Structure and reaction mechanism of pyrrolysine synthase (PylD)

Article abstract of DOI:10.1002/anie.201301164

The final step in the biosynthesis of the 22nd genetically encoded amino acid, pyrrolysine, is catalyzed by PylD, a structurally and mechanistically unique dehydrogenase. This catalyzed reaction includes an induced-fit mechanism achieved by major structural rearrangements of the N-terminal helix upon substrate binding. Different steps of the reaction trajectory are visualized by complex structures of PylD with substrate and product. Copyright

Full text of DOI:10.1002/anie.201301164

Angewandte
Chemie
DOI: 10.1002/anie.201301164
Pyrrolysine Biosynthesis
Structure and Reaction Mechanism of Pyrrolysine Synthase (PylD)**
Felix Quitterer, Philipp Beck, Adelbert Bacher, and Michael Groll*
Pyrrolysine (Pyl, 4), the recently discovered 22nd genetically
encoded amino acid, has almost instantaneously become
a hotspot of protein biochemistry,[1] although its natural
occurrence appears to be limited to just three proteins that
are involved in the breakdown of methylamines in a small
subgroup of archaea and bacteria.[2] The novel amino acid is
incorporated by a cognate pair of a tRNA and its synthetase
(specified as pylT and pylS, respectively) through recognition
of an amber stop codon (UAG).[2a,3]
Biosynthesis of 4 is accomplished from two molecules of
lysine (1) by sequential action of PylB, PylC, and PylD
(Scheme 1).[4] Specifically, the iron–sulfur S-adenosylmethio-
nine protein PylB catalyzes the conversion of 1 to (3R)-3-
methyl-d-ornithine (3MO, 2),[5] which is subsequently hooked
up ATP-dependently to the e amino group of a second lysine
molecule by PylC resulting in 3.[6] Dehydrogenation at the C5
position of the methylornithine moiety of the isopeptide and
subsequent ring closure catalyzed by PylD completes the
biosynthesis of the unusual amino acid pyrrolysine. The
present report on PylD describes the structural investigation
and implementation on the reaction mechanism of the
enzymes required for the biosynthesis of pyrrolysine and
may open novel opportunities for the harnessing of the system
for biotechnology purposes.[7]
We expressed the pylD gene of Methanosarcina barkeri
Fusaro in an Escherichia coli strain. The recombinant protein
was purified by metal-affinity chromatography and showed in
vitro catalytic activity with Km = 3.6 mm Æ 0.5 mm (Figure S1
in the Supporting Information) and kcat = 0.76 sÀ1 Æ 0.04 sÀ1
using the surrogate l-lysine-Ne-d-ornithine (LysNe-d-Orn,
3a) as substrate. PylD was crystallized together with 3a and/
or a pyridine nucleotide cofactor (NADH or NAD+). Crystals
diffracted to a maximum resolution of 1.8 ꢀ and starting
phases were obtained by a combination of single-wavelength
anomalous diffraction (SAD) methods using a selenomethio-
nine derivative and twofold noncrystallographic symmetry
(NCS) averaging. Real-space electron density map averaging
was performed with MAIN[8] in combination with CCP4
routines.[9] Model building was carried out with MAIN and
refinement was completed with REFMAC5[10] (see Table S1).
After structural elucidation of the selenomethionine-labeled
PylD holoenzyme (PylD:holo (peak), PDB ID: 4JK3), we
crystallized and determined the structure of native PylD in
the presence of NAD+ (PylD:holo, PDB ID: 4J43) to 2.2 ꢀ
resolution (Rfree = 20.1%, Table S1). Its molecular architec-
ture is shown schematically in Figure 1a: the C-terminal
segment (residues 139–259) resembles a Rossmann motif of
five parallel b strands (S6–S10) with 21345 topology, which is
N- and C-terminally flanked by helices H5 and H10,
respectively (secondary-structure nomenclature: see Fig-
ure S3), yielding in the DALI search[11] a highest Z-score of
15.6 for the Rhodospirillum rubrum Transhydrogenase
Domain I (PDB ID: 1L7D). The N-terminal segment (resi-
dues 1–138) comprises a b sheet of five strands (S1–S5) whose
overall orientation is orthogonal to that of the Rossmann
motif of the C-terminal part. Interestingly, the C-terminal
helix (H10) of PylD is wedged between the two b sheets
supporting the correct fold and orientation of the N-terminal
half of the dehydrogenase. In contrast to the Rossmann fold,
the DALI search for proteins resembling the topology of the
N-terminal segment resulted in only some similarities with the
tRNA binding domain of certain tRNA synthetases (Z-
score < 8).
Scheme 1. Biosynthesis of pyrrolysine. Note: PylB generates only 3MO
(2); PylC and PylD also catalyze the reactions of 2a and 3a,
respectively.
[*] F. Quitterer, P. Beck, Prof. Dr. A. Bacher, Prof. Dr. M. Groll
Center for Integrated Protein Science Munich (CIPSM) at the
Department Chemie, Lehrstuhl fꢀr Biochemie
Technische Universitꢁt Mꢀnchen
Lichtenbergstrasse 4, 85747 Garching (Germany)
E-mail: michael.groll@tum.de
[**] We thank the staff of the beamline X06SA at the Paul Scherrer
Institute, Swiss Light Source, Villigen (Switzerland) for their help
with data collection and Katrin Gꢁrtner for excellent technical
assistance. We acknowledge Dr. Anja List who participated in the
PylD project. This work was supported by the Hans-Fischer-
Gesellschaft, by Award No. FIC/2010/07 from the King Abdullah
University of Science and Technology (KAUST), and by the Deutsche
Forschungsgemeinschaft (DFG, grant GR1861/7-1).
The nicotinamide adenine dinucleotide coenzyme NAD+
is bound to PylD in an extended conformation, inside
a groove at the C-terminal pole of the Rossmann b sheet;
both furanose rings have C2’-endo conformation. A typical
VXGXGXXGXXXA motif[12] (residues 146–157) is part of
the coenzyme binding site and the backbone elements are
predominantly involved in a network of hydrogen bonds with
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201301164.
Angew. Chem. Int. Ed. 2013, 52, 7033 –7037
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7033

.
Angewandte
Communications
Figure 1. Overall structure of PylD. a) Topology with secondary structure elements shown for PylD in complex with 4a. The Rossmann motif is
highlighted against a dark gray background and the substrate-binding motif against a gray background. The N-terminal part and residues 52–61,
which participate in an induced-fit mechanism upon substrate binding, are presented in red and blue, respectively. b) Left panel: Ribbon drawing
of the PylD:holo structure (open conformation). The C-terminal Rossmann fold is shown in dark gray and the N-terminal substrate binding
segment is depicted in gray. The N-terminus (red) and a loop connecting S1 and H3 (blue) are flexible (indicated by dashed lines). Right panel:
Surface representation of the PylD:holoenzyme structure. Disordered regions are shown in accordance with the ribbon drawing. c) Left panel:
Ribbon drawing of PylD cocrystallized with 3a and NAD+ (closed conformation). The active site contains the furnished product 4a (C atoms of
the lysine moiety are shown in yellow, C atoms of the pyrroline moiety in green); further color coding is according to (b). Note: all residues of the
structure shown in (c) are well defined in the electron density map. Right panel: Surface representation of the PylD:4a structure.
NAD+ (Figure 2a). Both hydroxy groups of the adenosyl
moiety are hydrogen-bonded by the strictly conserved side
chain of Asp171. Notably, in the absence of substrate, the four
N-terminal amino acids and residues 56–59 of a loop con-
necting S1 and H3 are structurally disordered and thus are not
defined in the electron density map. Therefore, the coenzyme
complex of PylD reveals an open cavity at the interface
between the two b sheets that serves as the substrate binding
site (Figure 1b).
To investigate the reaction trajectory of PylD, we synthe-
sized the substrate analogue 3a (for details see the Supporting
Information) and determined the structure of PylD cocrystal-
Figure 2. a) Amino acid residues in contact with NADH (black) and 4a (pyrroline moiety: green, lysine moiety: yellow). The interaction distances
are shown in ꢂ. b) Connolly surface representation of the active site of PylD. The atoms of ligand 4a are presented as spheres with their
corresponding van der Waals radii. Solvent molecules are drawn as smaller spheres (water: red, ammonia: blue). Note: the ammonia is located
inside a channel leading to the protein surface. Orientation of the structure is according to Figure 1. The stereoview of Figure 2b is presented in
the Supporting Information (Figure S2).
