Substrate docking algorithms and prediction of the substrate specificity of cytochrome P450(cam) and its L244A mutant

Article abstract of DOI:10.1021/ja970349v

The substrate specificity of cytochrome P450, defined as the ability of a compound to promote NAD(P)H and O₂ utilization in the production of either organic or reduced oxygen metabolites, is largely determined by steric and hydrophobic interactions. P450 specificity may therefore be determined by the 'fit' of a compound within the active site. A receptor-constrained three- dimensional screening program (DOCK) has been used to select 11 compounds predicted to fit within the P450(cam) active site and 5 compounds predicted to fit within the L244A P450(cam) but not wild-type active site. The 16 compounds were evaluated as P450(cam) substrates by measuring (a) binding to the enzyme, (b) stimulation of NADH and O₂ consumption, (c) enhancement of H₂O₂ production, and (d) formation of organic metabolites. Seven of the compounds predicted to fit in the active site, and none of the compounds predicted not to fit, were found to be substrates. Compounds predicted to fit very tightly within the active site were poor or non-substrates. The L244A P450(cam) mutant was constructed, expressed, purified, and shown to readily oxidize some of the larger compounds incorrectly predicted to be substrates for the wild-type enzyme. The 5 ligands selected to fit the L244A but not wild-type sites were not detectable substrates, presumably because they fit too tightly into the active site. Retroactive adjustments of the docking program based on an analysis of the docking parameters, particularly variation of the minimum distance allowed between ligand and protein atoms, allow correct predictions for the activity of 15 of the 16 compounds with wild-type P450(cam). The DOCK predictions for the L244A mutant were also improved by changing the minimum contact distances to disfavor the larger compounds. The results indicate that ligands that fill the active site are marginal or non-substrates. A degree of freedom of motion is required for substrate positioning and catalytic function. If parameters are chosen to allow for this requirement, P450(cam) substrate predictions based on ligand docking in the active site can be reasonably accurate.

Full text of DOI:10.1021/ja970349v

J. Am. Chem. Soc. 1997, 119, 5489-5498
5489
Substrate Docking Algorithms and Prediction of the Substrate
Specificity of Cytochrome P450cam and Its L244A Mutant
James J. De Voss, Ole Sibbesen, Zhoupeng Zhang, and
Paul R. Ortiz de Montellano*
Contribution from the Department of Pharmaceutical Chemistry, School of Pharmacy, UniVersity
of California, San Francisco, California 94143-0446
ReceiVed February 3, 1997X
Abstract: The substrate specificity of cytochrome P450, defined as the ability of a compound to promote NAD(P)H
and O2 utilization in the production of either organic or reduced oxygen metabolites, is largely determined by steric
and hydrophobic interactions. P450 specificity may therefore be determined by the “fit” of a compound within the
active site. A receptor-constrained three-dimensional screening program (DOCK) has been used to select 11
compounds predicted to fit within the P450cam active site and 5 compounds predicted to fit within the L244A P450cam
but not wild-type active site. The 16 compounds were evaluated as P450cam substrates by measuring (a) binding to
the enzyme, (b) stimulation of NADH and O2 consumption, (c) enhancement of H2O2 production, and (d) formation
of organic metabolites. Seven of the compounds predicted to fit in the active site, and none of the compounds
predicted not to fit, were found to be substrates. Compounds predicted to fit very tightly within the active site were
poor or non-substrates. The L244A P450cam mutant was constructed, expressed, purified, and shown to readily
oxidize some of the larger compounds incorrectly predicted to be substrates for the wild-type enzyme. The 5 ligands
selected to fit the L244A but not wild-type sites were not detectable substrates, presumably because they fit too
tightly into the active site. Retroactive adjustments of the docking program based on an analysis of the docking
parameters, particularly variation of the minimum distance allowed between ligand and protein atoms, allow correct
predictions for the activity of 15 of the 16 compounds with wild-type P450cam. The DOCK predictions for the
L244A mutant were also improved by changing the minimum contact distances to disfavor the larger compounds.
The results indicate that ligands that fill the active site are marginal or non-substrates. A degree of freedom of
motion is required for substrate positioning and catalytic function. If parameters are chosen to allow for this
requirement, P450cam substrate predictions based on ligand docking in the active site can be reasonably accurate.
The cytochrome P450 enzymes are members of a large family
of hemoproteins1 that catalyze a wide range of oxidative and
reductive transformations, including carbon hydroxylation,
π-bond oxidation, heteroatom oxidation, hydrocarbon desatu-
ration, and halocarbon dehalogenation.2 The enzymes of the
cytochrome P450 family can be usefully classified in a variety
of ways. Classification of the enzymes according to their
phylogenetic origin shows that the best characterized enzymes
are of mammalian or bacterial origin, although knowledge of
the enzymes from plants and insects is rapidly growing.3 The
enzymes can also be classified according to whether a protein
with two flavin prosthetic groups or a binary system consisting
of a flavoprotein and an iron-sulfur protein transfers electrons
from NAD(P)H to the hemoprotein.4 The enzymes can also
be grouped according to their cytosolic or membrane location:
as a rule, eukaryotic P450 enzymes are membrane bound and
prokaryotic enzymes are soluble.4b,5 One consequence of this
difference in cellular location is that crystal structures are
available for a growing number of soluble bacterial P450
enzymes but none is yet available for a membrane-bound P450.6
Despite the categorical differences among P450 enzymes in
terms of origin, electron transfer partner(s), cellular localization,
substrate specificity, and reactivity, the available evidence
suggests that the active sites and catalytic mechanisms of P450
enzymes are remarkably similar. Catalytic turnover involves
(a) substrate binding, usually with a concomitant change in redox
potential that enables the heme to accept electrons from its
electron transfer partner, (b) reduction of the iron to the ferrous
state, (c) binding of molecular oxygen to give a ferrous dioxy
complex (Fe2+-O2), (d) reduction of the ferrous dioxy complex
to give a ferric peroxide (formally Fe3+-OOH) intermediate,
(e) conversion of the ferric peroxide to what is formally a ferryl
(Fe5+dO) species, (f) reaction of the ferryl species with the
substrate to give the enzyme-product complex, and (g) dis-
sociation of the product to regenerate the resting enzyme
(Scheme 1).2 The primary catalytic role of the enzyme is to
produce the ferryl species under controlled conditions: the role
of the enzyme in the reaction of the ferryl species with the
substrate appears to be limited to juxtaposing the substrate and
the ferryl species and restricting the sites on the substrate
X Abstract published in AdVance ACS Abstracts, May 15, 1997.
(1) Nelson, D.; Koymans, L.; Kamataki, T.; Stegeman, J.; Feyereisen,
R.; Waxman, D.; Waterman, M.; Gotoh, O.; Coon, M.; Estabrook, R.;
Gunsalus, I.; Nebert, D. Pharmacogenetics 1996, 6, 1-42.
(2) Ortiz de Montellano, P. R. In Cytochrome P450: Structure, Mech-
anism, and Biochemistry, 2nd ed.; Ortiz de Montellano, Ed.; Plenum
Press: New York, 1995; pp 245-304.
(5) von Wachgenfeldt, C.; Johnson, E. F. In Cytochrome P450: Structure,
Mechanism, and Biochemistry, 2nd ed.; Ortiz de Montellano, Ed., Plenum
Press: New York, 1995; pp 183-224.
(3) Schuler, M. Crit. ReV. Plant Sci. 1996, 15, 235-284.
