Crystal structures and enzymatic properties of a triamine/agmatine aminopropyltransferase from thermus thermophilus

Article abstract of DOI:10.1016/j.jmb.2011.03.025

To maintain functional conformations of DNA and RNA in high-temperature environments, an extremely thermophilic bacterium, Thermus thermophilus, employs a unique polyamine biosynthetic pathway and produces more than 16 types of polyamines. In the thermophile genome, only one spermidine synthase homolog (SpeE) was found and it was shown to be a key enzyme in the pathway. The catalytic assay of the purified enzyme revealed that it utilizes triamines (norspermidine and spermidine) and agmatine as acceptors in its aminopropyl transfer reaction; therefore, the enzyme was denoted as a triamine/agmatine aminopropyltransferase (TAAPT). We determined the crystal structures of the enzyme complexed with and without the aminopropyl group donor S-adenosylmethionine. Despite sequence and structural similarity with spermidine synthases from other organisms, a novel C-terminal β-sheet and differences in the catalytic site were observed. The C-terminal module interacts with the gatekeeping loop and fixes the open conformation of the loop to recognize larger polyamine substrates such as agmatine and spermidine. Additional computational docking studies suggest that the structural differences of the catalytic site also contribute to recognition of the aminopropyl/aminobutyl or guanidium moiety of the substrates of TAAPT. These results explain in part the extraordinarily diverse polyamine spectrum found in T. thermophilus.

Full text of DOI:10.1016/j.jmb.2011.03.025

doi:10.1016/j.jmb.2011.03.025
J. Mol. Biol. (2011) 408, 971–986
Contents lists available at www.sciencedirect.com
Journal of Molecular Biology
journal homepage: http://ees.elsevier.com.jmb
Crystal Structures and Enzymatic Properties of a
Triamine/Agmatine Aminopropyltransferase from
Thermus thermophilus
Mio Ohnuma1 †, Tadashi Ganbe2†, Yusuke Terui1, Masaru Niitsu3,
⁎
Takao Sato2, Nobuo Tanaka2, Masatada Tamakoshi1, Keijiro Samejima3,
Takashi Kumasaka2,4 and Tairo Oshima1,5
1Department of Molecular Biology, Tokyo University of Pharmacy and Life Science, Tokyo 192-0392, Japan
2Department of Life Science, Tokyo Institute of Technology, Yokohama 226-8501, Japan
3Faculty of Pharmaceutical Science, Josai University, Saitama 530-0248, Japan
4Japan Synchrotron Radiation Research Institute, Hyogo 679-5198, Japan
5Institute of Environmental Microbiology, Kyowa Kako Co., Ltd., Tokyo 194-0035, Japan
Received 25 December 2010;
received in revised form
10 March 2011;
accepted 13 March 2011
Available online
31 March 2011
To maintain functional conformations of DNA and RNA in high-
temperature environments, an extremely thermophilic bacterium, Thermus
thermophilus, employs a unique polyamine biosynthetic pathway and
produces more than 16 types of polyamines. In the thermophile genome,
only one spermidine synthase homolog (SpeE) was found and it was
shown to be a key enzyme in the pathway. The catalytic assay of the
purified enzyme revealed that it utilizes triamines (norspermidine and
spermidine) and agmatine as acceptors in its aminopropyl transfer
reaction; therefore, the enzyme was denoted as a triamine/agmatine
aminopropyltransferase (TAAPT). We determined the crystal structures of
the enzyme complexed with and without the aminopropyl group donor
S-adenosylmethionine. Despite sequence and structural similarity with
spermidine synthases from other organisms, a novel C-terminal β-sheet
and differences in the catalytic site were observed. The C-terminal
module interacts with the gatekeeping loop and fixes the open
conformation of the loop to recognize larger polyamine substrates such
as agmatine and spermidine. Additional computational docking studies
suggest that the structural differences of the catalytic site also contribute
to recognition of the aminopropyl/aminobutyl or guanidium moiety of
the substrates of TAAPT. These results explain in part the extraordinarily
diverse polyamine spectrum found in T. thermophilus.
Edited by G. Schulz
Keywords:
agmatine;
aminopropyltransferase;
crystal structure;
polyamine;
Thermus thermophilus
© 2011 Elsevier Ltd. All rights reserved.
*Corresponding author. 3-34-1 Nishiikebukuro, Toshima-ku, Tokyo 171-8501, Japan. E-mail address:
mioohnuma@rikkyo.ac.jp.
† M.O. and T.G. contributed equally to this work.
Present address: M. Ohnuma, Research Information Center for Extremophile, Rikkyo University, Tokyo 171–8501,
Japan.
Abbreviations used: ACAPT, agmatine/cadaverine aminopropyl transferase; AdoDATO, S-adenosyl-1,8-diamino-3-
thiooctane; dcSAM, decarboxylated S-adenosylmethionine; EDTA, ethylenediaminetetraacetic acid; SPDS, spermidine
synthase; SPMS, spermine synthase; 3D, three-dimensional; TAAPT, triamine/agmatine aminopropyltransferase;
TmSPDS, Thermotoga maritima SPDS; TtTAAPT, T. thermophilus TAAPT.
0022-2836/$ - see front matter © 2011 Elsevier Ltd. All rights reserved.

972
Crystal Structures and Properties of TtTAAPT
polyamines effectively stabilize double- and single-
stranded nucleic acids, respectively.5
Introduction
In the general biosynthetic pathway of poly-
amines, putrescine can be formed directly from
ornithine by the action of ornithine decarboxylase.
Plants and most bacteria possess an additional or
alternative pathway to form putrescine from
arginine;1,6,7 arginine is decarboxylated to form
agmatine by arginine decarboxylase, and agmatine
is subsequently converted to putrescine directly by
agmatine ureohydrolase or sequentially by agma-
tine deiminase and N-carbamoylputrescine amido-
hydrolase. Spermidine and spermine are
synthesized from putrescine and spermidine, re-
spectively, via spermidine synthase (SPDS) and
spermine synthase (SPMS) by the addition of an
aminopropyl group from decarboxylated S-adeno-
sylmethionine (dcSAM, dcAdoMet), leaving 5′-
methylthioadenosine (MTA) (Fig. 1).8
In our previous work, we proposed that T.
thermophilus employs a novel biosynthetic pathway;
that is, arginine is converted to agmatine followed
by aminopropylation to form N1-aminopropylag-
matine. N1-Aminopropylagmatine is then hydro-
lyzed to produce spermidine by the SpeB (agmatine
ureohydrolase) homolog.9 In this new pathway,
agmatine is aminopropylated by SpeE to form N1-
aminopropylagmatine. SpeE is the only SPDS
Polyamines are basic molecules containing more
than two amino nitrogen atoms and are involved in
cell proliferation through interactions with nega-
tively charged biomolecules such as DNA, RNA,
membrane lipids, and ATP.1 They are ubiquitously
distributed in prokaryotes and eukaryotes. The most
common polyamines in eukaryotes are putrescine,
spermidine, and spermine. Polyamine composition
varies between species: flowering plants and most
animal cells contain all three polyamines, while
many prokaryotes lack spermine. However, extreme
thermophiles often produce unusual polyamines;
for instance, Thermus thermophilus contains more
than 16 polyamines including long polyamines such
as caldopentamine [1,15-diamino-4,8,12-triazapen-
tadecane, NH2(CH2)3NH(CH2)3NH(CH2)3NH
(CH2)3NH2] and caldohexamine [1,19-diamino-
4,8,12,16-tetraazanonadecane, NH2(CH2)3NH
(CH2)3NH(CH2)3NH(CH2)3NH(CH2)3NH2],2 and
branched polyamines such as mitsubishine [tris(3-
aminopropyl)amine, [NH2(CH2)3]3N] and tetrakis
(3-aminopropyl)ammonium ([NH2(CH2)3]4N+).3
Cellular concentrations of these unusual polyamines
increase with elevation of growth temperature,
suggesting roles in biochemical reactions at high
temperature.4 The unusual long and branched
Fig. 1. Structural formula of polyamines and the reaction of aminopropyl transfer by aminopropyltransferases. The
aminopropyl moiety of dcSAM (shown in bold lines) is transferred to the amino terminus of substrates, and polyamine
products are obtained.

