Proc. Natl. Acad. Sci. USA
Vol. 94, pp. 11773–11776, October 1997
Chemistry
On roads not taken in the evolution of protein catalysts: Antibody
steroid isomerases that use an enamine mechanism
CHAO-HSIUNG LIN, TIMOTHY Z. HOFFMAN, PETER WIRSCHING, CARLOS F. BARBAS III, K IM D. JANDA*,
AND RICHARD A. LERNER†
Departments of Chemistry and Molecular Biology, The Scripps Research Institute and The Skaggs Institute for Chemical Biology, 10550 North Torrey Pines
Road, La Jolla, CA 92037
Contributed by Richard A. Lerner, August 21, 1997
mechanism was confirmed, model studies suggested that an
enamine mechanism should not be unequivocally ruled out
(9–11). It occurred to us that the antibodies from our previous
aldolase work could be used to rapidly assess the viability of an
enamine mechanism for the catalysis of an allylic rearrangement within the confines of a protein active site. In this way,
we could begin to examine the choice made by Nature as
opposed to a rational alternative.
ABSTRACT
Reactive immunization has emerged as a new
tool for the study of biological catalysis. A powerful application resulted in catalytic antibodies that use an enamine
mechanism akin to that used by the class I aldolases. With
regard to the evolution of enzyme mechanisms, we investigated
the utility of an enamine pathway for the allylic rearrangement
exemplified by D5-3-ketosteroid isomerase (KSI; EC 5.3.3.1).
Our aldolase antibodies were found to catalyze the isomerization of both steroid model compounds and steroids. The
kinetic and chemical studies showed that the antibodies
afforded rate accelerations up to a factor of 104 by means of
an enamine mechanism in which imine formation was the
rate-determining step. In light of our observations and the
enzyme studies by other workers, we suggest that an enamine
pathway could have been an early, viable KSI mechanism.
Although this pathway is amenable to optimization for increased catalytic power, it appears that certain factors precluded its evolution in known KSI enzymes.
MATERIALS AND METHODS
Preparation of mAbs. The antibodies used in this work
were obtained as previously described (5). Prior to use, they
were dialyzed into 100 mM 4-morpholinepropanesulfonic
acid (Mops), pH 7.5, and stored at 4°C. Under these conditions, there was no detectable loss of activity after at least
2 months.
Synthesis of Compounds. Compounds (S)-4 and (R)-4 (see
Fig. 2) were prepared by using an enantioselective Michaeltype alkylation followed by a base-induced ring closure (12)
and were purified by f lash chromatography (Rf 5 0.30, 90y10
hexaneyethyl acetate). The compounds were then isomerized with potassium tert-butoxide at room temperature, a
slight modification of a literature procedure (13). Purification by f lash chromatography (Rf 5 0.42, 90y10 hexaneyethyl
acetate) afforded substrates 1 and 2, respectively. Compound
5 was synthesized in racemic form according to known
procedures (14, 15) and purified by f lash chromatography (Rf
5 0.26, 90y10 hexaneyethyl acetate). Compound 5 could not
be cleanly deconjugated to substrate 3, therefore 3 was
prepared directly from 5,6,7,8-tetrahydro-2-naphthol (Aldrich) by Birch reduction (16) and purified by f lash chromatography (Rf 5 0.40, 90y10 hexaneyethyl acetate). Dideuterio substrate 6 was prepared from 1 by ref luxing in
CD3OD as described for a D5-steroid (17) and analyzed by
both 1H and 2H NMR. Steroid substrates 7 and 8 as well as
the corresponding D4-isomers were commercially available
(Steraloids, Wilton, NH).
Kinetics. The reactions were done at 22°C in 100 mM Mops,
pH 7.5, with 5% CH3CN as cosolvent in the presence or
absence of antibody. Assays for 1–3 were conducted using UV
spectrophotometry by observing the increase in absorbance
due to the formation of 4 («248 5 15,120 M21zcm21) or 5 («248
5 16,000 M21zcm21). Reactions were done singly in 1-ml
(1-cm) cuvettes using a Shimadzu UV2100 equipped with an
automatic cell changer and Peltier temperature control. The
reaction with 6 was done similarly and side-by-side with a
reaction using 1 to ensure accurate comparison. The purity of
substrates 1, 2, 3, and 6 was critical to avoid high initial
absorbance values. Activation parameters were determined as
Many chemical reactions can proceed by alternative mechanisms. Hence, if more than one energetically accessible pathway is available for a particular transformation, it might be
reasonable to expect that enzymes exist that operate via each
distinct route. Yet with few exceptions, most notably the
aldolases (1), Nature has apparently chosen to select and refine
one approach for the conversion of a given substrate to
product. One plausible explanation is that chemistry, and not
specificity, has been the most challenging problem to solve
during the course of enzyme evolution (2, 3). Once a clear
chemical solution is found, subtle active-site modifications
then occur that improve specificity and efficiency.
