Identification of a Histidine Metal Ligand in the argE-Encoded N-Acetyl-L-Ornithine Deacetylase from Escherichia coli

The H355A, H355K, H80A, and H80K mutant enzymes of the argE-encoded N-acetyl-L-ornithine deacetylase (ArgE) from Escherichia coli were prepared, however, only the H355A enzyme was found to be soluble. Kinetic analysis of the Co(II)-loaded H355A exhibited activity levels that were 380-fold less than Co(II)-loaded WT ArgE. Electronic absorption spectra of Co(II)-loaded H355A-ArgE indicate that the bound Co(II) ion resides in a distorted, five-coordinate environment and Isothermal Titration Calorimetry (ITC) data for Zn(II) binding to the H355A enzyme provided a dissociation constant (Kd) of 39 μM. A three-dimensional homology model of ArgE was generated using the X-ray crystal structure of the dapE-encoded N-succinyl-L,L-diaminopimelic acid desuccinylase (DapE) from Haemophilus influenzae confirming the assignment of H355 as well as H80 as active site ligands.

the production of polyamines involved in DNA replication and cell division, the L-NAO deacetylation step is critical for bacterial proliferation (Girodeau et al. 1986). Indeed, when an arginine auxotrophic bacterial strain void of L-NAO deacetylase activity was transformed with a plasmid containing argE, an Arg + phenotype resulted (Meinnel et al. 1992). However, when the start codon (ATG) of argE in the same plasmid was changed to the Amber codon (TAG), the resultant plasmid was unable to relieve arginine auxotrophy in the same cell strain indicating that ArgE is required for cell viability. Given the fact that ArgE is only found in prokaryotes and is required for bacterial cell growth and proliferation, it represents a potential enzymatic target for the development of a new class of antimicrobial agents (Girodeau et al. 1986).
At the present time, no X-ray crystallographic data is available for any ArgE enzyme but it shares sequence homology with several dinuclear Zn(II) metallopeptidases in the M28 family and biochemical studies revealed that the ArgE from E. coli is a Zn(II) containing enzyme (Javid-Majd and Blanchard 2000;McGregor et al. 2005;Holz 2002;Holz et al. 2003). Based on sequence alignments of the ArgE from E. coli with the aminopeptidase from Aeromonas proteolytica (AAP) (Desmarais et al. 2006), the dapE-encoded N-succinyl-L,L-diaminopimelic acid desuccinylase (DapE) (Nocek et al. 2010), the N-acetyl-L-citrulline deacetylase (ACD) from Xanthomonas campestris (Shi et al. 2007), and the carboxypeptidase from Pseudomonas sp strain-RS-16 (CPG2) (Rowsell et al. 1997), the residues that function as ligands in the dinuclear active site of AAP, DapE, ACD, and CPG 2 are strictly conserved in ArgE ( Figure 2) (Born et al. 1998;Chevrier et al. 1994;Rowsell et al. 1997). Since AAP, DapE, and CPG 2 possess a (μ-aquo)(μ-carboxylato)dizinc(II) core with one terminal carboxylate and one histidine residue at each metal site, a similar active site was proposed for ArgE (Tao et al. 2012).
In an effort to provide insight into the structural properties of each divalent metal ion in the active site of the ArgE form E. coli, we have prepared and purified the H355A, H355K, H80A, and H80K ArgE mutant enzymes. Figure 1 Reaction catalyzed by ArgE. Figure 2 Sequence alignment between the structurally characterized enzymes DapE from H. influenzae, ACD from X. campestris, and CPG 2 from Pseudomonas sp. strain RS-16 with the ArgE from E. coli.
Only the H355A mutant ArgE enzyme was expressed as a soluble enzyme and, consequently, it was the only ArgE mutant investigated via kinetic, electronic absorption, and isothermal titration calorimentry (ITC). A three-dimensional homology model of ArgE was also generated using the X-ray crystal structure of the DapE from Haemophilus influenzae. Combination of these data provides verification that H355 is an active site ligand and provide new insight into the divalent metal binding properties of co-catalytic metallohydrolase active sites.

Reagents
All chemicals used were purchased commercially and were of the highest quality available. All metal reagents were obtained from Strem Chemicals, Newburyport, MA with ≥99.999% purity. Stock solutions of Co (II) (50 mM) and Zn(II) (10 mM) were prepared by dissolving the metal chloride salts (CoCl 2 . 6H 2 O or ZnCl 2 ) in nanopure water.