Crystal Structures and Properties of TtTAAPT
973
homologue found in the genome of T. thermophilus
HB8. The amino acid sequence of SpeE exhibits
similarity to that of SPDS, but is more similar to that
of thermospermine synthase,10 and enzymatic char-
acterization revealed that the enzyme does not
accept putrescine as a substrate but does accept
agmatine and triamines [norspermidine (1,7-dia-
mino-4-azaheptane) and spermidine] as described
below (Fig. 1). Therefore, the SPDS homologue,
which is pivotal for the formation of spermidine and
formerly named agmatine/spermidine aminopro-
pyltransferase, is now renamed as a triamine/
agmatine aminopropyltransferase (TAAPT).
Our current research has determined the three-
dimensional (3D) structure and enzymatic proper-
ties of T. thermophilus TAAPT (TtTAAPT) in order to
clarify the molecular basis of its unique substrate
specificity. In some bacteria, archaea, and eukary-
otes, spermidine is generally synthesized from
putrescine by the addition of an aminopropyl
group from dcSAM. However, TtTAAPT does not
act on putrescine, but on agmatine or triamines.
Therefore, it is important to know the molecular
mechanism by which the thermophile enzyme
rejects putrescine as a substrate even though
agmatine and spermidine contain a putrescine
moiety in their structures.
was thought to be capable of accepting not only
agmatine but also various other polyamines as
substrates by analogy with the putrescine amino-
propyltransferase of the leguminous plant alfalfa,
which accepts not only putrescine but also spermi-
dine and spermine as a substrate.22 As shown in Fig.
2a, in addition to agmatine, norspermidine and
spermidine were efficiently aminopropylated by the
enzyme. When homospermidine [1,9-diamino-5-
azanonane, NH2(CH2)4NH(CH2)4NH2], mitsu-
bishine, or thermine [1,11-diamino-4,8-diazaunde-
cane, NH2(CH2)3NH(CH2)3NH(CH2)3NH2] was
used as a substrate, the enzyme showed low
activity. Only faint activities (b10% of spermidine)
were detected for diaminopropane [NH2(CH2)3NH2],
putrescine, spermine, and caldopentamine under the
assay conditions employed. Therefore, the enzyme
was named as a triamine/agmatine aminopropyl-
transferase (TAAPT). TtTAAPT effectively accepted
linear polyamines whose lengths are 8–12 Å.
Although the length of mitsubishine was calculated
to be 10 Å, the effect of steric hindrance by the
branched polyamine would decrease aminopropyl-
transferase activity. TmSPDS also accepts various
polyamines as substrates, such as putrescine,
diaminopropane, and spermidine.17 Putrescine is
the best substrate for TmSPDS. The difference in
substrate preference might result from the size of
the binding cavity, gatekeeping loop, and C-terminal
extra β-sheet as described below.
Results and Discussion
As shown in Table 1, we measured kinetic
parameters for dcSAM and polyamines whose
lengths are 8–12 Å. The affinity of TtTAAPT for
dcSAM was similar to that of E. coli SPDS15
(Km = 2.2 μM), human SPDS23 (0.9 μM), and
TmSPDS23 (0.75 μM) and higher than that of
Pyrococcus furiosus agmatine/cadaverine aminopro-
pyl transferase (ACAPT)24 (20 μM). The affinity of
TtTAAPT for spermidine (1.73 μM) and norspermi-
dine (0.57 μM) was 11- to 38-fold higher than that of
E. coli SPDS15 (12 μM), human SPDS23 (19 μM), and
TmSPDS23 (20 μM). Similarly, the affinity of
TtTAAPT for agmatine (0.77 μM) was 10-fold higher
than that of P. furiosus ACAPT24 for agmatine
(7.6 μM). Norspermidine is the best polyamine as a
substrate with the highest catalytic efficiency.
Spermidine was calculated to have the highest kcat
value among the polyamines. The kcat value of
TtTAAPT (0.37 s−1) for agmatine was fourfold
higher than that of P. furiosus ACAPT24 (0.1 s−1).
The catalytic efficiency (kcat/Km) for norspermidine
is only about twofold higher than that for agmatine
and spermidine. Therefore, these three polyamines
would be used as substrates in vivo. Based on the
activity of TtTAAPT, T. thermophilus produces
aminopropylagmatine, spermine, and thermine
from agmatine, spermidine, and norspermidine,
respectively. Aminopropylagmatine is a precursor
of spermidine.9 Spermine and thermine were the
Subunit structure
The molecular mass of TtTAAPT was estimated
with a calibrated Superdex 200 HR 10/30 column.
The enzyme was eluted from this column as a single
symmetrical peak corresponding to 130 kDa (data
not shown). Since the calculated molecular mass of
the TtTAAPT monomer is 36 kDa, we concluded that
TtTAAPT forms a homotetramer in solution. The
subunit structures of human SPDS, SPMS,11–13
Plasmodium falciparum SPDS,14 and Escherichia coli
SPDS15,16 are dimers. On the other hand, Thermotoga
maritima SPDS (hereinafter TmSPDS)17 is a tetramer.
Generally, thermophilic enzymes have a higher
oligomerization state than their mesophilic homo-
logues, leading to an enlarged buried surface area
and increased packing density.18–21 TtTAAPT would
appear to employ the same strategy to increase
thermostability.
Substrate specificity
TtTAAPT converts agmatine to N1-aminopropy-
lagmatine as previously reported.9 Despite the fact
that many unique polyamines are produced by the
thermophile, only one speE orthologue was found in
the T. thermophilus genome. Therefore, the enzyme

974
Crystal Structures and Properties of TtTAAPT
Fig. 2. (a) Substrate preference of
TtTAAPT. Activities are repre-
sented as percentage of activity of
the reaction using spermidine.
Agmatine (Agm) and linear poly-
amines (Dap, diaminopropane; Put,
putrescine; Nspd, norspermidine;
Spd, spermidine; Hspd, sym-homo-
spermidine; Thm, thermine; Spm,
spermine; Cdp, caldopentamine;
Cdh, caldohexamine), gray bars;
branched polyamine (Mb, mitsubi-
sine), white bar. The average of
duplicate reactions is shown. The
reaction was performed at pH 9 at
37 °C for 30 min with 10 ng of
TtTAAPT used in each reaction.
Calculated length of agmatine and
polyamines is shown in the graph
area. Minimized energy of poly-
amines was calculated using Chem-
Draw3D. Length of polyamines
was calculated using ViewerLite.
(b) Identification of reaction prod-
ucts by HPLC or gas chromatogra-
phy. Products of in vitro reactions
using (a) agmatine and (c) mitsu-
bishine as substrates were analyzed
by HPLC. Products formed by in
vitro reaction using (b) norspermi-
dine and (d) spermidine were ana-
lyzed by gas chromatography. The
900 -μl reaction mixtures consisted
of 0.1 M Tris–Cl buffer (pH 9),
5 mM dithiothreitol, 0.36 mM unla-
beled dcSAM, and polyamines
(1 mM norspermidine, 2 mM sper-
midine or agmatine, or 4 mM mitsubishine) as the substrate. Reaction mixtures were incubated with the purified enzyme
(4.5 μg for norspermidine, 9 μg for spermidine, agmatine, and mitsubishine) at 37 °C for 180 min. Aminopropylagm, N1-
aminopropylagmatine; Aminopropylthm, N4-aminopropylthermine.
major compounds when the cells were cultured at
60 and 80 °C, respectively.25 When mitsubishine
was used as a substrate, the kcat value was similar to
that when agmatine was used as a substrate;
however, catalytic efficiency (kcat/Km) was very
low due to very poor affinity of the enzyme for
mitsubishine. This is understandable for a branched
polyamine, which might be affected by steric
hindrance. Moreover, the fact that the reaction
product for mitsubishine [see below, Fig. 2b(c)]
was not found in T. thermophilus cells25 suggests that
mitsubishine would be unlikely to be used as a
substrate in vivo and that the aminopropylation
reaction of mitsubishine has no physiological
importance.
Identification of reaction products
TtTAAPT reaction products were determined by
HPLC and gas chromatography (Fig. 2b). Our
previous study suggested that the thermophile
SpeE converts agmatine to N1-aminopropylagma-
tine [Fig. 2b(a)].9 As shown in Fig. 2b(b), the
thermophile enzyme converted norspermidine to
thermine. When spermidine is aminopropylated,
spermine and/or thermospermine [1,12-diamino-
4,8-diazadodecane, NH2(CH2)3NH(CH2)3NH
Table 1. Km and kcat of TtTAAPT at 37 °C, pH 9.0
Km (μM)
kcat (s−1
)
kcat/Km (M−1 s−1
)
Spermidine
Norspermidine
Agmatinea
Mitsubishine
dcSAM
1.73
0.57
0.65
0.53
0.37
0.29
0.73
3.8×105
9.2×105
4.8×105
7.6×103
4.5×105
0.77
38.31
1.64
a
From Ohnuma et al.9

