Download Crystal structure of a membrane-bound l-amino acid

Journal of Structural Biology 195 (2016) 306–315
Contents lists available at ScienceDirect
Journal of Structural Biology
journal homepage: www.elsevier.com/locate/yjsbi
Crystal structure of a membrane-bound L-amino acid deaminase from
Proteus vulgaris
Yingchen Ju a, Shuilong Tong b,c, Yongxiang Gao b, Wei Zhao b,c, Qi Liu a, Qiong Gu a, Jun Xu a, Liwen Niu b,c,⇑,
Maikun Teng b,c,⇑, Huihao Zhou a,⇑
a
b
c
Research Center for Structural Biology and Drug Discovery, School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou 510006, China
Hefei National Laboratory for Physical Sciences at Microscale and School of Life Sciences, University of Science and Technology of China, 96 Jinzhai Road, Hefei, Anhui 230027, China
Key Laboratory of Structural Biology, Chinese Academy of Sciences, 96 Jinzhai Road, Hefei, Anhui 230027, China
a r t i c l e
i n f o
Article history:
Received 23 March 2016
Received in revised form 20 June 2016
Accepted 12 July 2016
Available online 13 July 2016
Keywords:
Crystal structure
Deaminase
L-amino acid oxidase
Membrane binding
Insertion module
Peroxide
a b s t r a c t
L-amino
acid oxidases/deaminases (LAAOs/LAADs) are a class of oxidoreductases catalyzing the oxidative
deamination of L-amino acids to a-keto acids. They are widely distributed in eukaryotic and prokaryotic
organisms, and exhibit diverse substrate specificity, post-translational modifications and cellular localization. While LAAOs isolated from snake venom have been extensively characterized, the structures
and functions of LAAOs from other species are largely unknown. Here, we reported crystal structure of
a bacterial membrane-bound LAAD from Proteus vulgaris (pvLAAD) in complex with flavin adenine
dinucleotide (FAD). We found that the overall fold of pvLAAD does not resemble typical LAAOs. Instead
it, is similar to D-amino acid oxidases (DAAOs) with an additional hydrophobic insertion module on protein surface. Structural analysis and liposome-binding assays suggested that the hydrophobic module
serves as an extra membrane-binding site for LAADs. Bacteria from genera Proteus and Providencia were
found to encode two classes of membrane-bound LAADs. Based on our structure, the key roles of residues
Q278 and L317 in substrate selectivity were proposed and biochemically analyzed. While LAADs on the
membrane were proposed to transfer electrons to respiratory chain for FAD re-oxidization, we observed
that the purified pvLAAD could generate a significant amount of hydrogen peroxide in vitro, suggesting it
could use dioxygen to directly re-oxidize FADH2 as what typical LAAOs usually do. These findings provide
a novel insights for a better understanding this class of enzymes and will help developing biocatalysts for
industrial applications.
Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction
L-amino acid oxidases/deaminase (LAAO/LAAD; EC 1.4.3.2) uses
flavin adenine dinucleotide (FAD) as cofactor to oxidatively deaminate L-amino acids to corresponding a-keto acids, and releases
ammonium and H2O2 as well. LAAO has multiple physiological
functions. The best-studied function is its antimicrobial activity,
in which membrane rupture caused by the produced H2O2 plays
Abbreviations: LAAO, L-amino acid oxidases; LAAD, L-amino acid deaminases;
DAAO, D-amino acid oxidases; FAD, flavin adenine dinucleotide; CTAB,
cetyltrimethylammonium bromide; SAD, single-wavelength anomalous dispersion;
DNP, 2,4-dinitrophenylhydrazine.
⇑ Corresponding authors at: School of Life Sciences, University of Science and
Technology of China, Hefei 230027, China (L. Niu, M. Teng), School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou 510006, China (H. Zhou).
E-mail addresses: [email protected] (L. Niu), [email protected] (M. Teng),
[email protected] (H. Zhou).
http://dx.doi.org/10.1016/j.jsb.2016.07.008
1047-8477/Ó 2016 Elsevier Inc. All rights reserved.
an essential role. Decomposition of H2O2 by adding catalase or
peroxidase significantly decreased the antimicrobial effect of LAAO
(Guo et al., 2012). Some snake venom LAAO can bind to bacterial
surface and generate high local concentration of H2O2, so a small
amount of LAAO is enough to inhibit bacterial growth (Zhang
et al., 2004a). The antimicrobial activity of LAAOs is employed by
rockfish Sebastes schlegeli to control the infection of Gramnegative bacteria in the skin mucus (Kitani et al., 2007), and also
by Streptococcus oligofermentans to compete with other bacteria
(Tong et al., 2008). Beyond antibiosis, LAAOs in some microorganisms have been reported to metabolize extra amino acids to provide nitrogen source for growing under certain conditions (Vallon
et al., 1993). The LAAO from Anacystis nidulans was suggested to
play a role in photosystem II on the thylakoid membrane
(Pistorius and Voss, 1982). Recently, more functions with biomedical application potentials were reported for LAAOs, such as
antitumor effects, induction of apoptosis, inhibition or activation
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
of platelet aggregation, and etc. (Izidoro et al., 2014; Yu and Qiao,
2012).
While most LAAOs are secreted or exist as cytoplasmic
flavoproteins, Proteus species express a class of membrane-bound
LAADs, which are anchored to cytomembrane surface through
their N-terminal transmembrane helix (Hossain et al., 2014b).
Some bacteria, such as Proteus mirabilis, was found to express
two types of LAADs, which share significant sequence similarity
but are with distinct substrate preferences: type I prefers aliphatic
and aromatic amino acids, while type II shows significant activity
for basic amino acids such as histidine and arginine (Baek et al.,
2011; Takahashi et al., 1999). These membrane-bound LAADs were
proposed to employ a noncanonical catalytic mechanism without
H2O2 production during deamination (Pantaleone et al., 2001; Yu
and Qiao, 2012). A widely accepted hypothesis is that the
membrane-bound LAADs are associated with the electron transport chain on the bacterial membrane, and electrons are finally
transferred to cytochrome oxidases to reduce O2 to H2O (Duerre
and Chakrabarty, 1975; Yu and DeVoe, 1981; Yu and Qiao, 2012).
It is a great advantage that membrane-bound LAAD does not produce H2O2 during catalysis, because it is helpful for maintaining
a mild environment suitable for industrial biocatalysis (Hou
et al., 2015a; Song et al., 2015). However, some studies reported
that the purified LAADs were still at least partially active in vitro
without adding any components from the electron transport chain
or other artificial electron acceptors (Hou et al., 2015a; Liu et al.,
2013).
The physiological roles of membrane-bound LAADs are not fully
understood so far, although it has been reported that they can produce a-keto acids as siderophores to relieve iron restriction
(Drechsel et al., 1993). Recently, membrane-bound LAAD has
attracted increasing attentions due to their potential biotechnological applications in industry. A series of LAAD-based technology
were developed to eco-friendly and efficiently transform L-amino
acids to a-keto acids, such as ketoglutaric acid, methylthiobutyric
acid, phenylpyruvic acid and etc (Hossain et al., 2014b,c; Hou
et al., 2015b; Liu et al., 2013). LAADs could also be used to produce
pure D-amino acids by removing L-amino acids from racemic mixtures (Takahashi et al., 1997). Other potential applications, such as
being used as biosensors of L-amino acid concentration in medical
and food industry, have also been proposed (Hossain et al., 2014a;
Yu and Qiao, 2012). Structure-guided protein engineering would be
helpful to decrease the product inhibition and improve biocatalytic
efficiency, specificity, stability and other properties important for
industrial applications (Hossain et al., 2014c; Hou et al., 2015b).
