Biohydroxylation reactions catalyzed by enzymes and whole-cell systems

Article abstract of DOI:10.1006/bioo.1999.1135

The biohydroxylation of a number of cyclic substrates (3-24) containing aromatic side chains was used to compare substrate specificity and selectivity of hydroxylation using microbial enzymes and whole-cell biocatalysts. In general, the regioselectivity of reaction was remarkably similar between the different catalysts in that little aromatic or benzylic, but significant aliphatic hydroxylation was observed. However, a more detailed investigation of isolated products showed complementary substrate specificity, functional group compatibility, and regioselectivity of hydroxylation. Substrate specificity and regioselectivity could be further modulated by small changes to the nature of the aromatic side chain, which appears to play an important role in substrate recognition.

Full text of DOI:10.1006/bioo.1999.1135

Bioorganic Chemistry 27, 81–90 (1999)
Article ID bioo.1999.1135, available online at http://www.idealibrary.com on
Biohydroxylation Reactions Catalyzed by Enzymes and
Whole-Cell Systems
Sabine L. Flitsch,1 Suzanne J. Aitken, Cathy S.-Y. Chow, Gideon Grogan,
and Adam Staines
Centre for Protein Technology, University of Edinburgh, Edinburgh, United Kingdom
Received August 24, 1998
The biohydroxylation of a number of cyclic substrates (3–24) containing aromatic side chains
was used to compare substrate specificity and selectivity of hydroxylation using microbial
enzymes and whole-cell biocatalysts. In general, the regioselectivity of reaction was remarkably
similar between the different catalysts in that little aromatic or benzylic, but significant aliphatic
hydroxylation was observed. However, a more detailed investigation of isolated products showed
complementary substrate specificity, functional group compatibility, and regioselectivity of
hydroxylation. Substrate specificity and regioselectivity could be further modulated by small
changes to the nature of the aromatic side chain, which appears to play an important role in
substrate recognition.
᭧ 1999 Academic Press
The hydroxylation of nonfunctionalized hydrocarbon centers using biological sys-
tems is recognized as an important biotransformation (1). This reaction has traditionally
been achieved using whole-cell systems, but as more isolated enzyme systems and
their three-dimensional structure are now becoming available, enzymes can be used
for catalysis and are particularly important for studying substrate recognition and
regio- and stereoselectivity of hydroxylation. The regioselectivity of hydroxylation
is of particular importance to these biotransformations, because substrates often have
a number of potential hydroxylation sites and thus a very high degree of selectivity
is required. In our studies, we have looked at a particular class of cyclic substrates
bearing aromatic side chains (3–24) and compared the selectivity of hydroxylation
in defined enzymatic systems with that of whole-cell systems.
BIOHYDROXYLATIONS USING ISOLATED ENZYMES
The best studied of the cell-free hydroxylase systems is the cytochrome P450cam
monooxygenase from Pseudomonas putida which catalyzes the hydroxylation of
camphor 1 to 5-exo-hydroxy camphor 2 (Scheme 1). This enzyme was first reported
in the 1960s by Gunsalus and colleagues and has been widely used as a model system
because it is easily obtained as a soluble protein from heterologous expression (2)
1 To whom correspondence should be addressed at Department of Chemistry, King’s Buildings, West
Mains Rd., Edinburgh EH9 3JJ, United Kingdom. Fax: (0)131-6504737. E-mail: s.flitsch@ed.ac.uk.
81
0045-2068/99 $30.00
Copyright ᭧ 1999 by Academic Press
All rights of reproduction in any form reserved.

