Structure and catalytic mechanistic insight into Enterobacter aerogenes acetolactate decarboxylase

Accepted Manuscript Title: Structure and catalytic mechanistic insight into Enterobacter aerogenes acetolactate decarboxylase Authors: Fangling Ji, Yanbin Feng, Mingyang Li, Feida Long, Yongliang Yang, Tianqi Wang, Jingyun Wang, Yongming Bao, Song Xue PII: DOI: Reference:

S0141-0229(18)30635-5 https://doi.org/10.1016/j.enzmictec.2019.03.005 EMT 9322

To appear in:

Enzyme and Microbial Technology

Received date: Revised date: Accepted date:

28 October 2018 20 February 2019 4 March 2019

Please cite this article as: Ji F, Feng Y, Li M, Long F, Yang Y, Wang T, Wang J, Bao Y, Xue S, Structure and catalytic mechanistic insight into Enterobacter aerogenes acetolactate decarboxylase, Enzyme and Microbial Technology (2019), https://doi.org/10.1016/j.enzmictec.2019.03.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Structure and catalytic mechanistic insight into Enterobacter aerogenes acetolactate decarboxylase Fangling Ji1,a,*, Yanbin Feng2,a, Mingyang Li1, Feida Long1, Yongliang Yang1, Tianqi Wang1, Jingyun Wang1, Yongming Bao1,3 and Song Xue2,* 1

School of Life Science and Biotechnology, Dalian University of Technology, Dalian,

2

IP T

Liaoning, 116024, P.R. China; Marine Bioengineering Group, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian, Liaoning, 116023, P. R. China;

School of Food and Environment Science and Engineering, Dalian University of

SC R

3

Technology, Panjin, Liaoning, 12422, P. R. China These two authors contributing equally to the paper.

U

a

N

* Correspondence: [email protected]; [email protected]; Tel.: +86-411-8470-6316



A

CC

EP



M



The three-dimensional structure of the acetolactate decarboxylase from Enterobacter aerogenes is first solved and reported. The overall structure of the acetolactate decarboxylase from Enterobacter aerogenes is highly similar to that of the ALDCs reported from other bacterial strains, except Arg150. The conserved Arg150 exhibited a unique flexible conformation oriented away, which distinguished this ALDC from the ALDCs from other bacterial strains. In the complexes of the acetolactate decarboxylase from Enterobacter aerogenes with two enantiomers, Arg150 maintains the conformation with weak contacts with the substrates.

TE D



A

Highlights

Abstract α-Acetolactate decarboxylase (ALDC) catalyses α-acetolactate into acetoin (3-hydroxy-2butanone, AC) and is considered to be the rate-limiting enzyme in the synthesis of 2,3butanediol. In this work, the enzymatic activity of ALDC from Enterobacter aerogenes

IP T

ALDC (E.a.-ALDC) was fully characterized with enzyme kinetics, indicating a Km of 14.83 ± 0.87 mM and a kcat of 0.81 ± 0.09 s-1. However, compared with the activities of ALDCs

SC R

reported from other bacteria, the activity of E.a.-ALDC was determined to present a relatively lower value of 849.08 ± 35.21 U/mg. The enzyme showed maximum activity at pH 5.5. In addition, the activity of E.a.-ALDC was promoted by Mg2+. The crystal structure of E.a.-

U

ALDC firstly solved by X-ray crystallography at resolution of 2.4 Å revealed a chelated zinc

N

ion with conserved His199, His201, His212, Glu70 and Glu259. In the active center, the

A

conservative Arg150 was particularly proven to deviate from the zinc ion of the active centre,

M

by adopting a flexible conformational change, resulting in a weak interaction network of the enzyme and the substrate. Further in silico docking of E.a.-ALDC with two enantiomers, (R)-

TE D

acetolactate and (S)-acetolactate, unaltered the interaction network of E.a.-ALDC from the apo structure, which confirmed the weakened role of Arg150 in the catalytic properties of

EP

E.a.-ALDC. Our results reveal a unique structure-function relationship of acetolactate

CC

decarboxylase and provide a fundamental basis for the enzymatic synthesis of acetoin.

A

Keywords: Enterobacter aerogenes; α-Acetolactate decarboxylase; X-ray crystal structure

1. Introduction Acetoin (3-hydroxy-2-butanone, AC) is widely used as a flavour and fragrance in food chemistry and as an important intermediate in chemical synthesis and the production of

IP T

multifunctional materials [1-4]. As a result of its extensive application and the substantial potential for large-scale industrial demand, it is listed as a promising biological platform-based

SC R

chemical. Therefore, the US Department of Energy [5] has given priority to its development and utilization. In bacterial metabolism, acetoin is formed from pyruvate through two enzymecatalysed steps: α-acetolactate synthase (ALS) and α-acetolactate decarboxylase (ALDC). To

U

generate one molecular α-acetolactate, the upper stream enzyme (ALS) catalyses two

N

molecules of pyruvate. Thereafter, ALDC (EC 4.1.1.5) catalyses the decarboxylation of α-

A

acetolactate to form acetoin (3-hydroxy-2-butanone) and carbon dioxide [6]. Furthermore, α-

M

acetolactate (AL) is spontaneously and slowly decarboxylated to diacetyl, which is converted into acetoin by diacetyl reductase (DAR) in aerobic conditions. Finally, acetoin is converted

TE D

into 2,3-BD by butanediol dehydrogenase (BDH), also known as acetoin reductase (AR) (Fig. 1). ALDC has been applied in the traditional beer brewing process since it could significantly

EP

increase the production rate and decrease the adverse flavour caused by diacetyl [7]. In addition to the catabolic degradation of α-acetolactate to acetoin in the 2,3-BD pathway, ALDC is also

CC

involved in the biosynthesis of BCAAs [8-10]. However, in many cases, the decarboxylation of α-acetolactate by ALDC is assumed to be a rate-limiting step in 2,3-BD biosynthesis [11,

A

12].

Figure 1 The gene encoding ALDC was first isolated from Enterobacter aerogenes [13], followed by the corresponding gene from Klebsiella [14]. E. aerogenes serves as an effective production

method for acetoin and 2,3-butanediol using broad ranges of carbon sources [15]. To date, few ALDCs

from Bacillus

subtilis

(B.s.-ALDC)

[16],

Bacillus

brevis

(B.b.-ALDC)

[17] Brevibacillus brevis [7] and Lactococcus lactis subsp. lactis [9] have been characterized structurally. The ALDCs adopt a two-domain α/β tertiary structure and a dimeric assembly, in which a seven-stranded mixed β-sheet extends into the other equivalent β-sheet and forms

IP T

fourteen-stranded β-sheets.

SC R

To obtain further catalytic mechanisms, the structures of a series of transition state analogues in complex with ALDC have been calculated, suggesting a mechanism that ALDC catalyses

both the decarboxylation of the favoured (S)-AL and the isomerization via carboxyl migration

U

and subsequent decarboxylation of the less-favoured (R)-substrate [18]. By modelling the

N

structure of B.b.-ALDC, two groups [19] elicited the catalytic mechanism though the hybrid

A

quantum mechanical/molecular mechanical simulations. In the models of Zhuang et al., two

M

initial models are proposed to simulate the direct decarboxylation of (S)-AL and the rearrangement of the non-natural substrate (R)-AL. In their hypothesis, the removal of the (S)-

TE D

AL carboxyl leads to the formation of an enediol intermediate and carbon dioxide, followed by the protonation reaction on the opposite side of the enediol intermediate to generate (R)-AC.

