Angewandte
Chemie
a problem with this particular molecule (see Table S3). In
addition if, as may be expected, the indole nitrogen proton
served in ThiH and HydG (see Figure S3). Conversely, NosL
Ser340 is found in a region which has significantly less
homology with these proteins. Taken together, these amino
acid sequence differences help define two zones, one that
recognizes the amino acid motif common to tryptophan and
tyrosine, and another that is likely to be specific for either
indole or phenol rings, thus strongly supporting equivalent
modes for substrate recognition between these proteins. It is
thus safe to conclude that H abstraction by ThiH and HydG
takes place at the amino group of tyrosine and not at its
phenolic OH group, as generally proposed but never observ-
from the positively charged skatole is transferred to
À
NHCCHCOOH to form a neutral NH CCHCOOH species
2
we find that DHG and the glycyl radical are equally probable
(
Table S2). Taken together, these results show that calcula-
tions strongly depend on both the protonation state of the
studied species and the solvent media used. Furthermore,
they confirm the central role the protein matrix plays in
controlling the reaction, and makes previous structural
knowledge essential. Indeed, if instead of considering the
[6a,7b]
protonation state of the amino acid fragment to be NH and
ed.
From an evolutionary standpoint, this appears to be
2
À
COO , which is charge-compensated by Arg323 as suggested
the most parsimonious solution because it involves a shared
binding mode and mechanism, centered at the common
amino acid motif.
+
from the structure, we optimized the geometry of CNH - and
2
À
-
COO originating from a zwitterionic tryptophan fragment,
and we observed spontaneous decarboxylation in both
ethanol and water.
Our findings will have a significant impact on current
[
6a,7b]
models for catalysis by ThiH and HydG.
For example, in
Our conclusion that a skatole radical and DHG are the
products of tryptophan CaÀCb cleavage by NosL (Fig-
our proposed mechanistic model, the 4-oxidobenzyl radical
(4-ODBC), for which Kuchenreuther et al. have provided
[
7b]
ure 3D) is also at odds with the assignment of a EPR signal
to a glycyl radical reported by Zhang et al. while character-
experimental evidence in HydG, results directly from the
homolytic cleavage of the CaÀCb bond of the CNH tyrosine
[
5]
izing NocL reaction products. However, their assignment is
not unambiguous. Indeed, the observed changes in the EPR
radical, without the formation of a tyrosyl radical intermedi-
ate. The same rational applies to ThiH because it would also
2
13
[6a,9]
spectrum, resulting from using either l-[ H ] or l-[1-C ]-
generate 4-ODBC and DHG, the substrate of ThiG.
The
8
labeled tryptophan, cannot be effectively used to discriminate
between a glycyl radical, a tryptophanyl radical centered on
the amino nitrogen atom, a MIA radical centered on C2, or
another unidentified species.
An intriguing feature of the NosL structure is the
presence of Tyr90 very close to C5’ (dOh-C5’ = 3.9 ꢀ) and in
direct interaction with the amino group of tryptophan
fate of this radical species, as the radical product in many
other radical SAM enzymes has not been thoroughly studied.
From a chemical standpoint H abstraction from the amino
nitrogen atom in both tryptophan and tyrosine allows the
facile cleavage of their CaÀCb bond. The orientation of
bound tryptophan in NosL, and presumably of bound tyrosine
in ThiH and HydG, positions the indole NH, or phenolic OH,
away from the reactive SAM. This positioning makes sense
because the formation of a stable radical will not lead to
effective CaÀCb bond breaking and production of the
(
Figure 2). This configuration raises the possibility of direct
H abstraction at the phenolic OH either by 5’-dAC or by the
resulting CNH amino radical. Tyrosyl radicals are known to be
[22]
stable in proteins, even for several days. The location of
Tyr90 is puzzling, to say the least. This residue, which is strictly
conserved in NosL and the tyrosine lyases, may play a role in
the catalytic mechanism during either CaÀCb bond cleavage
required products. In addition, aromatic-ring-delocalized
radical products should be good leaving groups. This feature
is particularly true for both HydG and ThiH because they
release p-cresol as a by-product.
or the termination reaction. As discussed in the Supporting
Information (see Figure S10), this situation is somewhat
reminiscent of what has been reported for spore photo-
product lyase (Spl), in which a tyrosine residue is proposed to
From our X-ray structure we conclude that neither the
[
4]
delocalized indole radical, as proposed for NosL, nor the
[
6a,7b]
tyrosyl OC radical, as postulated for ThiH and HydG,
are
plausible candidates to catalyze the CaÀCb bond cleavage in
[23]
serve as part of a relay to regenerate SAM. In fact, strictly
conserved tyrosine residues, able to play an equivalent
catalytic role, and located at similar positions, are also
observed in the active sites of the radical SAM enzymes
these enzymes. Conversely, the existence of a CNH tryptophan
radical species, its comparison with tyrosine lyases, and our
DFT calculations afford a solid basis for a shared CNH-based
catalytic mechanism for CaÀCb bond cleavage in NosL,
[
24]
[25]
[13]
LAM (Tyr290), RimO (Tyr227), HydE (Tyr303), and
ThiH, and HydG. The common arginine guanidinium moiety
fine-tunes the protonation state of the substrate, thus driving
the reaction towards CaÀCb bond cleavage instead of
[
14]
PylB (Tyr66). Although these are all intriguing features of
the NosL structure which may be extrapolated to other
enzymes, it is too early to propose a comprehensive mecha-
nism for MIA synthesis. Site-directed mutagenesis and
spectroscopic studies will be required to shed further light
on this process.
decarboxylation.
Surprisingly, the amino acid sequence of the radical SAM
CofH component of F -synthase, which also uses tyrosine as
0
a substrate, does not have residues equivalent to Tyr90,
Arg323, or Ser340. Consequently, this enzyme, which trans-
Given the amino acid sequence similarities, the related
nature of the substrates, and the sameness of the catalyzed
reaction, H abstraction from the amino group of tryptophan
in NosL may be safely extended to the tyrosine lyases ThiH
and HydG. Indeed, the invariant NosL residues Pro88, Tyr90,
and Arg323, which interact with tryptophan, are also con-
[
26]
fers the p-cresyl fragment of tyrosine to CofG, may have
a different way of cleaving its CaÀCb bond.
Finally, a thorough understanding of the structural basis
for MIA synthesis by NosL will open up the possibility of
modifying the enzyme to generate modified versions of the
Angew. Chem. Int. Ed. 2014, 53, 11840 –11844
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim