Mutational Analysis of a Highly Conserved PLSSMXP Sequence in the Small Subunit of Bacillus licheniformis γ-Glutamyltranspeptidase

A highly conserved 458PLSSMXP464 sequence in the small subunit (S-subunit) of an industrially important Bacillus licheniformis γ-glutamyltranspeptidase (BlGGT) was identified by sequence alignment. Molecular structures of the precursor mimic and the mature form of BlGGT clearly reveal that this peptide sequence is in close spatial proximity to the self-processing and catalytic sites of the enzyme. To probe the role of this conserved sequence, ten mutant enzymes of BlGGT were created through a series of deletion and alanine-scanning mutagenesis. SDS-PAGE and densitometric analyses showed that the intrinsic ability of BlGGT to undergo autocatalytic processing was detrimentally affected by the deletion-associated mutations. However, loss of self-activating capacity was not obviously observed in most of the Ala-replacement mutants. The Ala-replacement mutants had a specific activity comparable to or greater than that of the wild-type enzyme; conversely, all deletion mutants completely lost their enzymatic activity. As compared with BlGGT, S460A and S461S showed greatly enhanced kcat/Km values by 2.73- and 2.67-fold, respectively. The intrinsic tryptophan fluorescence and circular dichroism spectral profiles of Ala-replacement and deletion mutants were typically similar to those of BlGGT. However, heat and guanidine hydrochloride-induced unfolding transitions of the deletion-associated mutant proteins were severely reduced as compared with the wild-type enzyme. The predictive mutant models suggest that the microenvironments required for both self-activation and catalytic reaction of BlGGT can be altered upon mutations.


Introduction
N-terminal nucleophile (Ntn) hydrolases are a group of evolutionarily-related enzymes that can hydrolyze the amide bonds in peptides and proteins [1]. Notably, enzymes within this superfamily become catalytically active after an intramolecular autoproteolysis of the precursor polypeptides [2][3][4][5][6]. The intramolecular processing of the inactive precursors is generally believed to proceed through the nucleophilic attack of oxygen or sulfur atom in the side chains of Thr, Ser, or Cys residues on the carbonyl group of the immediate upstream backbone to form a tetrahedral intermediate [5][6][7][8][9][10]. of these residues near the essential environment for the self-activation of BlGGT and in the vicinity of the active cleft warrants their exploration as the critical residues for the enzyme ( Figure S1). In the current study, an attempt has been made to exploit the role of this conserved sequence by deletion and Ala-scanning mutagenesis. The data presented herein suggest that the conserved 458 PLSSMXP 464 sequence may indirectly be involved in the functionality of BlGGT, most likely through hydrogen-bonding and hydrophobic interactions with certain key amino acid residues to maintain the microenvironments required for the self-activation and catalysis.

Protein Expression and Purification
Wild-type enzyme and its deletion and Ala-replacement mutants were all expressed in E. coli M15 (pREP4), released from the recombinant cells by sonication, and purified by nickel-chelate chromatography, as described previously [31]. The imidazole elution fractions were pooled and resolved by 12% sodium dodecyl sulfate (SDS) polyacrylamide gels to assess the purity of each preparation. Bradford assays were routinely performed with the ready-to-use protein assay dye (Bio-Rad) and the protein concentration of each preparation was determined by comparing the assay response to a standard curve of bovine serum albumin.
A straightforward approach to calculate the level of self-activation was accomplished by quantifying the densitometry of each protein band in the 12% SDS-polyacrylamide gels using a computerized densitometer coupled with the GelAnalyzer 2010a software (http://www.gelanalyzer.com). The processing rate of each enzyme can be estimated by dividing the combined densitometric volume of L-and S-subunits by the total densitometric volume of the precursor and processed bands. Each value represents the mean of three independent determinations.

