Structural Evidence of a Productive Active Site Architecture for an Evolved Quorum-quenching GKL Lactonase

Abstract

The in vitro evolution and engineering of quorum-quenching lactonases with enhanced reactivities was achieved using a thermostable GKL enzyme as template, yielding the E101G/R230C GKL mutant with increased catalytic activity and broadened substrate range [Chow, J. Y., Xue, B., Lee, K. H., Tung, A., Wu, L., Robinson, R. C., and Yew, W. S. (2010) J Biol Chem 285, 40911-40920]. This enzyme possesses the (β/α)₈-barrel fold and is a member of the PLL (Phosphotriesterase-Like Lactonase)-group of enzymes within the amidohydrolase superfamily that hydrolyze N-acyl-homoserine lactones, which mediate the quorum-sensing pathways of bacteria. The structure of the evolved N-butyryl-L-homoserine lactone (substrate)-bound E101G/R230C GKL enzyme was solved, in the presence of the inactivating D266N mutation, to a resolution of 2.2 Å to provide an explanation for the observed rate enhancements; in addition, the substrate-bound structure of the catalytically-inactive E101N/D266N mutant of the manganese-reconstituted enzyme was obtained to a resolution of 2.1 Å; the ligand-free, manganese-reconstituted E101N mutant to a resolution of 2.6 Å; and the structures of ligand-free zinc-reconstituted wild-type, E101N, R230D, and E101G/R230C mutants of GKL to a resolution of 2.1 Å, 2.1 Å, 1.9 Å, and 2.0 Å, respectively. In particular, the structure of the evolved E101G/R230C mutant of GKL provides evidence for a catalytically productive active site architecture that contributes to the observed enhancement in catalysis. At high concentrations, wild-type and mutant GKL enzymes are differentially colored, with absorbance maximums in the range of 512 nm to 553 nm. The structures of the wild-type and mutant GKL provide a tractable link between the origins of the coloration and the charge-transfer complex between the α-cation and Tyr99 within the enzyme active site. Taken together, this study provides evidence for the modulability of enzymatic catalysis through subtle changes in enzyme active site architecture.

Quorum-sensing is an integral part of microbial interaction and is responsible for mediating virulence of pathogenic bacteria (1). Quorum-quenching, an attenuation of the quorum-sensing pathway, has been shown to be an effective anti-virulence strategy (2). We are interested in developing quorum-quenching lactonases as anti-virulence therapeutic agents to modulate quorum-sensing pathways of disease-causing microbes. Previously, we reported the in vitro evolution of a thermostable GKL (quorum-quenching lactonase from Geobacillus kaustophilus) enzyme with increased catalytic efficiencies and a broadened substrate range (3). This GKL enzyme is part of a group of Phosphotriesterase-Like Lactonases (PLLs) which hydrolyze quorum-sensing N-acyl-homoserine lactones (AHLs) (4, 5); members of this group are divergently related to other enzymes of the amidohydrolase superfamily.

The amidohydrolase superfamily is a homologous group of enzymes that catalyze hydrolytic reactions on a broad range of substrates bearing ester or amide functional groups with carbon or phosphorus centers (6). Members of the superfamily bear a conserved mononuclear or binuclear metal center within a (β/α)₈-barrel structural scaffold (7). This metal center is involved in the activation of a water molecule for subsequent nucleophilic attack on an activated scissile bond of the substrate; a resolution of the reaction intermediate results in subsequent hydrolysis. In our attempts to understand the basis for the observed rate enhancements (and broadened substrate range) of the evolved E101G/R230C GKL mutant (the mutations are localized on loops at the C-terminal ends of the third and seventh β-strand, respectively), especially toward the hydrolysis of N-butyryl-L-homoserine lactone (C4-HSL, a substrate that wild-type GKL previously could not hydrolyze), the structure of the evolved mutant was solved in the presence of C4-HSL bound in the enzyme active site and the inactivating D266N mutation. This report provides structural evidence of a catalytically competent active site architecture to account for the catalytic enhancements observed in the evolved mutant. Our study also suggests that the productive geometry of the enzyme active site can be positively modulated through subtle changes (suitable point mutations), reinforcing the need to consider the interplay between active site architecture and catalysis in future enzyme design and engineering efforts.

Materials and Methods

The substrates N-butyryl-DL-homoserine lactone (C4-HSL), N-(3-oxo-hexanoyl)-L-homoserine lactone (3-oxo-C6-HSL), and N-(3-oxo-octanoyl)-L-homoserine lactone (3-oxo-C8-HSL) were purchased from Sigma-Aldrich Co. (St. Louis, MO). All reagents were the highest quality grade commercially available.

Expression and protein purification of wild-type GKL and mutants

Wild-type GKL and mutants were expressed and purified as previously described (3), with the following modifications: briefly, the His-tagged proteins were expressed in E. coli strain BL21(DE3) in LB supplemented with 100 μg/mL of ampicillin; when the cells were grown to an OD_600nm of 0.1, 0.1 mM of 2′-2-bipyridal (Sigma) was added to the culture. The culture was grown to an OD_600nm of 0.6 and 0.1 mM IPTG was added for an additional 16 hr of induction at 37 °C. The cells were harvested and protein was purified by affinity chromatography using a column of chelating Sepharose Fast Flow (GE Healthcare Bio-Sciences Corp.) charged with Ni²⁺. The N-terminal His-tags were removed with thrombin (GE Healthcare Bio-Sciences Corp.) according to the manufacturer's instructions, and the proteins were purified to homogeneity on a Q Sepharose™ High Performance column (GE Healthcare Bio-Sciences Corp.).