82
FLITSCH ET AL.
FIG. 1. Active-site residues of cytochrome P450cam monooxygenase. (Left) Wild-type enzyme with
camphor bound to the active site (3); (right) mutant Y96A with 4-hydroxydiphenyl methane bound to
the active site (14).
and a high-resolution crystal structure is available (3). The reaction is catalyzed with
remarkable regio-and stereoselectivity in that only 1 of 10 possible hydroxylation
products is formed. This selectivity is due to a combination of the shape of the active
site which provides multiple van-der-Waals interactions between enzyme and substrate
and a single hydrogen bond between the camphor carbonyl group and tyrosine 96 of
the enzyme. The importance of the hydrogen bond has been established by mutagenesis
of tyrosine 96 and studies with substrate analogues lacking the carbonyl group (4).
In the context of applications to biocatalysis, P450cam suffers from very narrow
substrate specificity. It therefore provides an ideal target for redesign by site-directed
mutagenesis. Our goal was to broaden the substrate specificity but at the same time
retain regioselectivity of hydroxylation. We were particularly interested in aromatic
substrates, such as polyaromatic compounds and substrates with aromatic side chains,
because of potential applications in bioremediation (5) and in the synthesis of chiral
intermediates (6–10), respectively.
Investigation of the active site of P450cam (Fig. 1, left) suggests that removal of
tyrosine 96 would both enlarge the substrate pocket and perhaps even provide an
aromatic binding site, since tyrosine 96 in the wild-type structure is flanked by two
phenylalanine residues (Phe87 and Phe98, Figure 1), which might provide aromatic
SCHEME 1. Hydroxylation of camphor using P450cam monooxygenase from Pseudomonas putida.

CATALYZATION OF BIOHYDROXYLATION REACTIONS
83
stacking interactions. With this in mind, a number of single-site mutants at position
96 were generated, of which the alanine mutant Y96A was the most interesting and
hence was studied in most detail (11,12).
First, simple aromatic compounds such as naphthalene and diphenylmethane (3)
were tested as substrates. The hydroxylation of naphthalene was much more efficient
with the Y96A mutant than with wild-type enzyme (13), and diphenylmethane 3 was
almost exclusively converted to 4-hydroxy diphenylmethane (Table 1) by Y96A,
whereas no hydroxylation was observed with wild-type enzyme (11). No benzylic
hydroxylation was observed, which emphasizes that the mode of binding of the
substrate to the enzyme active site can be more important for selectivity than C–H
bond strength.
Thus, the type of substrate binding we observed with the Y96A mutant (Table 1)
suggested to us that removal of the tyrosine aromatic side chain had indeed created
a binding pocket for aromatic rings as predicted, where the substrate now takes
the place of the former protein tyrosine side chain. To verify this, we conducted
crystallographic studies on the Y96A mutant. We have recently obtained the structure
of the mutant Y96A with 4-hydroxy diphenylmethane bound to the active site to 1.9
˚
A resolution in collaboration with Malcolm Walkinshaw at Edinburgh University
(14). These preliminary studies have revealed that, perhaps not surprisingly, some
significant rearrangement of active-site residues occurs. Comparison of the wild-type
with the Y96A mutant shows that most active-site residues remain in the same place,
but very importantly that the phenylalanine 87 side chain has moved into the original
site of tyrosine 96 (Fig. 1, right). Thus, the mutation has increased the size of the
substrate binding pocket which now appears to be lined at the bottom, away from
the active site, with aromatic side chains (Phe 87 and 98). Thus, rather than, as
predicted, providing an aromatic pocket at the previous site of tyrosine 96, Phe 98
appears to be stacking against aromatic side chains of the substrate, resulting in
regioselectivity of hydroxylation. We are currently investigating cocrystals of Y96A
with other substrates to see if this change of active residue orientation is generally
observed.
Further evidence for this aromatic stacking interaction comes from studies of the
mutant Y96A with substrates 4–9 (Table 1), where no aromatic hydroxylation was
observed, but good regioselectivity of hydroxylation of the aliphatic part of the
molecule occurred. However, this can also be explained by favorable van-der-Waals
interactions of the aliphatic part of the substrate with aliphatic protein side chains
(Val 244, Leu 247, Val 295) closer to the active site.
In general, substrates 3–9 in Table 1 demonstrate that the substrate range for
hydroxylation has been greatly expanded by the single mutation of tyrosine 96, in
particular since none of these compounds bind to wild-type enzyme. The aromatic
side chain that is necessary for good binding to the enzyme does not need to be part
of the desired substrate structure, but can be a protecting group as in 5–9, which
can be cleaved off after hydroxylation. This allows for the enzyme to catalyze the
hydroxylation of functionalized cyclohexane (5,6,9) and piperidine (7,8) derivatives,
which themselves are not substrates.
Another advantage for using aromatic protecting groups in these biotransformations
is that they can be used as a further tool to manipulate regioselectivity of hydroxylation.

84
FLITSCH ET AL.
TABLE 1
Hydroxylation Products Isolated from Incubation of Substrates with the Y96A Mutant of P450cam
Monooxygenase from Pseudomonas putida
Note. The relative configuration of the stereocenters in products is as indicated. However, absolute
configurations have not been determined.