(4) (a) Strobel, H. W.; Hodgson, A. V.; Shen, S. In Cytochrome P450:
Structure, Mechanism, and Biochemistry, 2nd ed.; ortiz de Montellano, Ed.;
Plenum Press: New York, 1995; pp 225-245. (b) Mueller, E. J.; Loida, P.
J.; Sligar, S. G. In Cytochrome P450: Structure, Mechanism, and
Biochemistry, 2nd ed.; Ortiz de Montellano, Ed.; Plenum Press: New York,
1995; pp 83-124.
(6) (a) Poulos, T. L.; Finzel, B. C.; Howard, A. J. J. Mol. Biol. 1987,
195, 687-700; (b) Ravichandran, K. G.; Boddupalli, S. S.; Hasemann, C.
A.; Peterson, J. A.; Deisenhofer, J. Science 1993, 261, 731-736. (c)
Hasemann, C. A.; Ravichandran, K. G.; Peterson, J. A.; Deisenhofer, J. J.
Mol. Biol. 1994, 236, 1169-1185. (d) Cupp-Vickery, J. R.; Poulos, T. L.
Struct. Biol. 1995, 2, 144-153.
S0002-7863(97)00349-1 CCC: $14.00 © 1997 American Chemical Society

5490 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
De Voss et al.
Scheme 1. Catalytic Cycle of Cytochrome P450, Showing
a simple analysis of its ability to fit within the active site, is
sufficient to trigger catalytic turnover of the enzyme. According
to the conventional definition, a good substrate is one that is
efficiently converted to organic metabolites. However, the
formation of reduced oxygen metabolites is also a measure of
catalytic turnover because the partitioning between organic
versus oxygen metabolites occurs at a branch point distal to
the initiation of catalytic turnover. The ratio of organic to
reduced oxygen products is thus a measure of the quality rather
than the extent of catalysis. In elementary terms, a P450
substrate is a compound that initiates catalytic turnover regard-
less of the nature of the products. The intent of this study was
therefore to develop a practical approach for evaluating the
ability of potential substrates to bind within the P450cam active
site, to determine the extent to which predictions of fit correlate
with experimentally measured substrate activities, and to identify
and optimize parameters that influence this correlation. Parts
of the results of these studies have been communicated in
preliminary form.11
the Individual Steps at which Uncoupling Can Occura
a The iron of the heme group is indicated as an iron atom in brackets.
accessible to the ferryl species. In view of the diversity of P450
enzymes, the multiplicity of their substrates, and the breadth of
the reactions they catalyze, it is not surprising that the catalytic
cycle can be subverted. One example of such a diversion is
reductive metabolism, a reaction that competes with the binding
of oxygen to the ferrous intermediate.7 A more general
phenomenon is uncoupling of the consumption of oxygen and
NAD(P)H from oxidation of the organic substrate.7,8 This
uncoupling can stem from (a) dissociation of superoxide from
the ferrous dioxy complex, (b) dissociation of H2O2 from the
ferric peroxide complex, or (c) reduction of the final ferryl
species to water by electrons derived from NAD(P)H (Scheme
1). Uncoupled turnover is thus a catalytic process that produces
reduced oxygen metabolites rather than organic products.
The factors that control the substrate specificity of P450
enzymes are not well understood. In general, the binding and
orientation of potential substrates appears to be determined by
hydrophobic and steric interactions and only infrequently by
specific hydrogen bonding or ionic interactions. This conclusion
derives from the preference of the enzymes for lipophilic
substrates,9 the nature of the active sites in the four bacterial
enzymes for which crystal structures are available,6 and infer-
ences made from sequence alignments of P450 enzymes of
known and unknown tertiary structure.10 The specific site that
is oxidized, once the molecule is bound within the active site,
is largely determined by the intrinsic reactivity of the sites on
the molecule that are accessible to the iron-bound oxidizing
species.
A more refined understanding of P450 substrate specificity
is highly desirable for practical reasons. For example, it is
critical to the prediction of whether a specific compound will
be a substrate for a given P450 enzyme, for the design and
construction of P450 enzymes with tailored substrate specifici-
ties, and for the prediction of potential toxic interactions. Three
features make the application of computer algorithms to the
elucidation and prediction of P450 substrate specificity particu-
larly attractive: (a) specificity is primarily determined by
nonpolar interactions, (b) the enzyme has little active role in
the reaction of the ferryl species with the substrate, and (c) the
question of specificity can be dissected into distinct steps. The
premise tested in this work is whether binding of a compound
to P450cam without coordination to the iron atom, as judged by
Results
Prediction of P450cam Substrates. Cytochrome P450cam
(CYP101), a monooxygenase required by Pseudomonas putida
for growth on camphor as its sole carbon source, is the best
characterized P450 enzyme.4b Crystal structures are available
of P450cam in a variety of states, including as the camphor-
bound and camphor-free protein and as the ferrous-CO
complex.6a,12 The change in oxidation potential from -300 to
-170 mV associated with the binding of camphor to P450cam
makes possible its reduction by putidaredoxin, for which the
oxidation potential is -196 mV.13 The enzyme normally
catalyzes the 5-exo-hydroxylation of camphor, but recent work
has demonstrated that P450cam also oxidizes compounds other
than camphor, albeit with a higher degree of uncoupling of
substrate oxidation from NADH and O2 consumption.11,14
P450cam has been chosen for the present studies because (a) it is
a well-characterized system of known structure, (b) substrate
binding does not significantly alter the active site geometry,6a,12a
and (c) the system is experimentally versatile.
We have employed DOCK,15 a receptor-constrained three-
dimensional screening program, to evaluate the fit of potential
substrates within the P450cam active site. DOCK has been
successfully used to identify lead molecules for the development
of enzyme inhibitors and receptor ligands.15,16 The three steps
involved in the use of the program are the following: (a)
construction of a negative image of the molecular surface of
the active site,17,19a (b) screening a data base of three dimensional
structures to determine their fit within the active site volume
(11) (a) De Voss, J. J.; Ortiz de Montellano, P. R. J. Am. Chem. Soc.
1995, 117, 4185-4186. (b) De Voss, J. J.; Ortiz de Montellano, P. R.
Methods Enzymol. 1996, 272, 336-347.
(12) (a) Poulos, T. L.; Finzel, B. C.; Howard, A. J. Biochemistry 1986,
25, 5314-5322. (b) Raag, R.; Poulos, T. L. Biochemistry 1989, 28, 7586-
7592.
(13) (a) Sligar, S. G.; Gunsalus, I. C. Proc. Natl. Acad. Sci. U.S.A. 1976,
73, 1078-1082. (b) Sligar, S. G. Biochemistry 1976, 15, 5399-5406.
(14) (a) Fruetel, J. A.; Collins, J. R.; Camper, D. L.; Loew, G. H.; Ortiz
de Montellano, P. R. J. Am. Chem. Soc. 1992, 114, 6987-6993. (b) Jones,
J. P.; Trager, W. F.; Carlson, T. J. J. Am. Chem. Soc. 1992, 115, 381-387.
(c) Fruetel, J.; Chang, Y. T.; Collins, J.; Loew, G.; Ortiz de Montellano, P.
R. J. Am. Chem. Soc. 1994, 116, 11643-11648. (d) Filipovic, D.; Paulsen,
M. D.; Loida, P. J.; Sligar, S. G.; Ornstein, R. L. Biochem. Biophys. Res.
Commun. 1992, 189, 488-495. (e) Loida, P. J.; Sligar, S. G.; Paulsen, M.
D.; Arnold, G. E.; Ornstein, R. L. J. Biol. Chem. 1995, 270, 5326-5330.
(f) Gelb, M. H.; Malkonen, P.; Sligar, S. G. Biochem. Biophys. Res.