EP

Considering the non-natural substrate (R)-AL, the authors proposed that Glu253 induces the rearrangement reaction of (R)-AL to (S)-AL. Additionally, using the two enantiomers of

CC

acetolactate as substrates, Ji et al. investigated the substrate preference of ALDC from Bacillus subtilis by means of molecular docking and dynamic simulation in silico [16]. Overall, despite

A

sharing a similar active centre, the ALDC enzyme from different sources always presents noncommon substrate binding modes. Here, to investigate the structure-function relationship of E. a.-ALDC, we cloned, expressed and characterized the ALDC from E. aerogenes. We present the first crystal structure of E.a.-

ALDC resolved to 2.4 Å and compared it with the structures of ALDCs from Bacillus brevis and Bacillus subtilis. We also investigated the complexes of E.a.-ALDC with its enantiomer substrates. 2. Materials and Methods

IP T

2.1 Crystallization The initial crystallization trials of E.a.-ALDC were performed using a Mosquito

SC R

crystallization robot (TTP LabTech) to set up sitting drops composed of 0.1 μL protein solution mixed with an equal volume of reservoir solution equilibrated against 40 μL of the reservoir

U

solution. Small crystals were observed after 3 weeks in the plates at 4 °C. The final crystal for

N

the diffraction collection is picked from a crystal grown in 20% (v/v) 1,4-butanediol, 100 mM

A

sodium acetate/acetic acid, pH 4.5 at a protein concentration of 30 mg/mL at 4 °C.

M

2.2 Collection of crystal data and determination of the structure

TE D

Crystals of E.a.-ALDC grown in the above conditions were cryo-protected in the mother liquid plus 15% glycerol and flash-cooled on a CryoLoop in a liquid-nitrogen stream at 100 K. X-ray diffraction data were collected at 100 K on beamline BL18U1 of the Shanghai

EP

Synchrotron Radiation Facility (SSRF). All diffraction data were processed, integrated, and

CC

scaled using the HKL-3000 software package (Schrodinger, LLC. "The PyMOL Molecular Graphics System, Version 1.8." 2015) and converted to the MTZ format using the CCP4

A

package [20]. Initial indexing of the diffraction patterns indicated that E.a.-ALDC crystallized in the P41212 space group with the unit cell dimensions as follows: a=83.541 Å, b=83.541 Å and c=139.173 Å (α = β = 90° and γ = 120°). Two molecules are present per asymmetric unit with an estimated solvent content of ~33% (Vm=1.83 Å3/Da) based on the Matthews Probability Calculator [21]. The structure was solved using molecular replacement (MR) with

the coordinates 5XNE as the search model in PHASER [22]. After generation of the initial model, the model was improved manually as described by Coot [23], followed by automated refinement in PHENIX [24]. The final E.a.-ALDC structure was refined to 2.42 Å, with a final R work value of 0.173 and a free R value of 0.218. The stereochemical quality of the final structure was assessed with MolProbity [25]. A total of 96% and 4% of the residues are located

IP T

in the favoured and allowed regions of the Ramachandran plot, respectively, while no disallowed region residues were found. The data collection and refinement statistics are

SC R

generated in Table 1, and the coordinate files have been deposited in the Protein Data Bank with accession code 6J92. All structural figures were prepared using PyMOL [26].

U

Table 1

N

2.3 Determination of the enzymatic activity

A

Determinations of the properties of the E.a.-ALDC enzymatic activity were based on

M

procedures previously reported [17]. The (±)-AL substrate was freshly prepared by diluting

TE D

(±)-ethyl 2-acetoxy-2-methylacetoacetate with 0.5 M NaOH and then stirring for 20 min at room temperature. The pH of the mixture was adjusted to 6.0 and diluted to the desired concentration with 10 mM Tris-HCl buffer with pH 6.0. An (R)-AC stock solution was

EP

produced by the complete reaction of (±)-AL with E.a.-ALDC, and the solution was diluted from 0.5 to 15.0 mM to generate the standard curve of (R)-AC, (R)-AC concentrations versus

CC

CD signals at 278 nm. The standard curve of (R)-AC was determined by CD with a scanning range of 200-340 nm, a data pitch of 1 nm, a response of 0.5 sec, a bandwidth of 2 nm and a

A

scanning speed of 200 nm/min. An equal volume of substrate and E.a.-ALDC were mixed to start the reaction using a cuvette with an optical path length of 1 mm. The subsequent decarboxylation of (±)-AL into (R)-AC was monitored, and the CD signal at 278 nm was recorded. The amount of product was calculated by the (R)-AC standard curve, resulting in the E.a.-ALDC activity. Diluted (R)-AC stock solutions at concentrations of 0.53, 1.07 5.07, 10.14,

14.94 and 0.53 mM were used to generate the standard curve. One unit of E.a.-ALDC activity was defined as the amount of E.a.-ALDC required for the formation of 1 μmol AC per minute at 30 °C. The Km and kcat values of E.a.-ALDC were calculated by measuring the activity of the

IP T

enzyme at different substrate concentrations and fitting to a double reciprocal curve by Lineweaver-Burk. Various concentrations (53.4, 26.7, 17.8, 13.3, 10.7, 8.9, 7.6, 6.7, 5.9 and

SC R

5.3 mM) of the substrates were used, and all activity measurement experiments were carried out at 30 °C and the optimum pH as described.

U

2.4 Determination of the optimum pH and ion inhibition

N

To determine the optimal pH for E. a.-ALDC, 10 mM NaAc-NaOH and Tris-HCl buffer

A

ranging from pH 4.5 to 8.5 were pre-incubated with the enzyme at 4 °C for an hour; thereafter,

M

the activity of the enzyme was determined using the CD method at 30 °C. The PDB2PQR server [27] was used to prepare the structure for electrostatic calculations, and the

TE D

electrosurface potentials of E.a.-ALDC were calculated by APBS [28] using an ionic strength of 0.15 M. The figures were generated using PYMOL 1.5.0.3 [10].

EP

Metal ions are considered to play effective roles in the activities of decarboxylases; thus, the effect of different metal ions was evaluated for the relative activity E.a.-ALDC. The

CC

enzyme was first treated with 0.5 mg/mL EDTA, followed by the use of inductive coupled plasma emission spectrometry (ICP) to measure the residual ions. Common bivalence metal

A

ions, including Cu2+, Fe2+, Mn2+, Ba2+, Mg2+, Zn2+ and Ca2+, were supplied to the reaction buffer with a final concentration of 1 mM and incubated at 4 °C for an hour before determination of the relative activity. 2.5 Molecular docking and dynamics simulation

The docking procedure was carried out following the protocols used in our previous study [18].

