M. Basle et al.
Journal of Inorganic Biochemistry 220 (2021) 111446
+
hydrophobic pocket for an alcohol substrate and a metal binding pocket
for a zinc ion. In our initial design, we reasoned that the hydrophobic
catalytic site offers the necessary space to accommodate a non-biological
organometallic complex, without disturbing the nicotinamide cofactor
binding. Using this approach, our group recently reported a TbADH-
of the cysteine position within the NADP cofactor binding site of
TbADH on its ability to accommodate a non-natural functionality. The
mutant yielding the best bioconjugation results is then tested in the
+
+
reduction of NADP , NAD and 1-benzylnicotinamide, a hydrophobic
nicotinamide cofactor mimic. Docking studies are performed to shed
light on the mode of binding of the nicotinamide substrates to the arti-
ficial metalloenzymes.
+
based ArM for the reduction of NADP with rhodium bipyridine or
phenanthroline complexes [7]. This artificial formate dehydrogenase
was used in a coupled system with wild-type TbADH, for the in situ
recycling of NADPH during the reduction of 4-phenyl-2-butanone into
its corresponding (S)-alcohol. The ArM was developed using a covalent
binding approach: the single cysteine mutant TbADH 5M was devel-
oped, in which the cysteine at position 37 from the zinc binding site was
used as the anchoring point for the metal complexes, whilst the other
three native cysteines at positions 203, 283 and 295 were mutated into
alanine or serine residues. Two other alanine mutations were introduced
at positions 59 and 150 in order to remove the zinc binding site and to
free the space for the rhodium catalysts. The advantage of the ArMs over
the free Rh catalyst was demonstrated by the ability of the protein
scaffold to shield the metal complex from interacting with the native
TbADH during the recycling experiments, thus resulting in increased
stability of both the complex and the TbADH during ketone reduction
experiments.
2. Materials and methods
2.1. Computational docking
The Glide docking procedure within the Maestro software (Glide 6.7;
Maestro 11.6) was used to perform non-covalent and covalent docking,
with the default parameters as defined in the Schr o¨ dinger program [15].
From the crystal structure of wild-type TbADH (PDB 1YKF [16]), the
following mutations were created: H59A, D150A, C203S, C283A and
C295A. The catalytic zinc ion was removed from the file, to yield the
structure TbADH 5M. The cysteine in position 37 was mutated to alanine
to provide the scaffold TbADH 6M. Single cysteine mutations were
subsequently introduced at the required positions to create several
TbADH 7M scaffolds (174, 175, 178, 198, 203, 242, 243 and 266). Each
mutant was prepared by using the Maestro Protein Preparation Wizard
in the Schr o¨ dinger suite. Missing hydrogen atoms and side chains were
added to the structure by Prime-refinement throughout the pre-
processing [15]. During the refinement, water molecules with less
than three hydrogen bonds to other atoms were removed, which resulted
in no water in the binding site. The protonation/tautomer states and the
“flip” assignment of aspartate, glutamate, arginine, lysine and histidine
were adjusted at pH = 7.0 using PROPKA, in order to select the position
of hydroxyl and thiol hydrogen [17]. Finally, the structures were
geometrically optimized using the OPLS3 force field [18] with a RMSD
= 0.3 Å displacement of non‑hydrogen atoms as convergence parameter.
In our published work, the covalent bioconjugation of Rh complexes
to C37 resulted in mixtures of labelled and non-labelled protein, sug-
gesting that the thiol alkylation was not complete. Complete alkylation
of the C37 had previously been reported with iodoacetic acid [8], which
indicated that the lack of space around position 37 is likely to be
responsible for the incomplete labeling with the bulky organometallic
moieties. Moreover, the low activity observed with the resulting ArMs
also suggested that steric hindrance prevented nicotinamide binding in
the native position. The importance of the anchoring position within the
protein during the creation of ArMs by covalent modification was pre-
viously demonstrated. For example, when anchoring Cu(II)-
þ
þ
þ
phenanthroline catalysts to single cysteines in the
αRep scaffold, the
Ligands L1, L2, L3 benzylnicotinamide (BNA ), NAD and NADP
enantioselectivity of the Diels-Alder catalysis depended on the cysteine
position [9]. Similarly, incorporation of (2,2-bipyridin-5yl)alanine, a
metal-binding unnatural amino acid at different positions of the multi-
drug resistance regulators QacR, RamR or CgmR led to differences in
yields and enantioselectivities for Friedel-Crafts alkylation [10].