Commun. 1982, 104, 853-858. (g) Stevenson, J.-A.; Westlake, A. C. G.;
Whittock, C.; Wong, L.-L. J. Am. Chem. Soc. 1996, 118, 12846-12847.
(15) DesJarlais R. L.; Sheridan, R. P.; Seibel G. L.; Dixon J. S.; Kuntz
I. D.; Venkataraghavan, R. J. Med. Chem. 1988, 31, 722-729.
(7) Goeptar, A. R.; Scheerens, H.; Vermeulen, N. P. E. Crit. ReV. Toxicol.
1995, 25, 25-65.
(8) Archakov, A. I.; Bachmanova, G. I. Cytochrome P-450 and ActiVe
Oxygen; Taylor & Francis: New York, 1990; pp 129-138.
(9) (a) White, R. E.; McCarthy, M. Arch. Biochem. Biophys. 1986, 246,
19-32. (b) Duquette, P. H.; Erickson, R. R.; Holtzman, J. L. J. Med. Chem.
1983, 26, 1343-1348.
(10) Gotoh, O. J. Biol. Chem. 1992, 267, 83-90.

Substrate Docking Algorithms of Cytochrome P450cam
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5491
and assigning a score to the fit of each compound, and (c)
selection of compounds to be tested from the resulting list based
on additional criteria (see below). In effect, a 20 000 compound
subset of the Available Chemical Directory (ACD)19b was
screened with DOCK. Contact scoring, which assigns the
highest scores to compounds that maximize nonpenetrating
contacts between the substrate and the active site, was used to
evaluate the goodness of fit.18 High contact scores are
considered desirable for the identification of potential inhibitors,
but the suitability of high contact scores for the evaluation of
potential substrates had not been examined prior to this study.
Of the 500 compounds with the highest contact scores, a small
group of structurally diverse compounds was selected (Table
1) that did not have functionalities ionized at physiological pH,
were relatively simple structures, and represented a diversity
of structural types.11,19
To test the ability of this computer-assisted approach to
discriminate between substrates and nonsubstrates, a group of
compounds predicted not to fit within the P450cam active site
was also identified (Table 2). In order to make this a rigorous
test, the P450cam active site model based on the crystal
coordinates was modified by replacing the side chain of Leu-
244 with that of an alanine (i.e., by removing three carbons
from the crystal coordinates) while holding the rest of the atoms
fixed in their original positions. Dock was then run on the
“mutated” active site exactly as done for the native enzyme,
and a difference list was made of the compounds that fit within
the model L244A but not wild-type active site. This approach,
which requires that DOCK differentiate between ligands that
bind to sites that differ by only a small volume, generated a list
of predicted nonsubstrates closely related to the predicted
substrates (Table 2).11
search. Since better three-dimensional structures than those
provided by CONCORD can be generated by using the
molecular mechanics functions in the SYBYL modeling
package,19d the molecules selected from the initial DOCK
analysis of the ACD were minimized by using the default values
in SYBYL and were reDOCKed into the active site. Optimiza-
tion of the structures with the SYBYL molecular mechanics
programs resulted in two compounds (10, 11) migrating from
the nonsubstrate to the substrate list and one (12) migrating in
the opposite direction.
Dependence of Prediction on DOCK Parameters. Initial
efforts to predict P450cam substrates were carried out with the
default settings for the adjustable parameters in DOCK.
However, as these default settings are not necessarily optimal
for the identification of substrates, we have examined the effects
of varying some of these parameters. Studies of the minimum
distance allowed between an atom of the substrate and an atom
of the enzyme before the contact is classified as prohibitive have
been particularly fruitful. Two such distances are defined in
DOCK, one for polar and one for nonpolar interactions: the
default values for these distances are 2.3 and 2.8 Å, respectively.
These distances have been found to be critical determinants of
the substrate predictions. The results in Tables 1 and 2 are based
on the default values, but increasing the distances to 2.4 and
2.9 Å for polar and nonpolar contacts, respectively, shifts
compounds 8-10 from the predicted substrate to the nonsub-
strate column. The two contact distances have also been set to
the same value, and the DOCKing of the compounds, using
both the CONCORD and SYBYL-minimized structures, has
been reexamined as a function of this single minimum contact
distance. As shown in Figure 1, if the minimal contact distance
is set at 2.9 Å, all but one of the 16 compounds are correctly
predicted to be either substrates or nonsubstrates of P450cam
.
CONCORD Versus SYBYL Generated Structures. Three-
dimensional structures of the compounds in the Available
Chemicals Directory were generated with the program
CONCORD,19c which uses a set of empirical bond length and
bond angle rules to convert two-dimensional to three-dimen-
sional structures. After DOCK analysis of a 20 000 compound
subset of the ACD, the 500 compounds with the highest contact
scores were saved and visually screened. Ten compounds (1-
9, 12) were chosen as potential substrates from the wild-type
P450cam screen, and six compounds (10, 11, 13-16) were
chosen as probable nonsubstrates from the L244A P450cam
The exceptions are compound 3, using the CONCORD struc-
tures, and compound 6, using the SYBYL structures.
The best active site model to be used for the substrate docking
studies is also of concern. The ferric, camphor-bound enzyme
after subtraction of the camphor coordinates was used for the
above studies, but docking to the ferrous, ferrous dioxy, or
putative ferryl states of the protein could also be envisaged.
However, of these enzyme forms, coordinates are only available
for the ferrous state. Fortunately, although NMR and X-ray
crystallography indicate that ferric P450BM-3 undergoes a major
conformational change when the enzyme is reduced to the
ferrous state,21 the data on P450cam indicate that its active site
is little altered by changes in the redox and ligation states of
the protein.6a,12a The ferrous-CO complex, for which a crystal
structure is available,20 can therefore be used with some
confidence as a model for the ferrous dioxy complex, although
the CO and O2 ligands differ in that the first is essentially linear
and the second bent. As shown in Figure 1, correct predictions
can be made for 15 of the 16 compounds if the minimum contact
distance for binding to the Fe2+-CO complex is set to 2.7 Å
rather than the 2.9 Å found for docking to the ferric protein.
However, from the results, there is no obvious advantage to
selecting the ferrous-CO complex over the better defined ferric
site as the docking target.
(16) (a) DesJarlais, R. L.; Seibel, G. L.; Kuntz, I. D.; Furth, P. S.; Alvarez,
J. C.; Ortiz de Montellano, P. R.; DeCamp, D. L.; Babe´, L. M.; Craik, C.
S. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 6644-6648. (b) Shoichet, B. K.;
Stroud, R. M.; Santi, D. V.; Kuntz, I. D.; Perry, K. M. Science 1993, 259,
1445-1450. (c) Ring, C. S.; Sun, E.; McKerrow, J. H.; Lee, G. K.;
Rosenthal, P. J.; Kuntz, I. D.; Cohen, F. E. Proc. Natl. Acad. Sci. U.S.A.
1993, 90, 3583-3587. (d) Bodian, D. L.; Yamasaki, R. B.; Buswell, R. L.;
Stearns, J. F.; White, J. M.; Kuntz, I. D. Biochemistry 1993, 32, 2967-
2978.
(17) (a) Richards, F. M. Annu. ReV. Biophys. Bioeng. 1977, 6, 151-
176. (b) Connolly, M. L. Science 1983, 221, 709-713.
(18) Kuntz, I. D.; Meng, E. C.; Shoichet, B. K. Acc. Chem. Res. 1994,
27, 117-123.