In this study, we solved the crystal structure of a membranebound L-amino acid deaminase from Proteus vulgaris (pvLAAD), a
pathogenic bacterium that inhabits human intestinal tracts and
causes urinary tract and wound infections. The key residues
involved in substrate binding and selection were proposed and
further analyzed using mutagenesis. The specific insertion of
membrane-bound LAADs was proposed to function as an extra
membrane-binding site. Furthermore, in contrast to current
knowledge, the purified pvLAAD was found to be able to generate
a significant amount of H2O2 in vitro. These findings provide novel
insights into our understanding of a class of important
oxidoreductases.
307
The DNA fragments encoding the full length pvLAAD (FL-pvLAAD),
N-terminus truncated pvLAAD (res. 30-471, named DN-pvLAAD)
were subcloned into pQE-9 vector (Qiagen), with an
MRGSHHHHHHGS-tag at the N-terminus of pvLAADs. The
transformed Escherichia coli M15 (Qiagen) cells were grown to
OD600 = 0.7 in Luria-Bertani (LB) media supplemented with
100 lg/mL ampicillin, and then protein overexpression was
induced at 20 °C for 12 h by adding 0.5 mM IPTG. Cells were harvested and sonicated in washing buffer (500 mM NaCl, 20 mM Tris
pH 8.0, 0.05% w/v Triton X-100, 20 mM imidazole). Cell lysate was
centrifuged at 16000g for 30 min, and the supernatant was loaded
onto a nickel-chelating column (GE Healthcare) pre-equilibrated
with washing buffer. The column was washed with 20 column
volumes of washing buffer to remove the impurities, and then
the target protein was eluted with 30 mL elution buffer (500 mM
NaCl, 20 mM Tris pH 8.0, 500 mM imidazole). The protein was
further purified with HiLoad 16/60 Superdex 200 pg (GE healthcare). The pvLAAD proteins were desalted and concentrated, and
stored in 50 mM NaCl and 2 mM Tris pH 8.0 at 80 °C.
2.2. Crystallography
Because FL-pvLAAD is instable, we only chose DN-pvLAAD for
crystallization trials with sitting-drop vapor-diffusion method.
Each drop consisted of 1 lL protein (10 mg/mL) plus 1 lL reservoir
solution [5% m/v PEG20,000, 200 mM MES, and 1 mM
cetyltrimethylammonium bromide (CTAB)], and equilibrated
against 200 lL reservoir solution at 12 °C. Crystals appeared after
3 days, and was grown to a size of100X50X50 lm3. Diffraction data
was collected using a single crystal at 100 K at beamline 3W1A of
Beijing Synchrotron Radiation Facility (BSRF). The whole data set
contains 360 frames, and the oscillation angle is 1° for each frame.
The diffraction data were processed and scaled with HKL2000
(Otwinowski and Minor, 1997). The DN-pvLAAD crystal belonged
to P1 space group, and contained six DN-pvLAAD molecules in an
asymmetric unit.
Because of lacking known structure from close homologues,
molecular replacement method did not give a correct solution.
We next tried to produce selenomethionine (SeMet) derivative of
DN-pvLAAD, but the crystals did not give enough anomalous dispersion signal. The possible reason might be that selenomethionine
has quickly degraded by pvLAAD like methionine (Hossain et al.,
2014c; Takahashi et al., 1999). As a alternative strategy, we crystallized a further engineered pvLAAD [MRGSHHHHHHGS-(res.
30-325)-GGSS-(res. 375-471)] (DN-DINS-pvLAAD), which removed
the insertion sequence specific to the membrane-bound LAADs.
DN-DINS-pvLAAD presented much lower activity to selenomethionine, and its structure was solved using single-wavelength
anomalous dispersion (SAD) of selenium, and refined to 1.20 Å
with R/Rfree = 14.1%/15.7%. Then, the DN-DINS-pvLAAD structure
was used as the template of molecular replacement to solve the
structure of DN-pvLAAD. Six DN-pvLAAD molecules were found
in the P1 unit cell with program Molrep. Structure refinement
and model building were performed in Refmac5 (Murshudov
et al., 2011) and Coot (Emsley et al., 2010). The final model was
finally refined to 2.63 Å with R/Rfree = 22.1%/25.2%, and achieved
good stereochemistry quality assessed by program MolProbity
(Chen et al., 2010). The statistics parameters of data collection
and structure refinement were listed in Table 1.
2. Materials and methods
2.3. Liposome-binding assays
2.1. Protein expression and purification
The cDNA of full length L-amino acid deaminase from P. vulgaris
(pvLAAD, GeneBank Accession Number BAA90864) was kindly
provided by Dr. Eiji Takahashi (Tanabe Seiyaku Co. Ltd., Japan).
Liposome-binding assays were performed as reported (Zhu
et al., 2015), with some modifications. In brief, E. coli total lipid
extracts (Avanti Polar Lipids) were used to construct liposomes
to mimic the bacterial cytomembrane. For liposome preparation,
308
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
Table 1
Statistics of X-ray diffraction data collection and structure refinement.
Data collection
Wavelength (Å)
Resolution (Å)
Space group
Cell parameters a, b, c
(Å)
Cell parameters a, b, c (°)
Unique reflections
Redundancy
Completeness (%)
Average I/r(I)
Rmergeb (%)
Refinement
Resolution (Å)
Reflections for
refinement/test
Rworkc/Rfreed (%)
RMSD bond (Å)
RMSD angle (°)
Mean B factor (Å2)
Non-hydrogen protein
atoms
Water oxygen atoms
FAD atoms
CTAB atoms
Ethylene glycol
Ramachandran plot (%)
Favoured
Allowed
SeMet-DN-DINSpvLAAD
DN-pvLAAD
0.979
50.0–1.20(1.24–
1.20)a
P212121
a = 57.3, b = 63.7,
c = 125.0
a = b = c = 90
1.000
50.0–2.63(2.72–2.63)
139986(13965)
10.3(10.6)
97.8(98.8)
34.1(3.1)
7.2(55.6)
P1
a = 100.3, b = 104.6,
c = 105.4
a = 64.5, b = 73.1,
c = 61.2
97176(8581)
3.8(3.1)
97.2(86.2)
31.0(2.4)
4.1(36.2)
50.00–1.20(1.23–
1.20)
6840(507)
50.00–2.63(2.70–2.63)
14.1/15.7(20.1/20.7)
0.006
1.37
13.7
3032
22.1/25.2(34.9/36.9)
0.008
1.25
91.7
19661
514
53
0
36
106
318
300
0
98.0
100.0
96.5
99.8
92242/4861
a
Values in parentheses are for the highest resolution shell.
Rmerge = RhRl|I(h)l <I(h)>|/RhRlI(h)l, where I(h)l is the lth observation of the
reflection h and <I(h)> is the weighted average intensity for all observations l of
reflection h.
P
P
c
Rwork = h||Fobs(h)| |Fcal(h)||/ h|Fobs(h)|, where Fobs(h) and Fcal(h) are the
observed and calculated structure factors for reflection h respectively.
d
Rfree was calculated as Rwork using 5% of the reflections which were selected
randomly and omitted from refinement.
b
the 20 mg lipids were dissolved in 2 mL chloroform. Chloroform
was evaporated under a stream of N2 for 20 min, and lipids films
were further dried with a vacuum pump overnight. Lipid films
were hydrated at room temperature with constant mixing in
4 ml of binding buffer (100 mM NaCl and 50 mM Tris pH 8.0) for
1 h, and then lipid vesicles were sonicated briefly for 5 min on
ice. For liposome-binding assays, 20 lg of FL-pvLAAD, DN-pvLAAD,
DINS-pvLAAD, DN-DINS-pvLAAD and its variants were incubated
with liposomes (25–400 lg) at 4 °C for 30 min in 100 lL of binding
buffer. Then, the liposomes were pelleted by centrifugation at
70000g (Beckman OptimaTM L-100XP ultracentrifuge) for 1 h at
4 °C. The pellets were washed twice with binding buffer. Then,
the pellets were re-suspended in 30 lL of SDS-PAGE sample buffer,
and the proteins bound to liposomes were analyzed by SDS-PAGE.
2.4. Activity measurement
The catalytic activity of DN-pvLAAD was determined by measuring the production of a-keto acid using the assays as described
(Yu et al., 2014). In brief, each of the twenty L-amino acids was dissolved to a concentration of 10 mM in the buffer of 20 mM TrisHCl, and the final pH was adjusted to 8.0. 500 lL of each L-amino
acid was mixed with 20 lL of the enzyme (200 lM), and incubated
at 37 °C for 20 min. 450 lL of 20% trichloroacetic acid was added to
each reaction, then the reactions were kept at room temperature
for 30 min. Next, 200 lL of 20 mM 2,4-dinitrophenylhydrazine
(DNP, Sangon Biotech) was added and mixed, and the reactions
were further incubated at room temperature for 15 min to generate brownish-red dinitrophenylhydrazone. Next, 4 mL of 0.8 M
NaOH was added, and additional incubation of 15 min was applied
at room temperature. Finally, the mixture was centrifuged and the
supernatant was measured for the absorbance at 520 nm. The concentrations of a-keto acid were calculated according to the standard curve which was generated by adding DNP to different
concentrations (25, 100, 250, 500, 1000 and 2000 lM) of a-ketoc-methylthiobutyric acid (Sigma-Aldrich). Blank reactions without
adding any enzyme were used to subtract the background absorbance. Reactions were repeated for four times, and the data were
shown as mean ± standard deviation (SD).
Protein preparation and activity measurement of DN-pvLAAD
mutants (with single mutation of Q99A, R138A, R315A, Q278L or
I317F) were similar to the wild type enzyme. The differences
between wild type and mutant DN-pvLAADs were evaluated by
t-test using GraphPad Prism 6.0, and P < 0.05 was reported to be
significant (⁄P < 0.05, ⁄⁄P < 0.01; ⁄⁄⁄P < 0.001).
2.5. Detecting the peroxide production of pvLAAD variants
500 lL L-methionine (10 mM in 20 mM Tris pH 8.0) was mixed
with 20 lL of the enzyme (200 lM of FL-pvLAAD, DN-pvLAAD,
DINS-pvLAAD, DN-DINS-pvLAAD), and the reactions were incubated at 37 °C for 20 min. Then, 470 lL of the reactions were used
to quantify the catalytic activities with the assays described above,
and the rest 50 ll of the reactions were used to test the production
of H2O2 using FeIIIXO agar assay as described (Yu et al., 2013). In
brief, the ferrous-XO (FeIIXO) agar plates were poured with boiled
1.5% agar supplemented with 0.25 mM FeSO4, 6 mM H2SO4,
0.15 mM xylenol orange (XO) and 0.1 mM D-sorbitol (all chemicals
were purchased from Sangon Biotech). Then, circular wells were
made using a hole puncher whose diameter is 6 mm. 50 lL of each
reaction (see the above paragraph) were added to the wells, and
waited for 60 min at room temperature to allow red halos forming.
The diameters of the red halos reflect the concentration of H2O2
(Yu et al., 2013). The production of H2O2 was also measured using
Peroxide Test strip (Merck Millipore). According to instruction, the
test strips were immersed in the reactions for 1 s, and then taken
out and wait 15 s before read. A series of concentrations of H2O2
(0, 62.5, 125, 250, 500 and 1000 lM) were used to make a standard
color card. 10 lM of caLAAO (LAAO from Crotalus adamanteus
venom, Worthington Biochemical) was used as the positive
control.
2.6. Model the structure of pvLAAD with substrate L-methionine
A substrate L-methionine was put to the active site of
DN-pvLAAD according to the conformations of L-alanine and
L-phenylalanine in Rhodococcus opacus LAAO (PDB codes 2JB1 and
2JB2) (Faust et al., 2007), and the side chain of L-methionine was
manually adjusted to avoid clashes. This initial model of
DN-pvLAAD complexed with L-methionine was optimized using
molecular dynamics (MD) simulations with a restrained distance
between L-methionine and FAD isoalloxazine ring. The MD simulations were carried out using the AMBER12 molecular simulation
package (Case et al., 2005). Models were prepared using the ff99SB
force field and general AMBER force field (gaff) force field for
protein and substrates, respectively. Structures were collected by
sampling at 100 ps intervals, and were clustered according to the
root mean-square deviation (r.m.s.d.) of residues around the active
site with ptraj program in AMBER12.
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
2.7. Sequence analysis and structure presentation
Protein sequence alignments were done using program ClustalW (Chenna et al., 2003) and adjusted manually. The phylogenetic tree of phylum Proteobacteria LAADs were constructed
using neighbor-joining method in Mega5 (Tamura et al., 2011).
All protein structure illustrations were prepared with the program
PyMol (www.pymol.org).
3. Results and discussion
3.1. Overview of pvLAAD structure
Sequence alignment of pvLAAD to its related amino acid oxidoreductases revealed a conserved FAD-binding motif (with an
absolutely conserved -GxGxxG- sequence) in their N-terminal
region (Supplemental Fig. S1). In contrast to other amino acid oxidoreductases, pvLAAD contained an additional N-terminal peptide
before the FAD-binding motif and a specific insertion of about 50
residues (res. 321-375) (Supplemental Fig. S1). Previous studies
showed that the N-terminal peptide is a transmembrane element
(Pantaleone et al., 2001; Baek et al., 2011), while the function of
the specific insertion is unknown. We removed the N-terminal
transmembrane peptide (res. 1-29) during cloning to improve protein expression and stability. The DN-pvLAAD construct used in
crystallography contained the residues from Arg30 to Phe471
and an N-terminal hexahistidine tag. The structure of DN-pvLAAD
was solved to 2.63 Å resolution with R/Rfree factors = 22.1%/25.2%
(Table 1). An asymmetric unit contained six DN-pvLAAD chains
that are almost identical to each other. Thus the chain A was used
as the representative structure in later analysis unless otherwise
indicated.
The structure of DN-pvLAAD can be divided into three domains:
an FAD-binding domain (res. 30-98, 202-274, and 412-471), a
substrate-binding domain (res. 99-201, 275-320, 376-411), and
the specific insertion module (res. 321-375) (Fig 1a). The FADbinding domain consists of a mixed six-stranded b-sheet at the
center, which is sandwiched by a helix bundle on one side and
by a three-stranded antiparallel b-sheet and a single helix on the
other side. This structure greatly resembles the structures of LAAOs
309
and other GR (glutathione reductase) family flavoproteins (Dym
and Eisenberg, 2001). The substrate-binding domain of pvLAAD
consists of a mixed eight-stranded b-sheet flanked by two helices
on both sides. The active cavity locates between the FAD- and
substrate-binding domains, and the isoalloxazine ring of cofactor
FAD lies at the bottom of the active cavity, which is accessible to
the solvent (Fig. 1a). The specific insertion forms a mixed structure
module consisting of one b-strand and three a-helices on the surface. The b-strand of the specific insertion forms a two-stranded
parallel b-sheet with a b-strand from the substrate-binding
domain, and tethers the insertion module near to the activity cavity (Fig. 1a).
