338
Alpha Class Glutathione Transferases
steroid A ring that could stabilize a negatively
charged oxygen or promote its formation. It
therefore appears that GST A3-3 catalyzes the
double-bond isomerization by a mechanism differ-
ent from that of the bacterial isomerase.
various α/β barrel enzymes catalyze a broad variety
of chemical reactions and at least some of them are
1
8
evolutionarily related.
The current study provides a striking example of
how completely unrelated catalytic activities can
emerge or recede by changing very few residues in
the active site of the same protein scaffold. GST A2-2
and GST A3-3 have obviously arisen by gene
duplication from a common precursor, as evidenced
by the clustering of alpha class GST genes on the
Suggested reaction mechanism
1
2
Gu et al. attempted to rescue our previously
suggested isomerization mechanism involving an
9
19
enolate intermediate by invoking a water molecule
same chromosome. Both enzymes share the ability
stabilizing the negative charge of the C3 oxygen.
However, we do not see any water molecule in the
relevant position in our electron density maps. This
suggests that either the proposed water molecule is
not in a fixed position in our structure, or that the
enolate intermediate is not necessary for the reaction
to occur. At low contour levels, there is some
electron density connecting the sulfur atom of
GSH and the C4 atom of AD. Stronger density for
the sulfur atom of GSH is found close by, which
might indicate two alternative states: GSH in the free
form and GSH in linkage to AD. Binding of GSH to
AD could be due to a small fraction undergoing
reversible Michael addition to the activated double
to conjugate GSH with electrophilic toxic com-
pounds, but only GST A3-3 has a significant steroid
isomerase activity.
Materials and methods
The expression and purification of GSTs A1-1, A2-2 and
9,20
A3-3 have been described.
The crystals for all three
proteins were obtained by mixing 5 μl of reservoir solution
(100 mM Tris–HCl pH 7.8, 18% (v/v) PEG 4000, 2 mM
dithiothreitol) with 5 μl of protein solution (10 mg/ml in
10 mM Tris–HCl pH 7.8) and 1 μl of 200 mM spermine. For
GST A2-2/AD, and GST A3-3/AD complex co-crystalli-
4
4
zation, Δ -AD was added to a final concentration of 5 mM
bond of Δ -AD; i.e. the product of the reaction.
5
in the protein solution. The GST A2-2/GSH complex was
obtained with 0.8 mM GSH in the protein solutions.
GSH was not added to the crystallization mixture in the
co-crystallization of AD and GST A3-3. However, the
electron density map clearly revealed density for GSH.
Apparently, this reflects the purification conditions of GST
A3-3, which was purified on a GSH-Sepharose column and
eluted with free GSH.
Addition of GSH to C5 in the substrate Δ -AD was
1
7
considered by Armstrong. However, a Michael
addition would not facilitate the proton transfer of
the catalyzed reaction and cannot be part of the
reaction mechanism. Any such addition would be
nonproductive.
We therefore conclude that the sulfur of GSH
acts as an acid/base catalyst in the proton transfer
Due to the high concentration of PEG 4000 in the
droplets, the crystals obtained could be frozen in liquid
nitrogen without additional cryoprotectant. Data were
collected at the ESRF synchrotron, Grenoble, France, using
beamline ID14-2, and at the MAX-lab synchrotron, Lund,
Sweden, using beamline 9/11-2. The images were
9
from C4 to C6 (Fig. 1b), as proposed. However,
the role of Tyr9 has to be reconsidered; it probably
acts as an auxiliary that promotes the transfer of a
proton to C6 of AD by hydrogen bonding to the
protonated sulfur of GSH. We showed earlier that
GST A3-3 displays significant steroid isomerase
21
22
processed with MOSFLM and DENZO and scaled
22
23
with SCALEPACK and SCALA. Details of data
collection, scaling, and merging statistics are given in
Table 1.
9
activity in the absence of GSH. Tyr9 is too far
5
away from C4 in Δ -AD (Fig. 4), and we propose
that a water molecule could take the place of the
sulfur of GSH and serve as the acid/base catalyst.
Tyr9 might assist this alternative catalytic entity by
hydrogen bonding to the water oxygen, rather
than to sulfur as in the presence of GSH. In both
cases, the enzyme operates by a mechanism
involving concerted proton removal from C4 and
delivery to C6.
All structures were solved by molecular replacement,
1
1
using human GST A1-1 (PDB code 1GSE) and human
1
2
GST A3-3 (PDB code 1TDI) as search models in the
2
4
program MOLREP.
The crystals of the GST A2-2/AD complex contained
four dimers per asymmetric unit. The electron density for
AD was very clear (Fig. 3b) in four monomers, but was
almost absent from the other four. In monomers with very
low density for AD, no other recognizable density (e.g.
water or other small molecules) was observed. Non-
2
5
Structural redesign in enzyme evolution
crystallographic averaging in RAVE was tried over all
eight monomers, but the resulting electron density for AD
was less satisfactory than that in the unaveraged maps of
the four best monomers. Consequently, AD was modeled
in only four monomers. In all cases, AD was either present
or absent simultaneously in both monomers of the same
dimer. The reasons for the absence or presence of AD in the
respective dimers were not obvious, either from their
structures or from differences in crystal contacts. The
asymmetric unit of the GST A2-2/GSH structure contained
four dimers per asymmetric unit. The electron density was
satisfactory for bound GSH in all monomers, so no
averaging was attempted.
For enzymatic reaction to occur, functional
groups, topologically organized in a particular
manner in the active site, are required. The protein
fold provides the arrangement of the catalytic
residues, but completely unrelated folds could
evolve to support analogous arrangements and
similar chemical reactions, as exemplified by subtil-
isin and chymotrypsin. On the other hand, proteins
with similar folds and even similar sequences can
have very different activities. For example, the
Tars, Kaspars