Crystal Structures and Properties of TtTAAPT
975
HPLC analysis was used to identify the product
[Fig. 2b(c)]. TtTAAPT converted mitsubishine to N4-
aminopropylthermine, but not to tetrakis(3-amino-
propyl)ammonium.
pH and thermal dependency
As shown in Fig. 3a, the effect of pH on enzyme
activity was investigated over a pH range of 6–10 at
37 °C, and the enzyme showed highest aminopro-
pyltransferase activity at around pH 9.5. When the
reaction was performed at 60 °C, the optimum pH of
the reaction was shifted to 8.5. Enzyme activity was
barely affected by the presence or absence of 1 mM
magnesium ion and 1 mM ethylenediaminetetraa-
cetic acid (EDTA) at 60 °C, pH 8.5 (data not shown).
The rate of the reaction was significantly increased
by temperature elevation up to 80 °C (Fig. 3b).
To evaluate the pH and temperature stabilities of
the enzyme, CD spectra were measured. The
spectrum of the enzyme at pH 10.4 (Fig. 4a) was
nearly identical to that at pH 7.0 (data not shown).
This result suggests that the enzyme is stable at
around pH 10 at room temperature. Figure 4b shows
heat denaturation profiles at pH 10.4. The protein
(a)
200
150
100
50
Fig. 3. pH and temperature dependence of the
enzyme activity. 1 mM spermidine was used as a
substrate and 10 ng of the enzyme was used in each
reaction. (a) pH dependency. Buffers used were potas-
sium phosphate (at pH 6–7.4), Tris–HCl (pH 7.5–9), and
glycine–NaOH (pH 9.5 and 10). Reactions were per-
formed at 37 °C for 30 min (filled diamonds, left y-axis)
and 60 °C for 20 min (open circles, right y-axis). (b)
Temperature dependency. The reaction was performed
in Tris–HCl (pH 8) for 30 min at 37 °C, 20 min at 60 °C,
10 min at 70 °C, and 5 min at 80 °C.
0
-50
-100
-150
200 210 220 230 240 250
Wavelength (nm)
(b)
120
100
80
60
40
20
0
(CH2)4NH2] can be formed depending on the
reaction specificity of the enzyme. If the thermophile
enzyme aminopropylates the aminopropyl group of
spermidine, thermospermine can be synthesized. If
the enzyme specifically recognizes the aminobutyl
group of the spermidine, the product should be
spermine. If the enzyme possesses loose specificity,
then both thermospermine and spermine can be
formed. Gas chromatography analyses revealed that
the TtTAAPT specifically converts spermidine only
to spermine [Fig. 2b(d)]. When mitsubishine is used
as a substrate, tetrakis(3-aminopropyl)ammonium
and/or N4-aminopropylthermine can be formed.
Since heptafluorobutyrated tetrakis(3-aminopropyl)
ammonium decomposes spontaneously to mitsu-
bishine and heptafluorobutyrated monoamine,26
-20
40
60
80
100
Temperature (°C)
Fig. 4. pH and temperature stability of the enzyme. (a)
Far-UV CD spectra of native enzyme. (b) Heat denatur-
ation profile at 222 nm. The solution contained 0.2 mg/ml
enzyme (1.4 μM as tetramer enzyme) in 20 mM Na2B4O7 at
pH 10.4.

976
Crystal Structures and Properties of TtTAAPT
Fig. 5. Sequence alignment of characterized aminopropyltransferases. Group 1 (TAAPT/AtACL5): TthTAAPT:
TAAPT from T. thermophilus; AthACL5: ACL5 from A. thaliana;29 OsaSPMS2: SPMS from O. sativa; TpsSPMS: SPMS from
T. pseudonana.10 Group 2 (SPDS): AthSPDS: SPDS from A. thaliana;30,31 OsaSPDS: SPDS from O. sativa;32 TmaSPDS: SPDS
from T. maritima;17 EcoSPDS: SPDS from E. coli.33 The first line shows the secondary structural elements of TtTAAPT. The
GGG(D/E)G motif and (E/D)(I/V)D motif, which are important for binding of polyamine and dcSAM, are indicated by
green boxes. Residues D and E of GGG(D/E)G motif, critical to preference of polyamine, are indicated by a lime and pink
background, respectively. Residues assigned as homologous only in group 1 by the program ESPript are indicated by an
orange background.

Crystal Structures and Properties of TtTAAPT
977
was only gradually denatured upon elevation of
temperature and was not completely denatured at
least until 100 °C at pH 7.0 (not shown). The enzyme
proved to be a very high pH- and heat-stable
protein, as shown in Fig. 4b.
The amino acid sequence of the speE gene
homologue contains no cysteine residues. This is
consistent with the fact that the enzymes of aerobic
thermophiles tend to contain fewer cysteine resi-
dues, in particular, on the protein surface.27 It has
been reported that the Cβ–Sγ bond of cysteine
residues is labile at high temperature in aerobic
conditions.28
domains: the N-terminal domain (residues 1–58)
consisted of six β-strands (βA–βF), and the
C-terminal domain (residues 59–314) consisted of
10 α-helices (α1–α10) and nine β-strands (βG–βO)
(Figs. 5 and 6a).
The N-terminal βA and βB strands are the main
components of the tetramer interface and form an
eight-stranded antiparallel β-barrel by each chain.
From the two strands, Tyr6, Phe8, and Val18
protrude into the inside of the barrel and make a
hydrophobic core. Met1 and Tyr3 participate in the
hydrophobic core, and the N-terminal tail region
covers the barrel. The outside of the barrel is packed
with a four-stranded antiparallel β-sheet composed
of βC, βD, βE, and βF. This structure is similar to the
TmSPDS tetramer interface;17 therefore, we infer that
the oligomerization of thermophilic enzymes re-
duces the size of the molecular surface and contrib-
utes to the unusual thermostability of the enzyme.
The C-terminal domain possesses a Rossmann-
like fold that is mainly composed of a seven
stranded β-sheet (βG, βH, βI, βJ, βK, βL, and
βM). This motif accepts the substrate dcSAM.
Therefore, the C-terminal domain contained the
catalytic site, even though the catalytic cleft lay
between both domains. The cleft is large enough to
accept the additional polyamine substrate. For the
lid of the catalytic cleft, there is a loop structure
between βJ and α6, called the gatekeeping loop,
which is flexible in the apoenzyme of TmSPDS.17
Two β-strands closest to the C-terminus, βN and
βO, are found only in TtTAAPT and interact with
the gatekeeping loop (Fig. 6b, PDB ID 3ANX).
Including this region, no significant difference
between apoenzyme and binary complex can be
observed in their 3D structures.
Overall structure
The amino acid sequence of TtTAAPT exhibited
higher similarity with Arabidopsis ACL5, thermos-
permine synthase, than with SPDS (Fig. 5). The
crystal structure of TtTAAPT contained the two
independent subunits in an asymmetric unit. The
TmSPDS structure [Protein Data Bank (PDB) ID
1INL]17 is superimposed on TtTAAPT with a root-
mean-square deviation (RMSD) of 1.0 Å. The overall
structure of TtTAAPT also resembles other SPDS
structures. The RMSDs with other SPDS structures
are 1.3 Å from P. furiosus (PDB ID 1MJF),24 1.5 Å
from Bacillus subtilis (PDB ID 1IY9), 1.3 Å from
Arabidopsis thaliana (PDB ID 1XJ5), and 1.4 Å from
human (PDB ID 1ZDZ, replaced by 2O05).23 Since
the molecular mass of TtTAAPT was estimated to be
130 kDa by gel-filtration analysis as described, we
conclude that the biological assembly was a homo-
tetramer with a crystallographic 2-fold axis (Fig. 6a,
PDB ID 1UIR). Each subunit was composed of two
Fig. 6. Structure of TtTAAPT. (a) Ribbon diagram of TtTAAPT tetramer. Each subunit is colored differently. (b) Stereo
view of the TtTAAPT monomer in ribbon diagram representation. The adenosyl part of SAM is shown by blue sticks. The
two β-strands (βN, βO) of the C-terminal module are also shown. (c) Ribbon diagram of TmSPDS monomer represented
in the same orientation as in (b). Alternative conformation of the gatekeeping loop is colored in red. Multisubstrate
inhibitor AdoDATO is shown by green sticks.

978
Crystal Structures and Properties of TtTAAPT
Structure of the dcSAM binding site
From the structural and sequential alignment with
TmSPDS, Tyr63 and Asp158 in TtTAAPT also have
the same orientation as the corresponding residues
of TmSPDS (Tyr76 and Asp170, respectively), which
are thought to be catalytic residues, and TtTAAPT
Glu88 and His64 might interact with the aminopro-
pyl group of dcSAM. The motif for the dcSAM
binding scaffold is conserved in all members of the
SPDS/SPMS family consisting of the sequence (L/I/
V)GGG(D/E)G(G/A). The fifth residue of the motif
is variable, and there is a difference between
TtTAAPT and TmSPDS (Fig. 5). In the case of
TtTAAPT, the side chain of that residue, Glu88, is
bulkier than that of Asp101, the corresponding
residue in TmSPDS; therefore, it could not insert the
aminopropyl moiety of dcSAM into the deeper hole
of TtTAAPT. As a result, this might affect the
position of dcSAM, especially its Sγ atom, a key
atom of the aminopropyl transfer reaction.
We attempted to obtain TtTAAPT complexed
with SAM, dcSAM, polyamines, or inhibitors, but
we only succeeded in binding SAM with TtTAAPT.
In TtTAAPT complexed with SAM, the electron
density of SAM was observed only for its adenosyl
moiety (Fig. 7a). The methionyl moiety of SAM
showed ambiguous electron densities. SAM has a
carboxylate group that is absent in dcSAM; there-
fore, its negative charge might affect binding to the
catalytic cleft. The methionyl moiety might be
extended flexibly to the polyamine binding site,
and/or it could be digested by the hydrolytic
reaction producing MTA and homoserine. The
following discussion is based on the models
constructed using the TmSPDS–S-adenosyl-1,8-dia-
mino-3-thiooctane (AdoDATO) complex structure.
Binding of the adenine moiety is mostly similar to
the binding mode of AdoDATO to TmSPDS; the
adenine ring is sandwiched between Ile109 (corre-
sponding to Val122 in TmSPDS) and Leu159 (corre-
sponding to Ser171 in TmSPDS) by the stacking
interactions with these hydrophobic residues. A
hydrogen bond is formed between the N6 amino
group of SAM and Asp140. Although Val122 of
TmSPDS behaves in the same way as Ile109 of
TtTAAPT, Ser171 of TmSPDS does not contribute to
the interaction with the adenine ring in the way that
Leu159 of TtTAAPT does because of its polarity.
Instead, Leu182 of TmSPDS contributes to the
interaction with the adenine ring. On the other hand,
the equivalent residue of the TmSPDS Leu182 in
TtTAAPT is Leu172, which cannot make the hydro-
phobic interaction with the adenine. A hydrogen bond
between the N6 atom of adenine and Asp140 in
TtTAAPT is also observed in TmSPDS, and it consists
of the side chain of Asn152 (corresponding to Asp140
of TtTAAPT). Recognition of the ribose moiety is
performed by the interaction between the side chain of
Asp108 and the hydroxyl groups of the ribose in
positions 2′ and 3′, similar to the TmSPDS structure.
Structure of polyamine binding site
The interaction between polyamine and enzyme is
thought to be weak because of the failure of the co-
crystallization with norspermine, spermine, and
agmatine. This is consistent with the report that
the putrescine moiety of AdoDATO is not envel-
oped completely in the structure of TmSPDS. It
suggests that the aminopropyl enzyme was not
formed and the aminopropyl transfer reaction is a
single-displacement mechanism,34 not a sequential
ping-pong mechanism.
The polyamine binding site shows larger structur-
al differences with that of TmSPDS than the dcSAM
site. In particular, Asp161 of TtTAAPT makes an
interaction with the Gln194 side chain and shows a
positional difference of approximately 2 Å in Cα
atoms from the corresponding residue Asp173 of
TmSPDS, which might play a role in anchoring a
terminal amino group of putrescine in TmSPDS.
From the superimposition of AdoDATO onto
TtTAAPT, the carboxyl group of Asp161 may
Fig. 7. Substrate binding of TtTAAPT. (a) The complex structure around the substrate binding site. The protein portion
of TtTAAPT and the adenosyl part of SAM are shown in green and cyan, respectively. The omit Fo −Fc map contoured
with 3σ is also shown in magenta. (b) The docking models superimposed to the binary complex structure. The models
containing agmatine and spermidine moieties are shown in yellow and pink, respectively.

Crystal Structures and Properties of TtTAAPT
979
interact with the guanidium group of agmatine in a
bidentate manner. However, Asp161 and Phe234 are
relatively close to each other in the observed
structures, and there is not enough space for the
bulky guanidium group between the two residues. It
is also suggested by the fact that the two residues
repelled the guanidium group in the energy-mini-
mized binding model (data not shown). To make the
interaction, the side chain of Asp161 should break its
interaction with Gln194 and move to an open space,
the polyamine binding pocket, as with TmSPDS. In
the case of spermidine, its tail amino group might be
too close to interact with Asp161 because of its
length, and it might interfere with Met197.
might not contribute to recognition, and the
adaptation to agmatine recognition might be unique
to TtTAAPT. In the docking study, it is still not clear
how the central amine group in spermidine and
norspermidine is recognized. In the case of spermi-
dine, its molecular structure is not symmetric;
therefore, its substrate recognition is important to
the selectivity of the product as described before.
Hence, the candidates for substrate recognition
should be examined by site-directed mutagenesis
in future studies.
Structure of gatekeeping loop and C-terminal
extra β-sheet
Another conformational difference in the vicinity
of the active site is found in the region from Phe230 to
Leu232. TtTAAPT can accept substrates longer than
putrescine because this region is shorter by one
residue and forms a wider space than that of
TmSPDS. This is similar to E. coli SPDS, which has
a larger binding cavity.16 There is another similarity
to E. coli SPDS. Steric hindrance may be a contrib-
uting factor to the preference of polyamine substrate.
Human SPDS has high specificity for putrescine as
an amine acceptor, while TmSPDS can act on
spermidine even though activity on putrescine is
much higher than that on spermidine.23 In human
SPDS, a wall of the binding pocket formed by Trp28
and Pro242 side chains prevents the binding of an
additional aminopropyl group of longer substrates.
On the other hand, these residues are replaced by
Tyr23 and Pro240 in TmSPDS and provide more
space to accommodate larger substrates. The binding
pocket of TtTAAPT is even wider, because Glu9 is
the closest residue to the polyamine, while Trp28 and
Tyr23 are closest in human SPDS and TmSPDS,
respectively,23 and the rigid proline, which interferes
with the guanidium group of agmatine or the
aminopropyl group of spermidine in human SPDS
and TmSPDS, is replaced by Phe234 (Fig. 1).
To investigate other possible substrate-recognition
mechanisms, computational ligand docking was
performed (Fig. 7b). The results suggested that the
middle portion of the polyamine substrates was
accommodated by the movement of Asp161, and the
end portion was positioned toward the other side of
the active-site wall against Asp161. There are two
conserved acidic amino acids, Glu9 and Asp59. The
extended guanidium or N-terminal group interacts
with the former glutamate residue, which is
conserved in both SPDSs and SPMSs. Additionally,
the carbonyl oxygen of Val53 located at the same
wall may interact with the secondary amine group
of spermidine. TtTAAPT also recognizes norsper-
midine, which is shorter than spermidine by one
methylene group. However, the positional shift of
Asp161 might also be caused by the amino acid
insertion in the βj–η2 loop. This insertion is not
conserved with other SPMSs; therefore, Asp161
A large structural difference between TtTAAPT
and SPDSs is also found at the loop region βJ–α6
formed by residues 159–172. This region was
described as the gatekeeping loop in the TmSPDS
structure and is important for recognition of dcSAM
and putrescine. In the case of TtTAAPT, the main-
chain atoms of Pro162 O and Gly164 N of this loop
formed hydrogen bonds with the main-chain atoms
of Val301 in the novel C-terminal β-sheet, and there
is also a hydrophobic interaction among Pro162 and
Val163, Tyr173, Phe299, Val301. Accordingly, the
loop structure is fixed open, and is not changed
upon the SAM binding. In contrast, this loop in
TmSPDS becomes ordered from a disordered
conformation upon binding with AdoDATO in the
active site. As described below, several structural
analyses revealed that this structural change is
widely shared in SPDSs.
The structural rearrangement from SPDS to
TAAPT is achieved as follows: the position of
the carboxyl group of Asp161 of TtTAAPT is
occupied by Glu206 in the closed conformation of
TmSPDS. Gly196 of TtTAAPT, equivalent to the
glutamate, provides the room for the Asp161 side
chain. Here, Trp214 occupies the open structure of
TmSPDS. Since Asp173 of TmSPDS has to enter
the space to make an interaction with putrescine,
the χ1 angle of Trp214 rotates and opens the
space.
The structural change of the gatekeeping loop is
also essential to capture short molecules such as
putrescine in SPDSs. To interact with the substrate,
the gatekeeping loop has to be close to putrescine.
The closed conformation narrows the catalytic
cavity and encloses the bound dcSAM. On the
other hand, TtTAAPT fixes the loop structure by the
interaction with the extra C-terminal β-sheet, and
the fixed conformation is enough to allow interac-
tions with longer polyamines. Therefore, the gate-
keeper mechanism differs between TtTAAPT and
SPDSs. Recently, structures of human, nematode,
and apicomplexan SPDSs were determined.14,23,35
The human structure shows a structure similar to
the closed loop of the TmSPDS–AdoDATO complex.