We searched for the structural homologues of pvLAAD in Protein Data Bank using DALI (Holm and Rosenstrom, 2010), and the
best outputs included L-proline dehydrogenase I (PDH1; PDB code
1Y56-B; r.m.s.d. 2.1 Å for 359 Ca pairs; 23% sequence identity), glycine oxidase (GO; PDB code 1RYI-C; r.m.s.d. 2.2 Å for 354 Ca pairs;
21% sequence identity) and also D-amino acid oxidase (DAAO; PDB
code 1C0P; r.m.s.d. 2.8 Å for 300 Ca pairs; 15% sequence identity).
When superimposed DN-pvLAAD to these three enzymes, both
FAD-binding and substrate binding domain could be well superposed. The only major difference is that DN-pvLAAD contains the
insertion module, which is specific to membrane-bound LAADs
(Fig. 1a and b).
Typical LAAOs were not on the top list of structure homologues of
pvLAAD. Typical LAAOs consist of three domains: an FAD-binding
domain, a substrate-binding domain and a helical domain
(Pawelek et al., 2000; Zhang et al., 2004b). We found that the
FAD-binding domains in pvLAAD and typical LAAOs could be well
superimposed with each other. However, the core b-sheets in their
substrate-binding domains adopt significantly different orientations (Fig. 1c). Moreover, both the structure and the location are dissimilar between the insertion module of pvLAAD and the helical
domain of LAAOs (Fig. 1c).
In addition, the crystal structure of DN-DINS-pvLAAD was
solved to 1.20 Å resolution with R/Rfree factors = 14.1%/15.7%.
Structures of both FAD-binding domain and substrate-binding
domain are very similar between DN-DINS-pvLAAD and
DN-pvLAAD, and superimposition of two structures revealed an r.
m.s.d. of 1.22 Å for 375 Ca pairs (Supplemental Fig. S2). However,
a significant difference is observed at active site. In DN-pvLAAD,
Fig. 1. Structure of pvLAAD. (a) Cartoon representation of the overall structure of pvLAAD. pvLAAD consists of an N-terminal membrane-binding peptide (not included in this
structure model), an FAD-binding domain (colored cyan), a substrate-binding domain (colored yellow) and an insertion sequence (colored orange). Cofactor FAD is shown as
sticks (colored magenta). (b) Superimposition of pvLAAD, phPDH1 (PDB # 1Y56-B, colored green) and bsGO (PDB # 1RYI, colored blue) reveals significant structural similarity
for their FAD- and substrate-binding domains. pvLAAD has an extra structure module formed by its unique insertion sequence. (c) Superimposition of pvLAAD and crLAAO
(PDB # 1F8R, colored gray for FAD- and substrate-binding domains, and red for the helical domain) reveals different orientation of the b-sheet in their substrate-binding
domains. The insertion sequence of pvLAAD and the helical domain of crLAAO have different structure and location.
310
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
residues I317 to P320 forms a short b-strand, which further forming a two-stranded parallel b-sheet with the b-strand from the
specific insertion (Supplemental Fig. S2). After removing the
specific insertion, residues I316 to P319 become disorder in
DN-DINS-pvLAAD. A loop moves towards the active site for about
7 Å, and residues V316 and I317 occupy the substrate-binding site
(Supplemental Fig. S2). This conformational difference explains
why DN-DINS-pvLAAD is much less active.
The 1.20 Å density map of DN-DINS-pvLAAD clearly showed
that FAD binds to pvLAAD non-covalently and forms extensively
hydrogen-bonding interactions (Supplemental Fig. S3a). Interestingly, the isoalloxazine ring of FAD in DN-DINS-pvLAAD adopted
a bent conformation along the N5–N10 axis with a dihedral angle
about 10° (Supplemental Fig. S3b). It was suggested that the conformations of the isoalloxazine ring could reflect the redox states
of FAD, that the isoalloxazine rings adopt a planar conformation
in oxidized flavins and the bent conformations (the angle is up to
34°) in reduced flavins (Senda et al., 2009). So the bent conformation of FAD in DN-DINS-pvLAAD implied that the inactivation of
this double-truncated protein is due to inefficient re-oxidization
of cofactor FAD. However, recent studies reported that the conformational changes of flavin might be caused by strong X-ray radiation during data collection (Røhr et al., 2010).
Furthermore, we did not observe significant protein-protein
interface in either DN-pvLAAD or DN-DINS-pvLAAD crystals,
suggesting pvLAAD is monomeric, which is confirmed by sizeexclusion chromatography (Supplemental Fig. S4). Our biochemical analysis showed that the monomeric DN-pvLAAD was
enzymatically active (discussed later). While LAAOs predominately
form homodimers (Du and Clemetson, 2002; Yu and Qiao, 2012),
our findings suggested that pvLAAD represents a new class of
monomeric LAAOs.
3.2. The specific insertion of membrane-bound LAADs
Previous bioinformatics analysis identified a twin-arginine
translocation (Tat) signal peptide at the N-terminus of
membrane-bound LAAD, which was also found in other secreted
or membrane-bound proteins containing redox cofactors (Berks,
1996). Recently, researchers confirmed that the N-terminal Tat
peptide was indispensable for leading and anchoring LAADs to
the outer side of the cytomembrane (Hossain et al., 2014b). However, the transmembrane helix in the central region of Tat signal
peptide is less hydrophobic compared to typical transmembrane
helices. As a result, the Tat signal peptide-mediated membrane
binding is relatively weak, and these proteins are in a risk of dissociating from the cytomembrane (Bachmann et al., 2006). The
hydropathy plot was drawn using program MPEx (Snider et al.,
2009), and it revealed four hydrophobic segments (res. 8-26, 6482, 251-269 and 330-363) in pvLAAD sequence (Supplemental
Fig. S5). The first segment (res. 8-26) is corresponding to the
N-terminal Tat membrane-binding peptide. The second (res. 6482) and third (res. 251-269) segments are buried inside of
FAD-binding domain (Supplemental Fig. S5), which cannot
interact with the membrane. The fourth hydrophobic segment
(res. 330-363) is a part of the specific insertion (res. 321-375) on
pvLAAD surface.
Next, the surface hydrophobicity of pvLAAD was analyzed, and
we found that the specific insertion forms the largest hydrophobic
module on protein surface (Supplemental Fig. S6). Seven
hydrophobic residues (Val321, Phe326, Ile345, Leu347, Leu351,
I352 and Phe355) from the hydrophobic module were exposed to
solvent (Fig. 2a), and sequence alignments showed that all of them
are conserved among LAADs from genera Proteus and Providencia
(Fig. 2b). The hydrophobic insertion module in the six DN-pvLAADs
of the asymmetric unit captured fifteen CTAB detergents through
hydrophobic interactions during crystallization (Supplemental
Fig. S7). Some detergent molecules interacted with the insertion
modules of two or more DN-pvLAADs at the same time and are
involved in crystal packing. Base on these structural observations,
we hypothesize that the hydrophobic insertion module is another
membrane-binding site.