The process of reactive immunization allows the evolution in
real time of antibody catalysts with defined mechanisms (4, 5).
Hence, it is possible to test whether a chemical pathway other
than one found in Nature is feasible within a protein framework and how this new pathway compares with that optimized
by natural selection. In fact, this approach was used to procure
monoclonal antibodies (mAbs) that catalyze the aldol reaction
using a lysine-mediated enamine mechanism central to the
class I aldolase enzymes (5). The mechanism of these catalysts
contrasts with the class II aldolases, which invoke metal
ion–general base components.
The D5-3-ketosteroid isomerase (KSI; EC 5.3.3.1) provides
a significant opportunity to examine the question of mechanistic dichotomy. The extensively studied KSI from bacteria
makes use of a concerted interaction of a general acid and
general base to isomerize a b,g-unsaturated ketone to an
enone in various steroids and is nearly a catalytically ‘‘perfect
enzyme’’ (6–8) (Fig. 1). More than two decades ago, before the
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: KSI, D5-3-ketosteroid isomerase; D, deuterium.
*To whom reprint requests should be addressed. e-mail: kdjanda@
scripps.edu.
†To whom reprint requests should be addressed. e-mail: foleyral@
scripps.edu.
© 1997 by The National Academy of Sciences 0027-8424y97y9411773-4$2.00y0
PNAS is available online at http:yywww.pnas.org.
11773
11774
Chemistry: Lin et al.
FIG. 1.
Proc. Natl. Acad. Sci. USA 94 (1997)
Typical reaction catalyzed by KSI. Y14 and D38 are tyrosine and aspartic acid amino acid residues in the enzyme.
described above, using temperature-calibrated buffers over the
range 15–50°C. Assays for 7 and 8 were conducted using HPLC
by observing the formation of the corresponding enone product at 254 nm (7: 40% CH3CNy60% H2O, 0.1% trifluoroacetic
acid, 2.5 mlymin, anisole standard; 8: same solvent, 2.0 mly
min, benzophenone standard). Solubility limitations on 7 and
8 precluded the use of a wide range of substrate concentrations
to satisfy first-order kinetics, therefore the kcat value was
derived from the maximal velocity at the solubility limit of 7
(200 mM) or 8 (500 mM) under the assay conditions. Antibodycatalyzed reactions used 0.5 mM mAb with 1–3 and 5–20 mM
with 7–8. Two independent active sites were assumed per IgG
mAb. Background rate constants: 1 and 2, 5.10 3 1024 min21;
3, 4.32 3 1025 min21; 7, 2.27 3 1024 min21; 8, 2.70 3 1026
min21.
Trapping Experiments. Reactions using 700 mM 1 were
initiated as described above, and then various molar excesses
of either sodium borohydride (NaBH4) or potassium cyanide
(KCN) were added after 1 min. Control reactions did not
contain 1. After 15 min, the solutions were dialyzed (Slide-ALyzer cassettes; Pierce) against 100 mM Mops, pH 7.5 (four
changes, 200 ml each). The antibody concentrations were
determined [«280 5 1.35 (mgyml)21zcm21] and the residual
activity was measured using 1.
RESULTS AND DISCUSSION
The bicyclic compounds 1–3 (Fig. 2) were used as models of the
A–B ring portion of steroids, where the chemistry of the
isomerization takes place. All three compounds were found to
be substrates for the mAbs JW38C2 and JW33F12§ in which
Michaelis–Menten saturation kinetics were observed in the
formation of product 4 or 5. Interestingly, there was a strong
preference for the (S)-methyl group at C-10 (steroid numbering) in 1 corresponding to the natural (R)-configuration found
in steroids. Because both 1 and 2 were available for analysis,
it can be surmised that the diastereomeric activated complex
resulting from reaction of 1 and the L-amino acid antibody is
apparently the ‘‘matched pair’’ of intrinsically lower energy.