980
Crystal Structures and Properties of TtTAAPT
Nematode SPDS structure was determined in
ligand-free form, and its gatekeeping loop is
disordered even with the presence of proline
residues. The gatekeeping loop of the apicomplexan
SPDS is partially disordered and stabilization of the
loop was observed in the dcSAM complex. Hence,
this flexibility of the loop seems to be common in
SPDSs.
The flexibility of the loop may be related to
substrate specificities. Although the substrate spec-
ificity of TmSPDS changes with reaction
temperature,23 this was not the case with TtTAAP.
Furthermore, TtTAAPT does not recognize putres-
cine and the apparent order of its activity is
norspermidineNspermidine≥agmatine at either
60 or 37 °C (data not shown), suggesting that the
Fig. 8 (legend on next page)

Crystal Structures and Properties of TtTAAPT
981
gatekeeping loop is fixed independent of tempera-
ture change.
to minimal medium, the speBE double disruptant
could produce tetrakis(3-aminopropyl)ammonium
(M.O., unpublished data). Production of tetrakis(3-
aminopropyl)ammomium requires another amino-
propyltransferase in the cell. Nevertheless, only one
aminopropyltransferase homologue (speE) is present
in the T. thermophilus genome. These unique poly-
amines could be produced by other enzymes that
exhibit low homology to known aminopropyltrans-
ferases, or the enzymes may belong to a different
family. Elucidation of the biosynthesis of the long
and branched polyamines is the subject of our future
studies.
Biosynthesis of polyamines in T. thermophilus
The substrate specificity of TtTAAPT does not
fully explain the biosynthesis of the unique
polyamines such as homospermine [1,13-dia-
mino-4,9-diazatridecane, NH2(CH2)3NH(CH2)4NH
(CH2)4NH2], thermospermine, pentaamines, and
tetrakis(3-aminopropryl)ammonium, of which ther-
mospermine is a major component.25 TtTAAPT
accepted spermidine as a substrate, but the product
was only spermine and no thermospermine was
formed [Fig. 2b(b)]. TtTAAPT might provide some
portion of homospermine in the cell, since TtTAAPT
had low activity for homospermidine (Fig. 2a). The
thermophile enzyme did not produce thermosper-
mine in vitro [Fig. 2b(d)]. For the production of
thermospermine, at least two reactions are possible:
(1) another aminopropyltransferase might be present
in the cell to aminopropylate the aminopropyl group
of spermidine; (2) an aminobutylation capability
might be present in the cell to produce thermo-
spermine by reactions analogous to the production
of homospermidine from putrescine in some meso-
philic bacteria.36 Norspermidine (Caldine) has been
detected in thermophile cells in a relatively large
amount,25 and this triamine cannot be produced by
TtTAAPT. Moreover, when spermidine was added
Phylogenetic relationship of TtTAAPT
As shown in Fig. 5, polyamine aminopropyltrans-
ferase was classified into two groups. These corre-
spond to the ACL5 group (GGGEG motif) and the
SPDS/SPMS group (GGGDG motif), which Knott and
coworkers detected by comparison of various SPDSs
and SPMSs.10 A. thaliana ACL529 and the SPMS
homolog of Thalassiosira pseudonana, which produces
thermospermine instead of spermidine or spermine,10
also has a long C-terminal region, similar to TtTAAPT.
To examine the phylogenetic relationship of TtTAAPT,
we performed maximum likelihood analyses with 54
polyamine aminopropyltransferases. Based on the
phylogenetic tree shown in Fig. 8, polyamine amino-
propyltransferases can be classified into four groups
Fig. 8. Phylogenetic relationship of the polyamine aminopropyltransferase proteins. The polyamine aminopropyl-
transferase proteins, which exhibit sequence similarity to TtTAAPT, were found by using PSI-BLAST. Proteins that were
experimentally uncharacterized were included in these sequences. A multiple alignment obtained using ClustalW was
checked and manually edited. The phylogenetic tree was constructed by maximum likelihood methods using the RAxML
Web server37 and FigTree software (http://www.tree.bio.ed.ac.uk/software/figtree). Bootstrap support values obtained
from 100 pseudo-replicates of the data set were indicated only when the value was higher than 50. Branch lengths were
drawn proportional to the amount of difference based on genetic distance; the scale bar indicates the number of
substitutions per site. Human and mouse SPMS were used as an outgroup. These proteins are classified into four groups
according to the GGG(D/E)G motif and the C-terminal region (group 1 proteins have a GGGEG motif and a long C-
terminal region; group 2 proteins have a GGGDG motif and a short C-terminal region; group 3 proteins have a GGGEG
motif and a short C-terminal region; group 4 proteins have a GGGDG motif and a long C-terminal region). This
classification is described in detail in the Results and Discussion section. The asterisk represents thermophiles or
hyperthermophiles. Hypothetical, hypothetical protein. Aeropyrum pernix K1 (NP_147478), Aquifex aeolicus VF5
(NP_213033), A. thaliana (NP_001078605, NP_173794, NP_177188, NP_001078749), Aspergillus nidulans FGSC A4
(XP_658291), B. subtilis subsp. subtilis str. 168 (NP_391630), Caldivirga maquilingensis IC-167 (ZP_01710548), Clostridium
thermocellum ATCC 27405 (YP_001037122), E. coli K12 (NP_414663), Geobacillus thermodenitrificans NG80-2
(YP_001127432), Homo sapiens (NP_003123, NP_004586), Hyperthermus butylicus DSM 5456 (YP_001012598,
YP_001012280), Ignicoccus hospitalis KIN4/I (YP_001435222), Lyngbya sp. PCC 8106 (ZP_01619477), Metallosphaera sedula
DSM 5348 (YP_001192315), Methanococcus aeolicus Nankai-3 (YP_001324799), Mus musculus (NP_033298, NP_033240),
Neurospora crassa OR74A (XP_960907), Nitrococcus mobilis Nb-231 (ZP_01128528), O. sativa (NP_001046395,
NP_001057773, NP_001059438), Ostreococcus lucimarinus CCE9901 (XP_001420452, XP_001418368), P. falciparum 3D7
(XP_001347972), Pseudomonas aeruginosa PAO1 (NP_253462), Pyrobaculum arsenaticum DSM 13514 (YP_001152549),
Pyrococcus abyssi GE5 (NP_125890), P. furiosus DSM 3638 (NP_577856), Saccharomyces cerevisiae (NP_015394, NP_013247),
Salmonella enterica subsp. enterica serovar Typhi str. CT18 (NP_454779), Schizosaccharomyces pombe 972h- (NP_596015),
Shigella flexneri 2a str. 301 (NP_706074), Staphylothermus marinus F1 (YP_001041435), Sulfolobus acidocaldarius DSM 639
(YP_255324), Sulfolobus solfataricus P2 (NP_342261), Sulfolobus tokodaii str. 7 (NP_376216), Synechococcus sp. JA-2-3B'a(2-13)
(YP_476322), T. pseudonana CCMP1335 (XP_002294468), Thermoanaerobacter ethanolicus CCSD1 (ZP_05492908), Thermo-
coccus kodakarensis KOD1 (YP_182560), Thermofilum pendens Hrk 5 (YP_919533, YP_920782), T. maritima MSB8
(AAD35738), T. thermophilus HB8 (1UIR_A), Trichodesmium erythraeum IMS101 (YP_720760), Trypanosoma cruzi strain
CL Brener (XP_811725), Xenopus laevis (NP_001084875).