To test this hypothesis, we conducted liposome-binding assays
to evaluate the membrane-binding ability of pvLAAD in its intact
form (FL-pvLAAD), N-terminal membrane-binding peptidedeleted form (DN-pvLAAD), insertion sequence-deleted form
(DINS-pvLAAD) and the double deletion form (DN-DINS-pvLAAD).
The liposome prepared from the total lipid extract of E. coli was
used to mimic the bacterial cytomembrane. In the liposomebinding assays, purified FL-pvLAAD, DN-pvLAAD, DINS-pvLAAD
and DN-DINS-pvLAAD were incubated with various amounts of
liposomes. The liposomes were pelleted through ultracentrifugation, and the proteins associating with liposomes were analyzed
using SDS-PAGE. As shown in Fig. 2c, when the amount of
liposomes reduced, the liposome binding of FL-pvLAAD was still
maintained at a high level. In contrast, the binding of DINS-pvLAAD
was gradually decreased, and the band of DINS-pvLAAD became
very weak at the lowest concentration of liposomes. Consistently,
the liposome binding of DN-DINS-pvLAAD was found to be weaker
than that of DN-pvLAAD (Fig. 2c).
These results from liposome-binding assays supported the idea
that the specific hydrophobic module could enhance the membrane binding. We propose that the specific insertion of
membrane-bound LAADs is the second membrane-binding site,
which may be developed during the evolution to assist the Nterminal transmembrane peptide for membrane anchoring. With
the assistance of the specific insertion, LAADs could be tightly
anchored onto the cytomembrane to perform their physiological
functions. However, it is also possible that this hydrophobic module is a protein-protein interaction site and can stabilize the membrane association through interacting with other hydrophobic
membrane proteins.
We also found that the a1-loop-b1 of the insertion module partially covered the active cavity of pvLAAD. We analyzed the active
cavities of PDH1, DAAO and pvLAAD using program HOLLOW (Ho
and Gruswitz, 2008). PDH1 and DAAO both have one wide cavity,
and the isoalloxazine ring of FAD lies at the bottom of the cavity
(Fig. 2d and e). A loop functions as the ‘‘active site lid” in DAAO,
which switches between the ‘‘closed” conformation and ‘‘open”
state (Todone et al., 1997). In pvLAAD, the a1-loop-b1 of the insertion module divides the wide cavity into two parts: a smaller cavity
and a narrow channel (Fig. 2f). The narrow channel is hydrophobic,
consisting of residues from the insertion sequence as well as the
residues from the substrate-binding domain. A detergent molecule
is deeply inserted into the channel and reaches to the top of FAD
isoalloxazine ring, where the deamination reaction happens
(Fig. 2g). Because the hydrophobic module is proposed to participate
in membrane binding, this adjacent channel seems close to the
membrane too, but we believe there is still enough space for small
molecules to come in and out. The narrow hydrophobic channels
for O2 and/or H2O2 delivering have been identified in many LAAOs
(Moustafa et al., 2006). It will be interesting to see whether the
hydrophobic channel formed by the specific insertion sequence in
membrane-bound LAADs also plays a role in substrate or product
delivering.
3.3. The active site of pvLAAD
The active cavity of pvLAAD was between substrate-binding
domain and FAD-binding domain, and the isoalloxazine ring of
FAD is located at the bottom of the cavity (Fig. 1a). Previous studies
showed that the key residues involved in substrate binding were
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
311
Fig. 2. The unique insertion module of LAAD. (a) The insertion module is draw as cartoon on pvLAAD surface. The solvent accessible hydrophobic residues forming the
hydrophobic patch are shown as sticks. (b) Sequence alignments showed that all of the residues forming the hydrophobic patch are conserved in LAADs from genera Proteus
and Providencia (colored red). (c) pvLAAD and its truncates binding to liposome. 20 lg of each protein was incubated with 25–400 lg of liposomes constructed using E. coli
total lipid extracts, and the proteins associated with liposome were analyzed with SDS-PAGE. When liposome concentration reduced, the binding of full length pvLAAD
remained strong, but the binding of DN-pvLAAD and DINS-pvLAAD decreased gradually. Weak binding of DN-DINS-pvLAAD was only observed at the highest concentration of
the liposome. These results suggest both N-terminal Tat signal peptide and the insertion module facilitate membrane binding. (d)–(f) The active cavities are shown as the
green surface. PDH1 (PDB # 1Y56-B) and DAAO (PDB # 1C0P) each has a large cavity. However, this cavity is divided by the insertion module into a smaller cavity and a
narrow channel in pvLAAD. (g) The narrow channel is formed by hydrophobic residues from the insertion sequence (colored orange) and the substrate-binding domain
(colored yellow). A cetyltrimethylammonium bromide (CTAB) molecule (colored green) inserts to active center through this channel.
mirrored through a plane perpendicular to the isoalloxazine ring of
FAD between LAAOs and DAAOs so that the L-amino acids and
D-amino acids bound to their oxidases were also with the mirrored
geometry (Pawelek et al., 2000). We superimposed the
FAD-binding domain of pvLAAD to that of Rhodococcus opacus
LAAO (RoLAAO) complexed with L-alanine and L-phenylalanine
(PDB codes 2JB1 and 2JB2) (Faust et al., 2007). We manually put
an L-methionine to pvLAAD active site according to the position
and conformation of L-alanine and L-phenylalanine in RoLAAO.
The conformation of L-methionine side chain was manually optimized to avoid clashes. Then, the initial model of pvLAADmethionine complex was optimized using molecular dynamics
(MD) simulations with the restrained distance between
L-methionine and FAD isoalloxazine ring. The r.m.s.d trajectory
showed that both the backbones of whole DN-pvLAAD and the
backbones of the residues around the active cavity were stable
during simulations (Fig. 3a). In contrast, when side chain atoms
were counted, the r.m.s.d of the residues around the active cavity
fluctuated during simulations, indicating that the conformations
of the side chains dramatically changed. The conformations of
active cavity were clustered to five groups. Cluster 1, 2 and 4 are
dominant, and their representative structures were shown
(Fig. 3b). Q99 contributed a stable hydrogen bond to stabilize the
carboxyl group of L-methionine. The side chain of R315 was flexible, and its conformation dramatically changed during simulations.
Finally, it flipped into the active site, and formed a hydrogen bond
with L-methionine (Fig. 3b). In contrast, R138, another basic
residue in the active cavity, did not interact with L-methionine.
In the final model (in the cluster 4), the Ca-H of the
L-methionine pointed to FAD N5 atom, ready for hydrogen transfer
according to the hydride mechanism and the geometry observed in
LAAOs (Faust et al., 2007). The side chain of Q99 and R315 form
312
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
Fig. 3. Model a substrate to pvLAAD active site. (a) An L-methionine was manually modeled into pvLAAD active site, and the initial model was optimized using molecular
dynamics (MD) simulations with AMBER12 program package. The r.m.s.d-based trajectory analysis showed that both the backbones of whole DN-pvLAAD and the backbones
of the residues around the active cavity were quite stable during simulations. In contrast, the all-atoms r.m.s.d of the residues around the active cavity fluctuated during
simulation, indicating the possible side chain flipping during simulation. (b) The conformations of active cavity were clustered, and the representative structure was shown.