Similarly,

automated

ligand

docking

by

AutoDock

4.20

(http://www.autodock.scripps.edu/) recognized the binding sites of (R)/(L)-AL with E.a.ALDC using the empirical free energy function and the Lamarckian genetic algorithm [29, 30]. Input structures of E.a.-ALDC and the substrates for docking simulation were saved in AutoDock 4.20’s PDBQT file format. The ligand input files were prepared according to the standard protocols in the AutoDock manual. According to the two transition state analogues in

IP T

the PDB file (4BT4 and 4BT5), the coordinates of the ligand atoms were prepared on the basis

of the crystal structure of the E.a.-ALDC complex. The preparation of the protein target files

SC R

also followed the standard protocols in the AutoDock 4.20 manual. A default grid size of 60 x 60 x 60 points with a spacing of 0.375 Å was applied, corresponding to a cube with an edge length of 22.5 Å. The center of the grid box was defined as the center of the cocrystallized ligand. All the above procedures were carried out using Auto-Dock Tools [31, 32].

U

Molecular dynamics simulation was performed using GROMACS 4.6.7 [33] with the

N

AMBER ff99SB force field [34], and the parameters of ligands were calculated by Chimera

A

[35] and ACPYPE [36]. The complex was solvated in a cubic box of TIP3P [37] water, which was neutralized by sodium ions at a distance of 1 nm from the edge. After two energy

M

minimization phases, the temperature of the system was gradually heated to 300 K at 100 ps for a 5 ns NVT equilibration and a 5 ns NPT equilibration. Finally, a 10 ns molecular docking

TE D

simulation was performed at 300 K and 1 atm using the LINCS algorithm [38] to restrain hydrogen positions at their equilibrium distances, which allowed the integration time step of 2 fs to be used. The binding free energy was calculated by the Molecular Mechanics-Poisson-

EP

Boltzmann Surface Area (MM-PBSA) method with g_mmpbsa (a plugin of Gromacs).

CC

3. Results and Discussion 3.1. Enzymatic properties of E.a.-ALDC

A

Racemic AL presented no CD signal and was used as the substrate. In the enzymatic

reaction, spectral signals of (R)-AL and (R)-AC were immediately detected by the addition of the enzyme. First, the enzyme turned over the (S)-AL substrate, thus leaving an excess of (R)AL, and the signal for (R)-AL emerged (Fig. 2a). The generation of (R)-AL was observed at 278 nm, with an initial rapid rate that then decreased (Fig. 2b, left). The AL mix signal at 315

nm began at zero and was then followed by the formation of a negative peak due to an excess of (R)-AL as the enzyme turned over (S)-AL. Next, the peak began to disappear at a slow rate as the enzyme turned over the remaining (R)-AL (Fig. 2b, right). Due to the signal intensity for the product being stronger than that for the substrate, the reaction rates were monitored at 278 nm. The specific activity of E.a.-ALDC towards (±)-AL was calculated as 849.08 ± 35.21

IP T

U/mg.

SC R

Figure 2

Additionally, measuring reactions with (±)-AL at different concentrations using the Lineweaver-Burk double reciprocal curve, the Michaelis-Menten constant (Km) and turnover

U

number (kcat) values of E.a.-ALDC were calculated as 14.83 ± 0.87 mM and 0.81 ± 0.09 s-1,

N

respectively. Thus, the kcat/Km of E.a.-ALDC was 54.7 M-1s-1. In previous work studying B.s.-

A

ALDC, the Michaelis-Menten constant (Km) and turnover number (kcat) values of B.s.-ALDC

M

yielded Km = 0.25 ± 0.08 mM and kcat = 5.99 ± 0.95 s-1, respectively [39], and the kcat/Km value for B.s.-ALDC, which was 2.49 x 104 M-1s-1, represented a 438-fold difference, suggesting that

TE D

E.a.-ALDC was not an efficient enzyme compared to B.s.-ALDC. 3.2. Overall structure of E.a.-ALDC

EP

Similar to the previously reported crystal structures of B.s.-ALDC and B.b.-ALDC, E.a.-ALDC

CC

is a metalloprotein containing two domains of α/β tertiary structure, in which each domain contains a seven-stranded mixed β-sheet. These two β-sheets formed a nearly parallel

A

intramolecular surface (back-to-back) (Fig. 3A). On this surface, the shortest distance of all Cα atoms of the backbone chain was 6.2 Å between Ile72 and His201. Two ALDC molecules were identified in the asymmetric unit. Each of the monomers (chain A and chain B) was highly superimposed with a root mean square deviation (r.m.s.d) value of 0.26 Å, and the N-terminal domain consisted of a seven-stranded mixed β-sheet, whereas the C-terminal domain consisted

of a five-stranded β-sheet. The seven-stranded mixed β-sheet on the N-terminal domain of E.a.ALDC extended into the other equivalent β-sheet of the two-fold symmetry-related molecule, generating a fourteen-stranded β-sheet that spanned the physiologically relevant dimeric formation (Fig. 3a). Figure 3

IP T

The structure of the catalytic domain is essentially the same as the structures of B.b.-ALDC [40] and B.s.-ALDC [19], as previously described. The active centre of E.a.-ALDC is shown

SC R

in Fig. 3b with a refined electron density map including both chain A and chain B. Similarly,

a Zn2+ ion was coordinated by three highly conserved histidines (199, 201 and 202) in chain A

U

and chain B. A conserved glutamate (259) from the C-terminal tail in chain A also interacts

N

with the zinc ion, but in chain B twenty-four amino acid residues are missing at the C-terminal

A

(Fig. 3b). As a result, no such interactions could be observed. Likewise, the catalytic domain of B.s.-ALDC and the molecular docking results of B.s.-ALDC with substrates exhibited the

M

same conservative structure [18, 26]. A Zn2+ ion was coordinated by His191, His193 and

TE D

His201. Glu251, Glu62 and Arg142 interacted with the substrate through hydrogen bonds. Three histidine-coordination of a metal ion was found widely in many enzyme structures, such as manganese in oxalate decarboxylase [7], copper in quercetin 2,3-dioxygenase [41], and zinc

EP

in carbonic anhydrase [42].

CC

3.3. Structural comparisons of ALDCs

A

To demonstrate the sequence identities of B.s.-ALDC, B.b.-ALDC, and E.a.-ALDC,

multiple sequence alignments were conducted (Fig. S2). The sequence identity between E.a.ALDC and B.s.-ALDC was 38%, between E.a.-ALDC and B.b.-ALDC was 30%, and between B.s.-ALDC and B.b.-ALDC was 29%. The evolutionary tree of these three ALDCs is shown

in Fig. 4a. In particular, residues in the active site of these ALDCs were highly conserved, including histidines (199, 201 and 212), Glu70, Glu259, Arg150 and Thr63. Superposition of the overall structure of E.a.-ALDC chain A and chain B onto that of B.b.ALDC (PDB ID: 4BT2) yielded an average pairwise r.m.s.d. value of 0.641 Å and 0.689 Å,

IP T

respectively, for the backbone atoms (Fig. 4a, upper left, only chain A presented). The low r.m.s.d. value indicates that the overall structure of the E.a.-ALDC is similar to that of B.b.-

SC R

ALDC. We also superimposed the highly conserved amino acid residues in the active sites

(Fig. 4a, lower left). In addition to Arg150, most coordinates of the side chains of the remaining residues in chain A were superimposed effectively. Similar comparisons were carried out

U

between the structures of E.a.-ALDC and B.s.- ALDC (PDB ID: 5XNE). The superposition

N

yielded an average pairwise r.m.s.d. value of 0.661 Å (chain A) and 0.696 Å (chain B) for the

A

backbone atoms of E.a.-ALDC and B.s.-ALDC (Fig. 4a, upper right, only chain A presented).

M

Superimposing the highly conserved amino acid residues in the active site of E.a.-ALDC chain A and B.s.-ALDC resulted in comparable similarity with those of E.a.-ALDC and B.b.-ALDC

TE D

(Fig. 4a, lower right).

Figure 4

EP

The highly conserved Glu259 in E.a.-ALDC chain A exhibited a structure conformation similar to that of the Glu251 structure in B.s.-ALDC and B.b.-ALDC. This highly conserved

CC

glutamate played significant roles during the decarboxylation reaction. The distance between the zinc ion and the side chain carbonyl group of the side chain of Glu259 was 3.8 Å, while in

A

the structures of B.s.-ALDC and B.b.-ALDC, the distances were 3.5 Å and 4.0 Å, respectively. In the crystal structure of B.b.-ALDC in complex with transient intermediate analogues, a hydrogen bond was observed between the glutamate and the analogues [19]. In the docking models of B.b.-ALDC with (S)-AL, the carboxyl group of Glu254 formed a hydrogen bond with Arg145 and with the water molecule in the active site [8]. In the rearrangement process,

the hydrogen bond between Glu259 and the carboxyl of (R)-AL was crucial when the carboxyl rearranged to adjacent carbonyl carbon to form (S)-AL [43]. The orientation of the side chain of Arg150 was quite different in the structure of E.a.ALDC. Notably, the conformation of Arg150 in the chain A and chain B of E.a.-ALDC are

IP T

different when overlapped, bearing a 15-degree angle (Fig. 4b). In E.a.-ALDC chain A, when compared to arginines in B.s.-ALDC and B.b.-ALDC, the side chain of Arg150 tilted at a 50-

SC R

degree angle, orienting away from the centre of the active site. This specific flexible conformation of Arg150 might explain the low enzymatic activity, although all other residues in the active site of E.a.-ALDC were highly conservative. The distance between the zinc ion

U

and the NH2 group of the side chain of Arg150 was 8.0 Å, but in B.s.-ALDC and B.b.-ALDC,

N

this distance was extremely small, with distances of 4.0 Å and 3.9 Å, respectively. In fact, both

A

of the NH1 groups of the side chain of the arginine residue (142 in B.s.-ALDC and 145 in B.b.-

M

ALDC) formed hydrogen bonds with the oxygen atom of Glu251 in B.s.-ALDC and Glu254 in B.b.-ALDC. In the structure of E.a.-ALDC chain A, the NH1 group of Arg150 formed a

TE D

hydrogen bond with the oxygen atom of the backbone of Val258 and was also close to the carboxyl group of Glu259 with a distance of 4.5 Å. The NH1 and NH2 group of Arg150 were

EP

in close proximity to the carboxyl group of Asn260, with distances of 3.4 Å and 3.6 Å (Fig. 4c). Since chain B has lost the residues on its C-terminal, we cannot measure the distances in

CC

this region. The mobility of arginine side-chain was also observed in weak the binding of the inhibitors by partially blocking the S1’ cavity in the crystal structure of human matrix

A

metalloproteinase 9 [44]. In addition to the highly conserved Thr58 and Glu65 in B.b.-ALDC, Arg145 was also

highly conserved in the vicinity of the metal. In the transition state analogue structures of B.b.ALDC in complex with (2S,2S)-dihydroxy-2-methylbutanoic acids (PDB ID: 4BT4) and (2S,2R)-dihydroxy-2-methylbutanoic acids, Arg145 contacted the analogues through hydrogen

bonds, and the inhibitors adopted essentially identical conformations. The binding mode of the (R,R)-dihydroxy-2-methylbutanoic acids (PDB ID: 4BT5) isomer supported the proposed decarboxylation mechanism for the natural (S)-enantiomer, also involving transient binding of the departing CO2 to the zinc. The role of Zn(II), as a Lewis acid, is best understood in terms of the reverse reaction of acetoin carboxylation. Arg145 is well positioned to assist with

IP T

delivery of the proton, which creates the new chiral centre [19]. In the docking model of B.s.ALDC with (R)/(S)-AL, a hydrogen bond between the OH of (R)-AL and the side chain of

SC R

Arg142 was observed, while no hydrogen bond was identified in B.s.-ALDC with (S)-AL [17]. 3.4. The optimum pH and ion inhibition-activation of E.a.-ALDC activity

U

To obtain the optimum pH of E. a.-ALDC, the impact of pH on the enzymatic activity was

N

investigated and is presented in Fig. 5a. The enzyme maintained relatively high activity from

A

pH 4.5 to 8.0. The maximum activity was observed at pH 5.5 with 958 ± 38.32 U/mg using

M

(±)-AL as the substrate. In fact, the enzymatic decrease in the specific activity was exhibited

TE D

at a pH of 8.5 compared to that at pH values ranging from 4.5 to 8.0. As indicated by the results, E.a.-ALDC activity was more effectively inhibited at an alkaline pH than at an acidic pH, from 904 ± 34.35 to 760 ± 31.92 U/mg. To gain more insight into E.a.-ALDC, we also investigated

EP

its electrostatic surface properties (Fig. S3). The E.a.-ALDC revealed two different charged

CC

faces. One face was highly neutral with two negatively charged areas, while the second face was highly negatively charged. In a previous study on the ALDC from Bacillus subtilis,

A

maximum activity was observed at pH 5.0 [17]. Rasmussen et al. investigated the ALDC from Lactobacillus casei and showed relatively high activity at pH 4.5 to 5.0 [42]. The ALDC from Lactococcus lactis was found to have an optimal pH of 6.0 [45]. Site-directed mutagenesis was proposed by the ALDC from Staphylococcus aureus, and the modification dramatically improved the stability of S.a.-ALDC at low pH conditions with a value of 4.0 [46].

Since the function of many decarboxylases requires a divalent metal ion as a cofactor, the effects of various metal ions on E.a.-ALDC were investigated. Furthermore, in the crystal structure of E.a.-ALDC, the zinc ion was observed and directly interacted with the substrate. Thus, we investigated the effects of different ion types on E.a.-ALDC activity. To chelate the metal ions of E.a.-ALDC, EDTA was added and then measured using inductively coupled

IP T

plasma. As shown in Fig. 5b, Ba2+, Mg2+ and Ca2+ all improved E.a.-ALDC activity, while

Fe2+ and Cu2+ significantly inhibited the activity of E.a.-ALDC. Mn2+ decreased the activity of

SC R

E.a.-ALDC by almost 10%. However, we observed that Zn2+ inhibited the activity of E.a.ALDC by ~20%. In fact, before measuring the activity of E.a.-ALDC by adding different metal

U

ions, the enzyme was pre-incubated with EDTA to chelate the original Zn2+ ion in the active centre. We observed similar results when we expressed E.a.-ALDC in E. coli with and without

N

Zn2+, the former producing active E.a.-ALDC but the latter E.a.-ALDC exhibiting very low

M

A

activity, even after in vitro incubation with the solution of Zn2+. In contrast, the Zn2+ effect on B.s.-ALDC is different [17]. We believe that this difference is related to the stabilization role

TE D

of Arg150 on the substrate since the substrate interacts with Zn2+, as observed in the structure of B.b.-ALDC [19]. The ALDC from Lactococcus lactis DX was activated by Fe2+, Zn2+, Mg2+, Ba2+ and Ca2+ but was significantly inhibited by the addition of Cu2+ [47]. However, some of

EP

the metal ions do not function as catalytic centres, while others do. For instance, the Zn2+ ion

CC

in the biotin-dependent decarboxylases helps to properly orient the substrate, while the Mg2+ ion in the ThDP-dependent decarboxylases assists in binding the thiamine cofactor [48, 49].

A

For NAD(P)+-dependent decarboxylases, such as malic enzymes, a divalent metal ion (Mn2+ or Mg2+) plays an important role in both catalysis and structural stability [50]. Figure 5 3.5 Structures of E.a.-ALDC-substrate complexes

As we discovered in the apo structure of E.a.-ALDC, the Arg150 side-chain in the active site undergoes conformational changes, and our failure to obtain the co-crystallization of the complex of E.a.-ALDC with the substrates. The simulated structure of E.a.-ALDC chain A and chain B in complex with (S)-AL and (R)-AL are presented. As shown in Fig. 6, in the complex structure of (S)-AL and E.a.-ALDC chain A, the Zn2+ ion formed three coordinate bonds with

IP T

the histidine residues (His199, His201 and His212) near distances of 2.0, 2.1 and 2.1 Å. The

Zn2+ ion also formed two coordinate bonds with oxygen atoms on the carboxyl and hydroxyl

SC R

moieties of (S)-AL, and the distances were 2.2 and 3.5 Å, respectively. The above bonding networks indicate a strong coordination interaction of the Zn2+ ion with the three histidine

U

residues and the substrate. Similar interaction networks are also presented in the models of

N

B.s.-ALDC with (S)-AL obtained by Ji et al. [17] and B.b.-ALDC with (S)-AL obtained by

A

Zhuang and Zhao [8, 43] via the molecular docking method. Likewise, in the docking result of

M

E.a.-ALDC with (S)-AL, the side chain of Glu70 (Glu65 in B.s.-ALDC and Glu62 in B.b.ALDC) exhibited polar interactions with (S)-AL. In the complex of E.a.-ALDC chain B, the

TE D

interaction network of (S)-AL with the active centre is very similar as the networks in B.s.ALDC and B.b.-ALDC but with longer distances between Arg150 and (S)-AL. The NH2 group of Arg150 build interactions oxygen atoms on the carboxyl and hydroxyl moieties of (S)-AL

EP

with distances of 4.0 Å and 3.2 Å. This conformation could contribute to the enzyme activity

CC

of E.a.-ALDC.

In the crystal structure of B.b.-ALDC with the substrate analogues ((2S,2S)-dihydroxy¬2-

A

methylbutanoic acids (PDB ID: 4BT4) and (2S,2R)-dihydroxy-2-methylbutanoic acids, Arg145 forms four hydrogen bonds with the carboxyl groups of Glu62 and Glu251 as well as the hydroxyl groups of the substrate analogues. These hydrogen bonds functionally stabilized the Zn2+ ion and the substrate analogues. Similar hydrogen bond networks were observed in the model of B.b.-ALDC with (S)-AL obtained previously in the model of B.s.-ALDC [17]. As

we hypothesized, Arg150 did not change its conformation in the presence of (S)-AL and maintained its interactions with Val258, Glu259 and Asn260, as we observed in the apo structure of E.a.-ALDC. By adopting this tilted conformation, Arg150 in E.a.-ALDC chain A was not able to stabilize the Zn2+ ion and the substrate, resulting in low E.a.-ALDC enzymatic activity.

IP T

Figure 6

Additionally, the interactions of E.a.-ALDC chain A and chain B with (R)-AL were

SC R

calculated by the same molecular docking methodology. In the complex of E.a.-ALDC chain

A with (R)-AL, E.a.-ALDC showed a different binding mode of (S)-AL (Fig. 6). (R)-AL

U

adopted a perpendicular conformation to that of (S)-AL in E.a.-ALDC, with its polar groups

N

facing Glu259. The hydroxyl and carboxyl groups of (R)-AL were closer to Glu259 and formed

A

hydrogen bonds with the OE1 and OE2 groups of Glu259, with distances of 2.6 and 2.9 Å.

M

Thus, the Zn2+ ion only formed a tetracoordinated structure with His199, His201, His212 and the substrate, with binding distances of 2.0, 2.1, 2.1 and 2.2 Å, respectively. These hydrogen

TE D

bonds stabilized the substrate in the active site of the enzyme. In this model, Arg150 exhibited no conformational changes without interacting with (R)-AL. In chain B, in a different way, the NH2 group of Arg150 forms salt bridge with the hydroxyl group of (R)-AL, which is similar

EP

as the active form presented in B.s.-ALDC and B.b.-ALDC. So far, it is more possibly like the

CC

flexibility of Arg150 side-chain weaken the role of Arg150 in catalytic properties of E.a.ALDC. Here, in both complexes, we did not find that substrate binding altered the

A

conformation of Arg150 from its apo structure. However, further experimental cocrystallization studies are needed in the future to prove our hypothesis instead of just in silico studies. Considering that conformational changes can take place in the real-time environment, we further validated the docked complexes using molecular dynamic (MD) simulations to

understand the conformational changes. All complexes were subjected to a molecular dynamic simulation of up to 2 ns, and after the entire system was equilibrated, no significant RMSD variations were present during the remaining simulation, although a slight increase was exhibited (Fig. S4a). The results are very similar to those of the MD simulation carried out for B.s.-ALDC [18]. To specifically examine the conformational differences in Arg150, the frame

IP T

structures of E.a.-ALDC at the original state and the end state of the simulation were extracted

and compared (Fig. S4b). All structures were superimposed and yielded low RMSD values,

SC R

which were 0.654 Å for E.a.-ALDC with (S)-AL and 0.783 Å for E.a.-ALDC with (R)-AL, respectively. However, the side chain of Arg150 maintained its orientation away from the

U

active centre and was further away. In addition, we also calculated the energy contributions of

N

different enantiomers with E.a.-ALDC. In Table 2, it is shown that both substrates formed

A

stable complexes with E.a.-ALDC.

M

Table 2 Arg150 is very conservative in different sources of ALDC, especially in other sources of

TE D

ALDC structure information, such as B.b.-ALDC (PDB ID: 4BT7) [42] and B.s.-ALDC (PDB ID: 5XNE) [18]. Via site-directed mutagenesis, R142A and R142K mutants of B.s.-ALDC were expressed to manifest the role of arginine in the active centre. The R142A mutant showed

EP

very low activity, while R142K retained 40% of the activity of B.s.-ALDC. In the crystal structure, the corresponding Arg in B.s.-ALDC can form a stable hydrogen bond network with

CC

the amino acids of the attachment. Thus, mutations in Arg150 can directly affect the enzymatic

A

activity of the catalytic process. Additionally, for R142K, it might be true that the replacement of a specific residue would affect its neighbouring residues, which should not be ignored. Thus, atom-level structural details of the R142K mutant could better explain its activity change. In E.a.-ALDC, although Arg150 is highly conserved, Arg150 can only form hydrogen bonds with the carboxyl group of Glu259; thus, in E.a.-ALDC, Arg150 is a relatively active amino acid, as improved by the two conformations in chain A and B. Its B-factor (approximately 34.92 was

averaged from each atom in the pdb file) was similar to the B-factor of the overall molecule. However, the 2Fo - Fc map indicated (with maps) that in our structure (Fig. 3b), the threedimensional space of Arg150 deviated from the B-factor in other structures. For E.a.-ALDC, the conformation of Arg150 is what we observed; thus, it is possible that the unique flexible conformation of Arg150 affected the activity of E.a.-ALDC.

IP T

In addition to the participation of the conserved His199, 201 and 212 in direct coordination with the Zn2+ ion, the second coordination sphere, such as Arg150, Glu70 and Glu259, was

SC R

confirmed to be conserved. However, sequence alignment of E.a.-ALDC with B.s.-ALDC and

B.b.-ALDC revealed that 161-168 were not conserved (Fig. S2); in particular, the tertiary

U

structure of this region revealed superior deviation. It is quite possible that the region of 161-

N

168 caused the deviation of Arg150, which ultimately affected the activity of the enzyme.

M

A

4. Conclusions

In conclusion, the crystal structure of ALDC from Enterobacter aerogenes was first reported

TE D

and solved. The overall structure and the active centre of E.a.-ALDC preserved the conservative conformation, except that Arg150 exhibited a flexible tilted orientation when

EP

compared to the corresponding arginine residues of the ALDCs studied previously. The enzymatic parameters of E.a.-ALDC were characterized as well as the optimum pH and metal

CC

ion type. Subsequently, the molecular docking of (R)/(S)-AL into E.a.-ALDC revealed the binding profiles. The tilted angle of Arg150 weakened its direct interaction with the substrate,

A

which might explain the lower enzymatic activity of E.a.-ALDC compared to the frequently characterized ALDCs. To date, few high-resolution structures of ALDC have been reported. Our E.a.-ALDC structure will provide very important clues to the catalytic mechanism of ALDCs.

Acknowledgements: We would like to thank Mr. Peng Chu, who helped with the docking work. This research was funded by the National Science Foundation of China, grant number 21506025 and Dalian University of Technology Science Foundation, grant number of DUT8LK08.

IP T

Reference

[1] P.D. Adams, P.V. Afonine, G. Bunkóczi, V.B. Chen, I.W. Davis, N. Echols, J.J. Headd,

SC R

L.W. Hung, G.J. Kapral, R.W. Grosse‐Kunstleve, PHENIX: a comprehensive Python‐based

system for macromolecular structure solution, Acta Crystallogr D Biol Crystallogr 66(2) (2010)

U

213-221.

N

[2] R. Anand, P.C. Dorrestein, C. Kinsland, T.P. Begley, S.E. Ealick, Structure of oxalate

A

decarboxylase from Bacillus subtilis at 1.75 angstrom resolution, Biochemistry 41(24) (2002)

M

7659-7669.

[3] Z. Xiao, J.R. Lu, Strategies for enhancing fermentative production of acetoin: a review,

TE D

Biotechnology Advances 32(2) (2014) 492-503.

[4] Z. Xiao, J.R. Lu, Generation of Acetoin and Its Derivatives in Foods, Journal of Agricultural & Food Chemistry 62(28) (2014) 6487-97.

EP

[5] N.A. Baker, D. Sept, S. Joseph, M.J. Holst, J.A. Mccammon, Electrostatics of nanosystems:

CC

application to microtubules and the ribosome, Proc Natl Acad Sci U S A 98(18) (2001) 1003710041.

A

[6] K. Blomqvist, M.-L. Suihko, J. Knowles, M. Penttilä, Chromosomal Integration and Expression of Two Bacterial α-Acetolactate Decarboxylase Genes in Brewer's Yeast, Applied and Environmental Microbiology 57 (1991) 2796-2803.

[7] W. Buckel, Biotin-dependent decarboxylases as bacterial sodium pumps: Purification and reconstitution of glutaconyl-CoA decarboxylase from Acidaminococcus fermentans, 125 (1986) 547-558. [8] M.-H. Choi, S.-J. Kim, J.-W. Kim, Y.-C. Park, J.-H. Seo, Molecular cloning and expression of Enterobacter aerogenes α-acetolactate decarboxylase in pyruvate decarboxylase-deficient

IP T

Saccharomyces cerevisiae for efficient 2,3-butanediol production, Process Biochemistry 51(2) (2016) 170-176.

SC R

[9] I.W. Davis, A. Leaver-Fay, V.B. Chen, J.N. Block, G.J. Kapral, X. Wang, L.W. Murray,

W.B. Arendall, 3rd, J. Snoeyink, J.S. Richardson, D.C. Richardson, MolProbity: all-atom

U

contacts and structure validation for proteins and nucleic acids, Nucleic Acids Res 35(Web

N

Server issue) (2007) W375-83.

A

[10] W.L. Delano, The PyMOL Molecular Graphics System (2002), My Publications 30 (2002)

M

442-454.

[11] T.J. Dolinsky, J.E. Nielsen, J.A. Mccammon, N.A. Baker, PDB2PQR: an automated

TE D

pipeline for the setup of Poisson-Boltzmann electrostatics calculations, Nucleic Acids Research 32(Web Server issue) (2004) 665-7. [12] P. Emsley, B. Lohkamp, W.G. Scott, K. Cowtan, Features and development of Coot, Acta

EP

Crystallographica 66(4) (2010) 486–501.

CC

[13] Z. Fisher, J.P. Hernandez, C. Tu, D. Duda, C. Yoshioka, H. An, L. Govindasamy, D.N. Silverman, R.J.B. Mckenna, Structural and kinetic characterization of active-site histidine as a

A

proton shuttle in catalysis by human carbonic anhydrase II, 44(4) (2005) 1097. [14] F. Fusetti, K.H. Schroter, R.A. Steiner, P.I. van Noort, T. Pijning, H.J. Rozeboom, K.H. Kalk, M.R. Egmond, B.W. Dijkstra, Crystal structure of the copper-containing quercetin 2,3dioxygenase from Aspergillus japonicus, Structure 10(2) (2002) 259-68.

[15] N. Goupilfeuillerat, M. Cocaignbousquet, J.J. Godon, S.D. Ehrlich, P. Renault, Dual role of alpha-acetolactate decarboxylase in Lactococcus lactis subsp. lactis, Journal of Bacteriology 179(20) (1997) 6285-6293. [16] Y. Guo, D. Pan, H. Ding, Z. Wu, Y. Sun, X. Zeng, Purification and characterization of alpha-acetolactate decarboxylase (ALDC) from newly isolated Lactococcus lactis DX, J Sci

IP T

Food Agric 95(8) (2015) 1655-61.

[17] K. Ius, V.M. Serebrennikov, N.A. Zagustina, A.M. Bezborodov, Isolation and purification

Prikladnaia Biokhimiia I Mikrobiologiia 36(2) (2000) 131.

SC R

of acetolactate synthase and acetolactate decarboxylase from a Lactococcus lactis culture,

U

[18] F. Ji, M. Li, Y. Feng, S. Wu, T. Wang, Z. Pu, J. Wang, Y. Yang, S. Xue, Y. Bao, Structural

A

Microbiology & Biotechnology (2) (2018) 1-13.