Preparation of metal-reconstituted wild-type GKL and mutants

Purified wild-type GKL and mutants were dialyzed against storage buffer (100 mM NaCl, 20 mM Tris-HCl, pH 8.0) containing 0.1 mM 2′-2-bipyridal, followed by dialysis in storage buffer to remove excess chelator. Metal-reconstituted GKL was prepared by dialysis of 2′-2-bipyridal-treated GKL against storage buffer containing 100 μM metal ions (Fe³⁺, Zn²⁺, Mn²⁺, or combinations of two metals, respectively), followed by dialysis in storage buffer to remove unbound metal ions, before Inductively-Coupled Plasma Optical Emission Spectroscopy (ICP-OES) metal-analysis at the Elemental Analysis Laboratory, Department of Chemistry, National University of Singapore.

Kinetic assay of lactonase activities

The lactonase activity of GKL was assayed by a continuous spectrophotometric assay as previously described (5), using a UV-2550 Spectrophotometer (Shimadzu). Briefly, the assay (1 mL at 37 °C) contained GKL, 2.5 mM bicine buffer, pH 8.3, 0.08 mM cresol purple (577 nm, ε = 12,500 M⁻¹ cm⁻¹), 100 μM of metal ion (Zn²⁺ or Mn²⁺, respectively) and 0.025-5.0 mM AHL substrate (substrates were dissolved in DMSO, and regardless of substrate concentration, the final concentration of organic solvent DMSO was maintained at 1%). Initial rates (ν_o) were corrected for the background rate of spontaneous substrate hydrolysis in the absence of enzyme. Background rate of substrate non-enzymatic hydrolysis varies amongst different substrates and concentrations tested, but are typically less than 30 milliAbs per minute, and significantly below observed catalytic rates. Kinetic parameters were determined by fitting the initial rates to the Michaelis-Menten equation using Enzfitter (Biosoft).

Electron Paramagnetic Resonance (EPR) studies

Metal-reconstituted wild-type GKL (Zn²⁺), E101N (Zn²⁺ or Mn²⁺, respectively) and E101G/R230C (Zn²⁺) mutants were concentrated to 50 mg/mL, transferred to quartz EPR tubes and flash frozen in liquid nitrogen. The continuous-wave EPR measurements were performed using an X-band (∼ 9.05 GHz) Varian E-line 12″ spectrometer, equipped with a rectangular TM110 resonator and a helium cryostat. The microwave frequency and magnetic field were measured using the Varian internal frequency meter and field controller, respectively. Spectra were recorded at 10 K and 60 K. Different modulation amplitudes between 0.1 to 1.0 mT were used to detect the signal of the paramagnetic centers unambiguously. Microwave power was between 0.2 - 2 mW to avoid signal saturation.

Crystallization and Data Collection

Frozen stock of wild-type GKL and mutants were diluted to 50 mg/mL with their storage buffers, and C4-HSL was added to the catalytically inactive D266N mutants to a final concentration of 1.0 mM. Crystals grew within 2 weeks at 20 °C by the sitting-drop vapor diffusion method from a 1:1 (v/v) mixture with precipitant solution comprised of 100 mM Tris, pH 7.5, 20% (w/v) PEG 4000, 10% (v/v) glycerol, and 100 μM ZnCl₂ or MnCl₂. Crystals were harvested from the drop, and directly flash frozen in liquid nitrogen. X-ray diffraction data were collected at beamline BL13B1 on an Area Detection System Corporation Quantum-315 CCD detector at the National Synchrotron Research Center (Hsinchu, ROC), with different wavelengths (as summarized in Table 2) and constant temperature at 105K. Data were indexed, scaled and merged in HKL2000 (8).

Table 2.

Data collection, refinement and structure validation statistics.

	wild-type	E101G_R230C	E101G_R230C_D266N	E101N	E101N	E101N_D266N	R230D
PDB ID	4H9U	4H9V	4H9X	4H9Y	4H9Z	4H9T	4HA0
Metals	Zn	Zn	Zn	Zn	Mn	Mn	Zn
Ligand			HL4			HL4
Data collection
Space group	P 1 21 1	P 1 21 1	P 1 21 1	P 1 21 1	P 1 21 1	P 21 21 21	P 2 21 21
Cell dimension
a, b, c (Å)	47.3, 157.1, 4	9.9 51.3, 130.2, 51.3	51.6, 129.2, 51.6	51.5, 128.5, 5	1.5 51.2, 127.8, 5	1.5 78.2, 91.4, 95.7	70, 76.2, 134.8
α, β, γ (°)	90, 116.2, 90	90, 96.3, 90	90, 95.8, 90	90, 95.8, 90	90, 95.3, 90	90, 90, 90	90, 90, 90
Resolution	30 - 2.10	50 - 1.97	50 - 2.20	50 - 2.09	30 - 2.60	30 - 2.10	30 - 1.90
I/sigma	18.5 (2.6)	19.4 (3.8)	36.9 (3.8)	48.9 (6.0)	10.4 (3.7)	31.5 (3.7)	31.1 (3.5)
Rmerge (%)	6.2 (41.8)	6.2 (22.0)	3.0 (27.3)	2.1 (16.6)	11.2 (21.5)	5.2 (44.3)	10.4 (40.6)
Completeness (%)	98 (93)	92 (56)	99 (84)	97 (79)	89 (68)	100 (100)	98 (99)
Redundancy	3.6 (3.0)	3.6 (2.9)	3.7 (2.9)	3.7 (3.2)	3.0 (2.4)	6.0 (5.3)	5.9 (5.9)
Refinement
Resolution range (Å)	19.9 - 2.10	20.0 - 1.97	19.9 - 2.20	19.8 - 2.09	19.9 - 2.60	19.9 - 2.10	20.0 - 1.90
Highest resolution shell (Å)	2.16 - 2.10	2.02 - 1.97	2.27 - 2.20	2.14 - 2.08	2.74 - 2.60	2.15 - 2.10	1.93 - 1.90
Number of reflections	33812	38503	32403	38376	18030	40173	55872
Completeness (%)	89 (37)	82 (33)	95 (60)	97 (80)	89 (56)	98 (80)	97 (98)
Rwork (%)	17.5 (20.4)	18.5 (17.8)	17.8 (22.4)	19.2 (21.6)	18.7 (23.6)	19.1 (20.4)	26.1 (32.1)
Rfree (%)	21.7 (26.6)	22.2 (23.8)	21.5 (31.7)	22.7 (25.8)	24.6 (29.5)	22.3 (25.2)	30.3 (37.2)
Number of atoms
Protein	5086	5014	5014	5024	5084	5034	5080
Ions	6	6	6	6	4	6	6
Ligand			24			24
Water	272	233	189	244	18	341	283
B-factors
Protein	32.5	29.3	38.3	37.7	37.7	25.6	47.3
Ions	17.2	17.9	28.3	31.5	26.4	29.4	32.0
Ligand			53.4			43.1
Water	33.7	29.9	35.6	37.5	19.8	30.5	40.1
R.m.s deviations
Bonds (Å)	0.012	0.008	0.009	0.009	0.009	0.011	0.009
Angles (°)	1.288	1.102	1.174	1.162	1.162	1.263	1.194
Molprobity statistics
All-atom clashscore	6.76	7.55	10.04	8.23	21.27	9.5	15.63
Ramachandran plot
outliers	0.2%	0.0%	0.2%	0.0%	0.2%	0.0%	0.0%
allowed	2.3%	2.6%	2.1%	2.6%	5.1%	1.9%	2.7%
favored	97.5%	97.5%	97.8%	97.5%	94.7%	98.1%	97.4%
Rotamer outliers	1.3%	1.5%	2.5%	1.2%	4.0%	2.1%	3.0%
Cβ deviations	1	1	0	0	0	0	1