Q99 and R315 were found to could interact with the carboxyl group of L-methionine. Another basic residue in the active cavity, R138, did not interact with L-methionine. (c)
Activities of wild type, Q99A, R138A and R315A DN-pvLAAD were measured for twenty L-amino acids. Activity of wild type enzyme for L-methionine is defined as 100%. The
results were repeated four times, and the activity differences between wild type and each mutant were analyzed by t-test using GraphPad 6.0. The activities of both Q99A DNpvLAAD and R138A DN-pvLAAD significantly decreased for most of substrate amino acids. In contrast, R138A did not decrease the activity.
hydrogen bonds with the carboxyl group of L-methionine, and stabilizes the enzyme-substrate complex (Fig. 3b). To verify the
pvLAAD-methionine complex model, we mutated residues Q99,
R138 and R315 to alanine respectively, and measured the activities
of the three mutants and wild type DN-pvLAAD. The activity of
wild type enzyme for L-methionine was defined as 100% (Hossain
et al., 2014c; Takahashi et al., 1999). As shown in Fig. 3c, the purified wild type DN-pvLAAD exhibited significant activities for basic
residues of histidine, arginine and lysine. It also catalyzed other
eleven L-amino acids including alanine, asparagine, aspartate, isoleucine, leucine, methionine, phenylalanine, serine, tryptophan,
tyrosine and valine. In contrast, cysteine, glutamine, glutamate,
glycine, proline and threonine were not good substrates for DNpvLAAD. For most amino acids, both the mutation Q99A and mutation R315A could significantly decrease the activity of DN-pvLAAD.
But the decrease induced by R315A is less dramatically than Q99A
for some amino acids such as leucine, tryptophan and tyrosine
(Fig. 3c). In contrast, the activities of mutant R138A were similar
to the activities of wild type DN-pvLAAD for most L-amino acids
(Fig. 3c). These results supported the pvLAAD-methionine complex
model proposed above that Q99 and R315 bind with the carboxyl
group of substrate amino acids.
3.4. Two types of membrane-bound LAADs
We used pvLAAD to BLAST microbial genome database to search
for homologues in phylum Proteobacteria. Many species (such as
E. coli) did not encode any close homologues of pvLAAD. In some
bacteria from Alphaproteobacteria and Betaproteobacteria, they
have LAADs without the N-terminal Tat membrane-binding
peptide. Whether these LAADs function as intracellular soluble
proteins or they bind to cell membrane in other ways, such as
through the hydrophobic insertion sequence, is unknown.
P. mirabilis was reported to express two types of membranebound LAADs with different substrate preferences (Baek et al.,
2011). We found that all the species we looked at from genera
Proteus and Providencia encoded two LAAD proteins. In a phylogenetic tree of LAADs built by us, the two LAADs from each species
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
were equally clustered into two groups. One of the two LAADs was
clustered into a group containing type I P. mirabilis LAAD (pmLAAD)
and the other LAAD from the same specie was clustered into the
group containing type II pmLAAD (Fig. 4a). These results implied
that the existence of two types of LAADs is actually a common feature among genera Proteus and Providencia.
To understand the underlying mechanism of their different substrate preferences, we analyzed the residues around the substratebinding pockets of Type I and II LAADs. We found that residues
forming the left side of the pocket, such as Y97, Q99, I101, R138
and W438 (numbered according to pvLAAD), were conserved in
all twelve LAADs from genera Proteus and Providencia
313
(Fig. 4b and c). In contrast, some residues forming the right side
of the substrate-binding pocket were only conserved within the
LAADs belonging to the same type, but not conserved across type
I and type II LAADs. For examples, Q278, F301, S313 and I317 of
pvLAAD are conserved in all type II LAADs. But these residues are
substituted by leucine, histidine, alanine and phenylalanine in type
I LAADs (Fig. 4b and c).
In the pvLAAD-methionine complex model proposed above, the
side chain of L-methionine extended to the right side and to be
close to Q278 and I317 (Fig. 4b). Q278 and I317 are conserved in
all the six type II LAADs but not in type I LAADs, in which they
are substituted by leucine and phenylalanine (Fig. 4c). While
Fig. 4. Two types of LAAD with different substrate spectrum. (a) Phylogenetic tree for the LAADs from phylum Proteobacteria. (b) The substrate-binding pocket of the pvLAADs
is shown as semitransparent surface and sticks. Residues conserved in both Type I and II LAADs are colored blue, while residues conserved only in Type II LAADs are colored
green. An L-methionine is modeled in the pocket according to the mode of L-amino acid binding to LAAO. (c) Sequence alignments of residues forming the substrate-binding
pockets in Type I and II LAADs from genera Proteus and Providencia. Residues forming the substrate-binding pocket are highlighted blue if conserved in both Type I and II
LAADs, green if only conserved in Type II, or orange if only conserved in Type I. (d) Activities of wild type, Q278L and L317F DN-pvLAAD for twenty L-amino acids. Activity of
wild type enzyme for L-methionine is defined as 100%. The results were repeated four times, and the activity differences between wild type and each mutant were analyzed
by t-test using GraphPad 6.0. Both mutations significantly decreased the activities of DN-pvLAAD for most L-amino acids in vitro, but increased the activities for the three
aromatic amino acids.
314
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
Fig. 5. The purified pvLAAD produces H2O2 in vitro. (a) Activity of intact and truncated pvLAAD were tested by measuring the generation of a-keto acid using chromogenic
reaction with 2,4-dinitrophenylhydrazine (DNP). Reactions containing caLAAO, FL-pvLAAD and DN-pvLAAD became brownish-red, indicating the production of significant
amount of a-keto acid. The color was quantified by measuring absorption at 520 nm, and amounts of product a-keto acid were quantified by comparing with the standard
curve. The results are from three independent assays, and the error bars are SEM (standard error of the mean). (b) Detection of H2O2 by FeIIIXO agar assays. 50 lL of each
reactions were added to the wells of FeIIXO agar plate, and red halos were generated by reactions containing caLAAO, FL-pvLAAD and DN-pvLAAD, indicating the production of
H2O2. The sizes of the halos reflected the amounts of H2O2. (c) Production of H2O2 was also measured using Peroxide Test strip. The amount of H2O2 produced by caLAAO, FLpvLAAD and DN-pvLAAD were estimated by comparing with standard color card.
isoleucine in type II LAADs and it corresponding phenylalanine in
type I LAADs are both hydrophobic residues, the glutamine in type
II LAADs results in a more hydrophilic environment compared to
the corresponding leucine in type I LAADs. Consistently, the type
I LAADs prefer aliphatic and aromatic amino acids, while type II
LAADs show significant activities to basic amino acids (Baek
et al., 2011; Takahashi et al., 1999).
To study the roles of Q278 and I317 in substrate selectivity, we
expressed and purified the DN-pvLAADs containing Q278L or I317F
mutations and measured their activities for twenty L-amino acids.
For most L-amino acids, both mutants were significantly less active
than the wild type enzyme (P < 0.05), and the activity decrease
caused by Q278L was usually more significant than caused by
I317F (Fig. 4d). Interestingly, the activities of both mutants for
the three aromatic amino acids increased compared to wild type
DN-pvLAAD (Fig. 4d), supporting the idea that increasing the
hydrophobicity of the side chain-binding site pocket can increase
the activity of LAADs to aromatic substrates. However, this idea
does apply to other hydrophobic amino acids. Q278L mutation dramatically decreased the activity for leucine, isoleucine, valine and
alanine (Fig. 4d), suggesting that hydrophobicity is not the only
factor affecting substrate selectivity. These results confirmed that
Q278 and I317 are important in substrate selectivity, and they
are the prior candidates for protein engineering to develop biocatalysts with specificity to particular amino acid substrate.