982
Crystal Structures and Properties of TtTAAPT
according to the GGG(D/E)G motif, the binding site
for AdoDATO in TmSPDS,17 and the C-terminal
region. Group 1 proteins have a GGGEG motif and a
long C-terminal region, such as TtTAAT, A. thaliana
ACL5,29 and T. pseudonana SPMS.10 Group 2 proteins
have a GGGDG motif and a short C-terminal region.
Many conventional SPDSs and SPMSs found in
bacteria, plants, and animals can be classified into
group 2. Group 3 proteins have a GGGEG motif and a
short C-terminal region. Only polyamine aminopro-
pyltransferases from Crenarchaeota were classified
into group 3. Group 4 protein has a GGGDG motif and
a long C-terminal region. Oryza sativa_SPMS6 is the
single protein that was classified into group 4 among
54 analyzed proteins. The amino acid sequence of its
long C-terminal region was not conserved with that of
group 1. The function of the long C-terminal region of
Group 4 is unknown.
module interacts with the gatekeeping loop and fixes
the loop structure. Since TtTAAPT is similar to the
SPMS family in accepting longer substrates and also
the C-terminal extension in the primary structure,
this module might be shared with long polyamine
synthases in the thermophile. The binary complex
structure with SAM reveals a recognition scheme
similar to that of TmSPDS for the methylthioadeno-
sine portion, while the predicted recognition scheme
of the putrescine moieties of the longer polyamines,
such as agmatine and spermidine, reveals that the
moieties were located and recognized along the
opposite wall of the active-site cleft. The structure of
TtTAAPT is the first crystal structure of a group 1
polyamine aminopropyltransferase.
Groups 1 and 3 (GGGEG motif containing) pro-
teins are clearly separated from groups 2 and 4
(GGGDG motif containing) proteins. In group 2
proteins, bacterial and eukaryotic proteins are
clearly separated. In the eukaryotic clades, plants,
animals, and fungi formed individual clades. Groups
1 and 3 separated into distinct clades. In group 1
proteins, plant, cyanobacterial, and Chromalveolata
enzymes, which contain plastid, were grouped
within a clade.
In addition, there is a conserved MTA binding site at
Glu220 and Ile221 of human SPMS.12 The correspond-
ing amino acids of most aminopropyltransferases in
group 2 are Glu and Ile, respectively, with some
exceptions (e.g., TmSPDS Glu121 and Val122, B.
subtilis SPDS Asp107 and Ile108). We found that all
aminopropyl transferases in groups 1 and 3, which
have the GGGEG motif, have an Asp–Ile sequence
(e.g., TtTAAPT Asp108 and Ile109, A. thaliana ACL5
Asp130 and Ile131; Fig. 5). Sequences of both human
SPMS and P. furiosus agmatine/cadaverine aminopro-
pyltransferase are GGGDG and EID, even though they
accept spermidine and agmatine as a substrate,
respectively.12,24 TtTAAPT recognizes substrates dif-
ferently from human SPMS and P. furiosus agmatine/
cadaverine aminopropyltransferase. The GGG(D/E)G
and (E/D)(I/V)D motif is important for binding of
polyamine and dcSAM,12,17 and our results demon-
strate that the long C-terminal region is important for
recognition of longer polyamines as substrates. There-
fore, other proteins classified as group 1 polyamine
aminopropyltransferase could have altered substrate-
recognition mechanism and have important roles in
producing uncommon polyamines, which are often
found in thermophiles and plants.
Materials and Methods
Protein expression and purification
The SpeE (TtTAAPT) protein was overexpressed in E.
coli BL21 (DE3) cells carrying pESE89 by induction with
0.5 mM isopropyl β-D-thiogalactopyranoside (IPTG) at
37 °C. Cells were grown in Luria–Bertani medium for
3 h after IPTG induction and were harvested by
centrifugation at 4200g for 15 min at 4 °C. The cell
pellet was suspended in 20 mM Tris–HCl, 50 mM NaCl,
and 1 mM EDTA (pH 8.0) (buffer A) and was then
homogenized by sonication on ice. Crude lysate was
centrifuged at 36,000g for 30 min at 4 °C. After the
supernatant fraction was heated at 60 °C for 15 min and
cooled in an ice bath for 5 min, it was centrifuged at
36,000g for 30 min at 4 °C. The process was repeated
twice, and then the supernatant was fractionated with
ammonium sulfate. The precipitate from the salting out
at 30% saturated ammonium sulfate was discarded, and
the precipitate formed by gradual addition of ammoni-
um sulfate up to 70% saturation was collected by
centrifugation at 16,000g for 20 min at 4 °C. After the
precipitate was dissolved in buffer A, the solution was
dialyzed against the same buffer and was loaded onto a
DEAE ion-exchange column equilibrated with buffer A.
The column was washed with five volumes of buffer A
and eluted out with 20 mM Tris–HCl (pH 8.0), 200 mM
NaCl, and 1 mM EDTA. Eluted protein was precipitated
by 70% saturated ammonium sulfate and was collected
by centrifugation at 16,000g for 20 min. The precipitate
was dissolved in buffer A, and then the solution was
dialyzed against the same buffer. The solution was
loaded onto
a MonoQ ion-exchange column (GE
Healthcare, Buckinghamshire, UK) equilibrated with
buffer A and eluted with a 50 to 500 mM NaCl linear
gradient in buffer A. Fractions containing the protein
were stored at 4 °C. The resulting enzyme was
electrophoretically homogenous, with a specific activity
of 2.4 μmol/min per milligram protein at 60 °C (pH 9.0)
in the presence of 1 mM spermidine and purity of 51-
fold.9 Protein concentrations were estimated with BCA
Protein Assay Reagent Kit (Pierce, Rockford, IL) using
bovine serum albumin as a standard.
Conclusion: Molecular basis of substrate
specificity of TtTAAPT
The structure of TtTAAPT reveals a novel C-
terminal module composed of two β-strands. This

Crystal Structures and Properties of TtTAAPT
983
Gel filtration
the purified enzyme (4.5 μg for norspermidine, 9 μg for
spermidine, agmatine, and mitsubishine) at 37 °C for
180 min. The reaction was terminated with 5% perchloric
acid and the reaction product was identified by gas
chromatography26,41 or HPLC.2 Gas chromatography
was performed with a GC-9A gas chromatograph
(Shimadzu, Kyoto, Japan) after heptafluorobutyrization
of the concentrated reaction product. A Pyrex glass column
(2.1 m, 3 -mm inside diameter) was packed with 3% SE-30.
Helium was used as the carrier gas at the flow rate of
40 ml/min. The column oven temperature was pro-
grammed from 120 to 280 °C at a rate of 15 °C/min.
The molecular mass of the purified protein was
determined by gel-filtration analysis. Protein (0.4 mg/ml,
11 μM as monomer) was passed through a Sephadex 200
HR 10/30 column (GE Healthcare) at a flow rate of
0.2 ml/min in 100 mM Tris–Cl (pH 8.0) and 150 mM NaCl.
Ovalbumin (45 kDa, Sigma-Aldrich, St. Louis, MO) and
MW-GF-1000 Kit (Sigma-Aldrich) was used for molecular
weight markers. The kit contains Blue Dextran (2000 kDa),
apoferritin from horse spleen (443 kDa), β-amylase from
sweet potato (200 kDa), alcohol dehydrogenase from yeast
(150 kDa), and carbonic anhydrase from bovine erythro-
cytes (29 kDa).
Crystallization, data collection, and processing
CD measurements
The purified protein was dialyzed against 20 mM Tris–
HCl and 50 mM NaCl (pH 8.0) and concentrated to
approximately 8 mg/ml using a 100 -kDa cutoff Centricon
(Millipore, Billerica, MA). The initial crystallization trials
were carried out using Grid Screen PEG (polyethylene
glycol) 6000 and Grid Screen Ammonium Sulfate (Hamp-
ton Research, Aliso Viejo, CA) by the hanging-drop vapor-
diffusion method (1:1, protein solution to reservoir
solution) at 20 °C. Crystals were grown in two different
conditions: 30% PEG 6000, 0.1 M Tris–HCl (pH 8.0) or
1.6 M ammonium sulfate, 0.1 M 2-(N-morpholino)ethane-
sulfonic acid–NaOH (pH 6.0). Preliminary X-ray work
suggested that both crystals belong to the tetragonal space
group of either P41212 or P43212. After optimization of the
crystallization conditions, crystals used for structure
determination were obtained in 0.1 M 2-(N-morpholino)
ethanesulfonic acid–NaOH buffer (pH 6.5), containing
1.5 M ammonium sulfate and 3 mM norspermidine at
20 °C. The enzyme–SAM complex crystals were obtained
under the conditions of 9 mg/ml protein, 1.8 M ammo-
nium sulfate, 0.1 M 2-(N-morpholino)ethanesulfonic acid–
NaOH buffer (pH 5.5), and 1 mM SAM.
CD spectrum measurements were carried out with a
JASCO J-720C (JASCO, Hachioji, Japan) spectropolari-
meter. A 0.1 -cm cell was used for far-UV CD measure-
ments. The temperature of the sample solution in the cell
was controlled with a HAAKE circulating bath (Thermo
Fisher Scientific, Inc., Waltham, MA) and programmable
temperature controller. The temperature in the cell was
monitored with a thermocouple. Scan rate was 1 °C/min
and protein concentration of the enzyme was 0.2 mg/ml
(1.4 μM as tetramer).
Enzyme assay
The aminopropyltransferase assay was performed as
described,38 except as detailed below. Since putrescine was
inefficient as a substrate for the enzyme, spermidine was
used as a standard substrate instead. The pH of the
reaction mixtures was adjusted at the corresponding
reaction temperature. [methyl-14C]dcSAM was synthesized
according to the method of Samejima and colleagues.39 For
measurements of kinetic parameters, the enzymatic
reaction was performed in one tube and 100 μl of reaction
mixture was sampled at 0.5, 1, 5, 10, 15, and 20 min. Since
dcSAM decomposes spontaneously at high temperature
under alkaline conditions,40 kinetic parameters were
determined at 37 °C, pH 9. The Km for dcSAM was
determined by measuring initial velocities of the reaction
by changing the concentration (1.2–4.8 μM) of dcSAM in
the presence of a fixed concentration (1 mM) of spermidine
with 5 ng of TtTAAPT. Km values for polyamines were
measured by changing the concentration of the polyamine
(0.5–2 μM for spermidine and norspermidine, 10–60 μM for
mitsubishine) in the presence of a fixed concentration
(38 μM) of dcSAM. Km and kcat of spermidine, norspermi-
dine, and mitsubishine were measured with 10, 10, and
25 ng of TtTAAPT, respectively.
Table 2. Crystallographic data collection and refinement
statistics
Native
SAM complex
Data collection
Space group
Unit cell (Å)
Resolution (Å)
No. of reflections
Observations
Unique
Completeness
Averaged I/σI
Rmerge (%)
P43212
P43212
a=87.78, c=190.83
a=88.04, c=191.00
40.00–2.00 (2.11–2.00) 20.00–2.50 (2.59–2.50)
561,274
49,522
96.7 (85.0)
15.6 (5.1)
3.5 (15.9)
155,196
25,838
91.7 (78.3)
11.2 (3.2)
11.9 (35.4)
Refinement
R/Rfree (%)
RMSD from
ideal (Å, °)
Average B-factora
Ramachandran
plotb
19.0/23.8
0.008/1.5
18.3/26.8
0.030/2.7
Identification of reaction product
30.7/34.4/37.5/—
90.2/9.4/0.4/0.0
38.4/39.3/30.1/56.0
81.1/17.2/1.3/0.4
To identify the reaction product of TtTAAPT, 900 μl of
the reaction mixtures consisted of 0.1 M Tris–Cl buffer
(pH 9), 5 mM dithiothreitol, 0.36 mM unlabeled dcSAM,
and a polyamine (1 mM norspermidine, 2 mM spermi-
dine or agmatine, or 4 mM mitsubishine) as the
substrate. The reaction mixtures were incubated with
Values in parentheses are for the highest-resolution shell.
a
Main chain/side chain/water/ligand.
b
Most favored/additional allowed/generously allowed/
disallowed.