Values in parentheses are for the highest-resolution shell.

Structure Determination and Refinement

Structural determination was initiated by molecular replacement using GKL D266N mutant (PDB ID: 3OJG) as a model in PHASER (9). The solution was subjected to repetitive rounds of restrained refinement in PHENIX (10) and manual building in COOT (11). The C4-HSL ligand was then placed into the density in the active sites. The occupancies of the atoms in the ligand were refined as a group, while those of the two metal ions at the active site were refined individually. The chemical nature of the two metal ions was identified with the anomalous X-ray fluorescence method (12). TLS parameters generated by the TLSMD web server (13) were included in the final round of refinement. CCP4 program suite (14) was used for coordinate manipulations. The structures were validated with Molprobity (15). All the structure-related figures were prepared with the PyMOL Molecular Graphics System (DeLano Scientific LLC).

Results

Previously, we have determined that wild-type GKL is a metal-dependent quorum-quenching lactonase with an iron-zinc binuclear metal center within the active site (3). 2′-2-Bipyridal-treated GKL was found to contain 0.1 eq of iron and less than 0.01 eq of zinc per active site. 2′-2-Bipyridal -treated GKL, like non-2′-2-bipyridal-treated GKL, did not have any detectable lactonase activity when assayed in the absence of additional metal ions. GKL exhibited detectable lactonase activity only when reconstitution was made with both Fe³⁺ and Zn²⁺. Metal-reconstituted GKL contained 1.0 eq of iron and 1.1 eq of zinc per active site.

Identification of a Mn²⁺-dependent GKL mutant

When GKL mutants were screened for metal-dependency, we identified a GKL mutant (E101N) that exhibited lactonase activity when reconstituted with manganese. Apart from E101N, other mutants of GKL (including the E101G/R230C mutant) exhibited detectable lactonase activity only upon reconstitution with zinc. In comparison to zinc-reconstituted E101N, the manganese-reconstituted E101N mutant exhibited greater lactonase activity against 3-oxo-C8-HSL, and was able to hydrolyze 3-oxo-C6-HSL and C4-HSL when the zinc-reconstituted form could not (Table 1). Zinc-reconstituted E101N contained 1.0 eq of iron and 1.1 eq of zinc per active site, whilst manganese-reconstituted E101N contained 1.0 eq of iron, 0.6 eq of manganese and 0.1 eq of zinc per active site. Zinc-reconstituted E101G/R230C contained 0.9 eq of iron and 1.0 eq of zinc per active site.

Table 1.

Kinetic parameters of GKL and evolved mutants.

Substrate	GKL			E101G/R230C			R230D
	kcat (s−1)	KM (mM)	kcat/KM (M−1s−1)	kcat (s−1)	KM (mM)	kcat/KM (M−1s−1)	kcat (s−1)	KM (mM)	kcat/KM (M−1s−1)
C4-HSL	NDa	ND	ND	≤ 0.010	b	b	ND	ND	ND
3-Oxo-C6-HSL	c	c	4.0	1.1 ± 0.32	22 ± 12	50	ND	ND	ND
3-Oxo-C8-HSL	0.21 ± 0.10	5.5 ± 2.1	38	2.1 ± 0.07	3.0 ± 0.4	700	≤ 0.026	b	b
Substrate	E101N (Zn2+-reconstituted)			E101N (Mn2+-reconstituted)
	kcat (s−1)	KM (mM)	kcat/KM (M−1s−1)	kcat (s−1)	KM (mM)	kcat/KM (M−1s−1)
C4-HSL	ND	ND	ND	≤ 0.018	b	b
3-Oxo-C6-HSL	ND	ND	ND	c	c	5.3
3-Oxo-C8-HSL	≤ 0.019	b	b	0.21 ± 0.11	4.0 ± 3.2	53

^aND, no detectable activity.

^bApparent k_cat.

^cSaturation kinetics could not be attained.

Structure of GKL mutants E101G/R230C and E101G/R230C/D266N with C4-HSL ligand

Previously, we reported the E101G/R230C mutant of GKL, obtained via directed evolution experiments, which had enhanced quorum-quenching activity (3). This mutant exhibited a broadening of AHL substrate specificity (hydrolyzing C4-HSL, an AHL substrate that wild-type GKL did not hydrolyze), and increased catalytic activities against AHL substrates (kinetic parameters describing lactonase activities of wild-type and mutant GKL are detailed in Table 1). In an effort to obtain a structural explanation for the observed rate enhancements, we determined the structures of GKL mutant E101G/R230C, together with its catalytically inactive variant E101G/R230C/D266N with the quorum-molecule C4-HSL bound in the active site. To facilitate structure comparison, and to determine the chemical nature of metal ions in the binuclear center, the wild-type GKL structure was also determined. The statistics of data processing and refinement are listed in Table 2.