Enzyme Assays
The transpeptidation activity of BlGGT and its deletion and Ala-replacement mutants was measured by a colorimetric method as described elsewhere [32]. In a typical assay, the reaction mixture (0.5 mL) consisted of 25 mM Tris-HCl buffer (pH 9.0), 1.25 mM l-γ-glutamyl-p-nitroanilide (l-γ-Glu-p-NA), an excess of acceptor substrate Gly-Gly (30 mM), 1 mM MgCl 2 , and 0.1 mL of suitably diluted enzyme preparation (~6.0 µg/mL). After 10 min of reaction at either 40 • C for all enzymes, or 25 • C just only for the deletion mutants, a UV-visible spectrophotometer was employed to detect the released p-nitroaniline (p-NA) by measuring the absorbance at 410 nm. A standard curve of p-NA was established to determine the transpeptidation activity. One unit of the enzymatic activity is defined as the amount of enzymes that can liberate 1 µmol of p-NA from the chromogenic substrate per minute through the transpeptidation reaction. Data represent the mean of three independent assays.
Kinetic parameters of the wild-type and mutant enzymes were determined essentially as described previously [31]. To estimate the kinetic constants, a Lineweaver-Burk plot was established with data points derived from double-reciprocal transformation. Data represent the mean of three independent assays.

Spectroscopic Analyses
Fluorescence spectroscopic analysis of all protein samples was carried out at room temperature with a JASCO FP-6500 Spectrofluorometer (JASCO international Co., LTD., Tokyo, Japan). The excitation wavelength for the fluorescence spectroscopic analysis was set at 295 nm with a 1 nm bandwidth and the emission spectra of all protein samples were collected from 310 to 450 nm under a bandwidth of 10 nm. Prior to the spectroscopic study, all protein samples were diluted to approximately 12.5 µM with 25 mM Tris-HCl buffer (pH 9.0). Data acquisition and analysis were performed with JASCO's unique cross platform spectroscopy software supplied by the manufacturer. Each fluorescence experiment was done independently five times.
Circular dichroism (CD) spectra of all protein samples in the far-ultraviolet region were acquired from a JASCO-815 Spectrophotopolarimeter (JASCO international Co., LTD., Tokyo, Japan) with a 0.1 cm optical path cuvette and the spectral data were recorded over a wavelength range of 250 to 190 nm at room temperature. A scanning speed of 20 nm/min was carried out with an averaging time of 4 s and a wavelength step of 0.2 nm. Signal averaging over 10 scans or more were used to record and the data for each spectrum was acquired independently three times. Prior to the CD study, all protein samples were diluted to approximately 24.5 µM in 25 mM Tris-HCl buffer (pH 9.0). The CD spectra were corrected with the control curve of 25 mM Tris-HCl buffer (pH 9.0). The molar residue ellipticity (MRE) of the analyzed samples can be calculated by the following formula: [θ] = θ obs (in mdeg)/(the molar protein concentration × path length (in nm) × the total number of amino acid residues in the protein). The unit of molar residue ellipticity is deg · cm 2 · dmol −1 .
Thermal unfolding of all protein samples (~24.5 µM) in 25 mM Tris-HCl buffer (pH 9.0) was evaluated by monitoring the spectral change at 222 nm wavelength. Protein samples were heated from 20 to 90 • C with a constant heating rate of 1 • C/min. Melting transition temperature (T m ) was calculated as described previously [33].

Chemical Denaturation
The chemical stability of BlGGT and its deletion mutants was investigated by monitoring the changes in their fluorescence emission maximum of tryptophan as a function of guanidine hydrochloride (GdnHCl) concentrations. To obtain denaturation profiles, samples (1 mL) with increasing GdnHCl concentrations (0-2.5 M) were prepared by mixing 100 µL of enzyme preparation (~162 µM) with appropriate amounts of denaturant stock solution (3 M). The enzyme/GdnHCl mixtures were preincubated at room temperature for 30 min to allow for equilibration before spectroscopic analysis. Fluorescence emission spectra was determined by a JASCO FP-6500 Spectrofluorometer with an excitation wavelength of 295 nm and fluorescence emission was recorded from 310 to 500 nm. Both maximum wavelength shift and fluorescence intensity change were analyzed together to compute the average emission wavelength (AEW) [34].

Computer Modeling
Predictive models of the deletion and Ala-replacement mutants were fabricated by the Swiss-Model Server [35] using the three-dimensional structures of T399A-BlGGT (PDB code: 4Y23) [24] and BlGGT complexed with L-glutamate (PDB code: 4OTU) [20] as the templates. The mutant models were established by initially replacing or deleting the pertinent amino acids in the template structures and subsequently performing local side-chain minimization within 8 Å Cα-Cα distance from any residue of the interaction partner. Afterwards, the predictive models were subjected to an energy minimization (200 steps of steepest descent) with a partial implementation of the GROMOS force field accessible via the Swiss-pdb Viewer software v4.1.0 (http://spbdv.vital-it.ch/).