CATALYZATION OF BIOHYDROXYLATION REACTIONS
85
For example, the benzoyl ester of cyclohexanol 5 is selectively oxidized to the 4-
hydroxyl derivative, whereas the benzyl ether 9 leads to a mixture of products, or
the benzoyl protected cyclohexenol 6 gives the allylic hydroxylation product, whereas
incubation of the benzyl ether 10 with Y96A resulted in recovered starting material.
It appears that a certain distance between the aromatic side chain and the hydroxylation
site is important in these substrates to ensure good selectivity of hydroxylation as
comparison of 4 and 5 shows, where 4 gives 44% of 3-hydroxylated product and 5
gives almost exclusively 4-hydroxylated material.
A great practical problem with using isolated enzyme systems for biotransformations
is the requirement for coenzymes and cofactors. For example, in the case of cytochrome
P450cam monooxygenase discussed here, it is necessary to add the corresponding
ferredoxin, ferredoxin reductase, and NADH as a source of electrons. These reactions
are therefore only being conducted on a milligram scale and it would be expensive
to scale them up much further. In the future, it may be possible to bypass electron
transport proteins by driving the reaction electrochemically (15), but in the meantime
the use of whole-cell systems, whether recombinant (16) or in the natural microorgan-
ism, is much more practical for preparative-scale biohydroxylations. A few whole-
cell systems which we have investigated in our group are discussed below.
BIOHYDROXYLATIONS USING B. bassiana
The entomopathogen Beauveria bassiana ATCC 7159 (redesignated from Sporotri-
chum sulfurescens) has been used for a number of biotransformations. Of particular
interest to us were studies by Fonken and Johnson at Upjohn Company (6,10) on the
selective hydroxylation of N-benzoyl alkylpiperidines to give some optically active
hydroxy alkylpiperidines. This work led to the proposal of a stereochemical model
for hydroxylation of these amides, which placed particular importance on the distance
between the amide oxygen and the hydroxylation site (1). Interestingly, our studies
with the isolated enzyme, the cytochrome P450cam mutant Y96A, had given similar
products of hydroxylation of the Cbz protected piperidine derivatives 7 and 8 to those
observed by Johnson, but modeling studies of the substrates at the active site of the
mutant using the available structural data (17) would suggest that no protein side
chain is suitably placed to form a hydrogen bond with the carbonyl oxygen of the
Cbz group. Rather, the size and flexibility of the aromatic substrate protecting group
appear to determine regioselectivity of hydroxylation. The comparison of N-carboxy-
benzoyl protected piperidine derivatives with benzoyl protected piperidines using the
B. bassiana whole-cell hydroxylation system would therefore give us more information
on the importance of the aromatic side chain, since the carbonyl group in both substrate
classes is placed at the same distance to the different potential hydroxylation sites of
the ring. In addition, we were interested to see if the Cbz group was more compatible
with whole-cell biotransformations, since it is a much more convenient protecting
group for amines than the N-benzoyl group.
A comparison of the hydroxylated products from both series of substrates (Cbz
and benzoyl protected) is shown in Table 2 (17). The protected parent piperidine
derivatives 7 and 11 gave both 4-hydroxylated products, but with several of the alkyl
substituted piperidines a significant change in the regioselectivity of hydroxylation
was observed. Thus, while biohydroxylation of N-benzoyl 4-methyl piperidine 13

86
FLITSCH ET AL.
TABLE 2
Hydroxylation Products Isolated from Incubation of Alkylpiperidines with Beauveria bassiana
ATCC 7159