(19) (a) Molecular graphics visualization and molecular surface genera-
tion were carried out with the MidasPlus program from the Computer
Graphics Laboratory, University of California, San Francisco (supported
by NIH RR-01081): Ferrin, T. E.; Huang, C. C.; Jarvis, L. E.; Langridge,
R. J. Mol. Graphics 1988, 6, 13-27. (b) MDL Information Systems, Inc.,
San Leandro, CA, USA. (c) Rusinko, A.; Sheridan, R. P.; Nilakatan, R.;
Haraki, K. S.; Bauman, N.; Venkataraghavan, R. J. Chem. Inf. Comput.
Sci. 1989, 29, 251. (d) SYBYL 6.0 Tripos Inc., 1699 South Hanley Road,
St. Louis, MO 63144-2913 USA; (e) Molecular graphics images were
produced using the MidasPlus program from the Computer Graphics
Laboratory, University of California, San Francisco (supported by NIH RR-
01081): Ferrin, T. E., Huang, C. C., Jarvis, L. E., and Langridge, R., J.
Mol. Graphics 6, 13 (1988); (f) SPHGEN: Kuntz, I. D., Blaney, J. M.,
Oatley, S. L., Langridge, R., and Ferrin, T. E. (1982) J. Mol. Biol. 161,
269-288.
Binding of Compounds to P450cam
.
The binding of
compounds to P450cam was evaluated spectroscopically by
quantitating the low to high spin transition evidenced by a
decrease in the absorption maximum of the resting enzyme at
417 nm and the corresponding increase in the enzyme-substrate
(20) Raag, R.; Poulos, T. L. Biochemistry 1989, 28, 7586-7592.
(21) (a) Modi, S.; Sutcliffe, M. J.; Primrose, W. U.; Lian, L.; Roberts,
G. C. K. Nature Struct. Biol. 1996, 3, 414-417. (b) Li, H., and Poulos, T.
L. Biochimie 1996, 78, 695-699.

5492 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
De Voss et al.
Table 1. Active Site Docking Parameters and Experimental Activities of Potential P450cam Substrates
a DOCK results given as (i) the maximum binding energy score in “DOCK kcals” from DOCKing of each compound in SINGLE mode; (ii) the
number of orientations generated in the SINGLE DOCK run, and (iii) the DOCK rank of each compound after the SEARCH run of the ACD
subset. b Background NADH consumption is 4-6 nmol min-1 nmol-1, but the values have not been corrected for this background because it may
differ in the presence of substrates. c KI, determined by inhibition of camphor binding. d Not determined because the UV-visible absorption of
compound interfered with spectrophotometric assay. e These compounds were not ranked as they were not identified as substrates in the original
DOCK search. f An organic product is formed even though there is little stimulation of NADH consumption.
complex absorption maximum at 392 nm (a Type I spectroscopic
change).22 The results are expressed as the spectroscopic
binding constant Ks (Table 1) determined by plotting the peak
to trough amplitude of the difference spectrum versus the
reciprocal of the substrate concentration. The Ks values thus
obtained for compounds 1-6 and 11 ranged from 4 µM for 2
to 4400 µM for 4. These values are to be compared with Ks )
1.1 µM under similar conditions for the natural substrate
camphor.
If a normal spectroscopic change was not observed, the ability
of the test compound to inhibit the low to high spin transition
produced by camphor was examined and the data were used to
(22) Petersen, J. A.; Lorence, M. C.; Amarneh, B. J. Biol. Chem. 1990,
265, 6066-6073.

Substrate Docking Algorithms of Cytochrome P450cam
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5493
Figure 1. Correlation of the activity of compounds 1-16 as P450cam substrates with the value of the minimum distance allowed between substrate
and active site atoms. (A) DOCKing of Concord generated substrate structures within the ferric P450cam active site model, (B) DOCKing of Sybyl
minimized substrate structures within the ferric P450cam active site model, and (C) DOCKing of Concord generated substrate structures within the
Fe+2-CO P450cam complex as a model of the ferrous dioxygen intermediate. Solid bars indicate experimental substrates and open bars experimental
non-substrates. The vertical lines indicate the minimum interaction distance that gives the best fit of the computational predictions with the experimental
results.
Table 2. Active Site Docking Parameters and Experimental
Activities of Compounds Predicted Not To Be Substrates for
P450cam
a reaction that competes with actual oxidation of the organic
substrate (Scheme 1). The uncoupling estimated in these studies
(Tables 1-3) is thus a maximum, as formation of water by
reduction of the ferryl species consumes two NADH molecules
rather than the single NADH used when the substrate is
hydroxylated or oxygen is reduced to H2O2.
The catalytic oxidation of camphor is both rapid and tightly
coupled, less (possibly much less) than 8% of the O2 consumed
by the enzyme being funneled into reactions other than substrate
hydroxylation (Table 1). Of the unnatural substrates, 2 is turned
over the most effectively and its oxidation is only slightly less
well coupled than that of camphor. For the other compounds,
the consumption of NADH (and oxygen) decreases in the
following order: 7 > 4 > 5 > 3 > 1 > 6. The coupling within
this series of compounds with regard to H2O2 formation
decreases in essentially the same order (7 > 4 > 5 > 1, 3, 6),
with the caveat that the degree of uncoupling for the last three
very poor substrates cannot be differentiated. The finding that
substrates at the lower end of the O2 consumption scale appear
as the most uncoupled is to some extent artifactual because
P450cam maintains a low residual turnover rate in the absence
of a substrate, all of which results in uncoupled product
formation. If the substrate-mediated increase in O2 consumption
is small, a large error is introduced by the background reduction
of O2 to H2O2. It is difficult to correct for the background O2
consumption and H2O2 production because the values for these
parameters probably decrease in an unpredictable manner in the
presence of a substrate. The experimental values in Tables 1-3
have therefore not been corrected for the background values.
a Background NADH consumption is 4-6 nmol min-1 nmol-1, but
the values in the table are uncorrected because the background may
differ in the presence of substrates. b A decrease is seen in the Soret at
417 nm with a corresponding increase at 430 nm.
calculate a KI. Compounds 7-10, 13, 14, and 16 significantly
inhibited the spectroscopically-measured binding of camphor.
The best, and nearly equipotent, inhibitors of the binding of
camphor were 8, 13, and 16.
P450cam Metabolites. Organic products have been identified
in the reactions of the P450cam system with substrates 1, 2, 5,
6, and 7 (Scheme 2). The products have been identified by
tandem gas-liquid chromatography-mass spectrometry, com-
parison with authentic standards, and in some instances, NMR
characterization. Comparison with authentic material shows that
benzylic hydroxylation of 1 by P450cam is followed by elimina-
tion of the propionate moiety to give acetophenone (1a). Gas
chromatography-mass spectrometry shows that oxidation of the
adamantylamine derivative 2 produces two monohydroxylated
metabolites. Comparison with authentic standards identifies the
two products as 2a and 2b. Comparison of the product derived
Oxygen Consumption and Uncoupling. The activity of
compounds as P450cam substrates has been evaluated by
measuring their ability to stimulate NADH consumption. A
comparable but not necessarily identical estimate of substrate
activity is provided by the ability of the compounds to stimulate
O2 consumption. The extent of uncoupled turnover can be
partially evaluated by comparing the increase in NADH and
O2 consumption with the increase in the formation of H2O2. In
the absence of highly precise O2 consumption measurements,
it has not been possible to estimate the uncoupled production
of water. Water is formed by reduction of the ferryl species in

5494 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
De Voss et al.
