The acetohydroxyacid synthase (AHAS) holoenzyme catalyzes the first step of branch-chain amino acid biosynthesis and is essential for plants and bacteria. It consists of a regulatory subunit (RSU) and a catalytic subunit (CSU). The allosteric mechanism of the AHAS holoenzyme has remained elusive for decades. Here, we determined the crystal structure of the AHAS holoenzyme, revealing the association between the RSU and CSU in an A2B2 mode. Structural analysis in combination with mutational studies demonstrated that the RSU dimer forms extensive interactions with the CSU dimer, in which a conserved salt bridge between R32 and D120 may act as a trigger to open the activation loop of the CSU, resulting in the activation of the CSU by the RSU. Our study reveals the activation mechanism of the AHAS holoenzyme.

## Introduction

Acetohydroxyacid synthase (AHAS, EC 2.2.1.6) catalyzes the first step of branch-chain amino acid biosynthesis, playing a crucial role in the life cycle of plants and bacteria [1,2]. Due to its absence in animals, AHAS has become an interesting target for herbicides and for new antimicrobial and antifungal agents [3,4]. AHAS is currently a common target site of five globally used herbicide chemical groups, sulfonylurea (SU), imidazolinone (IMI), triazolopyrimidine (TP), pyrimidinyl-thiobenzoate (PYB) and sulfonyl-aminocarbonyl-triazolinone (SCT), which produce an annual revenue of ∼30 billion US dollars [5].

AHAS is composed of a catalytic subunit (CSU) and a regulatory subunit (RSU) [6,7]. Maximum activity is obtained for the holoenzyme, and the enzyme activity drops to below 15% of the holoenzyme in the absence of the RSU [8]. The catalytic reaction uses two molecules of the substrate pyruvate (or 2-ketobutyrate) to form the product 2-acetolactate (or acetohydroxybutyrate) and requires the participation of three cofactors, flavin adenine dinucleotide (FAD), thiamin pyrophosphate (ThDP or TPP) and Mg2+ [1,9,10]. The CSU sequence is largely conserved across all AHASs and has a molecular mass of ∼60 kDa. However, the RSU sequence has low conservation with different molecular masss ranging from 9 to 55 kDa. Interestingly, different RSU isoforms can activate the same CSU of the AHAS protein [7]. Another important feature of AHAS is feedback inhibition, in which the branch-chain amino acid product can negatively regulate the activity of the holoenzyme [11,12]. There is one exception. The RSU of the Escherichia coli (E. coli) AHAS isoform II activates the CSU but shows no feedback inhibition [13].

Multiple crystal structures of the RSU alone, RSU with the allosteric regulator (Val/Ile), the CSU alone and the CSU with different inhibitors have been reported [9,14–17]. The CSU typically forms a dimer, and three cofactors (FAD, TPP and Mg2+) are bound tightly inside the CSU without the requirement of an RSU [15]. Inhibitors are bound in the CSU, blocking the channel for substrate entry instead of occupying the substrate-binding site [5]. The RSU also forms a dimer with a ferredoxin-like fold [17,18]. Although the architectures of AHAS RSUs differ greatly, they share a conserved ACT domain, which is a conserved structural motif in a variety of proteins that are involved in metabolism, signal transduction, and solute transport, and has been proven to be sufficient to fully activate the corresponding CSU [7,19,20]. Due to the lack of the AHAS holoenzyme structure, a series of methods (yeast two-hybrid system, mutagenesis, multidimensional NMR spectroscopy and labeling scanning) have been used to verify the key residues for CSU–RSU interactions [18,21,22]. Our previous work indicated that a conserved arginine (R32 for E. coli AHAS isoform I) [23] and a negatively charged surface of the ACT dimer (e.g. D120, E124 and D126 for E. coli AHAS isoform I) were the major residues for subunit interactions, and our previous results indicated that the ionic strength could affect the AHAS subunit interactions and that the electrostatic interactions were an important driving force for these interactions [24].

To further understand the molecular mechanism of AHAS holoenzyme activation, we determined the crystal structure of the E. coli AHAS isoform I holoenzyme. Our structural analysis combined with biochemical evidence reveals that the coupling of the RSU to the CSU facilitates the opening of the substrate-binding pocket of the CSU, resulting in enzyme activation.

## Materials and methods

### Protein expression and purification

The genes ilvN and ilvB as well as DNA for a linker (amino acids: ASTAATSAAATSAAATSAAATSAAATSAAS) were cloned into the vector with a hexahistidine (6×His) and PreScission protease cleavage site at the N-terminal, resulting in a single-chain polypeptide of the AHAS full enzyme (referred to as AHAS-FULL hereafter). The plasmids were then transformed into E. coli BL21(DE3) and overexpressed by inducing with 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) at 18°C for 14–16 h.

The bacteria were harvested by centrifugation at 6000×g for 20 min, resuspended in buffer A (20 mM Tris–HCl pH 7.5 and 500 mM NaCl) and lysed by sonication. After centrifugation at 18 000×g for 40 min at 4°C, the supernatant was loaded onto a Ni-NTA affinity column for 1.5 h and then washed with different proportions of buffer A and buffer B (20 mM Tris–HCl pH 7.5, 500 mM NaCl and 500 mM imidazole). Finally, the target protein was eluted with buffer B, and the six-histidine tag was cleaved by PreScission protease at 4°C for 12 h. The protein was then purified by a HiTrap-Q column (GE Healthcare) with a salt gradient consisting of buffer QA (20 mM Tris–HCl pH 7.5, 1 mM EDTA and 1 mM DTT) and buffer QB (20 mM Tris–HCl pH 7.5, 1 mM EDTA, 1 mM DTT and 1 M NaCl). After that, protein peaks were collected and further purified by a HiLoad 26/60 Superdex 200 size-exclusion column (GE Healthcare) with running buffer (50 mM Tris–HCl pH 7.5, 200 mM NaCl, 1 mM DTT, 20 µM FAD, 0.1 mM ThDP and 1 mM MgCl2). The AHAS-FULL protein was concentrated to 4 mg/ml by Amicon Centrifugal Filter Devices (Millipore) for crystallization.

Genes for the AHAS CSUs and RSUs from E. coli isoforms I, II and III were cloned into the pET28a (+) vector with an additional N-terminal six-histidine tag. Then, plasmids were transformed into E. coli BL21(DE3). To improve the protein expression, we made the E14G mutant for the ilvM gene and truncated 86 residues from the C-terminus of the ilvH gene [7]. The mutants for assay experiments were created using a site-direct mutagenesis method. The resulting plasmids were overexpressed by inducing with 0.5 mM IPTG at 18°C for 14–16 h.

The bacteria containing different either the wild type or mutants of the E. coli AHAS isoform I RSU or CSU proteins were harvested by centrifugation at 6000×g for 20 min, resuspended in buffer A (20 mM potassium phosphate pH 7.5 and 500 mM NaCl) and then lysed by sonication in an ice bath. The supernatant was obtained by centrifugation at 18 000×g at 4°C for 40 min and loaded onto a Ni-NTA affinity column for 1 h. The column was washed and eluted with different proportions of buffer A and buffer B (20 mM potassium phosphate pH 7.5, 500 mM NaCl and 500 mM imidazole). The proteins were identified by SDS–PAGE and collected, concentrated and applied to a HiTrap-desalting column (GE Healthcare) with buffer (50 mM potassium phosphate pH 7.5 and 200 mM NaCl).

Wild type and mutants of the E. coli AHAS isoform II CSU or E. coli AHAS isoform III CSU proteins were expressed in E. coli and purified by a Ni-NTA affinity column. Different proportions of buffer A (20 mM Tris–HCl pH 7.5 and 500 mM NaCl) and buffer B (20 mM Tris–HCl pH 7.5, 500 mM NaCl and 500 mM imidazole) were applied to wash and elute the target protein. The protein samples were finally purified by a Superdex 200 size-exclusion column (GE Healthcare) with buffer (50 mM Tris–HCl pH 7.5, 200 mM NaCl, 1 mM DTT, 20 µM FAD, 0.1 mM ThDP and 1 mM MgCl2).

Wild type and mutants of the E. coli AHAS isoform II RSU or E. coli AHAS isoform III RSU proteins were also expressed in E. coli and purified by a Ni-NTA affinity column. The column was washed with buffer A (20 mM potassium phosphate pH 7.5, 500 mM NaCl and 20% (v/v) glycerol) for five column volumes, and the proteins were eluted using buffer B (20 mM Tris–HCl pH 7.5, 500 mM NaCl, 500 mM imidazole and 20% (v/v) glycerol). The protein samples were finally stored in buffer (50 mM potassium phosphate pH 7.5, 200 mM NaCl and 20% (v/v) glycerol).

### Crystallization and data collection

Wild-type AHAS-FULL protein (4 mg/ml in the buffer with 50 mM Tris–HCl pH 7.5, 200 mM NaCl, 1 mM DTT, 20 µM FAD, 0.1 mM ThDP and 1 mM MgCl2) was crystallized using the sitting-drop vapor-diffusion method mixed with a reservoir solution (0.1 M Tris–HCl pH 8.5, 0.2 M MgCl2·6H2O and 25% (w/v) PEG 3350). Crystals grew for 10 days at 20°C and were frozen in a cryoprotectant consisting of the reservoir solution with an additional 30% (v/v) glycerol.

All data were collected on the beamline BL-17A at the Photon Factory (Tsukuba, Japan) and the beamline BL19U1 at the Shanghai Synchrotron Radiation Facility (SSRF) and then were processed using HKL2000 software [25].

### Structure determination and refinement

The structure of AHAS-FULL was determined by molecular replacement with PHASER [26] using the CSU of yeast AHAS from PDB ID 1JSC [15] and the RSU of the E. coli AHAS from PDB ID 2LVW [18] as the search models. Four copies of each model were found in the asymmetric unit. The model was built in COOT [27] and refined using PHENIX software [28]. The AHAS-FULL structure was refined to a resolution of 2.85 Å, and the orientations of the amino acid side chains and bound water molecules were modeled on the basis of 2FobsFcalc and FobsFcalc difference Fourier maps. The final structure had an Rcrystal value of 16.7% and Rfree value of 24.3%. Detailed data collection and refinement statistics are summarized in Supplementary Table S1.

### Enzymatic assays

The AHAS enzyme was added to the reaction buffer, which contained 50 mM potassium phosphate pH 7.5, 10 mM MgCl2, 1 mM ThDP and 10 µM FAD, for 30 min at 37°C. Then, the reaction was stopped by adding H2SO4 to a final concentration of 300 mM at 60°C for 15 min. The resulting acetoin amount was further quantified by measuring the absorbance at 525 nm after incubating with 0.5% creatine and 5% (w/v) α-naphthol at 60°C for 15 min.

To determine the Vmax and Km, different concentrations of the substrate pyruvate were included in the reaction system. The parameters were calculated by fitting the data to the Hill equation: $V=Vmaxn/(1+Kmn/[S]n)$, where Vmax is the maximum velocity, Km is the Michaelis–Menten constant and [S] is the substrate concentration.

To determine the reconstituted holoenzyme activities, the CSU and RSU proteins (wild type and mutants) were preincubated in the reaction mixture buffer for 15 min at 37°C before starting the reaction. The protein concentrations are 18.5 nM EcCSUI and 390 nM EcRSUI for the E. coli AHAS isoform I, 74.1 nM EcCSUI and 2.67 µM EcRSUII for the E. coli AHAS isoform II, and 80 nM EcCSUIII and 851 nM EcRSUIII for the E. coli AHAS isoform III.