With substrates 1 and 3, the rate enhancements for both mAbs
over the spontaneous isomerizations were at or close to 104
(Table 1). Remarkably, the kcat value for the D5(10)-estrene
analogue 3 was less than an order of magnitude slower than
that for 1, whereas KSI demonstrates nearly a 1,000-fold
preference for D5(6) steroids (19, 20). Antibody-catalyzed
reactions showed time-dependent inhibition with 2,4pentanedione, which forms a stable enamine with the critical
lysine. In this regard, kinetic experiments using 1 and the
lysine-rich protein bovine serum albumin indicated no catalytic effects. Unexpectedly, the steroids 7 and 8 were also found
to be antibody substrates. The A–B rings of these large
structures are evidently able to enter the active site, although
steric hindrance was probably responsible for the slow rates.
Based on well established studies of Schiff-base chemistry
(21) and chemical models (9–11), a fundamental mechanism
for the mAb-catalyzed isomerization can be proposed (Fig. 3).
A key intermediate is the covalently bound dienamine, whose
counterpart is the noncovalently bound dienol in the KSI
reaction (22, 23). The mAb JW38C2 was used to evaluate
aspects of the kinetic and chemical mechanism. When substrate 1 was used, there was no detectable accumulation of
intermediates, judged from the lack of burst formation of
dienamine C at 278 nm and the absence of a lag in product
formation monitored at 248 nm. Other than the appearance of
the final product at 248 nm, no other spectral changes were
§Subsequent to the original work (5) these antibodies were found to
catalyze a large variety of aldol reactions (C.F.B., A. Heine, G.
Zhong, T.Z.H., S. Gramatikova, R. Björnestedt, B. List, J. Anderson,
E. A. Stura, I. A. Wilson, and R.A.L., unpublished results), including
intramolecular reactions that allowed for the synthesis of 4 (18).
Consequently, small model compounds were initially chosen to
investigate the kinetics of double-bond migration.
Table 1. Kinetic constants for catalytic antibodies JW38C2
and JW33F12
mAb
JW38C2
JW33F12
FIG. 2.
Structures of compounds under discussion. D, deuterium.
kcatyKm,
Substrate kcat, min21 Km, mM M21zmin21 kcatykuncat
1
2
3
7
8
1
2
3
7
8
ND, not determined.
4.13
0.27
1.03
0.0034
0.00066
4.42
0.32
0.54
0.0035
0.00049
271
280
2306
ND
ND
550
1097
2348
ND
ND
15,200
964
446
ND
ND
8,040
296
230
ND
ND
8,100
529
23,800
15
244
8,670
637
12,500
15
180
Chemistry: Lin et al.
Proc. Natl. Acad. Sci. USA 94 (1997)
FIG. 3.
11775
Enamine mechanism for antibody-catalyzed allylic rearrangement.
observed in the range 220 nm to 320 nm during the course of
the reaction. Substrate 6 showed no primary deuterium isotope
effect, so proton abstraction from B was not rate-determining.
Any b-secondary isotope effects, ,5% if present, could not be
determined with certainty. Although the solvent deuterium
isotope effect was not fully investigated because it could not be
analyzed in terms of a simple first-order process, the data
yielded an estimated value of (kcat)Hy(kcat)D 5 1.5–2. An effect
of this low magnitude suggested that proton transfer to C-6
(steroid numbering) of C did not limit the overall rate. Also,
it is known that dienamines usually undergo rapid rearrangement to yield iminium ions (9). The origin of the small solvent
isotope effect probably reflects the influence of D2O on the
equilibrium of 1 and B not mediated by buffer species, since no
buffer catalysis was observed. The loss of antibody activity
upon NaBH4 reduction in the presence of saturating concentrations of 1 established the intermediacy of covalent imineantibody species on the reaction pathway. However, imines B
and D were apparently short-lived relative to the rate of
trapping, because a maximum 50% decrease in activity was
obtained even at a molar ratio of NaBH4y1 5 100. No loss in
activity was observed upon exposure to NaBH4 of antibody
alone or antibody in the presence of 4. The use of cyanide
trapping with 1 supported the hydride result in that a concentration-dependent decrease in activity occurred, but up to only
27% at 1 mM KCN. The apparent rapid hydrolysis of imine D
deduced from the trapping results and UV spectral observations (see above) was also consistent with the fact that antibody
reactions with 1 could proceed to .95% completion. The
strongly favored equilibrium of D4-steroids and amines in the
direction of the enone was noted in model studies (10). Taken
together, the data indicate that catalytic throughput from
imine B to the final product has a large forward commitment
and is quite fast. Finally, the activation parameters for both the
spontaneous (DH‡ 5 15.1 kcalymol, DS‡ 5 214.0 calymolzK)
and antibody-catalyzed (DH‡ 5 12.5 kcalymol, DS‡ 5 25.4
calymolzK; for kcat) reactions were calculated from linear
Arrhenius plots. The data for the spontaneous isomerization
are in accord with expected values (24).
Given the above observations, a kinetically determined pKa
5 6.5 for the reactive lysine, and that carbinolamine dehydration is generally rate-determining in imine formation from
aliphatic ketones near physiological pH (21), suggested that
breakdown of A was the step that primarily controlled the
overall rate. The mechanism can now be rationalized in light
of a crystal structure of mAb JW33F12 (C.F.B., A. Heine, G.
Zhong, T.Z.H., J. Gramatikova, R. Björnestedt, B. List, J.
Anderson, E. A. Stura, I. A. Wilson, and R.A.L., unpublished
results). The data show a heavy-chain Lys-93 situated near the
bottom of a 13.5-Å hydrophobic cleft. The residue is isolated
with no hydrogen-bonding side chains in contact distance or
other acidic or basic amino acid chains in the vicinity. Unlike
the lysine in acetoacetate decarboxylase, in which neighboring
electrostatic effects are also important (25), the reduced pKa
of the antibody lysyl «-amine results principally from the low
dielectric medium of the active site. This medium might also
stabilize the formation of charge-neutral carbinolamines relative to charged intermediates. On the basis of the kinetic and
structural data, it is reasonable to conclude that the components of the chemical mechanism used by the antibody are
composed of a nucleophilic lysine and water.
The activation parameters support the premise that the
antibody and KSI lower the free energy of activation by
different thermochemical strategies. Whereas the antibody
restricts the entropy from being as unfavorable compared with
the spontaneous reaction, the enzyme facilitates the isomerization due to an extremely low enthalpy of activation (17).
Apparently, covalent catalysis provides a significant entropic
advantage in the antibody reaction. Consequently, the opportunity exists to impart a more favorable enthalpic component
to the enamine mechanism and achieve high catalytic power.
Since imine formation and a-deprotonation are subject to
general acid and general base catalysis, respectively, the addition of this machinery should enhance catalysis. Site-directed
mutagenesis studies of both aldolase (26, 27) and KSI (23, 28,
29) have shown that amino acids necessary for proton donationyabstraction contribute 104–105 to catalysis on an individual basis. Therefore, a protein catalyst operating along an
enamine pathway with general acid–base assistance might well
rival the efficiency of KSI.
While as yet unisolated mammalian steriod isomerases may
reveal the use of an enamine mechanism, precisely why two
enolization mechanisms have remained intact even in prokaryotic aldolases (30), but not in KSI, is unknown. One argument
could be mechanistic parsimony (31), since an optimized
enamine mechanism would likely require a catalytic triad
rather than the diad of KSI. However, recent elegant structural
studies uncovered a third amino acid that might be participatory in the KSI mechanism (32). A more concrete rationale
might be the place D5(6)-steroids occupy among the class of
relatively acidic ketones that makes enolization an accessible
pathway, whereas imine formation, most beneficial for nonacidic ketones, would require the extra condition of an amine
of reduced pKa (33, 34). In light of our findings, the results
from KSI mutagenesis experiments, and work by others where
a KSI-like catalytic antibody elicited using traditional hapten
design was inefficient (35), it is tempting to speculate that the
presence of an unrefined enamine pathway in the ancestral
progenote would have been competitive with other early KSI
mechanisms. Perhaps then, Nature determined long ago that
this mechanism could not be perfected to achieve the necessary velocities, and so it was a road not taken in the evolution
of these enzymes.
We thank Dr. Andreas Heine and Dr. Ian A. Wilson for analysis of
the crystal structure of mAb JW33F12 prior to publication. We also
thank Prof. Dr. Duilio Arigoni for helpful comments. This work was
supported in part by the National Institutes of Health (GM-43858,
K.D.J.) and The Skaggs Institute for Chemical Biology (K.D.S.).
11776
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Chemistry: Lin et al.
Rutter, W. J. (1964) Fed. Proc. Am. Soc. Exp. Biol. 23, 1248–1257.
Babbitt, P. C., Hasson, M. S., Wedekind, J. E., Palmer, D. R. J.,
Barrett, W. C., Reed, G. H., Rayment, I., Ringe, D., Kenyon,
G. L. & Gerlt, J. A. (1996) Biochemistry 35, 16489–16501.
Babbitt, P. C., Mrachko, G. T., Hasson, M. S., Huisman, G. W.,
Kolter, R., Ringe, D., Petsko, G. A., Kenyon, G. L. & Gerlt, J. A.