Local Environments Surrounding the Conserved 458 PLSSMXP 464 Sequence
In the molecular architecture of T399A-BlGGT [24], a precursor mimic of BlGGT, the side chain hydroxyl group of Thr417 is located within a competent position to act as a base to selectively deprotonate the OH group of Thr399 for an intramolecular nucleophilic attack, and the backbone atoms of this residue are held individually in their spatial positions by the positively charged guanidino group of Arg571 that, in turn, engages in a strong electrostatic-based interaction with the side-chain carboxyl group of Glu398. In addition, the correct spatial positions of the side-chain atoms of Thr399 are critical for the self-activation of BlGGT and this positioning is assured by the hydrophobic interactions among nonpolar groups of Thr415 and Met462 [24]. Giving the fact that the proper spatial position of Thr417 is linked to a hydrogen-bonding network involving one sodium ion, two water molecules, and amino acid residues Pro458, Ser460, and Ser461 ( Figure S1A), the amino acid residues within the conserved 458 PLSSMXP 464 sequence are likely to be involved in the autocatalytic processing of BlGGT.
With the support of evidence-based research, many of the amino acid residues critical for the transpeptidation activity of GGT enzymes were being implicated [36][37][38][39][40][41][42][43], and the identification of key functional residues in the active site of BlGGT has just been accomplished a few years ago by solving the crystal structure of this heterodimeric enzyme in complex with L-glutamate [20].
Notably, the substrate-binding site is deeply buried within a pocket that is lined by amino acid residues Arg109, Thr399, Thr417, Glu419, Glu438, Ser460, Ser461, Gly481, and Gly482 ( Figure S1B). Over the past years, several researchers were interested in exploring the role of these residues in BlGGT and the corresponding residues in its counterparts through the use of site-directed mutagenesis technique [23,36,38,[41][42][43][44][45]. The study of Ikeda and his coworkers has identified two serine residues, Ser451 and Ser452 (Ser460 and Ser461 in BlGGT), required for the catalysis of HsGGT [38]. Two previous investigations relevant to EcGGT have helped us further understand the functional role of Ser463 (Ser461 in BlGGT) and Met464 (Met462 in BlGGT) [41,43]. As shown in Figure 1A, the 458 PLSSMXP 464 sequence is highly conservative among GGT enzymes from a variety of species so that the amino acid residues within this peptide sequence may, therefore, play a role in the catalysis of BlGGT.