To determine the K0.5 (defined as the RSU concentration required for the half-maximum activity of AHAS), CSU was incubated with different concentrations of the RSU for 15 min at 37°C. Then, 50 mM pyruvate was included in the reaction system (50 mM potassium phosphate pH 7.5, 10 mM MgCl2, 1 mM ThDP and 10 µM FAD) to start the reaction. The CSU concentration (wild type and mutants) is 12.3 nM for E. coli AHAS isoform I, 23.9 nM for E. coli AHAS isoform II and 80 nM for E. coli AHAS isoform III. The values of K0.5 were calculated by fitting the data to the following equation: $V=V0+Vmaxn/(1+K0.5n/[C]n)$, where V0 is the reaction velocity of CSU alone, Vmax is the maximum velocity and [C] is the RSU concentration. The Ki values for valine data are fitted to the equation Vi= V+(V0+ V)/(1+[I]/Ki), where Vi and V0 represent the rates in the presence or absence of the inhibitor valine, respectively, V represents the final residual activity of the enzyme by adding inhibitor valine, [I] is the concentration of valine and Ki is the apparent inhibition constant, which means the concentration of inhibitor valine giving 50% inhibition. All experiments were performed with three technical replicates, and the data were processed by means of the nonlinear curve fit analysis by using Origin 8.0 software.

### Molecular dynamics simulation

Two simulation systems, the AHAS holoenzyme model and the CSU-alone model, were constructed. The structure of the AHAS-FULL was used as the initial structure of the AHAS holoenzyme model. The missing sidechains of amino acids were added using the tleap module of AMBER 14 [29]. The CSU-alone model was obtained by removing two RSUs from the AHAS holoenzyme model. Each system was solvated in a TIP3P water box [30] and neutralized by adding magnesium ions. The AHAS holoenzyme and CSU-alone models contain 139 023 atoms and 129 718 atoms, respectively.

The molecular dynamics (MD) simulations were performed using the PMEMD module of AMBER 14 [29]. The AMBER FF14SB force field [31] was used for proteins, and the general AMBER force field (GAFF) [32] was used for FAD and TPP. First, each system (i.e. AHAS holoenzyme and CSU-alone) was minimized for 10 000 steps. Second, each system was equilibrated by a 500 ps constant volume ensemble to heat the system from 0 to 310 K, and the Langevin thermostat [33] was used for the temperature control. Third, each system was equilibrated for 10 ns with the backbone of protein and ligands constrained at 10 kcal mol−1 Å−2. After equilibration, the 150 ns MD simulation was carried out in a constant pressure ensemble for each system. The particle-mesh Ewald (PME) algorithm [34] was used to treat long-range electrostatic interactions. All of the covalent bonds involving hydrogen atoms were constrained using the SHAKE algorithm [35].

To characterize the conformational changes of the activation loop for each simulated system, we used the potential of mean force (PMF) [36] analysis and generated 2D energy landscapes. The reaction co-ordinates of the PMF map were the distance between the activation loop and FAD (represented by the distance between the center of mass (COM) of the activation loop and the COM of the isoalloxazine ring of FAD) and the distance between the activation loop and TPP (represented by the distance between the COM of the activation loop and the COM of the aminopyrimidine of TPP). The energy landscape was calculated as$ΔG(x,y)=−kBTln⁡g(x,y)$, where kB is the Boltzmann constant, T is the temperature and g(x, y) is the normalized joint probability distribution [36].

## Results

### Overall structure of the AHAS holoenzyme

To obtain AHAS holoenzyme crystals, the RSU was covalently linked to the CSU using the sequence ASTAATSAAATSAAATSAAATSAAATSAAS as previously described [37]. The purified single chain of the AHAS full enzyme (AHAS-FULL) has similar enzymatic characteristics to the AHAS holoenzyme reconstituted by separating the E. coli AHAS I CSU (EcCSUI) and E. coli AHAS I RSU (EcRSUI) (Supplementary Table S2 and Supplementary Figure S1). The structure of AHAS-FULL was determined using molecular replacement at a resolution of 2.85 Å. The overall structure of AHAS-FULL is dimeric and consists of two RSUs and two CSUs with an A2B2 mode (Figure 1A and Supplementary Figure S2). The RSU monomer has a canonical ferredoxin-like fold with four β-sheets in the middle flanked by three α-helices (Figure 1B and Supplementary Figure S3A,B). The RSU monomer and dimer are similar to other RSU structures, as previously reported [17,18]. The CSU monomer can be further separated into three subdomains (Figure 1C): Domains I, II and III. Three cofactors (FAD, TPP and Mg2+) were unambiguously found in the CSU (Figure 1A,C). The CSU monomer and dimer are also similar to other reported CSU structures (Supplementary Figure S3C,D) [15,16].

#### Crystal structures of the AHAS holoenzyme.

Figure 1.
Crystal structures of the AHAS holoenzyme.

(A) A cartoon representation of the overall structure of the E. coli AHAS I holoenzyme dimer. CSU-A, CSU-B, RSU-A and RSU-B are colored green, orange, yellow and blue, respectively. FAD and TPP are shown as stick models and colored cyan and magenta, respectively. (B) Structure of the RSU monomer. The α-helices and β-sheets are colored red and yellow, respectively. (C) Structure of the CSU monomer. Domains I, II and III are colored orange, red and cyan, respectively. FAD and TPP are shown as stick models and colored green.

Figure 1.
Crystal structures of the AHAS holoenzyme.

(A) A cartoon representation of the overall structure of the E. coli AHAS I holoenzyme dimer. CSU-A, CSU-B, RSU-A and RSU-B are colored green, orange, yellow and blue, respectively. FAD and TPP are shown as stick models and colored cyan and magenta, respectively. (B) Structure of the RSU monomer. The α-helices and β-sheets are colored red and yellow, respectively. (C) Structure of the CSU monomer. Domains I, II and III are colored orange, red and cyan, respectively. FAD and TPP are shown as stick models and colored green.

### Interactions of the RSU with the CSU

The structural analysis of AHAS-FULL reveals that the entire AHAS-FULL dimer has a 2-fold noncrystallographic symmetry. The RSU dimer is sitting on the top of the CSU dimer (Figure 2A). The four helices from the RSU dimer form extensive interactions with the CSU dimer (Figure 2B). Specifically, the side chain of residue H18 and the main-chain carbonyl oxygen of residue P19 from the RSU-B monomer make hydrogen bonds with the side chain of residue R142 from the CSU-B monomer. The side chain of residue H24 from the RSU-B monomer makes hydrogen bonds with the side chains of residues Y128 and N138 from the CSU-B monomer. Moreover, the side chains of residues R31 and R32 from the RSU-A monomer form hydrogen bonds with the side chains of residues D126 and D120 from the CSU-B monomer, respectively. The reciprocal interactions between the RSU dimer and the CSU-A monomer also exist within the binding interface except the salt-bridge interaction (R32 of RSU-B and D120 of CSU-A) due to the disordered activation loop in CSU-A monomer.