7034 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 7033 –7037

Angewandte
Chemie
lized with NADH and 3a (PylD:3a, PDB ID: 4J4B) to 1.9 ꢀ
resolution (Rfree = 24.0%, Table S1). Remarkably, substrate
binding is accompanied by breaking of helix H1, resulting in
a major topological reorganization of the N-terminal segment
that is conducive to closure of the substrate cleft by the
proximal part of the broken helix (residues 1–11, colored in
red in Figure 1). As a consequence, the bound substrate is
almost entirely enclosed in a narrow, hydrophobic tunnel
formed by residues Leu3, Leu4, Ile60, Phe63, Ala103, and
Leu247 as shown in Figure 2a. The C- and N-terminal end of
the substrate and the isopeptide bond are hydrogen-bonded
to backbone elements and water molecules (Figure 2a). The
ornithine motif of the substrate analogue adopts a sickle-
shaped conformation, with the amino and amide nitrogen
atoms in close proximity (2.0 ꢀ). At a distance of 2.6 ꢀ to C4
of the pyridine ring of the cofactor, the pro-R hydrogen atom
of the aminomethylene group appears poised for abstraction
as hydride; thus, the orientation of NADH in the PylD:3a
complex structure towards the substrate suggests that PylD
acts as an R-type dehydrogenase (Figure 3a).
A further approach was to confirm that PylD crystallized
in an active state and that the induced-fit mechanism is part of
the catalytic reaction. We therefore soaked a PylD/NAD+
cocrystal with 3a (PylD:soak, PDB ID: 4J49); we observed
visible changes of the crystal morphology and transient
formation of bubbles inside the crystal. The substrate-treated
crystal still diffracted to a resolution of 2.2 ꢀ (Rfree = 20.1%,
Table S1). Elucidation of the structure revealed two mole-
cules in the asymmetric unit; whereas one molecule shows the
open conformation without bound substrate or product, the
electron density for the second molecule indicates that 3a had
been converted to the product pyrroline-carboxy-lysine (Pcl,
4a), though the ligand displays only partial occupancy as
indicated by an increased Debye–Waller factor. We thus
cocrystallized PylD in the presence of 3a and NAD+ and
determined the complex structure (PylD:4a, PDB ID: 4J4H)
to 1.8 ꢀ resolution (Rfree = 21.8%, Table S1). The overall
architecture of the structure does not differ from that of the
PylD:3a complex (Figure 3b); however, the differences
between the substrate and the furnished product 4a are
clearly depicted in the omit electron density maps (Fig-
ure 3c).[13] Furthermore, there is
a most notable change in the appear-
ance of a fixed small molecule that is
tentatively interpreted as ammonia
produced as the second reaction
product besides 4a (Figure 3b),
whereas all defined solvent mole-
cules embracing the ligands are
present in both the substrate and
product PylD crystal structures with
a
root-mean-square
deviation
(r.m.s.d.) of less than 0.2 ꢀ (Fig-
ure 3c).
Our crystallographic data illus-
trate that PylD is a member of the
large FAD/NAD(P)-binding Ross-
mann-fold superfamily. Neverthe-
less, no assignment to a specific sub-
group can be made on the basis of
structural arguments. With a length
of 259 amino acids, PylD is in the
250–300 residue range of the short-
chain
dehydrogenase/reductase
family (SDR). However, the Ross-
mann motif of SDR members is
typically located at the N-terminal
end,[14] as opposed to PylD where it
extends from positions 143–243 and
is followed by helix H10 which
interacts closely with the N-terminal
part housing the substrate-binding
site. Even more importantly, SDR
members whose structures so far
have been determined contain
a highly conserved tyrosine residue
which serves as a base for the
abstraction of a proton and thus
facilitates the hydride transfer.[15]
Figure 3. Stereoview of the active site of PylD. The 2FoÀFc omit electron density map is contoured
at 1.0 s. The N-terminus is shown in red and the loop connecting S1 and H3 is depicted in blue.