(1995) Science 267, 1159–1161.
Wirsching, P., Ashley, J. A., Lo, C.-H. L., Janda, K. D. & Lerner,
R. A. (1995) Science 270, 1775–1782.
Wagner, J., Lerner, R. A. & Barbas, C. F., III (1995) Science 270,
1797–1800.
Fisher, H. F. (1996) Chemtracts Biochem. Mol. Biol. 6, 69–75.
Schwab, J. M. & Henderson, B. S. (1990) Chem. Rev. 90,
1203–1245.
Batzhold, F. H., Benson, A. M., Covey, D. F., Robinson, C. H. &
Talalay, P. (1976) Adv. Enzyme Reg. 14, 243–267.
Pollack, R. M. & Kayser, R. H. (1976) J. Am. Chem. Soc. 98,
4174–4181.
Benisek, W. F. & Jacobson, A. (1975) Bioorg. Chem. 4, 41–57.
Kayser, R. H. & Pollack, R. M. (1975) J. Am. Chem. Soc. 97,
952–953.
Pfau, M., Revial, G., Guingant, A. & d’Angelo, J. (1985) J. Am.
Chem. Soc. 107, 273–274.
Moss, R. A. & Chen, E. Y. (1981) J. Org. Chem. 46, 1466–1469.
Duhamel, P., Dujardin, G., Hennequin, L. & Poirier, J.-M. (1992)
J. Chem. Soc. Perk. Trans. I, 387–396.
Sato, T., Wakahara, Y., Otera, J. & Nozaki, H. (1990) Tetrahedron Lett. 31, 1581–1584.
Holland, H. L. & Taylor, G. J. (1981) Can. J. Chem. 59,
2809–2819.
Jones, J. B. & Wigfield, D. C. (1969) Can. J. Chem. 47, 4459–
4466.
Zhong, G., Hoffmann, T., Lerner, R. A., Danishefsky, S. &
Barbas, C. F., III (1997) J. Am. Chem. Soc. 119, 8131–8132.
Proc. Natl. Acad. Sci. USA 94 (1997)
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
Pollack, R. M., Bounds, P. L. & Bevins, C. L. (1989) in The
Chemistry of Enones, eds. Patai, S. & Rappoport, Z. (Wiley,
NewYork), pp. 559–597.
Weintraub, H., Baulieu, E.-E. & Alfsen, A. (1980) Biochem. J.
185, 723–732.
Jencks, W. P. (1964) Prog. Phys. Org. Chem. 2, 63–129.
Zeng, B., Bounds, P. L., Steiner, R. F. & Pollack, R. M. (1992)
Biochemistry 31, 1521–1528.
Xue, L., Kuliopulos, A., Mildvan, A. S. & Talalay, P. (1991)
Biochemistry 30, 4991–4997.
Jones, J. B. & Wigfield, D. C. (1969) Can. J. Chem. 47, 4459–
4466.
Highbarger, L. A., Gerlt, J. A. & Kenyon, G. L. (1996) Biochemistry 35, 41–46.
Morris, A. J. & Tolan, D. R. (1994) Biochemistry 33, 12291–
12297.
Littlechild, J. A. & Watson, H. (1993) Trends Biochem. Sci. 18,
36–39.
Xue, L., Talalay, P. & Mildvan, A. S. (1991) Biochemistry 30,
10858–10865.
Kuliopulos, A., Talalay, P. & Mildvan, A. S. (1990) Biochemistry
29, 10271–10280.
Marsh, J. J. & Lebherz, H. G. (1992) Trends Biochem. Sci. 17,
110–113.
Hanson, K. R. & Rose, I. A. (1975) Acc. Chem. Res. 8, 1–10.
Wu, Z. R., Ebrahimian, S., Zawrotny, M. E., Thornburg, L. D.,
Perez-Alvarado, G. C., Brothers, P., Pollack, R. M. & Summers,
M. F. (1997) Science 276, 415–418.
Bruice, P. Y. (1990) J. Am. Chem. Soc. 112, 7361–7368.
Pollack, R. M., Zeng, B., Mack, J. P. G. & Eldin, S. (1989) J. Am.
Chem. Soc. 111, 6419–6423.
Khettal, B., de Lauzon, S., Desfosses, B., Ushida, S. & Marquet,
A. (1994) C. R. Acad. Sci. 317, 381–385.