#### Interactions of the RSU with the CSU.

Figure 2.
Interactions of the RSU with the CSU.

(A) Structure of the AHAS holoenzyme. A cartoon model of the RSU dimer (blue and yellow) is sitting on the top of the CSU dimer represented by the surface map (green and yellow). Four helices (α1A, α2A, α1B and α2B) of the RSU form the structure motif to mediate the interaction between the RSU and CSU. (B) Detailed interactions between the RSU dimer and the CSU-B monomer. It is noted that reciprocal interactions occur between the RSU dimer and the CSU-A monomer. Residues from the CSU-B monomer are colored orange, and residues from the RSU dimer (RSU-A and RSU-B) are colored yellow and blue, respectively. Two residues, R32 and D120, that form a salt bridge pair are colored red. (C) Sequence alignment of the AHAS proteins from various sources. Conserved residues among these AHASs are highlighted in magenta and red. Residues R32 and D120 are labeled with * symbols.

Figure 2.
Interactions of the RSU with the CSU.

(A) Structure of the AHAS holoenzyme. A cartoon model of the RSU dimer (blue and yellow) is sitting on the top of the CSU dimer represented by the surface map (green and yellow). Four helices (α1A, α2A, α1B and α2B) of the RSU form the structure motif to mediate the interaction between the RSU and CSU. (B) Detailed interactions between the RSU dimer and the CSU-B monomer. It is noted that reciprocal interactions occur between the RSU dimer and the CSU-A monomer. Residues from the CSU-B monomer are colored orange, and residues from the RSU dimer (RSU-A and RSU-B) are colored yellow and blue, respectively. Two residues, R32 and D120, that form a salt bridge pair are colored red. (C) Sequence alignment of the AHAS proteins from various sources. Conserved residues among these AHASs are highlighted in magenta and red. Residues R32 and D120 are labeled with * symbols.

Next, we used site-direct mutagenesis to verify these interactions. As we expected, mutating residues within the interface between the EcCSUI dimer and EcRSUI dimer caused decreased activities (Table 1). Mutating residues H18 and H24 of EcRSUI moderately decreased the activities, while mutating residue R32 of EcRSUI reduced the activity by 50%. Mutating residues D120, D126, Y128, N138 and R142 of EcCSUI greatly decreased the activities. The double mutant of residues D126 and Y128 of EcCSUI almost abolished activity. Moreover, mutating residues D120, D126 and Y128 attenuated the binding between the EcRSUI dimer and the EcCSUI dimer as reflected by the value of K0.5, the half-maximal concentration of EcCSUI activation by EcRSUI. Furthermore, the salt bridge between residue R32 of EcRSUI and residue D120 of EcCSUI seems to play a very important role in the activation of the CSU by the RSU since mutating one or both residues severely affected the activation of the reconstituted AHAS holoenzyme (Table 1 and Supplementary Figure S4A).

Table 1.
Steady-state kinetics of reconstituted enzymes of E.coli AHAS I wild type and mutants
Km (mM)Kcat (s−1)Kcat/Km (mM−1 s−1)K0.5 (μM)1Relative activity2
EcRSUI + EcCSUI (wild type) 0.99 ± 0.22 34.33 ± 3.77 34.68 0.02 ± 0.01 1.00
EcCSUI only 6.53 ± 0.15 6.75 ± 2.32 1.03 — 0.20
EcRSUI-H18A + EcCSUI 1.04 ± 0.27 27.73 ± 0.90 26.66 0.02 ± 0.01 0.81
EcRSUI-H18A-H24A + EcCSUI 1.16 ± 0.34 22.85 ± 1.60 19.70 0.15 ± 0.01 0.67
EcRSUI + EcCSUI-D126A-Y128A 9.00 ± 0.47 2.47 ± 0.84 0.27 16.28 ± 1.80 0.07
EcRSUI + EcCSUI-N138A 1.14 ± 0.22 21.49 ± 0.42 18.85 0.03 ± 0.01 0.63
EcRSUI + EcCSUI-N138A-R142A 2.27 ± 0.45 17.05 ± 1.04 7.51 0.27 ± 0.11 0.50
EcRSUI-R32A + EcCSUI 7.17 ± 4.28 16.07 ± 0.88 2.24 1.70 ± 0.74 0.47
EcRSUI + EcCSUI-D120A 7.24 ± 1.26 8.70 ± 5.25 1.21 13.30 ± 4.74 0.25
EcRSUI-R32A + EcCSUI-D120A 5.37 ± 0.57 5.21 ± 0.73 0.97 15.35 ± 2.94 0.15
Km (mM)Kcat (s−1)Kcat/Km (mM−1 s−1)K0.5 (μM)1Relative activity2
EcRSUI + EcCSUI (wild type) 0.99 ± 0.22 34.33 ± 3.77 34.68 0.02 ± 0.01 1.00
EcCSUI only 6.53 ± 0.15 6.75 ± 2.32 1.03 — 0.20
EcRSUI-H18A + EcCSUI 1.04 ± 0.27 27.73 ± 0.90 26.66 0.02 ± 0.01 0.81
EcRSUI-H18A-H24A + EcCSUI 1.16 ± 0.34 22.85 ± 1.60 19.70 0.15 ± 0.01 0.67
EcRSUI + EcCSUI-D126A-Y128A 9.00 ± 0.47 2.47 ± 0.84 0.27 16.28 ± 1.80 0.07
EcRSUI + EcCSUI-N138A 1.14 ± 0.22 21.49 ± 0.42 18.85 0.03 ± 0.01 0.63
EcRSUI + EcCSUI-N138A-R142A 2.27 ± 0.45 17.05 ± 1.04 7.51 0.27 ± 0.11 0.50
EcRSUI-R32A + EcCSUI 7.17 ± 4.28 16.07 ± 0.88 2.24 1.70 ± 0.74 0.47
EcRSUI + EcCSUI-D120A 7.24 ± 1.26 8.70 ± 5.25 1.21 13.30 ± 4.74 0.25
EcRSUI-R32A + EcCSUI-D120A 5.37 ± 0.57 5.21 ± 0.73 0.97 15.35 ± 2.94 0.15

E. coli AHAS I CSU and RSU abbreviated as EcCSUI and EcRSUI.

1

K0.5, concentration of RSU required for 50% maximum enzymatic activity;

2

Relative activity is with respect to that of mutants with EcRSUI + EcCSUI wild-type (1.00).

### Activation mechanism of the CSU by the RSU

To study the activation mechanism of the CSU by the RSU, first, we performed side-by-side comparisons of the structures of the RSU from various species. Although the sequences varied greatly among different RSU species, they all shared a similar structural feature of the four helical motifs that interacted with the CSU in our AHAS-FULL structure (Supplementary Figure S5A). We then compared the structures of the CSU from various species. One region (residues 103–145) also formed a similar motif that interacted with the RSU in our AHAS-FULL structure (Supplementary Figure S5B). These results possibly indicate that the activation mechanism of the CSU by the RSU is conserved in all AHAS holoenzymes. We further aligned the sequences of the salt bridge pairs (EcRSUI-R32/EcCSUI-D120) and found that these two residues were highly conserved among various species (Figure 2C). More importantly, the location of these two residues in three-dimensional structures among different species was also conserved (Supplementary Figure S5), suggesting that this salt bridge pair might be responsible for the activation of the AHAS holoenzyme. To test our hypothesis, we measured the activities of AHAS mutants from different isoforms as follows: residues EcRSUII R27A and EcCSUII D107A of E. coli AHAS isoform II and residues EcRSUIII R26A and EcCSUIII D111A of E. coli AHAS isoform III. As we expected, the mutants drastically decreased the holoenzyme activities (Table 2 and Supplementary Figure S4B,C).

Table 2.
Steady-state kinetics of reconstituted enzymes of salt bridge (R32/D120) mutant from various species
Km (mM)Vmax (Umg−1)kcat (s−1)kcat/Km (mM−1 s−1)K0.5 (μM)
E. coli AHAS II
EcCSUII only 3.21 ± 0.57 0.04 ± 0.02 0.04 ± 0.02 0.01 —
EcCSUII + EcRSUII (wild type) 1.21 ± 0.05 11.57 ± 0.14 12.11 ± 0.15 10.01 0.48 ± 0.07
EcCSUII-D107A + EcRSUII-R27A 6.23 ± 1.10 1.90 ± 0.18 2.00 ± 0.19 0.32 15.20 ± 0.82
E. coli AHAS III
EcCSUIII only 8.37 ± 0.81 0.23 ± 0.1 0.26 ± 0.11 0.03 —
EcCSUIII + EcRSUIII (wild type) 1.38 ± 0.06 10.65 ± 0.14 11.86 ± 0.16 8.59 0.51 ± 0.09
EcCSUIII-D111A + EcRSUIII-R26A 6.30 ± 0.37 1.85 ± 0.07 2.06 ± 0.08 0.33 4.81 ± 0.70
Km (mM)Vmax (Umg−1)kcat (s−1)kcat/Km (mM−1 s−1)K0.5 (μM)
E. coli AHAS II
EcCSUII only 3.21 ± 0.57 0.04 ± 0.02 0.04 ± 0.02 0.01 —
EcCSUII + EcRSUII (wild type) 1.21 ± 0.05 11.57 ± 0.14 12.11 ± 0.15 10.01 0.48 ± 0.07
EcCSUII-D107A + EcRSUII-R27A 6.23 ± 1.10 1.90 ± 0.18 2.00 ± 0.19 0.32 15.20 ± 0.82
E. coli AHAS III
EcCSUIII only 8.37 ± 0.81 0.23 ± 0.1 0.26 ± 0.11 0.03 —
EcCSUIII + EcRSUIII (wild type) 1.38 ± 0.06 10.65 ± 0.14 11.86 ± 0.16 8.59 0.51 ± 0.09
EcCSUIII-D111A + EcRSUIII-R26A 6.30 ± 0.37 1.85 ± 0.07 2.06 ± 0.08 0.33 4.81 ± 0.70

E. coli AHAS II CSU and RSU abbreviated as EcCSUII and EcRSUII;.

E. coli AHAS III CSU and RSU abbreviated as EcCSUIII and EcRSUIII.

There are four molecules in one asymmetric unit of the AHAS-FULL structure (Supplementary Figure S6). The activation loop (amino acids 114–126) is disordered in three molecules but ordered in one molecule, thus suggesting two different conformational states. The activation loop forms a cover for the catalytic pocket, which presumably regulates the entry of the substrate. In the conformational state I of the AHAS-FULL structure, the activation loop is flexible with an open position, resulting in an accessible catalytic pocket by the substrate (Figure 3A). In contrast, in conformational state II of the AHAS-FULL structure, the activation loop is ordered with a closed position, blocking substrate entry into the catalytic pocket (Figure 3B and Supplementary Figure S7). Furthermore, in the structures of the yeast CSU-alone system, the activation loop adopts an ordered conformation in which the catalytic pocket is blocked (Supplementary Figure S8). Therefore, the activation loop may adopt an open-closed motion during catalysis. The superimposition of two AHAS-FULL dimers from the same asymmetric unit gives r.m.s.d. value of 0.74 Å for 1222 Cα atoms, suggesting the subtle subunit motion during an open-closed state change.

#### Activation mechanism of the AHAS holoenzyme.

Figure 3.
Activation mechanism of the AHAS holoenzyme.

(A) The activation loop (dotted line in red) is in an open conformational state, resulting in a widely open pocket for substrate entry. It has been noted that the disordered activation loop is not in the pocket, so it has a surface representation effect similar to the deleted one. (B) The activation loop (solid line in red) is in a closed conformational state, resulting in a blocked pocket. (C) The potential of mean force (PMF) was calculated for the distance between the activation loop and FAD relative to the distance between the activation loop and TPP in the CSU-alone (left) and the AHAS holoenzyme systems (right). (D) The CSU superposition of the AHAS holoenzyme and CSU-alone systems. The activation loop is shown in as a cartoon (green for the AHAS holoenzyme system and magenta in the CSU-alone system), and FAD and TPP are displayed as a ball-and-stick representation.

