to the NH backbone and the
imidazole side chain of HisH96. This same imidazole also
packs against the 2-phenyl substituent, which is located
near the entrance of the cavity, through π-stacking
interactions.
for by somatic mutation of active site residue SerL34 in
the germline antibody to an asparagine in AZ-28.7 This
residue influences the conformation of the CDR H3 loop,
which in turn orients the 2-phenyl substituent of the
ligand.
By analogy with AZ-28, we expect that the conforma-
tions adopted by the flexible selenoxides 1 and sulfoxides
3 at the active sites of SZ-trans-28F8 and SZ-cis-39C11
will be determined primarily by extensive van der Waals
interactions with the aryl groups of the ligand. While
such interactions must be responsible for the observed
selectivities, they may limit catalytic efficiency by locking
the flexible substrates into a suboptimal ground state
conformation (as is apparently the case for AZ-28). This
may simply reflect imperfect mimicry of the transition
states for the corresponding [2,3]-sigmatropic rearrange-
ments by the haptens. Interestingly, unanticipated but
favorable interactions between the γ-methyl group in
compound 3b, which lacks a counterpart in hapten cis-
6, and the SZ-cis-39C11 active site apparently compen-
sate for deficiencies in hapten design to some extent and
catalytic effects in excess of the simple medium effect are
achieved.
Theoretical investigations of the allylic sulfoxide-
sulfenate rearrangement have noted the accelerating
effects of a hydrogen bond to the oxygen of the polarized
S-O bond in the transition state.32 The proline ring of
haptens cis-6 and trans-6 lacks a suitable counterpart
to this bond, so it is unlikely that the active sites of SZ-
trans-28F8 and SZ-cis-39C11 possess an analogue to
HisL96 in AZ-28 capable of hydrogen bonding with the
sulfoxide oxygen. It will be interesting to see whether,
with proper instruction, such an interaction might be
deliberately exploited within the antibody combining site
to further augment catalytic efficiency.
P er sp ective. Catalysis of the Evans-Mislow rear-
rangement extends the repertoire of antibody-catalyzed
reactions to a new pericyclic reaction for which natural
enzymes are unknown. The experiments presented here
show how conformational constraint imposed by an
antibody combining site can be exploited in combination
with a simple medium effect to achieve significant rate
accelerations and selectivities. Although these effects
exceed those attained by serum albumins, significant
improvements in hapten design and screening will be
necessary to identify catalysts with truly enzyme-like
properties.
Although hydrogen bonding to the hydroxyl group of
the substrate undoubtedly helps to position the flexible
substrate within the AZ-28 binding pocket and may
contribute to the 5000-fold rate acceleration in the
ensuing oxy-Cope rearrangement, the extensive van der
Waals contacts with the C-2 and C-5 aryl substituents
play a critical role in aligning the [4π + 2σ] orbitals of
the hexadiene for reaction.7 Because the C-2 and C-5
atoms of the substrate are sp2 rather than sp3 hybridized
as in the hapten, the ground-state conformation of the
substrate is unlikely to be optimal for reaction, however.
In fact, the germline precursor of AZ-28 binds the
transition state analogue 40-times less tightly but cata-
lyzes the reaction with 30-fold greater efficiency than the
mature antibody. These effects can be largely accounted
Ack n ow led gm en t. We thank Ning J iang for prepa-
ration and purification of the antibodies and Dr. G.
Schlingloff for help in determining enantiomeric ex-
cesses. This work was supported by the National
Institutes of Health (GM38273 to D.H.), Novartis Phar-
ma, and a fellowship of the scientific board of NATO by
the DAAD (to A.F.).
J O991299A