Table 3. Parameters for the Interaction of Compounds with the
L244A P450cam Mutant
Scheme 2. Structures of Products Formed in Cytochrome
P450cam Oxidation Reactions
spin
NADH,a
nmol-1
compd
no.
Ks,
µM
change, nmol min-1 % H2O2 metabolite
%
from O2
detection
Compounds Predicted To Be Substrates For Both
Wild-Type and L244A P450cam
camphor 4.9
85b
31
2
86 ( 3
24 ( 2
6 ( 1
7 ( 1
yes
yes
no
no
yes
8
9
10
11
3.3
66 ( 4
N/Ac
N/Ac
0.8
100
100
61 ( 4
<2
5 ( 1
16
23 ( 3
Compounds predicted to be substrates for L244A
but not wild-type P450cam
12d
13
N/Ac
1.1
<2
6 ( 2
100
98 ( 2
92 ( 4
100
100
no
no
no
no
no
36
<4
14
N/Ac
N/Ac
N/Ac
<4
<2
<2
6 ( 1
4 ( 1
6 ( 1
15d
16d
a These values have not been corrected for a background value of 6
nmol min-1 nmol-1 because it is likely to be lower in the presence of
substrates. b The maximum absorbance change in the L244A spectrum
caused by camphor binding is based on the assumption that the molar
absorbance coefficients for the spin state changes in the L244A and
wild-type proteins are the same. c N/A: no detectable spectroscopic
perturbation. d Compound precipitated at the usual 1 mM concentration,
so data were obtained at 0.15 mM substrate concentration.
studies indicate that saturation of the active site with camphor
converts 85% of the protein to the high-spin form, assuming
that the absorption coefficient for the spin state change is the
same for the wild-type and mutant enzymes. The L244A mutant
hydroxylates camphor to the normal 5-exo-hydroxycamphor
metabolite without the detectable formation of alternative
products (not shown). Measurements of the production of H2O2
indicate that the oxidation of camphor by the L244A mutant is
tightly coupled and produces no more H2O2 than turnover by
the wild-type enzyme. The rate of camphor hydroxylation is
150-250 nmol min-1 nmol-1, a value that is 2-3-fold lower
than that obtained for the wild-type enzyme under comparable
conditions. The L244A mutation thus appears to have only
minor consequences for the structure of the protein or its ability
to oxidize camphor.
from 5 with authentic materials identifies the metabolite as
2-(trifluoromethyl)benzoic acid (5a). Compound 6 is poorly
metabolized but gives a low yield of two products (6a, 6b)
shown by mass spectrometric analysis to be monohydroxylated
derivatives. The exact structures of these products have not
been established, although the mass spectrometric fragmentation
patterns indicate that the hydroxyl is not in the cyclohexane
ring (i.e., the mass increase is found in the fragment from which
the cyclohexyl ring has been eliminated).
The metabolism of 7 is unusual in that comparison with
authentic material identifies the product as the reduced com-
pound 7a. Due to the unusual nature of the product, control
incubations were carried out in the absence of the individual
components of the reaction mixture. In fact, the reductive
reaction takes place when 7 is incubated with NADH alone.
P450cam therefore does not appear to catalyze this transformation
of 7, although oxygen consumption experiments indicate that 7
triggers extensive coupled and uncoupled turnover of the enzyme
(see below). However, no other organic product has been
detected in the reaction mixture.
Binding of Compounds to L244A Mutant. Table 3
provides a summary of the binding of various compounds to
L244A P450cam. Two compounds (8 and 11) predicted to be
substrates for the wild-type enzyme, but which were not
detectable substrates, are shown by their Ks values to bind with
much higher affinity to the L244A mutant than to the wild-
type enzyme (Tables 1 and 3). The enhanced binding affinity
is paralleled by significant substrate-binding-mediated conver-
sion of the protein from the low to the high spin state (31 and
16%, respectively, for 8 and 11) even though neither compound
causes a detectable spin state change in the wild-type protein.
Furthermore, 8 and 11 cause substantial catalytic turnover, as
judged by NADH consumption of 23-24 nmol min-1 nmol-1
(Table 3), in contrast to the background level of NADH
consumption observed with the wild-type protein (Table 1). Of
the NADH consumed, 61-66% is used to reduce O2 to H2O2.
However, in the oxidation of 8, a part of the reducing equivalents
is used to hydroxylate the substrate to a product that migrates
as a single peak by GLC. The structure of this product has not
been determined. We have not succeeded in detecting an
organic product in incubations with 11, although the coupling
data suggest that one is formed. In contrast to the results with
compounds 8 and 11, compounds 9 and 10, which were not
substrates for the wild-type enzyme, are also not detectable
substrates for the L244A mutant (Table 3).
No organic products have been isolated from the incubations
with compounds 3, 4, or 8-11 even though compounds 3 and
particularly 4 consume a significant amount of NADH above
the background level.
L244A P450cam Mutant. Several compounds were predicted
to be unacceptable as substrates for P450cam based on differential
screening of their fit within the active site of P450cam and the
L244A model active site obtained by replacing the Leu by an
Ala side chain. To examine the predictive utility of identifying
compounds that fit within the L244A but not wild-type active
site, we constructed a cDNA coding for the L244A P450cam
mutant, expressed it in Escherichia coli, and purified the protein.
The L244A expressed at 20 °C is catalytically active and has
spectra identical with those of the wild-type enzyme (not
shown). The Ks value in 200 mM potassium phosphate buffer
for the binding of camphor to the mutant is 4.9 ( 1 µM, a
value approximately 4-fold higher than that for binding of
camphor to the wild-type enzyme (Ks ) 1.1 µM). Spectroscopic

Substrate Docking Algorithms of Cytochrome P450cam
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5495
Of the five compounds predicted with the DOCK default
parameters to bind to L244A but not wild-type P450cam (12-
16), only 13 gave a good binding spectrum (Ks ) 1.1 µM),
causing a 36% conversion from the low to the high spin state
(Table 3). However, NADH consumption remained low for
this compound and for the other four compounds predicted to
be substrates for the L244A mutant, and no organic products
were detected for any of the five compounds. Three of the
compounds (12, 15, 16) were relatively insoluble under the
reaction conditions and precipitated, so data for these compounds
were obtained with a lower (0.15 mM) substrate concentration
than that (1 mM) used for the other compounds.
that it was ranked by DOCK below the top 500 ligands for the
enzyme. The rank the DOCK contact score assigns to a ligand
is based exclusively on steric fit and is highest when the
substrate-enzyme surface contact is highest without intrusion
of the substrate into the protein structure. This contact scoring
system is designed for the evaluation of potential inhibitors,
for which maximum surface contact is desirable, so that the
ranking of a potential inhibitor increases in proportion to its
ability to fill the site. The present results suggest that filling
the site as tightly as possible is not a good indicator of activity
as a substrate (see below), so the DOCK rankings in Table 1
do not correlate with activity.
In view of the limitations, it is surprising that the DOCK-
based predictions are as good as they are. Of eleven compounds
identified with the DOCK default parameters as potential
P450cam ligands, seven either stimulate the consumption of
NADH and the production of either H2O2 or organic metabolites
or lead to the formation of organic products without a detectable
change in total consumption of NADH (Table 1). The seven
can therefore be classified as substrates. Impressively, none of
the five compounds identified as ligands for the L244A but not
wild-type enzyme are substrates for the wild-type enzyme (Table
2). Thus, the initial DOCK-based evaluation of the 16
compounds as potential ligands for wild-type P450cam correctly
predicted the behavior of the compounds as P450 substrates
75% of the time. Retrospective analysis indicates that even
better results are possible if the value for the minimum contact
distance between the protein and substrate atoms is optimized.
