Fig. 3 (A) Relative PABA conversion rates of AurF and mutated
variants. (B) HPLC profiles of in vitro biotransformation assay of
p-aminophenylguanidine (3) to the corresponding nitro compound 4
using (a) AurF mutant R96E/T100A/L202F; (b) native AurF;
(c) negative control (heat inactivated enzyme). Peak marked with
asterisk not related to enzyme reaction.
threonine and the amino group, whereas the L202F mutation
would hamper the close contact from PABA to L202-Cd.
Interestingly, these mutations resulted in a fourfold increase
of activity of the AurF variant compared to the native enzyme
(Fig. 3A). With this mutant at hand, we aimed at reexamining
the role of R96. One way to circumvent the complete shutdown
of activity was to inverse the functional groups of substrate
and its binding site in the sense of an umpolung. Specifically,
we aimed at exchanging R96 to glutamate in order to replace
the guanidinyl residue for a carboxylate. As a consequence, the
suitable substrate should bear a guanidinyl moiety instead of
the acidic functional group. Indeed, the triple mutant
(AurF-R96E/T100A/L202F) proved to be incapable of trans-
forming PABA into PNBA both in vivo and in vitro (Fig. 3A).
We next tested p-aminophenylguanidine (3), synthesized
from p-phenylenediamine and S-methylisothiourea sulfate, as
a substrate for the mutated AurF variant. HPLC and MS
analyses revealed that the mutant is capable of selectively
oxidising the amino group of 3 to yield the corresponding
nitro compound, p-aminophenylguanidine (4, Fig. 3B and 4).
The identity of 4 was confirmed by IR, spiking experiments
using a synthetic reference, and MSn, showing the M-46 peak
corresponding to the loss of the nitro group. The utilisation of
3 as a substrate and the rigorous chemoselectivity of the AurF
mutant in favor of the aniline amino group are remarkable. In
this context it may be noted that guanidinyl moieties are
readily oxygenated by various oxygenases that function as
NO-synthases,15–17 but here that functional group of 3
remains completely untouched by the AurF variant. We next
tested whether the structurally related p-aminobenzamidine (5)
is also accepted as a substrate, and indeed found that it can be
oxidised to the nitro derivative 6 (Fig. 4).
Fig.
4 Results from biotransformation experiments using
AurF-R96E/T100A/L202F and AurF, respectively.
refined substrate binding model. Obviously, the hydrogen
bond between R96 and the carboxy group of 1 is a prerequisite
for the conversion of 1 into PNBA. The mutant AurF-R96E/
T100A/L202F could not convert PABA because the carboxyl
of the substrate and Arg96Glu are not able to form hydrogen
bonds. The loss of this donor/acceptor relation is critical and
cannot be compensated by other possible hydrogen-bonds
(like to T167). To corroborate this, we also tested if wild type
and mutant were able to convert p-aminobenzylamine (7) or
p-aminoacetophenone (9), but as expected, no transformation
to nitro compounds 8 and 10 was observed (Fig. 4).
Obviously, two donors/acceptors have to be present to
warrant a snug placement of the substrate in the substrate
binding pocket. This scenario holds true for PABA in the
native N-oxygenase, as well as the alternative substrates,
p-aminophenylguanidine and p-aminobenzamidine, in AurF-
R96E/T100A/L202F, and the wild type enzyme. To illustrate
this, we modelled the binding of p-aminophenylguanidine to
both AurF and to the mutant AurF-R96E/T100A/L202F
(Fig. 5). In all cases, on the basis of the protein structure
and mutational analyses, plausible donor–acceptor relation-
ships can be elaborated, which position the substrate in a way
that the p-amino group points towards the oxygenated dimetal
cluster. Our refined model provided new insights into the
substrate binding and positioning, highlighting the crucial
roles of R96 and R96E, respectively. These residues, in
conjunction with flanking amino acids like Thr167 in the wild
type, place the substrates in the right position and at the
proper distance to react with the m-1,1 hydroperoxo metal
cluster.
Since both substrates, 3 and 5, are converted into nitro
compounds 4 and 6 by the AurF mutant, it is clear that the
R96E mutation does not have any detrimental impact on the
active site of the enzyme. One could argue that PABA would
be converted if R96 exerted solely spatial effects or alter the
conformation. However, much to our surprise we noted that
also the wild type enzyme can transform 3 and 5 into 4 and 6,
respectively. This intriguing observation served as the key to a
In sum, inspired by the AurF crystal structure and modelling
experiments, we have generated an AurF mutant with an
enlarged cavity at the active site, which resulted in a fourfold
increase in substrate turnover. To invert the molecular
recognition, we created a mutant in which arginine at position
96 was replaced for a glutamate. This N-oxygenase variant
proved to be incapable of transforming PABA, but accepted
c
This journal is The Royal Society of Chemistry 2010
Chem. Commun., 2010, 46, 7760–7762 7761