984
Crystal Structures and Properties of TtTAAPT
Diffraction data were collected with an ADSC Quantum
Accession numbers
4R CCD detector at beamline NW12 at the Photon Factory
Advanced Ring for the native crystals and an R-AXIS
detector at a laboratory X-ray source for the binary
complex crystals. Crystals were soaked with 20% glycerol
as a cryoprotectant and were mounted under a nitrogen
stream kept at −173 °C. Data were indexed and scaled
with Mosflm and Scala42 and CrystalClear/d⁎TREK
programs. Data processing statistics are presented in
Table 2.
The coordinates and structure factors of the native and
binary complex have been deposited in the PDB with
accession codes 1UIR and 3ANX, respectively.
Acknowledgements
Structure determination and refinement
The authors are grateful to Prof. Seiki Kuramitsu
of Osaka University for providing the nucleotide
sequence of TtTAAPT. We thank Daisuke Saito for
technical assistance, the synchrotron beamline staff
at SPring-8, the Photon Factory for crystallographic
data collection, Dr. Masaki Yoshida (University of
Tsukuba) for phylogenetic analyses, and Dr.
Anthony J. Michael (UT Southwestern Medical
Center, Dallas, TX) for critical reading of the
manuscript. This work was supported in part by
Grants-in-Aid for Scientific Research (No. 11794038
and 16657040), for Promoting Bioventures in
Private Universities, and for National Project on
Protein Structural and Functional Analyses from
the Ministry of Education, Culture, Sports, Science
and Technology, the Japanese Government.
The structure was solved by molecular replacement
with the program EPMR.43 TmSPDS17 (PDB ID 1INL) has
about 34% identical amino acid sequence with TtTAAPT
and was used to construct a model by substitution of all
amino acids into serine residues except for alanine and
glycine residues. After several cycles of calculations, the
best correlation coefficient between observed and calcu-
lated structure amplitudes was estimated to be 0.45 with
two independent molecules in an asymmetric unit in the
space group P43212.
The converged molecules were mutated so as to have
the same amino acid sequence with TtTAAPT and were
refined with the CNS program.44 In each cycle of the
CNS refinement, each model was rebuilt with the
program XtalView/Xfit.45 The final model was refined
at 2.0 Å resolution to R and free R-factors of 19.9 and
23.9%, respectively. The refined model had geometries
with RMSD of 0.008 Å and 1.5° from ideal values for
bond lengths and angles, respectively, and was qualified
by PROCHECK.46 There are two residues (Asp50 of both
chains) in the disallowed region; however, they fitted
well into the electron density map and are considered to
have this conformation actually. The binary complex
structure was determined using the native structural
model and the inclusion of SAM molecular models as
MTA molecules. Refinement statistics are summarized in
Table 1.
References
1. Tabor, C. W. & Tabor, H. (1984). Polyamines. Annu.
Rev. Biochem. 53, 749–790.
2. Oshima, T. (1983). Novel polyamines in Thermus
thermophilus: isolation, identification, and chemical
synthesis. Methods Enzymol. 94, 401–411.
3. Oshima, T., Hamasaki, N., Senshu, M., Kakinuma, K.
& Kuwajima, I. (1987). A new naturally occurring
polyamine containing a quaternary ammonium nitro-
gen. J. Biol. Chem. 262, 11979–11981.
4. Oshima, T., Hamasaki, N. & Uzawa, T. (1988).
Biochemical properties of unusual polyamines found
in an extreme thermophile, Thermus thermophilus. Adv.
Exp. Med. Biol. 250, 633–642.
5. Terui, Y., Ohnuma, M., Hiraga, K., Kawashima, E. &
Oshima, T. (2005). Stabilization of nucleic acids by
unusual polyamines produced by an extreme thermo-
phile, Thermus thermophilus. Biochem. J. 388, 427–433.
6. Bell, E. & Malmberg, R. L. (1990). Analysis of a cDNA
encoding arginine decarboxylase from oat reveals
similarity to the Escherichia coli arginine decarboxylase
and evidence of protein processing. Mol. Gen. Genet.
224, 431–436.
7. Sekowska, A., Bertin, P. & Danchin, A. (1998).
Characterization of polyamine synthesis pathway in
Bacillus subtilis 168. Mol. Microbiol. 29, 851–858.
8. Pegg, A. E., Shuttleworth, K. & Hibasami, H. (1981).
Specificity of mammalian spermidine synthase and
spermine synthase. Biochem J. 197, 315–320.
Structural analyses
The structural modeling for substrate recognition was
performed using the binary complex structure in
comparison with the 3D structure of TmSPDS–Ado-
DATO complex structures.17 Superimposition of Ado-
DATO to SAM was done by least-square minimization
of their adenosyl moiety. Instead of the putrescine
moiety in AdoDATO, agmatine and spermidine moie-
ties were constructed using modeling programs
MOLDA and Chimera. The model structures were
docked into the TtTAAPT structure with the program
DOCK version 6.1 with anchor-and-grow mode,47
where the adenosyl moieties of the models were treated
as the rigid substructure and anchored into the protein
model with the least-square result, and the structures of
the other polyamine portions were extended as flexible
ligands.
Molecular graphics in the figures were drawn with the
program PyMOL. Superimposition was carried out with
TOP.48 Sequence alignments were performed with the
program ClustalW49 and displayed with ESPript.50
9. Ohnuma, M., Terui, Y., Tamakoshi, M., Mitome, H.,
Niitsu, M., Samejima, K. et al. (2005). N1-Aminopro-
pylagmatine, a new polyamine produced as a key