By using a single value of 2.9 Å for this minimum contact
distance, all but one of the compounds (94%) are correctly
accounted for as substrates or nonsubstrates (Figure 1).
Discussion
A major question examined here is whether the ability to fit
within the P450cam active site, as determined by computer-
assisted docking of rigid 3-dimensional structures into a rigid,
crystallographically determined P450cam active site model, can
be used to identify compounds that are substrates for the
enzyme. In addition to fit, the predictions are subject to the
additional constraints that the compounds must be largely un-
ionized and soluble under the assay conditions. Furthermore,
for the purpose of these studies, substrates are defined in their
broadest sense as compounds that trigger catalytic turnover and
lead to the production of either organic or reduced oxygen
metabolite(s).
It is important to understand the limitations of a rigid-body
docking approach whether one uses DOCK, as done here, or
an alternative docking algorithm. The most significant limitation
is that single energy-minimized conformers of potential sub-
strates are docked as rigid structures into an inflexible active
site model. No allowance is made for the existence of other
energy-accessible conformations or for conformational alter-
ations mediated by interactions with the protein. The same is
true of the active site into which the substrates are docked, as
it is derived from the enzyme crystal structure and is held
invariant during the docking studies. No allowance is made
for the adjustment of active site residues to accommodate
individual substrates. Furthermore, it is not clear which P450cam
active site structure is the most appropriate for docking
predictions. Substrates bind initially to the ferric state of the
protein, but catalysis requires reduction to the ferrous state,
oxygen binding to give the ferrous dioxy complex, and
formation of a putative ferryl complex (Scheme 1). The ferric
enzyme has been primarily used for the present studies.
Although docking to other enzyme intermediates must be
considered, docking to the Fe2+-CO complex as a model of
the Fe2+-O2 complex did not improve the predictions, sug-
gesting that for P450cam the ferric state is an adequate docking
target. Finally, in the case of the L244A mutant, the docking
studies were carried out with a model of the enzyme derived
from the crystal structure of the native rather than mutant
protein.
The predictions for the L244A P450cam model obtained with
the default DOCK parameters are less accurate. Of the five
compounds predicted to be substrates for the L244A mutant,
based on docking to a structure model created by replacing the
Leu-244 side chain with an alanine side chain (Table 3), only
one (14) is possibly a marginal substrate. The other four (12,
13, 15, 16), as judged by the failure to enhance NADH
consumption and the absence of detectable organic products,
are not substrates for the L244A mutant. One factor that must
be considered in this context is that the docking studies were
done with a model of the L244A structure derived from the
crystal coordinates of the wild-type enzyme rather than of the
mutant itself. The model assumes that the L244A mutation
decreases the volume of the residue at position 244 without
causing significant reorganization of the active site. This
assumption is valid for some but not all mutations, and the
accuracy of the results would clearly be improved if a model
based on crystal coordinates of the mutant itself were used for
the docking studies.
Important information with respect to the L244A mutant is
provided by the compounds that were predicted to be, but were
not, detectable substrates for the wild-type enzyme. Two of
these compounds (8, 11) are good substrates for the L244A
mutant (Table 3). Both compounds are predicted with the
default DOCK parameters to bind to the L244A mutant site,
but to do so without filling the cavity as completely as they fill
that of wild-type P450cam. As already noted, comparison of
the DOCK predictions with the experimental results reveals that
compounds that completely fill the active site are either very
poor or nonsubstrates. An imperfect measure of the tightness
of fit is provided by the number of orientations which DOCK
finds for a given molecule within the active site (Table 1).
Compound 2, the best substrate other than camphor, can be
docked in 56 orientations by using the default DOCK param-
Additional limitations are provided by (a) differences between
the true active site volume and the negative image of it created
by SPHGEN (a DOCK subroutine)19f and (b) the fact that we
have chosen to use DOCK 3.0 without considering electrostatic
or hydrogen bonding interactions. This is a serious limitation
in some contexts, but appears not to be critical here because
the binding of most substrates to P450 enzymes is controlled
by nonpolar interactions. Nevertheless, binding of camphor,
the natural substrate, to P450cam involves a hydrogen bond
between Tyr-96 and the camphor ketone group. This hydrogen
bond contributes to the fact that camphor is an excellent
substrate and is converted to a single product despite the fact

5496 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
De Voss et al.
eters. Of the other six compounds judged to be substrates, all
but two have many docking modes. The two exceptions are 3
and 6, neither of which is more than a marginal substrate. On
the other hand, the four compounds predicted to be ligands but
not found to be detectable substrates (8-11) can only be fit by
DOCK into the active site in one or two orientations. This
relationship between tightness of fit and substrate activity is
reinforced by the finding that compounds 8 and 11, which are
not detectable substrates for the wild-type enzyme, are good
substrates for the L244A mutant.
Although the volume of a compound does not correlate in a
simple manner with its activity as a substrate, it is clear from
the data in Tables 1 and 2 that compounds with volumes much
larger than ∼210 Å3 are unlikely to be substrates. This holds
for compounds (8-11) predicted by DOCK to be P450cam
substrates but which are not, as well as compounds actually
predicted to be nonsubstrates (12-16). The only exception is
compound 6, with a volume of 246 Å3, but it is a very poor
substrate. Although poor substrates can be found among the
smaller compounds (i.e., compound 1, volume 194 Å3), the best
indicator of a poor or nonsubstrate is a size that barely allows
the compound to fit into the active site. Retrospectively, this
observation rationalizes both the activity of 8 and 11 with the
L244A mutant and the inactivity of compounds 12-16 as
substrates for the L244A mutant. Compounds 8 and 11 fill the
P450cam active site too tightly to be more than trace substrates,
but are sufficiently well accommodated in the larger L244A
site to be acceptable substrates for that protein. Compounds
12-16, chosen to be too large for the wild-type site, by
definition barely fit into the L244A site. It is therefore NOT
surprising that they are not acceptable substrates for the mutant
enzyme. One approach to the selection of substrates for a
mutant enzyme suggested by these results is to employ a
differential DOCK procedure using a relatively long minimum
contact distance between the protein and the substrate atoms,
as this would bias the selection of substrates toward compounds
that do not absolutely fill the active site cavity. Retrospective
docking of the 16 CONCORD structures in the model of the
L244A active site, using a minimum contact distance of 2.9 Å,
correctly predicts that compounds 9 and 13-16 are substrates
(Figure 2). The two incorrect predictions are compounds 10
and 12, which are predicted to be substrates. Thus, 14 of the
16 predictions are correct. Approximately the same accuracy
is achieved with the Sybyl minimized structures (Figure 2).
Figure 2. Prediction of the activity of compounds 1-16 as substrates
for L244A P450cam as a function of the minimum distance allowed
between substrate and active site atoms. (A) DOCKing of Concord
generated substrate structures within the ferric L244A P450cam active
site model and (B) DOCKing of Sybyl minimized substrate structures
within the ferric L244A P450cam active site model. Solid bars indicate
experimental substrates and open bars experimental non-substrates. The
vertical lines are drawn at a value of 2.9 Å for the minimum interaction
distance.
by NADH. We have not been able to isolate a major product
from the oxidation of compound 7, which is shown by the
NADH consumption studies to be a good, well-coupled substrate
(Table 1). The oxidation of 7 apparently produces a polymeric
or water-soluble product that is not readily extracted and/or
characterized by the methods used in this study.
Conclusions
In the structure-based design of enzyme inhibitors, efforts
are usually made to fill the active site and to maximize
inhibitor-protein contacts. The present results indicate that the
same criteria are not appropriate for the selection of P450
substrates. Allowance must be made for a degree of active site
plasticity to allow (i) optimal positioning of the substrate within
the site and (ii) the active site and substrate motions required
for catalytic turnover. The present rigid body docking approach
makes no allowance for either global or localized conformational
flexibility, but an averaged degree of freedom can be introduced
into the DOCK predictions by varying the minimum distance
allowed between the substrate and the protein atoms. A short
value for this distance increases the acceptance of substrates
that are tightly wedged into the active site, whereas a long value
allows for an averaged looseness of fit and therefore, indirectly,
for a small degree of conformational and translational freedom.
Retrospective analysis of the data in Tables 1 and 2 suggests
that a single contact distance of 2.9 Å correctly rationalizes the
activity of 15 of the 16 compounds with P450cam (Figure 1). A
modestly lower accuracy is obtained in rationalizing the data
for the L244A mutant (Figure 2).
The DOCK approach predicts fairly well whether or not a
compound will initiate catalytic turnover but, not unexpectedly,
it is not a good predictor of the extent to which catalytic turnover
results in oxidation of the substrate rather than uncoupled
reduction of molecular oxygen. Minimal uncoupling, as with
camphor, is observed with compounds 2 and 7, intermediate
uncoupling with 4 and 5, and essentially complete uncoupling
with compounds 1 and 3 (Table 1). Compound 6 is a very
poor substrate but gives both H2O2 and hydroxylated products.
Likewise, the products formed from the organic substrates
cannot be predicted by the present approach, as it requires a
sophisticated analysis of both the intrinsic reactivity of the
substrate sites and the time averaged location and orientation
of the substrate within the active site. Predictions of this nature
have been made with the help of molecular dynamics simula-
tions for the epoxidation of substituted styrenes, sulfoxidation
of substituted thioanisoles, oxidation of ethylbenzene, and other
reactions.14 All the isolated and characterized products (1a, 5a,
2a,b, 6a,b) in the present studies are the result of straightforward
carbon hydroxylation reactions with the exception of 7a, a minor
product that is an artifact due to direct reduction of the substrate
Experimental Section
Materials. The cytochrome P450cam, putidaredoxin, and putidare-
doxin reductase clones in pIB125 were kindly provided by Dr. Julian
Peterson (University of Texas Southwestern Medical Center). Cam-
phor, 1 ((()-1-phenyl ethylpropionate), 2 (N-(1-adamantyl)acetamide),
3 (1-isopropyl-1-(m-tolyl)urea), 4 (methyl 4-acetyl-5-oxohexanoate),
7 (2-methylene-1,3,3-trimethylindoline), 10 (diethyl diallylmalonate),

Substrate Docking Algorithms of Cytochrome P450cam
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5497
11 (dibenzosuberenol), 12 (isodrin), 13 (DL-4-chloro-2-(R-methylben-
zyl)phenol), 14 (triethyl 1,1,2-ethanetricarboxylate), 15 (methyl 1-fluo-
renecarboxylate), 16 (flavone), and H2O2 were purchased from Aldrich.
Compound 5 (2-(trifluoromethyl)benzaldehyde) was purchased from
Lancaster, 6 (9-cyclohexylbicyclo[3.3.1]nonan-9-ol) from Janssen, and
8 (cedrol) from K&K Laboratories. Compound 9 was prepared by
acetylation of 9-fluorenol (from Lancaster) with acetic anhydride.
Putidaredoxin reductase was heterologously expressed in E. coli and
purified as reported in the literature.22
Analytical Methods. Absorption measurements were made on an
SLM-Aminco DW-2000 or a Hewlett-Packard Model 8452A diode
array UV-vis spectrophotometer. Gas-liquid chromatography was
performed on a Hewlett-Packard Model 5890 instrument equipped with
a flame ionization detector and interfaced to a Hewlett-Packard 3365
ChemStation. Tandem gas chromatography-mass spectrometry was
performed on a Hewlett-Packard Model 5890 gas chromatograph
interfaced with a VG-70 mass spectrometer.
Construction of a P450cam pCW Vector and Expression and
Purification of the Protein. An NdeI site was introduced by
polymerase chain reaction (PCR) at the 5′ end of the P450cam gene in
the previously prepared pBScam plasmid.23 The 5′ PCR primer
contained a SacI sequence and an NdeI site that spanned the ATG start
codon of the gene. The 3′ primer was complementary to a region of
the P450cam gene approximately 450 bp from the 5′ end of the gene.
The PCR product was ligated directly into the PCRII vector (In
Vitrogen, San Diego, CA) according to the manufacturer’s instructions.
The 5′ end of the gene was then isolated from a SacI/SphI digest of
the pCRII vector and ligated into similarly cut pBScam to give a new
plasmid pBScam1. This was sequenced to ensure that the wild-type
sequence was maintained. Digestion of pBScam1 with NdeI/XhoI gave
the gene which was ligated into NdeI/SaII cut pBACE. The gene was
then excised from pBACE as an NdeI/XbaI fragment and ligated into
a similarly cut pCW vector to give pCWcam. The expression system
transfected into DH5RF′ cells and plated with lawn cells in 0.6% 2×
YT top agar on 2× YT plates. The plates were incubated overnight at
37 °C.
Plaques were lifted onto nitrocellulose disks (BioRad Laboratories),
air dried, and baked for 3 h at 80 °C in a vacuum oven. After the
disks were rinsed in 2 × SSC, they were hybridized overnight at 40
°C in 5 × Denhardt’s solution, 5 × SSPE, 0.1% SDS, and 50 µg/mL
of tRNA with the 32P-kinase-labeled mutagenic oligomer.26 The filters
were then washed in 6 × SSC twice at 4 °C for 20 min each, followed
by washes at 65 °C in a tetramethylammonium chloride buffer to detect
one base-pair mismatches.27 X-ray film was exposed to the washed
filters and the plaques with the mismatch were isolated and re-screened
in a second round of plaque purification. The positive plaques from
the second round were isolated and amplified, and the DNA was isolated
and sequenced. A SphI/BstEII fragment was cleaved out of the repli-
cative form, recloned into the expression vector to give pBScamL244A,
and transformed into DH5R cells for expression. Individual colonies
were picked and sequenced.
A BspEI/HindIII fragment from digestion of pBScamL244A was
purified and ligated into a similarly cut pCWcam vector. Expression
yielded approximately 1000 nmol/L of the mutant protein, which was
purified as described for wild-type P450cam
.
Construction of a Putidaredoxin pCW Vector and Expression
and Purification of the Protein. To place the Pd gene provided in
the pIBI25 vector by J. Peterson into the pCW expression vector, it
was necessary to introduce an NdeI site at the 5′ end of the gene and
an XbaI site at the 3′ end. This was done by PCR, employing primers
that matched either the 3′ or 5′ end of the gene and also carried the
appropriate restriction sites. The PCR product was ligated directly into
the PCRII vector according to the manufacturer’s instructions and the
construct was sequenced. The gene was then removed as an NdeI/
XbaI fragment and was ligated into similarly cut pCW. The yield of
the protein expressed at 30 °C and purified as reported in the literature28
was approximately 2500 nmol/L.
yielded approximately 1500 nmol/L of P450cam
.