Figure 3.
Activation mechanism of the AHAS holoenzyme.

(A) The activation loop (dotted line in red) is in an open conformational state, resulting in a widely open pocket for substrate entry. It has been noted that the disordered activation loop is not in the pocket, so it has a surface representation effect similar to the deleted one. (B) The activation loop (solid line in red) is in a closed conformational state, resulting in a blocked pocket. (C) The potential of mean force (PMF) was calculated for the distance between the activation loop and FAD relative to the distance between the activation loop and TPP in the CSU-alone (left) and the AHAS holoenzyme systems (right). (D) The CSU superposition of the AHAS holoenzyme and CSU-alone systems. The activation loop is shown in as a cartoon (green for the AHAS holoenzyme system and magenta in the CSU-alone system), and FAD and TPP are displayed as a ball-and-stick representation.

To test this hypothesis, we took advantage of the MD simulation methodology. We constructed two systems (i.e. the AHAS holoenzyme and CSU-alone systems) and performed MD simulations for the AHAS-FULL and the CSU dimer. The dynamic behaviors of the activation loop and the activation mechanism of the CSU induced by the RSU were explored. Our 150 ns MD simulations show that the CSU structures in the AHAS holoenzyme system deviated significantly from the structure in the CSU-alone system (Supplementary Figure S9). Furthermore, we calculated the potential of the mean force (PMF) for the distance between the activation loop and FAD vs. the distance between the activation loop and TPP in the AHAS holoenzyme and CSU-alone systems (Figure 3C). The PMF map clearly demonstrates two different conformational states for the activation loop: open and closed states. In the open state (which correlates to the AHAS holoenzyme system), the distance between the activation loop and FAD/TPP increases from 14 Å/18 Å (initial state) to 19 Å/19 Å, forming an accessible catalytic pocket for the substrate. However, in the closed state (which correlates to the CSU-alone system), the distance between the activation loop and FAD/TPP decreased to 11.5 Å/14 Å, resulting in the blockage of substrate entry into the catalytic pocket. The structure of the final snapshot of each system is shown in Figure 3D.

## Discussion

In this study, we determined the crystal structure of the E. coli AHAS isoform I holoenzyme. Our structural analysis demonstrates that the dimer of the RSU closely associates with the dimer of the CSU in an A2B2 mode. The interaction interface consists of a bundle of four helices in the RSU dimer and a structural motif containing the activation loop in the CSU dimer. Surprisingly, a structural comparison indicates that the interaction interface is highly conserved in the three-dimensional structure, although the amino acid sequence varies greatly among different species. This structural feature explains the biochemical data from a previous report that different RSU isoforms can activate the same CSU from the AHAS protein [7]. More importantly, the activation loop of the CSU is a part of the interface, providing a good basis for the RSU directly regulating the activities of the CSU in the AHAS holoenzyme.

Our structure of the AHAS holoenzyme reveals details of extensive interactions within the binding interface between the RSU and CSU. Mutations of these residues indeed greatly reduced the activities of the AHAS holoenzyme. A highly conserved salt bridge between R32 of the EcRSUI dimer and D120 of the activation loop of the EcCSUI dimer was identified. Disruption of this salt bridge facilitates the dissociation of the RSU from the CSU and thereby significantly (50–80%) reduces the activity of the AHAS holoenzyme among the three E. coli AHAS isoforms, indicating that this salt bridge may play an important role in the activation the CSU by the RSU. In addition, our crystal structure of the AHAS holoenzyme captured two different conformations (disordered and ordered) of the activation loop, which presumably represents different intermediate states of the activation loop during catalysis. In the ordered conformation of the activation loop, the salt bridge of R32 of EcRSUI and D120 of EcCSUI is nicely formed. However, the activation loop blocks the entry channel of the active site, representing the closed conformational state of the catalytic pocket. The closed state of the catalytic pocket was found in many previously determined structures of the AHAS CSU-alone system [5,10,15,16]. In the disordered conformation of the activation loop, the active site is fully exposed, presumably allowing substrate entry during catalysis and representing the open conformational state of the catalytic pocket. Thus far, the open state has been only captured in our holoenzyme structure. These results suggest that the activation loop may adopt an open-closed motion during catalysis. This open-closed motion can be found in many allosteric enzymes, such as 1-acyl-sn-glycerophosphate acyltransferase from Thermotoga maritima [38] and insulin-degrading enzyme [39]. Our MD simulation results confirm that the activation loop prefers the open state in the AHAS holoenzyme and the closed state in the CSU-only enzyme. Taken together, the RSU activating the CSU facilitates the opening of the activation loop.

Based on the above results, we propose a model to describe the activation mechanisms of the AHAS holoenzyme (Figure 4). The activation loop in the CSU alone is in a closed conformation state. Its occasional opening results in a low activity of the CSU alone (Phase I). When the RSU dimer associates with the CSU dimer, the salt bridge between R32 and D120 forms (Phase II) and accelerates the opening rate of the activation loop (Phase III). Therefore, the AHAS holoenzyme is activated (Phase IV, Phase V). Overall, our results uncover an allosteric regulation mechanism of the AHAS holoenzyme and provide important information for designing green herbicides and developing new generations of antibiotics.

### Model of the activation mechanism of the AHAS holoenzyme.

Figure 4.
Model of the activation mechanism of the AHAS holoenzyme.

The activation loop in the CSU alone is in the closed state. The association of the RSU to the CSU facilitates the open state of the activation loop, resulting in substrate entry and catalysis.

Figure 4.
Model of the activation mechanism of the AHAS holoenzyme.

The activation loop in the CSU alone is in the closed state. The association of the RSU to the CSU facilitates the open state of the activation loop, resulting in substrate entry and catalysis.

## Data Availability

Atomic co-ordinates and structure factors for the reported crystal structures have been deposited in the Protein Data Bank with accession code 6LPI for AHAS-FULL.

## Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

## Funding

This work was supported by the National Key R&D Program of China (grant 2017YFD0200500 to Z.X., grant 2017YFA0504801 to Y.S., grant 2017YFC1104400 to J.L.), the Natural Science Foundation of China (grants 21740002 and 21837001 to Z.X., grants 31570750 and 31870834 to Y.S., grants 91842302 and 31870736 to X.Y.).

## Author Contributions

Y.Z. performed protein purification, protein crystallization and assay experiments. X.Y. performed structure determination and structure refinement. X.Liu, X.Li did mutants purification. Y.L., J.S. and J.L. performed MD simulations. Y.Z., X.Y., Z.X. and Y.S. designed the study and wrote the paper. All authors discussed the results and commented on the manuscript.

## Acknowledgements

We are grateful to Mrs Xiaomin Ji for help in protein purification and crystallization.

## Abbreviations

• AHAS

acetohydroxyacid synthase

•
• COM

center of mass

•
• CSU

catalytic subunit

•

•
• IPTG

isopropyl β-D-1-thiogalactopyranoside

•
• MD

molecular dynamics

•
• PMF

potential of mean force

•
• PMF

potential of mean force

•
• RSU

regulatory subunit

•
• SU

sulfonylurea

•
• TP

triazolopyrimidine

## References

References
1
Barak
,
Z.
and
Chipman
,
D.M.
(
2012
)
Allosteric regulation in acetohydroxyacid synthases (AHASs)–different structures and kinetic behavior in isozymes in the same organisms
.
Arch. Biochem. Biophys.
519
,
167
174
2
Chipman
,
D.M.
,
Duggleby
,
R.G.
and
Tittmann
,
K.
(
2005
)
Mechanisms of acetohydroxyacid synthases
.
Curr. Opin. Chem. Biol.
9
,
475
481
3
Duggleby
,
R.G.
,
McCourt
,
J.A.
and
Guddat
,
L.W.
(
2008
)
Structure and mechanism of inhibition of plant acetohydroxyacid synthase
.
Plant Physiol. Biochem.
46
,
309
324
4
Baig
,
I.A.
,
Moon
,
J.Y.
,
Kim
,
M.S.
,
Koo
,
B.S.
and
Yoon
,
M.Y.
(
2014
)
Structural and functional significance of the highly-conserved residues in Mycobacterium tuberculosis acetohydroxyacid synthase
.
Enzyme Microb. Technol.
58–59
,
52
59
5
Garcia
,
M.D.
,
Nouwens
,
A.
,
Lonhienne
,
T.G.
and
Guddat
,
L.W.
(
2017
)
Comprehensive understanding of acetohydroxyacid synthase inhibition by different herbicide families
.
114
,
E1091
E1100
6
Eoyang
,
L.
and
Silverman
,
P.M.
(
1984
)
Purification and subunit composition of acetohydroxyacid synthase I from Escherichia coli K-12
.
J. Bacteriol.
157
,
184
189
7
Zhao
,
Y.
,
Niu
,
C.
,
Wen
,
X.
and
Xi
,
Z.
(
2013
)
The minimum activation peptide from ilvH can activate the catalytic subunit of AHAS from different species
.
Chembiochem
14
,
746
752
8
Lee
,
Y.T.
and
Duggleby
,
R.G.
(
2002
)
Regulatory interactions in Arabidopsis thaliana acetohydroxyacid synthase
.
FEBS Lett.
512
,
180
184
9
Gedi
,
V.
and
Yoon
,
M.Y.
(
2012
)
Bacterial acetohydroxyacid synthase and its inhibitors–a summary of their structure, biological activity and current status
.
FEBS J.
279
,
946
963
10
Lonhienne
,
T.
,
Garcia
,
M.D.
,
Fraser
,
J.A.
,
Williams
,
C.M.
and
Guddat
,
L.W.
(
2017
)
The 2.0 A X-ray structure for yeast acetohydroxyacid synthase provides new insights into its cofactor and quaternary structure requirements
.
PLoS One
12
,
e0171443
11
Umbarger
,
H.E.
and
Brown
,
B.
(
1958
)
Isoleucine and valine metabolism in Escherichia coli. VIII. The formation of acetolactate
.
J. Biol. Chem.
233
,
1156
1160
PMID:
[PubMed]
12
McCourt
,
J.A.
and
Duggleby
,
R.G.
(
2006
)
Acetohydroxyacid synthase and its role in the biosynthetic pathway for branched-chain amino acids
.
Amino Acids
31
,
173
210
13
Hill
,
C.M.
,
Pang
,
S.S.
and
Duggleby
,
R.G.
(
1997
)
Purification of Escherichia coli acetohydroxyacid synthase isoenzyme II and reconstitution of active enzyme from its individual pure subunits
.
Biochem. J.
327
(
Pt 3
),
891
898
14
Lonhienne
,
T.
,
Nouwens
,
A.
,
Williams
,
C.M.
,
Fraser
,
J.A.
,
Lee
,
Y.T.
,
West
,
N.P.
et al (
2016
)
Commercial herbicides can trigger the oxidative inactivation of acetohydroxyacid synthase
.
Angew. Chem. Int. Ed. Engl.
55
,
4247
4251
15
Pang
,
S.S.
,
Duggleby
,
R.G.
and
Guddat
,
L.W.
(
2002
)
Crystal structure of yeast acetohydroxyacid synthase: a target for herbicidal inhibitors
.
J. Mol. Biol.
317
,
249
262
16
McCourt
,
J.A.
,
Pang
,
S.S.
,
King-Scott
,
J.
,
Guddat
,
L.W.
and
Duggleby
,
R.G.
(
2006
)
Herbicide-binding sites revealed in the structure of plant acetohydroxyacid synthase
.
103
,
569
573
17
Kaplun
,
A.
,
Vyazmensky
,
M.
,