Crystal Structures and Properties of TtTAAPT
985
intermediate in polyamine biosynthesis of an extreme
thermophile, Thermus thermophilus. J. Biol. Chem. 280,
30073–30082.
agmatine/cadaverine aminopropyl transferase: bio-
chemical and structural characterization of an enzyme
involved in polyamine biosynthesis in the hyperther-
mophilic archaeon Pyrococcus furiosus. J. Bacteriol. 189,
6057–6067.
10. Knott, J. M., Römer, P. & Sumper, M. (2007). Putative
spermine synthases from Thalassiosira pseudonana and
Arabidopsis thaliana synthesize thermospermine rather
than spermine. FEBS Lett. 581, 3081–3086.
11. Kajander, E. O., Kauppinen, L. I., Pajula, R. L.,
Karkola, K. & Eloranta, T. O. (1989). Purification and
partial characterization of human polyamine
synthases. Biochem. J. 259, 879–886.
12. Wu, H., Min, J., Zeng, H., McCloskey, D. E., Ikeguchi,
Y., Loppnau, P. et al. (2008). Crystal structure of
human spermine synthase: implications of substrate
binding and catalytic mechanism. J. Biol. Chem. 283,
16135–16146.
13. Pegg, A. E. & Michael, A. J. (2010). Spermine synthase.
Cell. Mol. Life. Sci. 67, 113–121.
14. Dufe, V. T., Qiu, W., Müller, I. B., Hui, R., Walter, R. D.
& Al-Karadaghi, S. (2007). Crystal structure of
Plasmodium falciparum spermidine synthase in com-
plex with the substrate decarboxylated S-adenosyl-
methionine and the potent inhibitors 4MCHA and
AdoDATO. J. Mol. Biol. 373, 167–177.
15. Bowman, W. H., Tabor, C. W. & Tabor, H. (1973).
Spermidine biosynthesis. Purification and properties
of propylamine transferase from Escherichia coli. J. Biol.
Chem. 248, 2480–2485.
16. Zhou, X., Chua, T. K., Tkaczuk, K. L., Bujnicki, J. M. &
Sivaraman, J. (2010). The crystal structure of Escher-
ichia coli spermidine synthase SpeE reveals a unique
substrate-binding pocket. J. Struct. Biol. 169, 277–285.
17. Korolev, S., Ikeguchi, Y., Skarina, T., Beasley, S.,
Arrowsmith, C., Edwards, A. et al. (2002). The crystal
structure of spermidine synthase with a multisub-
strate adduct inhibitor. Nat. Struct. Biol. 9, 27–31.
18. Dams, T., Auerbach, G., Bader, G., Jacob, U., Ploom,
T., Huber, R. & Jaenicke, R. (2000). The crystal
structure of dihydrofolate reductase from Thermotoga
maritima: molecular features of thermostability. J. Mol.
Biol. 297, 659–672.
25. Oshima, T. (1989). Polyamines in thermophiles. In The
Physiology of Polyamines (Bachrach, J. & Heimer, Y. M.,
eds), vol. 2, pp. 36–46. CRC Press, Inc., Boca Raton, FL.
26. Niitsu, M., Samejima, K., Matsuzaki, S. & Hamana,
K. (1993). Systematic analysis of naturally occurring
linear and branched polyamines by gas chromatog-
raphy and gas chromatography–mass spectrometry.
J. Chromatogr. 641, 115–123.
27. Kagawa, Y., Nojima, H., Nukiwa, N., Ishizuka, M.,
Nakajima, T., Yasuhara, T. et al. (1984). High
guanine plus cytosine content in the third letter of
codons of an extreme thermophile. DNA sequence of
the isopropylmalate dehydrogenase of Thermus
thermophilus. J. Biol. Chem. 259, 2956–2960.
28. Jaenicke, R. & Bohm, G. (1998). The stability of
proteins in extreme environments. Curr. Opin. Struct.
Biol. 8, 738–748.
29. Hanzawa, Y., Takahashi, T., Michael, A. J., Burtin, D.,
Long, D., Pineiro, M. et al. (2000). ACAULIS5, an
Arabidopsis gene required for stem elongation, en-
codes a spermine synthase. EMBO J. 19, 4248–4256.
30. Hashimoto, T., Tamaki, K., Suzuki, K. & Yamada, Y.
(1998). Molecular cloning of plant spermidine
synthases. Plant Cell Physiol. 39, 73–79.
31. Hanzawa, Y., Imai, A., Michael, A. J., Komeda, Y. &
Takahashi, T. (2002). Characterization of the spermi-
dine synthase-related gene family in Arabidopsis
thaliana. FEBS Lett. 527, 176–180.
32. Imai, R., Ali, A., Pramanik, H. R., Nakaminami, K.,
Sentoku, N. & Kato, H. (2004). A distinctive class of
spermidine synthase is involved in chilling response
in rice. J. Plant Physiol. 161, 883–886.
33. Tabor, C. W. & Tabor, H. (1987). The speEspeD operon
of Escherichia coli. Formation and processing of a
proenzyme form of S-adenosylmethionine decarbox-
ylase. J. Biol. Chem. 262, 16037–16040.
19. Maeda, N., Kanai, T., Atomi, H. & Imanaka, T.
(2002). The unique pentagonal structure of an
archaeal Rubisco is essential for its high thermosta-
bility. J. Biol. Chem. 277, 31656–31662.
20. Salminen, T., Teplyakov, A., Kankare, J., Cooperman,
B. S., Lahti, R. & Goldman, A. (1996). An unusual
route to thermostability disclosed by the comparison
of Thermus thermophilus and Escherichia coli inorganic
pyrophosphatases. Protein Sci. 5, 1014–1025.
21. Vieille, C. & Zeikus, G. J. (2001). Hyperthermophilic
enzymes: sources, uses, and molecular mechanisms
for thermostability. Microbiol. Mol. Biol. Rev. 65, 1–43.
22. Bagga, S., Rochford, J., Klaene, Z., Kuehn, G. D. &
Phillips, G. C. (1997). Putrescine aminopropyltransfer-
ase is responsible for biosynthesis of spermidine,
spermine, and multiple uncommon polyamines in
osmotic stress-tolerant alfalfa. Plant Physiol. 114,
445–454.
23. Wu, H., Min, J., Ikeguchi, Y., Zeng, H., Dong, A.,
Loppnau, P. et al. (2007). Structure and mechanism of
spermidine synthases. Biochemistry, 46, 8331–8339.
24. Cacciapuoti, G., Porcelli, M., Moretti, M. A., Sorrentino,
F., Concilio, L., Zappia, V. et al. (2007). The first
34. Orr, G. R., Danz, D. W., Pontoni, G., Prabhakaran,
P. C., Gould, S. J. & Coward, J. K. (1988). Synthesis
of chirally deuteriated (S-adenosyl-S-methylsulfo-
nio)propylamines and spermidines. Elucidation of
the stereochemical course of putrescine aminopro-
pyltransferase (spermidine synthase). J. Am. Chem.
Soc. 110, 5791–5799.
35. Dufe, V. T., Luersen, K., Eschbach, M. L., Haider, N.,
Karlberg, T., Walter, R. D. et al. (2005). Cloning,
expression, characterisation and three-dimensional
structure determination of Caenorhabditis elegans sper-
midine synthase. FEBS Lett. 579, 6037–6043.
36. Yamamoto, S., Nagata, S. & Kusaba, K. (1993).
Purification and characterization of homospermi-
dine synthase in Acinetobacter tartarogenes ATCC
31105. J. Biol. Chem. 114, 45–49.
37. Stamatakis, A., Hoover, P. & Rougemont, J. (2008). A
rapid bootstrap algorithm for the RAxML Web
servers. Syst. Biol. 57, 758–771.
38. Raina, A., Eloranta, T. & Pajula, R. (1983). Rapid assays
for putrescine aminopropyltransferase (spermidine
synthase) and spermidine aminopropyltransferase
(spermine synthase). Methods Enzymol. 94, 257–260.

986
Crystal Structures and Properties of TtTAAPT
39. Dejima, H., Kobayashi, M., Takasaki, H., Takeda, N.,
Shirahata, A. & Samejima, K. (2003). Synthetic
decarboxylated S-adenosyl-L-methionine as a sub-
strate for aminopropyl transferases. Biol. Pharm. Bull.
26, 1005–1008.
40. Zappia, V., Galletti, P., Oliva, A. & de Santis, A.
(1977). New methods for preparation and analysis
of S-adenosyl-(5′)-3-methylthiopropylamine. Anal.
Biochem. 79, 535–543.
41. Hamana, K., Niitsu, M., Samejima, K., Itoh, T.,
Hamana, H. & Shinozawa, T. (1998). Polyamines of
thermophilic eubacteria belonging to the genera
Thermotoga, Thermodesulfovibrio, Thermoleophilum,
Thermus, Rhodothermus and Meiothermus, and the
thermophilic archaebacteria belonging to the genera
Aeropyrum, Picrophilus, Methanobacterium and Metha-
nococcus. Microbios, 94, 7–21.
Crystallography & NMR system: a new software
suite for macromolecular structure determination.
Acta Crystallogr., Sect. D: Biol. Crystallogr. 54,
905–921.
45. McRee, D. E. (1999). XtalView/Xfit—a versatile
program for manipulating atomic coordinates and
electron density. J. Struct. Biol. 125, 156–165.
46. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &
Thornton, J. M. (1993). PROCHECK—a program to
check the stereochemical quality of protein structures.
J. Appl. Crystallogr. 26, 283–291.
47. Brooijmans, N. & Kuntz, I. (2003). Molecular recogni-
tion and docking algorithms. Annu. Rev. Biophys.
Biomol. Struct. 32, 335–373.
48. Lu, G. (2000). TOP: a new method for protein structure
comparisons and similarity searches. J. Appl. Crystallogr.
33, 176–183.
42. Collaborative Computational Project, Number 4 (1994).
The CCP4 suite: programs for protein crystallography.
Acta Crystallogr., Sect. D: Biol. Crystallogr, 50, 760–763.
43. Kissinger, C. R., Gehlhaar, D. K. & Fogel, D. B.
(1999). Rapid automated molecular replacement by
evolutionary search. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 55, 484–491.
49. Thompson, J. D., Higgins, D. G. & Gibson, T. J.
(1994). CLUSTAL W: improving the sensitivity of
progressive multiple sequence alignment through
sequence weighting, position-specific gap penalties
and weight matrix choice. Nucleic Acids Res. 22,
4673–4680.
50. Gouet, P., Courcelle, E., Stuart, D. I. & Metoz, F.
(1999). ESPript: analysis of multiple sequence align-
ments in PostScript. Bioinformatics, 15, 305–308.
44. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano,
W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).