The active-site residues and ligands are presented as stick models and solvent molecules are drawn
as balls (red: water, blue: ammonia). Hydrogen bonds are indicated by dashed lines. a) PylD in
complex with 3a; b) PylD in complex with 4a; c) superposition of 3a (dark green, blue mesh) and
4a (green, red mesh). Orientation of the structure is according to Figure 1.
Angew. Chem. Int. Ed. 2013, 52, 7033 –7037
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7035

.
Angewandte
Communications
PylD has neither a conserved tyrosine nor any other amino
acid in its active site cavity that could act as an auxiliary base
to support proton release from the amino nitrogen. We also
note that to our knowledge dehydrogenases have not been
described to undergo the type of induced-fit motility observed
with PylD. Therefore, PylD appears to be without known
precedent beyond the general assignment to the Rossmann-
fold superfamily nor can any analogy to a specific family be
based on its unique primary structure. Rather, it must be
emphasized that the allowed sequence space for PylDꢁs
functionality is, per se, very large. In a pairwise comparison of
the 22 hypothetical PylD sequences in the Universal Protein
Resource (UniProt) database, identities as low as 22% are
observed (Table S2).
complex pathway, for example through the formation of an
aldehyde intermediate.
Overall, the PylBCD biosynthetic machinery essentially
encompasses three chemical reactions, each representing
a challenge for the synthetic chemist. Whereas PylB and
PylC catalysis is strictly dependent on their substrates 1, 2,
and 2a, active PylD literally harbors enough room for
innovation beyond natureꢁs scope of substrates and products.
In particular, its binding cavity for the ornithine moiety is
large enough to accommodate a variety of different substrate
analogues. The presented results provide new directions for
future investigations on pyrrolysine derivates synthesized and
incorporated into a defined primary sequence by means of the
PylDST machinery in vivo.
Based on the crystal structures of PylD:holo, PylD:3a,
and PylD:4a, we propose the reaction mechanism shown in
Scheme 2. The observed topology of the substrate analogue
Received: February 8, 2013
Revised: March 9, 2013
Published online: May 29, 2013
Keywords: amino acids · dehydrogenases ·
.
Methanosarcina barkeri · PylD · pyrrolysine
[1] C. Hertweck, Angew. Chem. 2011, 123, 9712 – 9714; Angew.
Chem. Int. Ed. 2011, 50, 9540 – 9541.
[2] a) G. Srinivasan, C. M. James, J. A. Krzycki, Science 2002, 296,
1459 – 1462; b) B. Hao, W. Gong, T. K. Ferguson, C. M. James,
J. A. Krzycki, M. K. Chan, Science 2002, 296, 1462 – 1466; c) J. A.
Soares, L. Zhang, R. L. Pitsch, N. M. Kleinholz, R. B. Jones, J. J.
Wolff, J. Amster, K. B. Green-Church, J. A. Krzycki, J. Biol.
Chem. 2005, 280, 36962 – 36969.
[3] a) A. Mahapatra, A. Patel, J. A. Soares, R. C. Larue, J. K. Zhang,
W. W. Metcalf, J. A. Krzycki, Mol. Microbiol. 2006, 59, 56 – 66;
b) S. K. Blight, R. C. Larue, A. Mahapatra, D. G. Longstaff, E.
Chang, G. Zhao, P. T. Kang, K. B. Green-Church, M. K. Chan,
J. A. Krzycki, Nature 2004, 431, 333 – 335; c) C. Polycarpo, A.
Ambrogelly, A. Berube, S. M. Winbush, J. A. McCloskey, P. F.
Crain, J. L. Wood, D. Sçll, Proc. Natl. Acad. Sci. USA 2004, 101,
12450 – 12454.
[4] a) M. A. Gaston, L. Zhang, K. B. Green-Church, J. A. Krzycki,
Nature 2011, 471, 647 – 650; b) S. E. Cellitti, W. Ou, H. P. Chiu, J.
Grunewald, D. H. Jones, X. Hao, Q. Fan, L. L. Quinn, K. Ng,
A. T. Anfora, S. A. Lesley, T. Uno, A. Brock, B. H. Geierstanger,
Nat. Chem. Biol. 2011, 7, 528 – 530.
[5] F. Quitterer, A. List, W. Eisenreich, A. Bacher, M. Groll, Angew.
Chem. 2012, 124, 1367 – 1370; Angew. Chem. Int. Ed. 2012, 51,
1339 – 1342.