CATALYZATION OF BIOHYDROXYLATION REACTIONS
87
resulted in a 2:1 mixture of the 4-hydroxylmethyl- and the 4-hydroxy-4-methyl-
piperidine derivatives, the Cbz compound 12 gave only the 4-hydroxy-4-methyl-
piperidine in 48% yield. The 3-methyl derivatives 14 and 15 and the 2-methyl deriva-
tives 16 and 17, on the other hand, resulted in the same product ratios. The most
significant change of regioselectivity was observed with the dimethyl-piperidines 8
and 18 and ethyl-piperidines 19 and 20. The Cbz derivatives 8 and 19 led exclusively
to the 4-hydroxylated materials, whereas the benzoyl derivative 18 gave only 3-
hydroxylated product and 20 gave a mixture of isomers.
One concern with whole-cell biotransformations is the lability of certain functional
groups to other enzymes in the microorganism. Esters, for example, are likely to be
hydrolyzed by microbial enzymes and compounds such as 5 and 6 would therefore
not be suitable for whole-cell biotransformations. The Cbz group is potentially labile
either to microbial carbamate hydrolases (18) or to benzylic hydroxylation (19). The
results presented in Table 2 suggest that degradation of the Cbz group is generally
low. However, we found one exception in substrate 8 where very little hydroxylated
product (5%) was isolated together with a large amount (75%) of benzoic acid (19),
indicating benzylic hydroxylation.
BIOHYDROXYLATIONS USING Rhodoccocus spp.
The genus Rhodococcus is a group of gram-positive bacteria that is widely distrib-
uted in both soil and aquatic environments and is known for their broad metabolic
diversity which includes hydrocarbons, many aromatic compounds, and larger mole-
cules such as steroids. A number of Rhodococcus strains have been investigated for
their ability to catalyze hydroxylation (20,21) and have demonstrated that this genus
of bacteria has great potential to provide useful hydroxylases. After investigating a
number of strains, we decided to study Rhodococcus rhodochrous ATCC 19067 and
Rhodococcus sp NCIMB 9784 in the context of our studies described above.
R. rhodochrous ATCC 19067 had been reported by Cardini and Jurtshuk (22,23)
to contain an inducible P450 monooxygenase (P450oct) which performed NADH-
dependent oxidation of n-octane to 1-octanol. We have found that this whole-cell
system can also accept a wide range of cyclic substrates such as illustrated in Table
3. Interestingly, the substrates were again exclusively hydroxylated in the aliphatic
rather than the aromatic ring. Cbz protected piperidine 7 was hydroxylated at the 4-
position as observed by B. bassiana and the P450cam mutant Y96A hence suggesting
similarities in substrate recognition. However, the Rhodococcus showed complemen-
tary catalytic activity to the P450cam and the B. bassiana systems with a number of
substrates: benzylether 9 led to a mixtures of products with the Y96A system, yet
gave only a single product with R. rhodochrous. Compounds 10, 21, and 22 were
hydroxylated in good yield by R. rhodochrous, yet were not substrates for the Y96A
mutant. Cbz protected dehydropiperidine 24 was not a B. bassiana substrate but could
be oxidized by R. rhodochrous. In only one of the cases studied (compound 23) did
we find benzylic hydroxylation.
The other strain we studied, Rhodococcus sp NCIMB 9784, contains a hydroxylase
which has been reported to generate 6-endo-hydroxycamphor from camphor (24),
although evidence for this reaction has been indirect because of further metabolism
of the alcohol product. Because of our interest in the substrate specificity of the

88
FLITSCH ET AL.
TABLE 3
Hydroxylation Products Isolated from Incubation of Substrates with Rhodococcus rhodochrous 9703
Note. The relative configuration of the stereocenters in products is as indicated. However, absolute
configurations have not been determined.
cytochrome P450cam hydroxylase from P. putida, we have studied this enzymatic
activity in more detail. The hydroxylase appears to be remarkably similar in its narrow
substrate specificity to the wild-type P. putida enzyme in that they are limited to
molecules of similar size to camphor and terpenoid substrates. Larger compounds
such as the Cbz piperidine 7 are not hydroxylated.