3.5. Production of peroxide by purified pvLAAD in vitro
Previous studies have suggested that membrane-bound LAADs
employed an unusual mechanism for L-amino acid deamination
without producing peroxide (Pantaleone et al., 2001; Yu and Qiao,
2012). We measured the in vitro enzyme activities of full-length
and truncated pvLAADs by measuring the production of a-keto acid.
L-methionine was used as the substrate. A highly active snake venom
LAAO from Crotalus adamanteus (caLAAO) was used as the positive
control. We confirmed that both purified FL-pvLAAD and DNpvLAAD were at least partially active in vitro, while the other two
pvLAAD truncates without the specific insertion sequence were
much less active (Fig. 5a). Then we measured the production of
H2O2 using FeIIIXO agar assays. The results showed that only caLAAO
and the active FL-pvLAAD and DN-pvLAAD produced H2O2, but not
for the inactive DINS-pvLAAD and DN-DINS-pvLAAD (Fig. 5b). The
production of H2O2 was further confirmed using peroxide test strips
and semi-quantified by comparing with standard color card
(Fig. 5c). Roughly, FL-pvLAAD, DN-pvLAAD and caLAAO each pro-
duced about 250–500 lM of H2O2 (Fig. 5c), which were similar to
the amount of a-keto acids produced during the catalysis (Fig. 5a).
Our results suggested that pvLAADs are at least partially active without any components from electron transport chain, and they also
posses the ability of producing H2O2 in vitro.
4. Summary
pvLAAD represents a class of membrane-bound LAAOs/LAADs.
The structure of pvLAAD is dissimilar to typical LAAOs, but more
like DAAOs plus a specific insertion. This specific insertion formed
a hydrophobic insertion module on the surface, and it could play
an assistant role on stabilizing the membrane association of LAADs.
In the substrate-binding cavity, Q99 and R315 were suggested to
interact with the carboxyl group of substrate amino acids, while
Q278 and L317 might be important for the substrate selectivity.
The purified pvLAAD was partially active in vitro, and could produce H2O2 like typical LAAOs. These findings will improve the
understanding of this family of enzymes and will help engineering
better biocatalysts.
Accession numbers
The structure factors and atomic coordinates have been
deposited to Protein Data Bank with the accession number 5I39
for DN-DINS-pvLAAD and 5HXW for DN-pvLAAD.
Author contributions
L.N., M.T., and H.Z. designed the research, Y.J., S.T., W.Z., Q.L. and
H.Z. performed the experiments, Y.G., Q.G., J.X., L.N., M.T., and H.Z.
analyzed the data, Y.J. and H.Z. wrote the manuscript. All the
authors reviewed the results and approved the final version of
the manuscript.
Conflict of interest
The authors declare that they have no conflicts of interest with
the contents of this article.
Acknowledgements
We thank Dr. Eiji Takahashi (Tanabe Seiyaku Co. Ltd., Japan) for
kindly providing pvLAAD cDNA. We thank staff on beamline 3W1A
Y. Ju et al. / Journal of Structural Biology 195 (2016) 306–315
of the Beijing Synchrotron Radiation Facility (BSRF) for the support
in X-ray diffraction data collection. This research was supported by
the Chinese Ministry of Science and Technology (Nos.
2012CB917200 and 2011CBA00800), the Chinese National Natural
Science Foundation (Nos. 31130018 and 31170726), and also by a
start-up funding to H.Z. from Sun Yat-sen University (No. 3600018811200).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jsb.2016.07.008.
References
Bachmann, J., Bauer, B., Zwicker, K., Ludwig, B., Anderka, O., 2006. The Rieske protein
from Paracoccus denitrificans is inserted into the cytoplasmic membrane by the
twin-arginine translocase. FEBS J. 273, 4817–4830.
Baek, J.O., Seo, J.W., Kwon, O., Seong, S.I., Kim, I.H., Kim, C.H., 2011. Expression and
characterization of a second L-amino acid deaminase isolated from Proteus
mirabilis in Escherichia coli. J. Basic Microbiol. 51, 129–135.
Berks, B.C., 1996. A common export pathway for proteins binding complex redox
cofactors? Mol. Microbiol. 22, 393–404.
Case, D.A., Cheatham 3rd, T.E., Darden, T., Gohlke, H., Luo, R., Merz Jr., K.M., Onufriev,
A., Simmerling, C., Wang, B., Woods, R.J., 2005. The Amber biomolecular
simulation programs. J. Comput. Chem. 26, 1668–1688.
Chen, V.B., Arendall 3rd, W.B., Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J.,
Murray, L.W., Richardson, J.S., Richardson, D.C., 2010. MolProbity: all-atom
structure validation for macromolecular crystallography. Acta Crystallogr. D
Biol. Crystallogr. 66, 12–21.
Chenna, R., Sugawara, H., Koike, T., Lopez, R., Gibson, T.J., Higgins, D.G., Thompson, J.
D., 2003. Multiple sequence alignment with the Clustal series of programs.
Nucleic Acids Res. 31, 3497–3500.
Drechsel, H., Thieken, A., Reissbrodt, R., Jung, G., Winkelmann, G., 1993. Alpha-keto
acids are novel siderophores in the genera Proteus, Providencia, and Morganella
and are produced by amino acid deaminases. J. Bacteriol. 175, 2727–2733.
Du, X.Y., Clemetson, K.J., 2002. Snake venom L-amino acid oxidases. Toxicon 40,
659–665.
Duerre, J.A., Chakrabarty, S., 1975. L-amino acid oxidases of Proteus rettgeri. J.
Bacteriol. 121, 656–663.
Dym, O., Eisenberg, D., 2001. Sequence-structure analysis of FAD-containing
proteins. Protein Sci. 10, 1712–1728.
Emsley, P., Lohkamp, B., Scott, W.G., Cowtan, K., 2010. Features and development of
Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501.
Faust, A., Niefind, K., Hummel, W., Schomburg, D., 2007. The structure of a bacterial
L-amino acid oxidase from Rhodococcus opacus gives new evidence for the
hydride mechanism for dehydrogenation. J. Mol. Biol. 367, 234–248.
Guo, C., Liu, S., Yao, Y., Zhang, Q., Sun, M.Z., 2012. Past decade study of snake venom
L-amino acid oxidase. Toxicon 60, 302–311.
Ho, B.K., Gruswitz, F., 2008. HOLLOW: generating accurate representations of
channel and interior surfaces in molecular structures. BMC Struct. Biol. 8, 49.
Holm, L., Rosenstrom, P., 2010. Dali server: conservation mapping in 3D. Nucleic
Acids Res. 38, W545–W549.
Hossain, G.S., Li, J., Shin, H.D., Du, G., Liu, L., Chen, J., 2014a. L-amino acid oxidases
from microbial sources: types, properties, functions, and applications. Appl.
Microbiol. Biotechnol. 98, 1507–1515.
Hossain, G.S., Li, J., Shin, H.D., Chen, R.R., Du, G., Liu, L., Chen, J., 2014b. Bioconversion
of L-glutamic acid to alpha-ketoglutaric acid by an immobilized whole-cell
biocatalyst expressing L-amino acid deaminase from Proteus mirabilis. J.
Biotechnol. 169, 112–120.