N

and enzymatic characterization of acetolactate decarboxylase from Bacillus subtilis, Applied

M

[19] M.-Y. Jung, B.-S. Park, J. Lee, M.-K. Oh, Engineered Enterobacter aerogenes for efficient utilization of sugarcane molasses in 2,3-butanediol production, Bioresource technology 139

TE D

(2013) 21-27.

[20] H. Sone, T. Fujii, K. Kondo, F. Shimizu, J. Tanaka, T. Inoue, Nucleotide sequence and expression of the Enterobacter aerogenes alpha-acetolactate decarboxylase gene in brewer's

EP

yeast, Applied & Environmental Microbiology 54(1) (1988) 38.

CC

[21] C. Vogel, J. Pleiss, The modular structure of ThDP-dependent enzymes, Proteins 82(10) (2014) 2523-37.

A

[22] T.A. Werpy, J.E. Holladay, J.F. White, Top Value Added Chemicals From Biomass: I. Results of Screening for Potential Candidates from Sugars and Synthesis Gas, Synthetic Fuels (2004).

[23] G.N. Murshudov, P. Skubák, A.A. Lebedev, N.S. Pannu, R.A. Steiner, R.A. Nicholls, M.D. Winn, F. Long, A.A. Vagin, REFMAC5 for the refinement of macromolecular crystal structures, Acta Crystallographica 67(4) (2011) 355-367. [24] P.D. Adams, P.V. Afonine, G. Bunkoczi, V.B. Chen, I.W. Davis, N. Echols, J.J. Headd, L.W. Hung, G.J. Kapral, R.W. Grosse-Kunstleve, A.J. McCoy, N.W. Moriarty, R. Oeffner, R.J.

IP T

Read, D.C. Richardson, J.S. Richardson, T.C. Terwilliger, P.H. Zwart, PHENIX: a

comprehensive Python-based system for macromolecular structure solution, Acta Crystallogr

SC R

D Biol Crystallogr 66(Pt 2) (2010) 213-21.

[25] S. Qian, C. Wang, H. Wang, F. Yu, C. Zhang, H. Yu, Synthesis and characterization of surface-functionalized paramagnetic nanoparticles and their application to immobilization of

N

U

α-acetolactate decarboxylase, Process Biochemistry 50(9) (2015) 1388-1393.

[26] C. Zhao, H. Su, Y. Liu, Catalytic mechanism of acetolactate decarboxylase from

M

A

Brevibacillus brevis towards both enantiomers of α-acetolactate, Rsc Advances 6(84) (2016). [27] C.J. Zhu, S. Tao, L. Dong, J.L. Wu, C. Yong, L.F. Wang, G. Kai, H.J. Ying, P.K. Ouyang,

TE D

Production of liquid hydrocarbon fuels with acetoin and platform molecules derived from lignocellulose, Green Chemistry 18(7) (2016) 2165-2174. [28] C.B. Zhuang, Q.C.J.R.A. Zheng, QM/MM calculations and MD simulations of

EP

acetolactate decarboxylase to reveal substrate R/S-acetolactate binding mode and

CC

stereoselective catalytic mechansim, 6(94) (2016). [29] A.M. Rasmussen, R.M. Gibson, S.E. Godtfredsen, M.J.C.R.C. Ottesen, Purification of α-

A

acetolactate decarboxylase from Lactobacillus casei DSM 2547, 50(2) (1985) 73-82. [30] M.F. Sanner, Python: a programming language for software integration and development, J Mol Graph Model 17(1) (1999) 57-61.

[31] D. Chen, G. Menche, T.D. Power, L. Sower, J.W. Peterson, C.H. Schein, Accounting for ligand-bound metal ions in docking small molecules on adenylyl cyclase toxins, Proteins 67(3) (2007) 593-605. [32] X. Hu, W.H. Shelver, Docking studies of matrix metalloproteinase inhibitors: zinc parameter optimization to improve the binding free energy prediction, J Mol Graph Model 22(2)

IP T

(2003) 115-26.

[33] M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E.J.S. Lindahl,

SC R

GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers, s1–2(C) (2015) S2352711015000059.

U

[34] V. Hornak, R. Abel, A. Okur, B. Strockbine, A. Roitberg, C. Simmerling, Comparison of

N

multiple Amber force fields and development of improved protein backbone parameters,

A

Proteins 65(3) (2006) 712-25.

M

[35] E.F. Pettersen, T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C. Meng, T.E. Ferrin, Journal of Computational Chemistry, UCSF Chimera--a visualization system for

TE D

exploratory research and analysis, (2004).

[36] A.W.S.D. Silva, W.F.J.B.R.N. Vranken, ACPYPE - AnteChamber PYthon Parser interfacE, 5(1) (2012) 367-367.

EP

[37] W.L. Jorgensen, W.L.J.J.o.C.P. Jorgensen, Erratum: Monte Carlo simulation of n‐butane

CC

in water. Conformational evidence for the hydrophobic effect [J. Chem. Phys. 77, 5757 (1982)], 78(9) (1983) 5846-5846.

A

[38] B. Hess, P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation, J Chem Theory Comput 4(1) (2008) 116-22. [39] V.A. Marlow, D. Rea, S. Najmudin, M. Wills, V. Fueloep, Structure and mechanism of acetolactate decarboxylase, ACS Chem. Biol. 8(10) (2013) 2339-2344.

[40] S. Li, X. Gao, N. Xu, L. Liu, J. Chen, Enhancement of acetoin production in Candida glabrata by in silico-aided metabolic engineering, Microbial Cell Factories 13(1) (2014) 55. [41] M.C. Renna, N. Najimudin, L.R. Winik, S.A. Zahler, Regulation of the Bacillus subtilis alsS, alsD, and alsR genes involved in post-exponential-phase production of acetoin, J Bacteriol 175(12) (1993) 3863-75.

IP T

[42] V.A. Marlow, D. Rea, S. Najmudin, M. Wills, V. Fülöp, Structure and Mechanism of Acetolactate Decarboxylase, ACS Chemical Biology 8(10) (2013) 2339-2344.

SC R

[43] K.A. Kantardjieff, B. Rupp, Matthews coefficient probabilities: Improved estimates for

unit cell contents of proteins, DNA, and protein-nucleic acid complex crystals, Protein Sci 12(9)

U

(2003) 1865-71.

N

[44] A. Tochowicz, K. Maskos, R. Huber, R. Oltenfreiter, V. Dive, A. Yiotakis, M. Zanda, T.

A

Pourmotabbed, W. Bode, P. Goettig, Crystal structures of MMP-9 complexes with five

M

inhibitors: contribution of the flexible Arg424 side-chain to selectivity, J Mol Biol 371(4) (2007) 989-1006.

TE D

[45] S. Kisrieva Iu, V.M. Serebrennikov, N.A. Zagustina, A.M. Bezborodov, Isolation and purification of acetolactate synthase and acetolactate decarboxylase from a Lactococcus lactis culture, Prikl Biokhim Mikrobiol 36(2) (2000) 131-7.

EP

[46] C. Monnet, M. Nardi, P. Hols, M. Gulea, G. Corrieu, V. Monnet, Regulation of branched-

CC

chain amino acid biosynthesis by alpha-acetolactate decarboxylase in Streptococcus thermophilus, Letters in Applied Microbiology 36(6) (2010) 399-405.

A

[47] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, A.J. Olson, Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function, Journal of Computational Chemistry 19(14) (2015) 1639-1662.