[6] F. Quitterer, A. List, P. Beck, A. Bacher, M. Groll, J. Mol. Biol.
2012, 424, 270 – 282.
[7] a) E. Kaya, M. Vrabel, C. Deiml, S. Prill, V. S. Fluxa, T. Carell,
Angew. Chem. 2012, 124, 4542 – 4545; Angew. Chem. Int. Ed.
2012, 51, 4466 – 4469; b) M. J. Gattner, M. Vrabel, T. Carell,
Chem. Commun. 2013, 49, 379 – 381; c) B. H. Geierstanger, W.
Ou, S. Cellitti, T. Uno, T. Crossgrove, H. P. Chiu, J. Grunewald,
X. Hao, International patent application WO/2010/048582, 2010;
d) P. R. Chen, D. Groff, J. Guo, W. Ou, S. Cellitti, B. H.
Geierstanger, P. G. Schultz, Angew. Chem. 2009, 121, 4112 –
4115; Angew. Chem. Int. Ed. 2009, 48, 4052 – 4055; e) M. G.
Hoesl, N. Budisa, Angew. Chem. 2011, 123, 2948 – 2955; Angew.
Chem. Int. Ed. 2011, 50, 2896 – 2902.
Scheme 2. Proposed reaction mechanism of PylD.
3a and the pyridine nucleoside coenzyme (in its reduced, that
is, nonreactive form) suggests that the pro-R hydrogen at the
terminal aminomethylene group of 3a is well positioned for
hydride abstraction by the coenzyme in the oxidized form.
The resulting, protonated and positively charged imine
carbon is perfectly suitable for a nucleophilic attack by the
a amino group of the ornithine moiety, thus enabling the
closure of the five-membered pyrrolidine ring. Alternatively,
the abstraction of the hydride ion and the attack by the
a amino group could be achieved in a concerted process, thus
explaining the absence of a base for proton abstraction from
the N-terminal nitrogen of the substrate. Notably, the absence
of a proton acceptor in the active site sets PylD apart from
other alcohol dehydrogenases and amine dehydrogenases.
Subsequent to the formation of the 2-aminopyrrolidine
intermediate 5, the transfer of a proton from the ring nitrogen
to the amino group at C5 of 5 could be mediated by a network
of hydrogen-bonded water molecules (Figure 2b and
Figure 3). Thus, the proton transponders could set the stage
for the elimination of the amino group affording the pyrroline
ring. In conclusion, our structural data provide snapshots of
different steps of the PylD reaction trajectory and strongly
support the proposed short mechanism rather than a more
[8] D. Turk, PhD Thesis, Technische Universitꢂt Mꢃnchen 1992.
[9] M. D. Winn, C. C. Ballard, K. D. Cowtan, E. J. Dodson, P.
Emsley, P. R. Evans, R. M. Keegan, E. B. Krissinel, A. G. Leslie,
A. McCoy, S. J. McNicholas, G. N. Murshudov, N. S. Pannu, E. A.
7036 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 7033 –7037

Angewandte
Chemie
Potterton, H. R. Powell, R. J. Read, A. Vagin, K. S. Wilson, Acta
Crystallogr. Sect. D 2011, 67, 235 – 242.
[10] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr.
Sect. D 1997, 53, 240 – 255.
[11] L. Holm, P. Rosenstrçm, Nucleic Acids Res. 2010, 38, W545 –
549.
[12] G. Kleiger, D. Eisenberg, J. Mol. Biol. 2002, 323, 69 – 76.
[13] The asymmetric unit contains four PylD molecules, each in
complex with the product. The r.m.s.d. of Ca atoms among
subunits is less than 0.2 ꢀ.
[14] B. Persson, Y. Kallberg, U. Oppermann, H. Jornvall, Chem.-Biol.
Interact. 2003, 143, 271 – 278.
[15] U. Oppermann, C. Filling, M. Hult, N. Shafqat, X. Wu, M. Lindh,
J. Shafqat, E. Nordling, Y. Kallberg, B. Persson, H. Jornvall,
Chem.-Biol. Interact. 2003, 143 – 144, 247 – 253.
Angew. Chem. Int. Ed. 2013, 52, 7033 –7037
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7037