Hossain, G.S., Li, J., Shin, H.D., Du, G., Wang, M., Liu, L., Chen, J., 2014c. One-step
biosynthesis of alpha-keto-gamma-methylthiobutyric acid from L-methionine
by an Escherichia coli whole-cell biocatalyst expressing an engineered L-amino
acid deaminase from Proteus vulgaris. PLoS One 9, e114291.
Hou, Y., Hossain, G.S., Li, J., Shin, H.D., Liu, L., Du, G., 2015a. Production of
phenylpyruvic acid from L-phenylalanine using an L-amino acid deaminase
from Proteus mirabilis: comparison of enzymatic and whole-cell
biotransformation approaches. Appl. Microbiol. Biotechnol. 99, 8391–8402.
Hou, Y., Hossain, G.S., Li, J., Shin, H.D., Du, G., Liu, L., 2015b. Combination of
phenylpyruvic acid (PPA) pathway engineering and molecular engineering of Lamino acid deaminase improves PPA production with an Escherichia coli
whole-cell biocatalyst. Appl. Microbiol. Biotechnol. 100, 2183–2191.
315
Izidoro, L.F., Sobrinho, J.C., Mendes, M.M., Costa, T.R., Grabner, A.N., Rodrigues, V.M.,
da Silva, S.L., Zanchi, F.B., Zuliani, J.P., Fernandes, C.F., Calderon, L.A., Stabeli, R.G.,
Soares, A.M., 2014. Snake venom L-amino acid oxidases: trends in
pharmacology and biochemistry. Biomed. Res. Int. 2014, 196754.
Kitani, Y., Tsukamoto, C., Zhang, G., Nagai, H., Ishida, M., Ishizaki, S., Shimakura, K.,
Shiomi, K., Nagashima, Y., 2007. Identification of an antibacterial protein as Lamino acid oxidase in the skin mucus of rockfish Sebastes schlegeli. FEBS J. 274,
125–136.
Liu, L., Hossain, G.S., Shin, H.D., Li, J., Du, G., Chen, J., 2013. One-step production of
alpha-ketoglutaric acid from glutamic acid with an engineered L-amino acid
deaminase from Proteus mirabilis. J. Biotechnol. 164, 97–104.
Moustafa, I.M., Foster, S., Lyubimov, A.Y., Vrielink, A., 2006. Crystal structure of
LAAO from Calloselasma rhodostoma with an L-phenylalanine substrate:
insights into structure and mechanism. J. Mol. Biol. 364, 991–1002.
Murshudov, G.N., Skubak, P., Lebedev, A.A., Pannu, N.S., Steiner, R.A., Nicholls, R.A., Winn,
M.D., Long, F., Vagin, A.A., 2011. REFMAC5 for the refinement of macromolecular
crystal structures. Acta Crystallogr. D Biol. Crystallogr. 67, 355–367.
Otwinowski, Z., Minor, W., 1997. Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol. 276, 307–326.
Pantaleone, D., Geller, A., Taylor, P., 2001. Purification and characterization of an Lamino acid deaminase used to prepare unnatural amino acids. J. Mol. Catal. B
Enzym. 11, 795–803.
Pawelek, P.D., Cheah, J., Coulombe, R., Macheroux, P., Ghisla, S., Vrielink, A., 2000.
The structure of L-amino acid oxidase reveals the substrate trajectory into an
enantiomerically conserved active site. EMBO J. 19, 4204–4215.
Pistorius, E.K., Voss, H., 1982. Presence of an amino acid oxidase in photosystem II of
Anacystis nidulans. Eur. J. Biochem. 126, 203–209.
Røhr, A.K., Hersleth, H.P., Andersson, K.K., 2010. Tracking flavin conformations in
protein crystal structures with Raman spectroscopy and QM/MM calculations.
Angew. Chem. Int. Ed. Engl. 49, 2324–2327.
Senda, T., Senda, M., Kimura, S., Ishida, T., 2009. Redox control of protein
conformation in flavoproteins. Antioxid. Redox Signaling 11, 1741–1766.
Snider, C., Jayasinghe, S., Hristova, K., White, S.H., 2009. MPEx: a tool for exploring
membrane proteins. Protein Sci. 18, 2624–2628.
Song, Y., Li, J., Shin, H.D., Du, G., Liu, L., Chen, J., 2015. One-step biosynthesis of
alpha-ketoisocaproate from L-leucine by an Escherichia coli whole-cell
biocatalyst expressing an L-amino acid deaminase from Proteus vulgaris. Sci.
Rep. 5, 12614.
Takahashi, E., Furui, M., Seko, H., Shibatani, T., 1997. D-methionine preparation from
racemic methionines by Proteus vulgaris IAM12003 with asymmetric degrading
activity. Appl. Microbiol. Biotechnol. 47, 173–179.
Takahashi, E., Ito, K., Yoshimoto, T., 1999. Cloning of L-amino acid deaminase gene
from Proteus vulgaris. Biosci. Biotechnol. Biochem. 63, 2244–2247.
Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5:
molecular evolutionary genetics analysis using maximum likelihood,
evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28,
2731–2739.
Todone, F., Vanoni, M.A., Mozzarelli, A., Bolognesi, M., Coda, A., Curti, B., Mattevi, A.,
1997. Active site plasticity in D-amino acid oxidase: a crystallographic analysis.
Biochemistry 36, 5853–5860.
Tong, H., Chen, W., Shi, W., Qi, F., Dong, X., 2008. SO-LAAO, a novel L-amino acid oxidase
that enables Streptococcus oligofermentans to outcompete Streptococcus mutans
by generating H2O2 from peptone. J. Bacteriol. 190, 4716–4721.
Vallon, O., Bulte, L., Kuras, R., Olive, J., Wollman, F.A., 1993. Extensive accumulation
of an extracellular L-amino-acid oxidase during gametogenesis of
Chlamydomonas reinhardtii. Eur. J. Biochem. 215, 351–360.
Yu, E.K., DeVoe, I.W., 1981. L-cysteine oxidase activity in the membrane of Neisseria
meningitidis. J. Bacteriol. 145, 280–287.
Yu, Z., Qiao, H., 2012. Advances in non-snake venom L-amino acid oxidase. Appl.
Biochem. Biotechnol. 167, 1–13.
Yu, Z., Wang, J., Zhou, N., Zhao, C., Qiu, J., 2013. A highly sensitive method for
quantitative determination of L-amino acid oxidase activity based on the
visualization of ferric-xylenol orange formation. PLoS One 8, e82483.
Yu, Z., Zhou, N., Qiao, H., Qiu, J., 2014. Identification, cloning, and expression of Lamino acid oxidase from marine Pseudoalteromonas sp. B3. Sci. World J. 2014,
979858.
Zhang, H., Yang, Q., Sun, M., Teng, M., Niu, L., 2004a. Hydrogen peroxide produced
by two amino acid oxidases mediates antibacterial actions. J. Microbiol. 42,
336–339.
Zhang, H., Teng, M., Niu, L., Wang, Y., Liu, Q., Huang, Q., Hao, Q., Dong, Y., Liu, P.,
2004b. Purification, partial characterization, crystallization and structural
determination of AHP-LAAO, a novel L-amino-acid oxidase with cell
apoptosis-inducing activity from Agkistrodon halys pallas venom. Acta
Crystallogr. D Biol. Crystallogr. 60, 974–977.
Zhu, Y., Wu, B., Zhang, X., Fan, X., Niu, L., Li, X., Wang, J., Teng, M., 2015. Structural
and biochemical studies reveal UbiG/Coq3 as a class of novel membranebinding proteins. Biochem. J. 470, 105–114.