Expression and purification were carried out essentially as described
for the P450cam-putidaredoxin-putidaredoxin reductase fusion pro-
tein.24 Bacterial cells collected by centrifugation were resuspended in
50 mL of cold (4 °C) buffer (pH 7.5) containing 50 mM Tris/HCl, 50
mM KCl, 1 mM ethyldiaminetetraacetic acid, 0.1 mM phenylmethyl-
sulfonyl fluoride, 10 mM â-mercaptoethanol, 100 µM camphor, and
10 mg of lysozyme. After the mixture was stirred at 4 °C for 2 h, the
suspension was frozen in liquid nitrogen, thawed at 4 °C, and sonicated
with a Branson sonicator (medium power output) until the viscous
consistency due to the nucleic acids was disrupted (∼2 min). The
soluble fraction after centrifugation at 100 000g for 25 min was loaded
onto an anion exchange column (DEAE Fast Flow) equilibrated in
buffer A (50 mM KPi, 50 mM KCl, 100 µM camphor, and 10 mM
â-mercaptoethanol, pH 7.5), washed extensively with buffer A, and
eluted with a 0-500 mM KCl gradient in buffer A. The fractions
containing the P450cam, identified by absorption at 418 nm, were
combined and concentrated in an Amicon ultrafiltration cell equipped
with a YM-30 membrane prior to gel filtration chromatography (S-
200 Sepharose) in buffer A plus 200 mM KCl (but no camphor).
Residual camphor was removed by passage through a PD-10 (Phar-
macia) column. The purified proteins were stored at -70 °C.
Construction and Expression of the L244A P450cam Mutant. In
order to carry out the site specific mutation by a standard procedure,25
pBScam was digested with SphI and SaII, and the resulting 650 base-
pair bp fragment was isolated from low melting temperature agarose
and ligated into M13mp19, yielding the vector M13ps. The following
complementary 23 base-pair oligomer with the appropriate sequence
mismatch (CAG f GGC) was synthesized by the Biomolecular
Resource Center of the University of California, San Francisco:
CGACCAGTAAGGCGCCACACATC. The oligomer was used to
prime the polymerization reaction, using as a template single-stranded
DNA isolated from phage particles present in the media of M13ps-
infected DH5RF′ E. coli cells. The double-stranded DNA was
Binding of Compounds to P450cam. Increasing amounts of the test
compound in ethanol were added to the sample cuvette, and an equal
amount of ethanol was added to a matched reference cuvette, both of
which contained 1 µM camphor-free P450cam in 400 µL of 0.2 M
potassium phosphate buffer (pH 7.0). The ethanol concentration did
not exceed 1%. Dimethyl sulfoxide was used instead of ethanol with
compound 12. The solutions were mixed after each aliquot and the
difference spectrum recorded. The value of Ks was determined by
plotting 1/∆A vs 1/[S], the x-intercept of which yields -1/Ks from the
relationship 1/[S] ) [E]∆ꢀ/(Ks∆A) - 1/Ks, where [S] is the concentra-
tion of the test compound, [E] is the concentration of P450cam, ∆ꢀ is
the difference in molar absorptivity of the free and bound P450cam, Ks
is the spectroscopic binding constant of the compound, and ∆A is the
peak-to-trough absorbance in the difference spectrum.29,30
KI Measurement. KI values were obtained by determining the Ks
value of camphor as described above in the absence and in the presence
of at least three concentrations of the inhibitor. The four or more points
thus obtained were plotted against the inhibitor concentration, and the
KI value was determined from the graph.
Oxygen Consumption, NADH Consumption, and H2O2 Produc-
tion. The reaction was carried out in a custom made quartz cell in
which the O2 concentration and the decrease in the NADH absorption
maximum at 340 nm can be simultaneously monitored. The NADH
consumption rate was obtained from the resulting data by a zero-order
calculation. The reaction mixture contained, in a final volume of 1.1
mL, 200 mM potassium phosphate buffer (pH 7.0), 1 µM P450cam, 2
(26) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A
Laboratory Manual, 2nd ed.; Cold Spring Harbor Laboratory Press: Cold
Spring Harbor, New York, 1989; p 1.98-1.99.
(27) Wood, W. I.; Gitschier, J.; Lasky, L. A.; Lawn, R. M. Proc. Natl.
Acad. Sci. U.S.A. 1985, 82, 1585-1588.
(28) Gunsalus, I. C.; Wagner, G. C. In Methods in Enzymology; Fleischer,
S.; Packer, L., Eds.; Academic Press: New York, 1978; Vol. 52, Part C,
pp 166-188.
(23) Tuck, S. F.; Graham-Lorence, S.; Peterson, J. A.; Ortiz de
Montellano, P. R. J. Biol. Chem. 1993, 268, 269-275.
(29) Estabrook, R. W.; Peterson, J.; Baron, J.; Hildebrandt, A. In Methods
in Pharmacology; Chignell, C. F., Ed.; Appleton-Century-Crofts: New York,
1972; Vol. 2, pp 303-350.
(24) Sibbesen, O.; De Voss, J. J.; Ortiz de Montellano, P. R. J. Biol.
Chem. 1996, 271, 22462-22469.
(30) Rao, S.; Wilks, A.; Hamberg, M.; Ortiz de Montellano, P. R. J.
Biol. Chem. 1994, 269, 7210-7216.
(25) Zoller, M. J.; Smith, M. Methods Enzymol. 1983, 100, 468-500.

5498 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
De Voss et al.
µM PdR, 4 µM Pd, 250 µM NADH, and the substrate added in 10 µL
of ethanol. Substrates were present at a concentration 10-fold higher
than their Ks values or just below their solubility limits. The NADH
was added last to initiate the reaction. After 5 min of incubation, a
0.3-mL aliquot of the incubation mixture was added to 0.2 mL of 0.3
M H2SO4 and the mixture was stored on ice until the H2O2 assay was
carried out essentially as described previously (see below).14c â-Mer-
captoethanol was removed from the Pd prior to use by passage through
a G-25 gel filtration column (100 mM Tris, pH 7.4) because the thiol
interferes with the peroxide assay and, at higher concentrations, raises
the background NADH and O2 consumption values. To determine the
background consumption values, ethanol alone was added to the reaction
mixture. The O2 electrode was calibrated in distilled water at 25 °C,
in which the O2 concentration was taken to be 261 µM.
To quantitate H2O2, 0.125 mL of 6 mM ferrous ammonium sulfate
and 0.125 mL of 6 M ammonium thiocyanate were added to the
acidified aliquot of the reaction mixture taken above. The mixture was
centrifuged at 12 000 rpm for 2 min to pellet precipitated protein and
the absorbance was measured at 480 nm. The amount of H2O2 formed
was calculated from a standard curve made from a stock H2O2 solution.
Metabolite Analysis and Characterization. A 1-mL solution
containing 1 µM camphor-free P450cam, 2 µM putidaredoxin reductase,
8 µM putidaredoxin, 1 µM catalase (to remove any H2O2 formed in
the reaction), 1 mM test compound (added in 10 µL of ethanol), and
5 mM NADH was incubated at 25 °C for 1.5 h. A suitable amount of
organic solvent (e.g. chloroform or diethyl ether) was added to the above
reaction solution with 1-bromoadamantane as the internal standard. The
extracted material was characterized by gas chromatography, gas
chromatography-mass spectrometry, and nuclear magnetic resonance
spectroscopy.
Acknowledgment. We thank Sandra Graham-Lorence for
her help in expression of the proteins of the wild-type P450cam
system. This work was supported by grant GM25515 from the
National Institutes of Health and by a fellowship to Ole Sibbesen
from the Danish Veterinary and Agricultural Research Council.
Mass spectrometry was carried out at the University of
California, San Francisco, Biomedical Bioorganic Mass Spec-
trometry Facility (A. Burlingame, director) supported by
National Institutes of Health Grants RR 01614 and Liver Core
Center 5 P30 DK26743.
JA970349V