Furthermore, the question arose whether the ADH-based artificial
metalloenzymes would be able to reduce other nicotinamide derivatives.
Synthetic nicotinamide cofactor biomimetics are less costly and more
stable versions of their natural counterparts and are accepted by a range
of oxidoreductases [11]. Chemical catalysts developed for the regener-
ation of the natural cofactors NADH and NADPH in the presence of
formate [12] have already been shown to reduce other nicotinamide-
containing compounds, albeit with lower efficiency (about 2.5 times
lower) [13]. Ward and co-workers have also developed ArMs based on
biotinylated Ir(III)-N-sulfonyl-ethylenediamine complexes incorporated
into streptavidin, for the regeneration of NAD(P)H and their mimics
when combined with ene reductases, oxidases, oxygenases and glucose
dehydrogenase [14]. Our previously published results indicated that the
were prepared using the Ligprep tool from the Schr o¨ dinger suite, with
the OPLS3 force field. Generation of all possible protonation and ion-
isation states combinations was performed by using Epik in aqueous
solution at pH of 7.0 +/ꢀ 2.0. The metal complex Cp*Rh(L3)H was built
from the L3 structure by using the Maestro interface to build a pyramidal
metal centre, substituted by a pentamethyl cyclopentadienyl moiety
(Cp*) and hydrogen. The structure was then minimized through
“minimized selected atoms” task on Maestro workspace. Ligand L1 and
the complex Cp*Rh(L3)H were covalently docked into TbADH 7M using
Glide SP procedure in Maestro, with the grid centre for the docking
defined by the corresponding cysteine residue at the centre of the grid
and using the template of a nucleophilic substitution between the bro-
mide functionality and the corresponding thiol. For this step the re-
ceptor was kept rigid. The structure with the best Glide refined by Prime
score was used as the receptor for the covalently bound ligand poses and
for the following non-covalent docking.
þ
þ
þ
For non-covalent docking of BNA , NAD and NADP the grid for
the docking site was defined from the optimized protein structure at the
centroid of the active site (10 Å radius around the co-crystallized
NADPH ligand). The standard settings of a van der Waals scaling fac-
tor of 1.0 for nonpolar atoms was conserved and no constraints were
added. Nonpolar atoms were defined with absolute value of partial
atomic charges ≤0.25 [e]. The structure was first docked with Glide SP
score then with the more accurate Glide XP score, ranking the affinity
(or binding free energy) of ligands for the enzyme. For each non-
covalent docking, the structure with the best Glide XP score was used
for analysis.
+
specificity for NADP observed with the wild-type TbADH was not al-
ways translated to the ArMs, with similar activities being measured for
+
+
the reduction of NADP and NAD . Again, this suggested that the
nicotinamide substrate was not bound at its native binding site.
With this in mind, the aim of the current work is to gain a better
understanding of nicotinamide reduction catalysed by the TbADH-based
artificial metalloenzymes. In particular, an understanding of the sub-
strate scope and its binding to the active site of the ArMs is needed in
order to engineer these entities towards better functionality. We hy-
+
pothesized that the NADP cofactor site may offer more space than the
hydrophobic substrate pocket, for the binding of the organometallic
moiety. With the non-native catalyst positioned in the cofactor site,
there would in turn be more space for nicotinamide mimics to bind to
the hydrophobic substrate pocket. Therefore, we first evaluate the effect
2.2. Bioconjugation of ligands and complexes to TbADH variants
2.2.1. Bioconjugation with ligands L1, L2, L3
The optimized procedure was as follows. Ligand L1, L2 or L3 (100
2
Basle, Mattias