Enzyme purification
The aminopeptidase from Aeromonas proteolytica (AAP) was purified according to the previously published procedure (Chen et al. 1997). Protein concentrations were estimated from the absorbance at 280 nm using an extinction coefficient ε 280 = 41,800 M -1 cm -1 (Bennett and Holz 1997). Wild-type recombinant ArgE from E. coli was expressed and purified, as previously described, from a stock culture of E. coli BL21 Star™ cells (Invitrogen, Carlsbad, CA) kindly provided by Professor John Blanchard (Javid-Majd and Blanchard 2000). Pure ArgE enzyme (> 98%), as determined by SDS-PAGE gel electrophoresis, was stored at −80°C until used. The concentration of ArgE was estimated from the absorbance at 280 nm using an extinction coefficient ε 280 = 41,250 M -1 cm -1 (McGregor et al. 2005). Enzyme concentrations determined by molar absorptivity were in close agreement to that obtained using a Bio-Rad assay.

Site-Directed Mutagenesis
The pET-27a(+) plasmid containing the argE gene was extracted and transformed into E. coli BL21 Star™ cells for expression of protein variants, which were purified in the same manner as WT ArgE. The Quick Change™ Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) and the following primers: 5'-GGC TCA ATT AAT CAG GCT XXX CAA CCT GAT GAA TAT CTG G-3'and 5'-C CAG ATA TTC ATC AGG TTG YYY AGC CTG ATT AAT TGA GCC-3' with GCT and AAA for XXX and AGC and TTT for YYY were employed for production of the H355A and H355K variants respectively. Likewise, 5'-G CTG GCG GGG XXX ACC GAT ACG GTG CC-3' and 5'-GG CAC CGT ATC GGT YYY CCC CGC CAG C-3' with GCT and AAA for XXX and AGC and TTT for YYY were employed for production of H80A and H80K variants respectively. Products from these Site-Directed Mutagenesis reactions were transformed into E. coli XL1-Blue competent cells which were subsequently spread on LB-agar plates containing 100 μg/mL of Ampicillin. Following over-night incubation at 37°C, one of many colonies from each plate was selected for growth in 50 mL liquid LB medium, containing 100 μg/mL of Ampicillin, for ≥10 hours at 37°C and shaking at 225 rpm. Plasmids were isolated from the resultant cultures using the Qiaprep® Spin Miniprep kit (Qiagen, Valencia, CA) and incorporation of the variant gene sequence was confirmed by DNA sequencing.

Enzymatic assay of ArgE
ArgE activity was measured by the method of Javid-Majd and Blanchard (Javid-Majd and Blanchard 2000). In this assay, the hydrolysis of a 2 mM L-NAO solution, in 50 mM Chelex-100 treated phosphate buffer at pH 7.5, was measured spectrophotometrically at 25°C as the decrease in absorbance at 214 nm (Δε 214 = 103 M -1 cm -1 ), corresponding to the cleavage of the L-NAO amide bond. Protein concentrations were determined using the theoretical value ε 280 = 41,250 M -1 cm -1 (Gill and von Hippel 1989). The specific activity of purified wild-type ArgE with L-NAO was typically found to be 2,000 units per mg of enzyme. One unit was defined as the amount of enzyme that releases 1 μmole of ornithine in 60 sec at 25°C. Initial rates were fit directly to the Michaelis-Menten equation to obtain the catalytic constants K m and k cat .

Preparation of Apo-enzymes
Apo-enzyme samples were prepared for each purified enzyme by methods previously described in the literature (Prescott et al. 1983;Javid-Majd and Blanchard 2000). Briefly, AAP was dialyzed for 72 h at 4°C against 10 mM 1,10-phenanthroline monohydrochloride in 50 mM HEPES buffer, pH 7.5, and then exhaustively dialyzed against Chelex-treated (Chelex-100 column) 50 mM HEPES buffer, pH 7.5. (Bennett and Holz 1997). Apo-ArgE enzyme samples were prepared by adding 15 mM EDTA and incubating for 24 h at 4°C. The EDTA was removed by exhaustive dialysis against Chelextreated (Chelex-100 column) 50 mM HEPES buffer, pH 7.5, containing 150 mM KCl at 4°C for 36 h. Metal-free buffers were prepared by passing the buffers through a Chelex-100 column. The resulting apo-enzymes were inactive, and were found to contain no detectable metals ions via Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES).

Spectroscopic measurements
Electronic absorption spectra were recorded on a Shimadzu UV-3101PC spectrophotometer equipped with a constant temperature holder and a Haake (Type 423) constant temperature-circulating bath. All solutions were degassed prior to performing an experiment. Electronic absorption spectra were normalized for protein concentration and the absorption due to uncomplexed Co(II) (ε 512 = 6.0 M -1 cm -1 ) (D'souza et al. 2000).

Isothermal Titration Calorimetry
Isothermal Titration Calorimetry (ITC) measurements were carried out on a MicroCal OMEGA ultrasensitive titration calorimeter at 25 ± 0.2°C. The divalent metal ion titrants and apo-enzyme solutions were prepared in 25 mM Chelex-100 treated HEPES buffer at pH 7.5. Stock buffer solutions were thoroughly degassed before each titration. The enzyme solution (70 μM) was placed in the calorimeter cell and stirred at 200 rpm to ensure rapid mixing. Typically, 4-6 μL of titrant was delivered over 7.6 s with 6 min. intervals between injections to allow for complete equilibration. Each titration was continued until 4.5-6 equiv. of M(II) had been added to ensure that no additional complexes were formed in excess titrant. A background titration, consisting of the identical titrant solution but only the buffer solution in the sample cell, was subtracted from each experimental titration to account for heat of dilution. Individual AAP solutions were prepared by diluting stock enzyme solutions with 50 mM Chelexed HEPES buffer (pH 7. 5), and ArgE solutions were prepared by diluting stock enzyme solutions with 25 mM Chelexed HEPES buffer (pH 7.6) containing 150 mM KCl.

Molecular modeling
A three-dimensional homological structure of the ArgE from E. coli was developed using the X-ray crystal structure of the DapE from H. influenzae (PDB code:3IC1) as a template (Figure 3) (Rowsell et al. 1997;Eswar et al. 2003;Krissinel and Henrick 2004;Nocek et al. 2010). Sequence analysis of the target and template show that these two proteins are of similar length, share good sequence identity (~24%), and exhibit no significant sequence gaps. The ModWeb-based homology-building server was used to construct a structural homology model of the ArgE from E. coli. Comparison of the energy minimized ArgE homology model to the X-ray crystal structure of DapE, using the MathMaker in UCSF Chimera, reveals that the ArgE homology model displays the typical dimerization and catalytic domains of DapE with a Needleman-Wunsch (Needleman and Wunsch 1970) score of 490.6 and an average RMSD of 0.974 Å for the core atom pairs (Figure 3).

Protein production
Of the four ArgE mutant plasmids produced (H355A, H355K, H80A and H80K) and transformed into E. coli BL21 Star™ cells, only ArgE H355A was produced as a soluble enzyme. SDS-PAGE analyses of protein from E. coli cultures hosting plasmids containing the other three ArgE variants indicated that the enzymes were produced in abundance, but in an insoluble form. Various attempts were made to refold inclusion bodies of the three insoluble ArgE variants, but none proved fruitful.

Kinetic studies on ArgE H355A
Based on ICP-AES data, the as-purified H355A ArgE mutant enzyme contains no Zn(II) ions. These data imply that the remaining metal binding site in ArgE has a much higher K d value and is unable to bind Zn(II) in the absence of saturating conditions, similar to DapE, and consistent with weak cooperative binding as suggested by EXAFS (Tao et al. 2012;Gillner et al. 2009). The hydrolysis of L-NAO by H355A was monitored in the presence of three equivalents of Zn(II) or Co(II), after a one-hour incubation period, at pH 7.5 ( Figure 1). Interestingly, no enzymatic activity was detected for H355A ArgE in the presence of three equivalents of Zn (II). The lack of measurable activity for the H355A mutant enzyme may be the result of the reaction conditions used. Since the amide bond cleavage of L-NAO is monitored at 214 nm, enzyme concentrations must be < 10 nM so as not to overwhelm the amide bond absorption of L-NAO with background protein absorption. The severely reduced/eliminated enzymatic activity is consistent with the impaired ability of this variant to bind Zn (II). This further substantiates the hypothesis that this point mutation has knocked out a key metal ligand.
Further verification of this hypothesis comes from examining the hydrolysis of L-NAO by the Co (II)loaded H355A variant. The WT Co (II)-loaded ArgE enzyme is 2.4-times more active than the WT Zn (II)loaded form. For Co (II)-loaded H355A ArgE, a k cat value of 10 ± 1 s -1 was obtained in the presence of three equivalents of Co(II) using L-NAO as the substrate (K m = 0.5 ± 0.1 mM). Comparison of the kinetic parameters for the Co(II)-loaded WT and H355A ArgE enzymes indicate that the decrease in activity of H355A is due to a~380fold decrease in k cat . Interestingly, the K m value determined for the Co (II)-loaded H355A mutant enzyme is smaller than Co(II)-loaded WT ArgE by~2-fold. Calculation of the catalytic efficiency (k cat / K m ) for the Co(II)loaded H355A mutant enzyme results in a value~160-fold less than the Co(II)-loaded WT ArgE. The histidine to alanine mutation at this position, while having some impact on substrate binding, has primarily affected a key aspect of the active site of the enzyme (i.e. metal cofactor binding).

Electronic absorption spectra of Co(II)-loaded ArgE
Insight into whether H355 acts as an active site Zn(II) ligand was also investigated by examining the electronic absorption spectra of the Co(II)-substituted forms of the enzyme (Figure 4). The addition of one equivalent of Co (II) to apo-H355A ArgE (in Chelex-100 treated 50 mM HEPES buffer at pH 7.5) results in a broad absorption band with maxima at 560 nm (ε 560 =~95 M -1 cm -1 ) along with shoulders at 515 nm (ε 515 =~75 M -1 cm -1 ) and 620 nm (ε 515 =~55 M -1 cm -1 ) suggesting that the Co(II) ion bound to H355A ArgE resides in a pentacoordinate geometry (Bertini and Luchinat 1984;Horrocks et al. 1980). The addition of a second equivalent of Co(II) to H355A ArgE did not alter the overall shape of the observed absorption spectrum but did approximately double the molar absorptivity, consistent with a coordination number of five. In contrast, the WT ArgE from E. coli in the presence of one equivalent of Co (II) exhibited three distinct peaks at 560, 619, and 705 nm with ε values of ε 560 = 114 M -1 cm -1 , ε 619 = 119 M -1 cm -1 , and ε 705 = 52 M -1 cm -1 , respectively (McGregor et al. 2005). The addition of a second equivalent of Co(II) increased the intensity of each of the observed absorption bands, providing ε values of ε 560 = 229 M -1 cm -1 , ε 619 = 290 M -1 cm -1 , and ε 705 = 121 M -1 cm -1 , respectively ( Figure 4). Thus, the UV-vis spectrum of H355A ArgE is significantly different from WT ArgE in the presence of one and two equivalents of Co(II) (Figure 4) and the loss of distinct d-d transitions in the UV-vis spectrum of H355A is likely the result of a more flexible distorted fivecoordinate geometry with water molecules occupying non-protein coordination sites. Taken together, these data are consistent with the loss of an active site histidine ligand.
The altered ability of the H355A ArgE to bind Co(II) was also investigated by titrating Co(II) into apo-H355A ArgE samples and monitoring the molar absorptivity of the λ max for the Co(II) d-d bands, which allowed the dissociation constant (K d ) for this divalent metal binding site to be obtained by fitting these titration data to equation 1 (Winzor and Sawyer 1995): where p is the number of sites for which interaction with Co(II) is governed by the intrinsic dissociation constant K d , C s is the free metal concentration and, r is the binding function calculated by conversion of the fractional saturation (f a ) (Winzor and Sawyer 1995). Values for K d and p were obtained by fitting these data via an iterative process that allowed K d and p to vary ( Figure 5). The best fits obtained for Co(II) binding to the H355A ArgE provided a p value of 0.9 and a K d value of 1.6 ± 0.3 μM, which is 4-fold larger than that observed for WT ArgE (K d = 0.4 μM) (McGregor et al. 2005) consistent with a loss of an active site ligand. Since alteration of H355 will greatly perturb and potentially alter the ability of metal binding, this K d value likely corresponds to a single metal binding event. Moreover, the K d value obtained for Co(II) binding to H355A is~4 times stronger than that obtained for Zn(II) binding to WT ArgE. This difference is typically related to Zn(II)'s preference for four or five-coordinate geometries vs. Co(II)'s preference for five or six-coordinate structures (Holz 2002). These data, taken together with the lack of observed activity for H355A ArgE in the presence of Zn(II), suggest that H355 is in fact an active site ligand.

Isothermal Titration Calorimetry
Additional insight into whether H355 is a ligand in the active site of ArgE and which metal binding site it resides in, were gleaned by obtaining K d values and thermodynamic profiles using ITC ( Figure 6). These data were analyzed with a one, two, or three-site binding model after subtraction of the background heat of dilution, via an interactive process using the Windowsbased Origin software package supplied by MicroCal. This software package uses a nonlinear least-squares algorithm, which allows the concentrations of the titrant and the sample to be fit to the heat-flow-per-injection to an equilibrium binding equation. The K a value, enzymemetal stoichiometry (n), and the change in enthalpy (ΔH o ) were allowed to vary during the fitting process. The equilibrium binding constant, K a , and the enthalpy change ΔH, were used to calculate ΔG and ΔS using the Gibbs free energy relationship (equation 2): where R = 1.9872 cal mol -1 K -1 . The relationship between K a and K d is defined as: For the H355A ArgE mutant, the best fits provided n values of 2 for two non-interacting sites with K d values of 39 and 330 μM ( Figure 6; Table 1). The ITC data do not suggest cooperative binding whereas the EXAFS data do; however, weak cooperative binding is typically undetectable via ITC. The ΔH o value for the first Zn (II) binding event is endothermic with a large positive entropy consistent with the loss of water ligands from the Zn(II)-hexaauqo complex upon Zn(II) binding to ArgE. On the other hand, the second Zn(II) binding event is exothermic with an entropy change near zero (Table 2).  Comparison of these data with previously reported ITC data for Zn(II) binding to WT ArgE and DapE provide insight into which metal binding site is being populated first in the H355A ArgE mutant enzyme (McGregor et al. 2005;Davis et al. 2006). Both ArgE and DapE bind two Zn(II) ions in non-interactive binding sites with K d values for the first Zn(II) binding event for ArgE and DapE being 2.7 and 4.4 μM, respectively, whereas the observed K d values for the second metal binding event in ArgE and DapE are 51 and 13.6 μM, respectively. The ΔH o values observed for the first Zn(II) binding event in both ArgE and DapE are endothermic while the second is exothermic (Table 2). While K d for AAP has been reported previously by one of us (Bzymek et al. 2005), we report here a high quality ITC data set for Zn(II) binding to AAP, which provided K d values for the first and second metal binding event of 0.008 and 0.4 μM, respectively (Figure 7). The energy change observed for both AAP binding sites indicates an endothermic-type reaction with the second metal binding event providing a more negative ΔH o value than the first. Taken together, the thermodynamic data obtained for H335A are similar to those observed for the first binding events in WT ArgE, DapE, and AAP, suggesting that the H335 residue resides in the second  metal binding site similar to that observed for DapE (Gillner et al. 2009;Nocek et al. 2010). The weaker binding affinity for the first Zn(II) binding event in H355A ArgE vs. WT ArgE is likely the result of weak cooperative binding, which would be lost upon mutation of H355.

Modeling of the ArgE structure
Since no three-dimensional X-ray crystal structure has been reported for the ArgE form E. coli, a threedimensional homological structure was developed using the X-ray crystal structure of the DapE from H. influenzae (PDB code:3IC1) as a template (Figure 3) (Rowsell et al. 1997;Eswar et al. 2003;Krissinel and Henrick 2004;Nocek et al. 2010). The catalytic domain of ArgE contains the dinuclear Zn(II) site with a Zn-Zn distance of~3.3 Å and within 4.0 Å of the Zn(II) binding sites of ArgE, and all of the amino acid residues are nearly identical to those observed in DapE confirming the assignment of H80 and H355 as active site ligands in ArgE (Figure 3) (Rowsell et al. 1997;Eswar et al. 2003;Krissinel and Henrick 2004). The active site in ArgE is located in a long and open cleft of the catalytic domain; the cleft is parallel to and in the vicinity of the "hinge" region between the catalytic domain and the dimerization domain (Figure 8). With such a location, conformational changes upon metal and/or substrate binding are likely and have been observed for DapE (Nocek et al. 2010). Both Zn(II) ions are solvent accessible and have four protein centered ligands including H80 and H355 as well as two terminal glutamate residues (E145 and E160) and a bridging aspartate residue (D122). A bridging water provides the fifth ligand for both metal ions, which forms a hydrogen bond to a second water molecule that is in turn hydrogen bonded to E144, a candidate for the general acid/base during catalysis (Davis et al. 2006). Therefore, our data supports the hypothesis that H355 is an active site Zn (II) ligand, and as this residue corresponds to H349 in DapE, which has been shown to be the second metal binding site, it is likely that H355 resides in the second metal binding site of ArgE.

Summary
The kinetic, spectroscopic and thermodynamic data presented herein provide evidence that H355 is intimately involved in catalysis and serves as a metal binding ligand. While these data do not conclusively indicate which metal binding site H355 resides in, they are consistent with H355 residing in the second metal binding site, similar to DapE. Therefore, the ArgE from E. coli forms a (μ-aquo)(μ-carboxylato)dizinc(II) core where H80 and H355 function as active site metal ion ligands similar to DapE, AAP, and CPG 2 and consistent with EXAFS data (Tao et al. 2012). While the data reported here provide a structural starting point for elucidating the catalytic mechanism of ArgE, which may lead to the design and synthesis of small molecule inhibitors that specifically target ArgE, an X-ray crystal structure of the ArgE from E. coli with both one and two metal ions bound is desperately needed.