The overall structures of E101G/R230C, E101G/R230C/D266N with bound C4-HSL, and the wild-type are similar to the previously reported D266N mutant of GKL (3) and organophosphorus hydrolase from Geobacillus stearothermophilus (PDB ID: 3F4D) (16). The biological assembly is a dimer that occupies the asymmetric unit. A major difference lies in the β3-loop region, which is visible in the wild-type and the D266N mutant, but becomes disordered in E101G/R230C and E101G/R230C/D266N. In both the D266N mutant and the wild-type, the β3-loop is stabilized by the interaction between the side-chain of Glu101 with those of Thr154 and Gln110. The E101G mutation abolishes these interactions, and hence, the loop is more mobile.

As observed in the D266N mutant, within the active site, which is located at the C-terminal end of the barrel, there are two metal ions bound at the α- and β-sites, corresponding to the more solvent occluded and the more solvent exposed cation positions, respectively. Using X-rays to scan for characteristic fluorescence absorption edges and inspection of anomalous difference Fourier electron density maps (together with ICP-OES and EPR data presented in the later section), we have determined that the α-cation is occupied by Fe³⁺, and is coordinated by His23 and His25 at the end of the first β-strand, the carboxylated Lys145 at the end of the fourth β-strand, and Asp266 (Asn266 for the corresponding E101G/R230C/D266N GKL mutant) at the end of the eighth β-strand. The β-cation position is occupied by Zn²⁺, and is coordinated by His178 and His206 at the ends of the fifth and sixth β-strands, respectively, and by the carboxylated Lys145 at the end of the fourth β-strand. A bridging hydroxide ion serves as an additional ligand for both cations at about equal interatomic distances (Figure 1).

Figure 1.

Structures of GKL wild-type, E101G/R230C, and E101G/R230C/D266N with bound C4-HSL. (a) Comparison between ligand-bound E101G/R230C/D266N with previously published D266N (PDB ID: 3OJG), with D266N in green and E101G/R230C/D266N in magenta. The α-cation and the β-cation are depicted as cyan and orange spheres, respectively. The bridging hydroxide and the bound C4-HSL ligand are depicted as a sphere and sticks, respectively, with their colors following that of the peptide chain. The β3-loop in D266N is highlighted blue, which becomes disordered in E101G/R230C/D266N. (b) Close-up of the active site, highlighting changes in the bound ligand, and residues 99, 230, and 266. (c, d) Interatomic distances and facial selectivity in D266N (c) and E101G/R230C/D266N (d). The difference electron density map (F_o-F_c) in (d), calculated by omitting the ligand and contoured at 3σ, is shown as a green mesh. (e, f) Bijvoet difference Fourier maps of wild-type GKL (e) and E101G/R230C (f) corresponding to data collected at 1.258 (yellow, with strong Zn-fluorescence), 1.311 (red, without Zn-fluorescence), 1.696 (blue, with strong Fe-fluorescence), and 1.794 Å (no visible peak, without Fe-fluorescence). All the maps are contoured at 3σ. These results clearly indicate that in both the wild-type and E101G/R230C the α- and β-cation sites are mainly occupied by Fe and Zn ions, respectively.

A change in the position and orientation of the bound C4-HSL becomes obvious when comparing E101G/R230C/D266N with previously published D266N. In E101G/R230C/D266N, the C4-HSL is positioned closer to the binuclear metal center, and the lactone ring is rotated by almost 90°. This change can be at least partly attributed to the R230C mutation. In the D266N mutant, the side chain of Arg230, which adopts the same conformation as in the wild-type, is involved in positioning the bound C4-HSL by hydrogen-bonding to the carbonyl-oxygen on the lactone ring. With the loss of this interaction in E101G/R230C/D266N, the bound C4-HSL is less confined in its orientation. Consequently, the distance from the carbonyl-carbon of the lactone ring to the attacking hydroxide nucleophile and the β-cation are both shortened (3.4 and 5.4 Å, respectively, in D266N versus 3.1 and 5.0 Å in E101G/R230C/D266N), and the nucleophilic attack angle (between the nucleophile and the plane of the lactone ring formed by the carbonyl-carbon and the oxo-ring prior to the formation of the tetrahedral intermediate) changes from 28.9° in D266N to 51.0° in E101G/R230C/D266N.

Structure of GKL E101N mutants and Mn²⁺-reconstituted E101N/D266N with C4-HSL

Mn²⁺-reconstituted E101N mutant of GKL exhibited enhanced quorum-quenching activity over wild-type GKL; this mutant, like the E101G/R230C mutant, exhibited a broadening of AHL substrate specificity, hydrolyzing C4-HSL, an AHL substrate that wild-type GKL did not hydrolyze. To elucidate the observed enhanced quorum-quenching activity, we solved the structures of both Mn²⁺- and Zn²⁺-reconstituted E101N, and Mn²⁺-reconstituted, catalytically inactive mutant E101N/D266N with bound C4-HSL (Figure 2). Overall, these structures are similar to the other structures reported in this study. Similar to the E101G mutation, replacing Glu101 in the β3-loop with Asn increases the mobility of this loop in the Zn²⁺-reconstituted E101N, rendering it invisible in the electron density maps. Interestingly, the same loop becomes ordered in the Mn²⁺-reconstituted E101N, but regains its mobility when C4-HSL is bound, as observed in the structure of the Mn²⁺-reconstituted E101N/D266N with bound C4-HSL. A closer look at the binuclear center of Mn²⁺-reconstituted E101N reveals dramatic changes in metal coordination. When Zn²⁺ occupies the β-cation position, the bridging hydroxide is located approximately equidistant (2.2 - 2.6 Å) from the two metal ions. While in the Mn²⁺-reconstituted E101N, it resides 2.9 and 3.5 Å away from the α- and β-cation. Furthermore, the side chain of Tyr99 moves closer to the β-cation, with a distance of 2.9 Å between the hydroxyl-oxygen of Tyr99 and the Mn²⁺ (Table 3). Upon C4-HSL binding, this distance becomes 6.3 Å, as seen in the E101N/D266N with bound C4-HSL.