CATALYZATION OF BIOHYDROXYLATION REACTIONS
89
This hydroxylase will therefore not so much be of interest as a new complementary
hydroxylation system, but rather be useful for comparison to the P450cam from P.
putida in order to allow us to further understand factors underlying regioselectivity
of hydroxylation. Thus, we have been able to purify a cytochrome P450 monooxygen-
ase from Rhodococcus species NCIMB 9784 which accepts camphor as a substrate.
The first 16 N-terminal amino acids have been determined by sequencing and little
homology was found in this short sequence to the cytochrome P450cam monooxygen-
ase of P. putida. Studies toward the cloning and further characterization of this enzyme
are currently in progress in our laboratories.
CONCLUSIONS
Comparison of the substrate specificity of the different monooxygenases from
microbial sources studied here suggests some surprising similarities in selectivity of
hydroxylation, in that the 3- and 4-positions of the aliphatic rings are the most likely
to be hydroxylated. However, the substrate selectivity and the regioselectivity of
hydroxylation can be fine-tuned either by change of substrate structure (e.g., using
the Cbz instead of benzoyl-derivate) or by change of biocatalyst. With cell-free and
purified enzyme systems becoming more and more available, we can start to rationalize
regioselectivity of hydroxylation and redesign the active site of hydroxylases to suit
the particular requirements of reaction selectivity.
ACKNOWLEDGMENTS
The studies with P450cam monooxygenase were conducted in collaboration with Dr. Luet-Lok Wong
at the University of Oxford and crystallographic studies with Professor Malcolm Walkinshaw at the
University of Edinburgh. This work was funded by the BBSRC and the Edinburgh Centre for Protein
Technology.
REFERENCES
1. Holland, H. L. (1998) in Biotransformations (D. R. Kelly, Ed.), pp. 475–533, Wiley–VCH, Weinheim.
2. Koga, H., Rauchfuss, B., and Gunsalus, I. C. (1985) Biochem. Biophys. Res. Commun. 130, 412–417.
3. Poulos, T. L., Finzel, B. C., and Howard, A. J. (1987) J. Mol. Biol. 195, 697–700.
4. Atkins, W. M., and Sligar, S. G. (1988) J. Biol. Chem. 263, 18842–18849.
5. Kulisch, G. P., and Wilker, V. L. (1991). Biotechnol. Prog. 7, 93–98.
6. Johnson, R. A., Herr, M. E., Murray, H. C., Reineke, L. M., and Fonken, G. S. (1968) J. Org. Chem.
33, 3195.
7. Furstoss, R., Archelas, R., and Waegell, B. (1981) Tetrahedron Lett. 22, 445.
8. Carruthers, W., Prail, J. D., Roberts, S. M., and Willetts, A. J. (1990) J. Chem. Soc. Perkin Trans.
I, 2854.
9. Palmer, C. F., Webb, B., Broad, S., Casson, S., McCague, R., Willetts, A. J., and Roberts, S. M.
(1997) Bioorg. Med. Chem. Lett. 7, 1299.
10. Johnson, R. A., Murray, H. C., Reineke, L. M., and Fonken, G. S. (1969) J. Org. Chem. 34, 2279.
11. Fowler, S., England, P. A., Westlake, A. C. G., Rouch, D. R., Nickerson, D. P., Blunt, C., Braybrook,
D., West, S., Wong, L.-L., and Flitsch, S. L. (1994) J. Chem. Soc. Chem. Commun., 2761–2762.
12. Wong, L.-L. (1998) Curr. Opin. Chem. Biol. 2, 263–268.
13. England, P. A., Harford-Cross, C. F., Stevenson, J.-A., Rouch, D., and Wong, L.-L. (1998) FEBS
Lett. 424, 271–274.
14. Staines, A., Taylor, P., Flitsch, S. L., and Walkinshaw, M. Unpublished results.
15. Estabrook, R. W., Faulkner, K. M., Shet, M. S., and Fisher, C. W. (1996) Methods Enzymol. 272, 44–51.
16. Sibbesen, O., De Voss, J. J., and Ortiz de Montellano, P. R. (1996) J. Biol. Chem. 271, 22462–22469.

90
FLITSCH ET AL.
17. Aitken, S. J., Grogan, G., Chow, C. S.-Y., Turner, N. J., and Flitsch, S. L. (1998) J. Chem. Soc.
Perkin Trans. 1 20, 3365–3370.
18. Gotor, V., Quiros, M., Liz, R., Frifola, J., and Fernandez, R. (1997) Tetrahedron 53, 6421–6432.
19. Holland, H. L., Conn, M., Chinna Chenchaiah, P., and Brown, F. M. (1988) Tetrahedron Lett. 29, 6393.
20. Banci, L., Bertini, I., Eltis, L. D., and Pierattelli, R. (1993) Biophys. J. 65, 806–813.
21. Nagy, I., Compernolle, F., Ghys, K., Vanderleyden, J., and De Mot, R. (1995) Appl. Environ.
Microbiol., 2056–2060.
22. Cardini, G., and Jurtshuk, P. (1968) J. Biol. Chem. 243, 6070–6072.
23. Cardini, G., and Jurtshuk, P. (1970) J. Biol. Chem. 245, 2789–2796.
24. Chapman, P. J., Meerman, G., Gunsalus, I. C., Srinivasan, R., and Rinehart, K. L. (1966) J. Am.
Chem. Soc. 88, 618–619.