.
Angewandte
Communications
[
10]
[11]
enzyme HydG
and F420 biosynthesis protein CofH,
the hydrogen abstracted by the dAdo radical is from the
indole nitrogen atom. Incubation of ABPA in the assay
mixture containing reconstituted NosL, SAM, and dithionite
resulted in a product with a protonated molecular ion at
m/z = 176.0553 in LC-MS analysis, and this product was
absent in the control assay in which the supernatant of boiled
enzyme was used (Figure 2B). The suggested molecule
respectively, also showed that a dAdo radical abstracted the
solvent-exchangeable hydrogen atoms in the reactions. HydG,
CofH, and thiamine biosynthesis protein ThiH cleave the
CaÀCb bond of l-Tyr (Figure 1C), and together with NosL
and NocL that cleave the CaÀCb bond of l-Trp, these
enzymes form a phylogenetically related group (Figure 1D).
Likely because of the prevalence of tyrosyl and tryptophanyl
+
formula of the product is C H O ([M + H] calc. 176.0552,
10
8
3
[12]
radicals in biochemistry,
radical-mediated hydrogen abstraction was from the phenolic
position of l-Tyr for HydG, CofH, and ThiH, and from
the indole nitrogen atom of l-Trp for NosL. However,
a recent structural study on NosL showed that the l-Trp
amino group is placed spatially close to dAdoH in the enzyme
active site, indicating the hydrogen abstraction may possibly
it was believed that the dAdo
0.6 ppm error), which is consistent with an MIA analogue 3-
methyl-2-benzofuranic acid (MBA). The identity of MBAwas
further confirmed by co-eluting with the synthetic standard in
HPLC analysis (Figure 2B). The 3-methylindole (MI) ana-
logue 3-methylbenzofuran (MB) was also observed in the
reaction mixture (Figure 2C), and was confirmed by GC-MS
analysis (Figure S1) and by co-eluting with the synthetic
standard (Figure 2C). These results clearly demonstrated that
the hydrogen abstracted by the Ado radical in MIA biosyn-
thesis is not from the indole nitrogen but likely from the
[13]
[14]
[15]
[7]
[16]
be from the amino group instead of the indole nitrogen.
To biochemically validate the site of the dAdo radical-
mediated hydrogen abstraction in NosL catalysis, we synthe-
sized 2-amino-3-(benzofuran-3-yl)propanoic acid (ABPA)
[16]
amino group of Trp, as suggested recently. Because of the
phylogenetically close relationship (Figure 1D) and similar
catalytic activities of NosL-related enzymes, our results
suggested that the Ado-radical-mediated hydrogen abstrac-
tion in the catalysis of ThiH, HydG, and CofH is not from the
phenolic group but likely from the amino group of l-Tyr.
Surprisingly, we observed a major peak (shown by the red
asterisks in Figure 2B) in HPLC analysis of the assay mixture,
and this peak did not appear in the control reaction (Fig-
ure 2B). LC-MS analysis showed that the compound exhib-
ited a protonated molecular ion at m/z = 162.0918, indicating
(
see the Supporting Information, SI) and used it as a mech-
anistic probe. ABPA is structurally very similar to Trp but
lacks a solvent-exchangeable hydrogen atom on the aromatic
ring (Figure 2A). This molecule can preclude the reaction if
+
the molecule formula C H NO ([M + H] calc. 162.0919,
10
12
0
.6 ppm error). MS/MS analysis suggested that this compound
may be 2-(benzofuran-3-yl)ethanamine (BEA), a decarboxy-
lated product of ABPA (Figure S2). The identity of this
compound was further confirmed by co-eluting with the
authentic BEA standard in HPLC (Figure 2B). Production of
BEA is time-dependent, and its yield is about 20-fold and 8-
fold higher than that of MBA and MB, respectively (Fig-
ure 2D), suggesting that NosL was transformed to a non-
oxidative decarboxylase by ABPA. To data, BlsE involved in
the biosynthesis of the peptidyl nucleoside antibiotic Blasti-
cidin S is the only known radical SAM enzyme that catalyzes
Figure 2. Tuning NosL chemistry by using an unnatural substrate
ABPA. A) ABPA can be converted to MBA and 3-methybenzofuran
[17]
a non-oxidative decarboxylation reaction.
The sequence
similarity between BlsE and NosL is barely detectable,
demonstrating the remarkable catalytic versatility of radical
SAM superfamily enzymes. We failed to detect tryptamine in
the NosL assay with Trp, indicating that the benzofuran ring
may play a pivotal role in tuning enzyme conformation for the
cryptic decarboxylase activity of NosL.
(
MB) by CaÀCb bond cleavage (minor pathway, ca. 15% conversion),
or to BEA by decarboxylation (major pathway, ca. 85% conversion).
B) HPLC analysis of the ABPA reaction mixture, using the supernatant
of boiled NosL as a negative control (trace i), or using reconstituted
NosL as the catalyst (trace ii). The NosL reaction mixture (trace ii) was
also co-injected with either the synthetic MBA standard (trace iii) or
BEA standard (trace iv) to confirm the production of these two
compounds in the reaction. C) Production of MB in NosL-catalyzed
reaction. Trace i and ii are the same from (B), whereas trace iii is the
reaction mixture (trace ii) co-injected with the synthetic MB standard.
HPLC analyses in (B) and (C) were performed using different eluting
programs detailed in the SI. Red, blue, and green asterisks indicate
BEA, MBA, and MB, respectively; the solid circle in (C) indicates an
unknown compound that is independent of NosL-catalysis. D) Quanti-
tative production of MBA, MB, and BEA in the NosL-catalyzed
reaction. Assay was carried out by incubating 300 mm ABPA with ca
To study the dAdo radical-mediated hydrogen abstraction
in APBA decarboxylation, we performed the assay in a buffer
containing 67% D O. LC-MS analysis clearly showed incor-
2
poration of deuterium into dAdoH. Although the majority of
labeled dAdoH is singly deuterated, doubly deuterated
dAdoH is also observable (Figure 3A,B), demonstrating
that the hydrogen abstracted by the dAdo radical in NosL
catalysis is reversibly from the solvent-exchangeable site
[10]
[11]
similar to the catalysis of HydG and CofH. In addition,
we observed significant deuterium incorporation into BEA.
This compound is singly deuterium-labeled and the doubly
8
0 mm reconstituted protein, 500 mm SAM, and 2 mm of sodium
dithionite in 50 mm MOPS buffer (pH 8.0). The time scale is not
shown proportionally.
9
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 9021 –9024