Figure 2.

Structures of Mn²⁺-reconstituted GKL mutants E101N and E101N/D266N with bound C4-HSL. (a) Overlay of chain A from ligand-free E101N (green) with the two chains from ligand-bound E101N/D266N. For clarity, the whole chain B of E101N/D266N is shown and colored magenta, while only the β7-loop and the bound-ligand from chain A are shown and colored blue. The α-cation and the β-cation are depicted as cyan and orange spheres, respectively. The ligand and water molecules near the active center are depicted as sticks and spheres, respectively, with their colors following that of the peptide chain. The β3-loop in E101N is highlighted blue. (b) Close-up of the active site, highlighting conformational changes in residues Tyr99 and Arg230, as well as the bound ligands. (c, d) Difference electron density map (green mesh) of the bound ligands in chains B (c) and A (d) of E101N/D266N. The maps are calculated by omitting the ligand and are contoured at 3σ.

Table 3.

Color properties and active site architecture of wild-type and mutants of GKL.

Mutation (reconstitution)	Absorbance maximum (nm) / Molar Extinction Coefficient (M−1. cm−1)	Color	β3-loop	Distance (Å)			Occupancy
				α - β	Y99 - α	Y99 - β	α	β
Wild-type GKL (Fe3+-Zn2+)	545 / 337	Purple	ordered	3.6 (Fe-Zn)	6.2	4.6	1.0	1.0
E101G / R230C (Fe3+-Zn2+)	512 / 376	Purple	disordered	3.4 (Fe-Zn)	6.6	5.4	1.0	0.6
E101G / R230C / D266N (Fe3+-Zn2+) + C4-HSL	N.D.	Purple	disordered	3.6 (Fe-Zn)	7.2	5.6	1.0	0.7
E101N (Fe3+-Zn2+)	535 / 435	Dark Brown	disordered	3.6 (Fe-Zn)	6.7	5.2	1.0	0.9
E101N (Fe3+-Mn2+)	531 / 733	Dark Brown	ordered	3.2 (Fe-Mn)	4.8	2.9	1.0	1.0
E101N / D266N (Fe3+-Mn2+) + C4-HSL	N.D.	Dark Brown	disordered	3.5 (Fe-Mn)	7.9	6.2	0.9	0.5
R230D (Fe3+-Zn2+)	553 / 1066	Dark Purple	ordered	3.4 (Fe-Zn)	5.0	3.5	1.0	0.8

α: α-cation; β: β-cation; N.D.: not determined

The two monomers within the dimeric biological assembly are almost identical in all the structures reported here, except for the Mn²⁺-reconstituted E101N/D266N with bound C4-HSL, where the β7-loop adopts two different conformations - a “closed” conformation that is seen in most of the structures, and a “open” conformation that is unique to one of the monomers. Correspondingly, the bound C4-HSL in the monomer with the “closed” β7-loop resembles that in E101G/R230C/D266N, while the substrate in the monomer with the “open” β7-loop is more similar in conformation to the bound C4-HSL in D266N. Contrary to what is observed in the D266N mutant, and independent of the conformation of the β7-loop, the side chain of Arg230 adopts a conformation that is away from the binuclear center, and does not have strong interaction with the carbonyl-oxygen on the lactone ring of C4-HSL.

Charge-transfer in the wild-type and mutants of GKL

Wild-type and mutants of GKL exhibited intense coloration at high concentration (at approximately 50 mg/mL) (3). We previously attributed this phenomenon to a charge-transfer complex between the active site tyrosine (Tyr99) and the bound iron (α-cation). This property was observed for Dr0930 (17), purple acid phosphatase (18) and uteroferrin (19), in which a charge-transfer complex between an active site tyrosine residue and the iron cation (in previously reported complexes) within the binuclear metal center was reportedly responsible for the observed purple coloration. In our attempts to arrive at a structural explanation for the observed charge-transfer complexes, the structure of GKL mutant R230D was solved in addition to the structures mentioned earlier. Wavelength scans of the various GKL mutants are presented in the Supporting Information.

A superposition of all the structures reveals remarkable (and expected) similarity in the overall structures. A close inspection of the active sites of the superposition, however, reveals significant perturbation to the position of the active site Tyr99, and hence, to the overall geometry of the charge-transfer complex (Figure 3). As will be further discussed, mutations to amino acids 101 and 230, as well as substitution of zinc for manganese at the β-site, resulted in changes to the overall geometry of the active sites, contributing to the observed changes in the charge-transfer mutants. The properties of these charge-transfer mutants, including interatomic distances of the charge-transfer complexes, are listed in Table 3.

Figure 3.

Superposition of active site of wild-type GKL (light green), E101G/R230C (yellow), E101N (green), R230D (light blue), Mn²⁺-reconstituted E101N without (pink) and with (magenta) bound ligand, and the hydrolase from Geobacillus stearothermophilus (blue, PDB ID: 3F4D). The α-cation and the β-cation are depicted as cyan and orange spheres, respectively. Modulation of the charge-transfer complexes between Tyr99 and the β-cation is localized to the (re)positioning of Tyr99.