[48] S. Najmudin, J.T. Andersen, S.A. Patkar, T.V. Borchert, D.H. Crout, V. Fülöp, Purification, crystallization and preliminary X‐ray crystallographic studies on acetolactate decarboxylase, Acta Crystallographica 59(6) (2003) 1073–1075. [49] N. Niedzwiecka, J. Gronczewska, E.F. Skorkowski, NAD-preferring malic enzyme: localization, regulation and its potential role in herring (Clupea harengus) sperm cells, Fish

IP T

Physiol Biochem 43(2) (2017) 351-360.

[50] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in oscillation

A

CC

EP

TE D

M

A

N

U

SC R

mode, Method Enzymol 276 (1997) 307-326.

Legends Figure 1. Scheme of the acetolactate pathway starting with pyruvate. In the figure: ALS, α-acetolactate synthase; ALDC, α-acetolactate decarboxylase; BDH, 2,3-butanediol dehydrogenase; AR, acetoin reductase; DAR, diacetyl reductase and NOD, nonoxidizing

IP T

non-enzymatic oxidative decarboxylation. Figure 2. Enzymatic properties of E.a.-ALDC. (a) E.a.-ALDC catalysed reaction of (±)-

SC R

AL monitored by continuous wavelength scan circular dichroism from 190 nm to 340 nm.

The entire spectra are shown starting from (±)-AL (blue) to (R)-AC (purple). (b) AC signal at 278 and 315 nm. At 278 nm, there is no signal for (S)-AL, and the change in the CD

U

signal is due to the formation of (R)-AC. At 315 nm, there is no signal for (R)-AC, and the

N

change in the CD signal is due to the disappearance of (S)-AL. The specific activity of the

M

then converted to units of μmol/mg/min.

A

enzyme was converted from units of mdeg/s into units of mM/s using molar ellipticity and

TE D

Figure 3. Crystallographic structure of E.a.-ALDC. (a) The overall tertiary and quaternary structure of E.a.-ALDC, chain A instead of chain B was presented. (b) The electron density of residues in contact with a zinc ion within a 5 Å distance, including His199, His201 and

EP

His212; Glu70 and Glu255; and Arg150 (chain A, left and chain B, right).

CC

Figure 4. Comparison between the crystal structures of E.a.-ALDC with B.b.-ALDC and B.s.-ALDC. (a) Upper: superposition of the crystal structure coordinates of E.a.-ALDC

A

chain A (grey) onto those of B.b.-ALDC (cyan, PDB ID: 4BT7) and B.s.-ALDC (green, PDB ID: 5XNE). Lower: active site residues superimposed between E.a.-ALDC chain A and B.b.-ALDC (cyan, PDB ID: 4BT7) and between E.a.-ALDC and B.s.-ALDC (green, PDB ID: 5XNE), including Thr63; His199, His201 and His212; Glu70 and Glu255; and Arg150. (B.b.-ALDC: Thr58; His194, His196 and His207; Glu65 and Glu254; and

Arg145) (B.s.-ALDC: Thr55; His191, His193 and His204; Glu62 and Glu251; and Arg142). (b) Superimpose of active centres of E.a.-ALDC chain A (grey) and chain B (dark grey) onto B.b.-ALDC (cyan) and B.s.-ALDC (green). (c) Intramolecular networks of Arg150 in E.a.-ALDC chain A interacting with Val258, Glu259 and Asn260.

IP T

Figure 5. Effects of pH (a) and different metal ions (b) on the enzymatic activity of E.a.ALDC. Error bars indicate the standard deviations of three independent experiments.

SC R

Figure 6. Energy-optimized active site structures of E.a.-ALDC chain A (upper) and chain B (lower) in complex with (R)-AL and (S)-AL. Selected key distances are shown in angstroms. Hydrogen bonds are shown in black dashed lines, and coordination bonds are

A

CC

EP

TE D

M

A

N

U

shown in red dashed lines.

EP

CC

A TE D

IP T

SC R

U

N

A

M

Figures

Figure 1.

EP

CC

A TE D

(a) (b)

IP T

SC R

U

N

A

M

Figure 2.

Figure 3. Figure 3. A

B

Arg150 Glu70 His212

His201

Zn Glu259

Cl

IP T

His199

(a)

E.a.-ALDC

U

SC R

E.a.-ALDC

A

CC

EP

TE D

M

A

N

(b)

TE D

M

A

(a)

N

U

SC R

IP T

Figure 4.

A

CC

EP

(b)

Arg150

3.6 Å

Asn260

Arg150

Glu70 His201

His212 3.4 Å 4.5 Å 3.2 Å

Zn Val258 Glu259 E.a.-ALDC PDB ID: 6J92

Thr63

(c)

His201

His212 Asn260

Zn

Cl His199

Glu70

Val258 Glu259

Thr63 Cl His199

E.a.-ALDC PDB ID: 5YHK

Figure 6. Figure 5.

Figure 5.

A

250

E.a.-ALDC

E.a.-ALDC

200

1,000

Relative activity

Specific activity (U/mg)

1,200

800

600

150

100

400

4

5

6

7

8

0

9

Cu2+

pH

Fe2+ Mn2+ Ba2+ Mg2+ Zn2+ Ca2+ Control Ion type

(a)

(b) 180

o

SC R

B

CC

EP

TE D

M

A

N

U

E.a.-ALDC

A

IP T

50

EP

CC

A TE D

IP T

SC R

U

N

A

M

Figure 6.

Tables Table 1. Data Collection and Refinement Statistics for E.a.-ALDC.

IP T

Data Collection Space group

P 41 21 2

Cell dimensions

83.541 83.541 139.173

SC R

a, b, c (Å) α, β, γ (°)

90.00 90.00 120.00

Total reflections

245067 (23615) 19453 (1906)

U

Unique reflections Resolution (Å)

45.03-2.42

(2.51-2.42)

0.104

(0.585)

30.16

(7.21)

99.88

(98.90)

12.2

(11.7)

0.999

0.93

0.999

0.982

Resolution (Å)

45.03-2.42

(2.46-2.42)

No. reflections

19445

(1898)

Rwork/Rfree

0.173/0.219

(0.228/0.316)

Clashscore

7.14

N

Rmerge

CC1/2

TE D

CC*

M

Completeness (%)

A



EP

Refinement

A

CC

Ramachandran favored (%)

96

outliers (%)

0.23

No. atoms Protein

3501

Water

116

B-factors Protein

38.60

Water

40.20

R.m.s. deviations

Bond lengths (Å)

0.009

Bond angles (°)

1.21

Values for the highest resolution bin are given in parentheses.

Table 2. Binding energy components of E.a.-ALDC in complex with (S)-AL and (R)-AL

IP T

calculated from AMBER MM-PBSA and g_mmpbsa. The energy unit is kcal/mol. (S)-AL

(R)-AL

∆EvdW

-73.86 ± 1.33

-87.04 ± 0.88

∆Eele

-77.26 ± 2.83

∆GPB

118.67 ± 1.70

∆GSA ∆Gbinding

-8.9 ± 0.06

-120.70 ± 2.00 130.79 ± 0.96

U

-9.34 ± 0.06

A

CC

EP

TE D

M

A

N

-41.43 ± 2.15

SC R

Contribution

-86.09 ± 1.52