Zherdev
,
Y.
,
Belenky
,
I.
,
Slutzker
,
A.
,
Mendel
,
S.
et al (
2006
)
Structure of the regulatory subunit of acetohydroxyacid synthase isozyme III from Escherichia coli
.
J. Mol. Biol.
357
,
951
963
18
Karanth
,
N.M.
and
Sarma
,
S.P.
(
2013
)
The coil-to-helix transition in IlvN regulates the allosteric control of Escherichia coli acetohydroxyacid synthase I
.
Biochemistry
52
,
70
83
19
Chipman
,
D.M.
and
Shaanan
,
B.
(
2001
)
The ACT domain family
.
Curr. Opin. Struct. Biol.
11
,
694
700
20
Liberles
,
J.S.
,
Thorolfsson
,
M.
and
Martinez
,
A.
(
2005
)
Allosteric mechanisms in ACT domain containing enzymes involved in amino acid metabolism
.
Amino Acids
28
,
1
12
21
Mitra
,
A.
and
Sarma
,
S.P.
(
2008
)
Escherichia coli ilvN interacts with the FAD binding domain of ilvB and activates the AHAS I enzyme
.
Biochemistry
47
,
1518
1531
22
Kyselkova
,
M.
,
Janata
,
J.
,
Sagova-Mareckova
,
M.
and
Kopecky
,
J.
(
2010
)
Subunit-subunit interactions are weakened in mutant forms of acetohydroxy acid synthase insensitive to valine inhibition
.
Arch. Microbiol.
192
,
195
200
23
Zhao
,
Y.
,
Wen
,
X.
,
Niu
,
C.
and
Xi
,
Z.
(
2012
)
Arginine 26 and aspartic acid 69 of the regulatory subunit are key residues of subunits interaction of acetohydroxyacid synthase isozyme III from E. coli
.
Chembiochem
13
,
2445
2454
24
Xie
,
Y.
,
Wen
,
X.
,
Zhao
,
D.
,
Niu
,
C.
,
Zhao
,
Y.
,
Qi
,
H.
et al (
2018
)
Interactions between the ACT domains and catalytic subunits of acetohydroxyacid synthases (AHASs) from different species
.
Chembiochem
19
,
2387
2394
25
Otwinowski
,
Z.
and
Minor
,
W.
(
1997
)
Processing of X-ray diffraction data collected in oscillation mode
.
Methods Enzymol.
276
,
307
326
26
McCoy
,
A.J.
,
Grosse-Kunstleve
,
R.W.
,
,
P.D.
,
Winn
,
M.D.
,
Storoni
,
L.C.
and
,
R.J.
(
2007
)
Phaser crystallographic software
.
J. Appl. Crystallogr.
40
,
658
674
27
Emsley
,
P.
and
Cowtan
,
K.
(
2004
)
Coot: model-building tools for molecular graphics
.
Acta Crystallogr. D Biol. Crystallogr.
60
,
2126
2132
28
Zwart
,
P.H.
,
Afonine
,
P.V.
,
Grosse-Kunstleve
,
R.W.
,
Hung
,
L.W.
,
Ioerger
,
T.R.
,
McCoy
,
A.J.
et al (
2008
)
Automated structure solution with the PHENIX suite
.
Methods Mol. Biol.
426
,
419
435
29
Lee
,
J.
,
Cheng
,
X.
,
Swails
,
J.M.
,
Yeom
,
M.S.
,
Eastman
,
P.K.
,
Lemkul
,
J.A.
et al (
2016
)
CHARMM-GUI input generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM simulations using the CHARMM36 additive force field
.
J. Chem. Theory Comput.
12
,
405
413
30
Jorgensen
,
W.L.
,
Chandrasekhar
,
J.
,
,
J.D.
,
Impey
,
R.W.
and
Klein
,
M.L.
(
1983
)
Comparison of simple potential functions for simulating liquid water
.
J. Chem. Phys.
79
,
926
935
31
Maier
,
J.A.
,
Martinez
,
C.
,
Kasavajhala
,
K.
,
Wickstrom
,
L.
,
Hauser
,
K.E.
and
Simmerling
,
C.
(
2015
)
ff14SB: Improving the accuracy of protein side chain and backbone parameters from ff99SB
.
J. Chem. Theory Comput.
11
,
3696
3713
32
Wang
,
J.
,
Wolf
,
R.M.
,
Caldwell
,
J.W.
,
Kollman
,
P.A.
and
Case
,
D.A.
(
2004
)
Development and testing of a general amber force field
.
J. Comput. Chem.
25
,
1157
1174
33
Pastor
,
R.W.
,
Brooks
,
B.R.
and
Szabo
,
A.
(
1988
)
An analysis of the accuracy of Langevin and molecular dynamics algorithms
.
Mol. Phys.
65
,
1409
1419
34
Darden
,
T.
,
York
,
D.
and
Pedersen
,
L.
(
1993
)
Particle mesh Ewald: An N log(N) method for Ewald sums in large systems
.
J. Chem. Phys.
98
,
10089
35
Ryckaert
,
J.P.
,
Ciccotti
,
G.
and
Berendsen
,
H.J.C.
(
1977
)
Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes
.
J. Comput. Phys.
23
,
327
341
36
Roux
,
B.
(
1995
)
The calculation of the potential of mean force using computer simulations
.
Comput. Phys. Comm.
91
,
275
282
37
Vyazmensky
,
M.
,
Engel
,
S.
,
Kryukov
,
O.
,
Berkovich-Berger
,
D.
and
Kaplun
,
L.
(
2006
)
Construction of an active acetohydroxyacid synthase I with a flexible linker connecting the catalytic and the regulatory subunits
.
Biochim. Biophys. Acta
1764
,
955
960
38
Robertson
,
R.M.
,
Yao
,
J.
,
Gajewski
,
S.
,
Kumar
,
G.
,
Martin
,
E.W.
,
Rock
,
C.O.
et al (
2017
)
A two-helix motif positions the lysophosphatidic acid acyltransferase active site for catalysis within the membrane bilayer
.
Nat. Struct. Mol. Biol.
24
,
666
671
39
Shen
,
Y.
,
Joachimiak
,
A.
,
Rosner
,
M.R.
and
Tang
,
W.J.
(
2006
)
Structures of human insulin-degrading enzyme reveal a new substrate recognition mechanism
.
Nature
443
,
870
874