X-band EPR spectroscopy of wild-type and mutants of GKL

Frozen solutions of metal-reconstituted wild-type GKL (Zn²⁺), E101N (Zn²⁺ or Mn²⁺, respectively) and E101G/R230C (Zn²⁺) mutants were studied by X-band EPR spectroscopy to ascertain the type and oxidation state of the metal ions in the respective binuclear centers. EPR spectra of Zn²⁺-reconstituted wild-type GKL and mutants (E101G/R230C and E101N, respectively) reveal an intense signal for a framework Fe³⁺ site with g = 4.3 and several other much weaker signals (Figure 4). The position and shape of the g = 4.3 signal is characteristic of uncoupled, high-spin Fe³⁺ centers in rhombically distorted environments and for Fe³⁺ in binuclear Fe³⁺-Zn²⁺ centers (20-23), and results from the $m_s=\pm \stackrel{3}{}/\underset{2}{}$ state. The minor signals at g ∼ 5.2 and 8.5 can be produced by the other state of the iron or cluster (24). The weak resonances in the g ∼2.0-2.27 area are also part of Fe³⁺ in a different environment (such as varying rhombic distortion within the binuclear center). The later resonances overlap with the multicomponent signal pertaining to Mn²⁺. It is well observed in the Zn²⁺-reconstituted samples of wild-type GKL and the E101N mutant (six lines with the splitting of ∼7.8-8.8 mT), but not the E101G/R230C mutant. We attribute this signal to “contaminating” Mn²⁺ in the β-site (we cannot rigorously exclude other metals, including Mn²⁺, from our buffers; however, we have shown that wild-type and mutants of GKL, except E101N, are only active upon reconstitution with Zn²⁺).

Figure 4.

X-band EPR spectra of Zn²⁺-reconstituted wild-type GKL (a, b), E101N GKL mutant (c, d) and E101G/R230C GKL mutant (e, f). Temperature: 11 K (a, b) and 10 K (c-f); microwave frequency 9.0419 GHz (a), 9.0459GHz (b), 9.0416 GHz (c), 9.0444 GHz (d), 9.0494 GHz (e), and 9.0434 GHz (f); microwave power 2mW (a-f).

The spectrum of the Mn²⁺-reconstituted E101N mutant of GKL also displayed the framework Fe³⁺ site signals (Figure 5). However, the shape of the line at g = 4.3 has changed and can be considered as a superposition of narrow and broad components. Another noteworthy peculiarity of this spectrum is the appearance of new intensive signals in the g ∼ 2.0 area. The spectrum shows hyperfine structure consisting of ∼17 components with the splittings ∼8.3 mT at the low- and high-field edges. Six lines in the middle of the spectrum possess higher intensity, and are indicative of the presence of Mn²⁺ (six line hyperfine structure is from ⁵⁵Mn with nuclear spin $\mathit{\text{I}}=\stackrel{5}{}/\underset{2}{}$). These six lines remained in the spectrum when the temperature was increased to 60 K, while other components of the hyperfine structure were not observed at this temperature. Our observations suggest that the multicomponent spectrum can be attributed to an overlap of the Mn²⁺ spectrum (six-line pattern) and the spectrum of the binuclear Mn cluster. Comparison of the hyperfine structure and total width of the cluster contribution to the spectrum with data available in the literature suggested a mixed-valence Mn²⁺-Mn³⁺ binuclear cluster (25). In addition, the presence of a broad background line with width ∼200 mT may indicate the presence of a Mn²⁺-Mn²⁺ cluster in the sample (26).

Figure 5.

X-band EPR spectrum of the Mn²⁺-reconstituted E101N mutant of GKL (a), with high-field (b) and low-field (c) part of this spectrum. Temperature: 10 K and 60 K; microwave frequency 9.0423 GHz at 10 K, and 9.0414 GHz at 60 K; microwave power 2mW.

In summary, our EPR spectroscopy data suggest that upon Zn²⁺- or Mn²⁺-reconstitution, several types and combinations of mono- and dinuclear sites (including mixed and EPR-silent sites) may exist within the active sites of the respective wild-type and mutant GKL enzymes.

Discussion

In our efforts to develop suitable anti-virulence therapeutic agents for use in biomedical applications to address nosocomial bacterial infections, we obtained an in vitro evolved thermostable quorum-quenching lactonase (E101G/R230C mutant of GKL) that exhibited enhanced catalytic activities and broadened substrate range over the wild-type template (3). The E101G/R230C mutant exhibited an increase in catalytic activity for AHL substrates; substrates tested include N-butyryl-DL-homoserine lactone (C4-HSL), N-hexanoyl-DL-homoserine lactone (C6-HSL), N-(3-oxo-hexanoyl)-L-homoserine lactone (3-oxo-C6-HSL), N-octanoyl-DL-homoserine lactone (C8-HSL), N-(3-oxo-octanoyl)-L-homoserine lactone (3-oxo-C8-HSL), and N-decanoyl-DL-homoserine lactone (C10-HSL). The chemical structures of these substrates are represented in the Supporting Information (Figure S1). In comparison to wild-type GKL, this double mutant had increased k_cat values, decreased K_M values, and overall increases in catalytic efficiencies (k_cat/K_M) of 1.2 to 32-fold (for C6-HSL and 3-oxo-C12-HSL, respectively). Most noteworthy, the double mutation resulted in a broadening of AHL substrate specificity, allowing the quorum-quenching lactonase to hydrolyze C4-HSL, an AHL substrate that the wild-type GKL did not previously hydrolyze. We sought to obtain a structural explanation for the observed enhanced reactivity of the E101G/R230C mutant to facilitate future catalytic design and engineering efforts.

EPR spectroscopy of wild-type and mutants of GKL identifies different types of dinuclear metal sites

EPR spectroscopy can provide information about the type and oxidation state of paramagnetic metal ions as well as the spin-coupling between metals in binuclear centers. For Zn²⁺-reconstituted wild-type and mutants of GKL (including E101G/R230C), the α-cation position is occupied by high-spin Fe³⁺, while the β-cation position is presumably occupied by Zn²⁺ (we do not have “direct” EPR evidence as zinc is not paramagnetic), and not by another paramagnetic metal ion such as Mn²⁺. Together with ICP-OES and EXAFS data, the binuclear active site of Zn²⁺-reconstituted GKL enzymes is represented by a Fe³⁺-Zn²⁺ center in $\mathit{\text{S}}=\stackrel{5}{}/\underset{2}{}$ high-spin state, as illustrated in Figure 1. For the E101N-Mn mutant, our EPR data, in addition to high-spin Fe³⁺ state(s), reveal the presence of new spectra which could be assigned to Mn³⁺-Mn²⁺ and Mn²⁺-Mn²⁺ binuclear centers, based on previously reported studies of Mn-Mn binuclear centers in proteins and model complexes. The presence of the EPR silent, strongly antiferromagnetically coupled Fe³⁺-Mn²⁺ center with S=0 (27) can also be suggested (and can be detected by other methods). Although the extent of contribution of the respective Fe-Mn and Mn-Mn binuclear centers to activity is unclear, the fact remains that the effect of the E101N mutation, manifested through a change in the productive geometry of the charge-transfer complex, resulted in Mn²⁺-dependent lactonase activity (a metal-dependency not observed with wild-type GKL or with the evolved E101G/R230C mutant).

Structural evidence for rate enhancement of evolved GKL mutants

The hydrolysis of AHL substrates by GKL is mediated by both the orientation of the substrate within the active site and the catalytic cycle that ensues: upon substrate binding, the carbonyl oxygen of the lactone ring would be polarized by interaction with the β-cation. A facial selectivity of the re-face of the lactone ring, consistent with observed facial orientations of the amide or ester bonds of substrates of other amidohydrolase superfamily members (such as dihydroorotase and isoaspartyl dipeptidase), would orientate the scissile bond toward the hydroxide nucleophile that bridges the two metal ions. Nucleophilic attack by the bridging hydroxide would result in the formation of a tetrahedral intermediate, followed by the subsequent transfer of a proton from the hydroxide to Asp266, to yield the hydrolyzed acylhomoserine, resulting in a quenching of quorum-signaling.

The catalytic efficiency of the cycle is dependent on a number of factors: these include 1) the ease of binding of the substrate within the active site; 2) the nucleophilicity of the bridging hydroxide; 3) the productive geometry presented within the active site architecture vis a vis the orientation of the scissile bond toward the attacking nucleophile; and 4) the resolution of the tetrahedral intermediate and subsequent ease of product release from the active site. Upon inspection of the pre-evolved GKL structure (represented by the C4-HSL-bound D266N GKL structure, PDB ID: 3OJG), an explanation for the observed low catalytic AHL lactonase activities (or no detectable lactonase activity in the case of C4-HSL) is suggested. Successful formation of the tetrahedral intermediate during hydrolysis depends upon the productive geometry presented during nucleophilic attack; an efficient catalysis would involve an orthogonal (107°) angle of nucleophilic attack on the plane of the scissile bond (28). In the case of wild-type GKL, the nucleophilic attack angle (represented by the angle formed by the nucleophile, the carbonyl carbon C2 of the lactone ring, and the carbonyl oxygen O6 of the lactone) is 82.7°, a large deviation from the “perfect” orthogonal (107°) angle (Figure 1C); in addition, the position of the proposed nucleophile is almost in the plane of the lactone ring, and thus, the inability of wild-type GKL to hydrolyze C4-HSL is thus attributed to this unproductive geometry presented in the active site.

Previously, we proposed that the rate enhancements observed in AHL reactivity in the evolved mutant of GKL (we were unable to obtain suitable crystals of the E101N/R230I GKL mutant for structural studies; we report the catalytically equivalent E101G/R230C GKL mutant in this study) could be attributed to the combined effects of an altered positioning of the lactone ring and a modulation of the attacking nucleophile toward the catalytically required/preferred orthogonal positioning to the scissile bond (Figure 1). An inspection of the evolved E101G/R230C mutant GKL structure (represented by the C4-HSL-bound E101G/R230C/D266N mutant GKL structure) corroborated the proposal for rate enhancement: an altered positioning of the lactone ring and a modulation of the position of the attacking hydroxide nucleophile led to an observed change in the nucleophilic attack angle of 114.5° (Figure 1D). Although this angle defers from the orthogonal (107°) angle, it is considerably closer to the orthogonal angle, relative to the geometry presented in wild-type GKL! A superposition of the structure of the active E101G/R230C mutant with the inactive E101G/R230C/D266N mutant showed a virtually identical active site (rmsd of 0.248 Å); in addition, an observed modulation in the position of the proposed attacking nucleophile, from an almost in-plane position relative to the lactone ring (for wild-type GKL) to a more productive position of 51.0° relative to the plane of the lactone ring (for the evolved mutant), suggested that the observed C4-HSL-bound E101G/R230C/D266N mutant GKL structure is a good approximation of the evolved E101G/R230C mutant GKL structure.

A change in facial selectivity of the lactone ring vis a vis orientation of the scissile bond toward the hydroxide nucleophile, from re-face (observed in wild-type GKL) to si-face (observed in the evolved E101G/R230C mutant), was noted (Figure 1). This facial selectivity is opposite to the observed facial orientations of the amide or ester bonds of substrates or inhibitors in the active sites of other amidohydrolase superfamily members such as dihydroorotase (29) and isoaspartyl dipeptidase (30). Although we are unable to comment on the effect of facial selectivity on catalysis, we noticed that the E101G/R230C mutation brought about a total translation of 79.9° (28.9° re-face and 51.0° si-face) for the attacking nucleophile, compared to wild-type (pre-evolved) GKL. A switch in facial selectivity of the lactone ring was also observed for the catalytically-competent Mn²⁺-reconstituted E101N mutant of GKL (si-face, with an average angle of 29.1°).

Based on the work reported here and in our earlier paper (3), as well as homologous structures (16, 17, 31) reported by others, a few key residues are highlighted. One such residue identified through random mutagenesis is Glu101, a remote residue from the active center (3). As mentioned earlier, its side chain interacts with those of Gln110 and Thr154 from the same polypeptide chain. Substitution of this residue by Asn or Gly will abolish the former or both of these interactions, making the β3-loop more mobile, as evidenced by the absence of this loop in most of the structures bearing such mutations. The only exception observed so far is the Mn²⁺-reconstituted E101N, with the loop adopting the same conformation as the wild-type. However, this is accompanied by a much shorter distance between the hydroxyl group of Tyr99 and the β-cation. By considering all the data listed in Table 3, a coupling between this distance and the flexibility of the β3-loop is evident. The ordering of the β3-loop in the Mn²⁺-reconstituted E101N can thus be attributed to the tight tethering of Tyr 99 by the β-cation, Mn²⁺ in this case, rather than the built-in stabilizing effect by the Glu101, which it harbors.

Another key residue identified by our previously reported random mutagenesis approach is Arg230 (3). It has been shown previously that mutation at corresponding residues in some homologous proteins alters their enzymatic activities, such as the R228A mutation of Dr0930 (31), and the H254G mutation of phosphotriesterase (32). In the wild-type GKL, the side chain of Arg230 extends close to the active center, with its Nη2 nitrogen sitting 3.8 Å away from the β-cation, and forms a hydrogen bond to the catalytically critical residue Asp266 with its Nε nitrogen (Figure 1). Arg230 is also involved in positioning the bound C4-HSL in an unfavorable orientation, which it does without the need of changing conformation, in the D266N mutant. These factors may contribute to wild-type GKL's non-detectable activity towards C4-HSL (Table 1). In contrast, in mutants that are active towards C4-HSL, this residue is either mutated, as in the E101G/R230C (Figure 1) and E101N/R230I (no structure available), or moves away from the active center, as in the Mn²⁺-reconstituted E101N without or with bound C4-HSL (Figure 2). It is perceivable that such changes would be beneficial to the enzyme's activity based on the catalysis mechanism mentioned earlier, where the β-cation polarizes the carbonyl oxygen of the lactone ring and makes the carbonyl carbon more susceptible for nucleophilic attack by the bridging hydroxide. Removing the side chain of Arg230, which acts as a hydrogen-donor to Asp266 and probably an electron-donor to the β-cation, from the active center would enhance the polarization effect of the β-cation on the carbonyl oxygen, and promote the proton transfer from the bridging hydroxide to Asp266.

Tyr99 is the most perturbed residue at the active center (Figure 3). The distance between the hydroxyl group of Tyr99 and the β-cation depends on the chemical nature of the β-cation, the flexibility of the β3-loop next to Tyr99, and the presence/absence of ligand at the active center (Table 3). A similar trend has been observed for the highly homologous Dr0930 (17, 31). For the enzymatically active mutants, binding of C4-HSL invariably coincides with an increase in this characteristic distance, which is more pronounced when the β-cation position is occupied by Mn²⁺. In analogy to the effect of the conformational change of Arg230, removal of the electron-rich hydroxyl group of Tyr99 from the β-cation would enhance its polarization effect on the carbonyl oxygen.

As has been clearly demonstrated (Table 1), the presence of the “correct” cation within the binuclear center is critical for the enzymatic activity of GKL towards AHL. A survey of homologous structures in the PDB reveals a common feature that the occupancy of the two metal cations, especially that of β-cation site, is less than unity, which is also the case for the structures reported here (Table 3). Interestingly, for the mutants that lose their Arg230, the occupancy of the β-cation is lower than those without such mutation. Furthermore, among the ligand-free structures, the least occupied β-site is from the E101G/R230C mutant, which also has the largest distance between Tyr99 and the β-cation (Table 3). All these suggest that prior to ligand binding, both Arg230 and Tyr99 interact with the β-cation, acting as either sensors or determinants of the chemical nature and the occupancy of the β-cation.

Taken together, a structural explanation for the observed enzymatic activities of GKL and its mutants towards homoserine lactones can be drafted. In the wild-type GKL, Arg230 makes the reaction less favorable by interacting with Asp266, the β-cation, and by positioning the ligand in a less optimal orientation. This effect can be alleviated by mutating this residue, or by changing the chemical nature and the occupancy of the β-cation. Another determinant for the enzymatic activity is the ease of movement of Tyr99 away from the β-cation, which is again affected by the identity and the occupancy of the β-cation, as well as the flexibility of the adjacent β3-loop that in turn can be enhanced by mutating Glu101 within the loop. These two kinds of modulations are singly not sufficient for optimal enzymatic activities towards AHL, as manifested by the Zn²⁺-reconstituted E101N or R230D mutants. When combined together, however, they result in enhanced activity of GKL, as exemplified by the E101G/R230C, E101N/R230I, and Mn²⁺-reconstituted E101N mutants.

Although this study provides unequivocal evidence for a catalytically productive active site architecture of an evolved GKL that resulted in the observed increase in catalysis, the exact nucleophilicity of the bridging/attacking hydroxide is still unclear and the extent of the contributions of the binuclear metal center on overall catalysis is still open to question. We are currently performing high-resolution EPR studies to obtain further insights into the role of the binuclear metal center (and the nucleophilicity of the attacking hydroxide) on catalysis.

Implications for enzymatic design and catalysis

This study provides evidence that enhancement of enzyme activity can be brought about by subtle (two point mutations), “remote” changes (Glu101 is located on a loop remote from the central catalytic barrel) that favor catalytically productive geometries during catalysis. The challenge in future catalytic design and engineering efforts will involve the incorporation of predictive algorithms into rational design protocols to fashion suitable modulations of the active site architecture that take into account the productive geometries required for chemical catalysis.

Abbreviations

AHL

N-acyl-homoserine lactone

GKL

lactonase from Geobacillus kaustophilus

Dr0930

lactonase from Deinococcus radiodurans

IPTG

isopropyl D-thiogalactopyranoside

ICP-OES

inductively coupled plasma optical emission spectroscopy

EPR

Electron Paramagnetic Resonance