Self-Activation and Catalytic Activity of the Wild-Type Enzyme and its Mutants
Site-directed mutagenesis techniques are one of the principal tools of molecular biology to study the sequence-structure-function relationships of a protein [46]. To probe the role of the amino acids within the conserved 458 PLSSMXP 464 sequence of BlGGT, four deletion (∆M462, ∆S461-M462, ∆S460-M462, and ∆P464) and six Ala-replacement (P458A, L459A, S460A, S461A, M462A, and P464A) mutants were generated by site-directed mutagenesis. The verified plasmids pQE-BlGGT/∆M462, pQE-BlGGT/∆S460-M462, pQE-BlGGT/∆S461-M462, pQE-BlGGT/∆P464, pQE-BlGGT/P458A, pQE-BlGGT/L459A, pQE-BlGGT/S460A, pQE-BlGGT/S461A, pQE-BlGGT/M462A, and pQE-BlGGT/P464A were separately transformed into the competent E. coli M15 (pREP4) cells by calcium chloride (CaCl 2 ) heat shock method. Following recombinant gene expression in host cells and single-step protein purification, the freshly prepared enzyme samples were subjected to purity and molecular size analysis by 12% SDS-PAGE. As shown in Figure 2A, three protein bands with the individual molecular masses of about 72, 50, and 22 kDa were found in the freshly prepared BlGGT sample. These bands were consistently present in Ala-replacement mutant preparations, while the precursor form was predominantly observed in the deletion-related mutant samples (Figure 2A). The purified wild-type and mutant proteins were also subjected to native gel electrophoresis. As shown in Figure 2B, just a single major band was observed in all enzyme preparations, indicating that the BlGGT integrity remains unchanged after the mutations.
The extent of autocatalytic processing of each enzyme preparation was subsequently quantified by densitometry analysis of the protein bands on an SDS-PAGE gel stained with Coomassie Brilliant Blue R-250. The results of such analysis revealed that the freshly prepared BlGGT was able to achieve a processing rate of 95.5% and three Ala-replacement mutant enzymes (L459A, S462A, and P464A) displayed a slight impairment in the capability to self-process (Table 1), with 4.5% to 13.1% reduction in the processing rate. Based on these observations, it can be concluded that the individual residues within the conserved 458 PLSSMXP 464 sequence are nevertheless absolutely essential for the autocatalytic processing of BlGGT. Consistently, site-specific replacements of Asn450, Gly481, and Gly482 by other amino acids have shown only a certain degree of influence on the autocatalytic processing of BlGGT [31,44]. Minor contribution of the aforementioned residues to the self-activation of BlGGT may be due solely to the facts that they are just in close spatial proximity to the self-activating site and do not play a direct role in the autocatalytic processing of the enzyme. However, the deletion-associated mutations had a serious impact on the self-activation with processing rates of less than 8.3% (Table 1). This result indicates that detrimental changes in the microenvironment required for the autocatalytic processing of BlGGT may probably occur as a result of the deletion mutations.  To verify whether the existing functionality was working as expected, the GGT activity of freshly prepared BlGGT, P458A, L459A, S460A, S461A, M462A, P464A, ∆M462, ∆S460-M462, ∆S461-M462, and ∆P464 was accordingly determined. As shown in Table 1, the freshly prepared BlGGT, P458A, L459A, M462A, and P464A had a specific activity of 14.4, 17.8, 14.6, 19.5, and 6.7 U/mg, respectively. The specific activity of S460A and S461A was significantly increased by 1.1-and 1.4-fold with respect to that of the wild-type enzyme. In contrast, single substitutions at residues Ser451 and Ser452 (Ser460 and Ser461 in BlGGT) of HsGGT with Ala yielded the mutant enzymes with only about 1% activity of the wild-type enzyme [38]. Our results seem to contradict the findings of Ikeda et al. [38] so that further experimental clarification is required prior to make a good conclusion.
Steady-state kinetic parameters of BlGGT and its Ala-replacement mutants were determined by a series of measurements of the initial rate of the transpeptidation reaction. As shown in Table 2, most of the Ala-replacement mutants displayed K m -values close to that of the wild-type enzyme. P458A had an apparent k cat value of 16.0 s −1 , which resembles the turnover number (16.6 s −1 ) of BlGGT. The apparent k cat values of S460A, S461A, and M462A showed an increase to a certain extent, while L459A and P464A had more than 18% reduction. Notably S460A, S461A, and M462A exhibited 273%, 267%, and 51% higher catalytic efficiency (k cat /K m ) over BlGGT, respectively. Although all previously solved structures of GGT enzymes share a common overall architecture of four-layer αββα sandwich, they do not have similar primary structures. For example, the amino acid sequence of BlGGT only exhibits 38.6% and 31.9% identity with those of EcGGT and HsGGT, respectively. In the crystal structures of EcGGT and HpGGT [18,47], there is a "lip loop" (Pro438-Gly449 in EcGGT and Pro427 to Gly438 in HpGGT) that covers most of the γ-glutamyl binding portion of the active site and that may block or limit the binding of acceptor molecules ( Figure 3A,B). The molecular architecture of HsGGT also shows that the anchoring termini of the lid loop are situated in comparable positions relative to those of EcGGT and HpGGT ( Figure 3C), with its lip loop rotated away from the active site [21]. It has been documented by Castellano and Merlino [48] that the lid loop has a well-defined position and shields the enzymatic pocket from the solvent upon the binding site of a substrate or inhibitor. However, the analogous residues forming the lip loop are absent in BlGGT [20], leading to the formation of a more open substrate channel in the active site ( Figure 3D). Notably, the side chain of an exposed tyrosine (Tyr444 in EcGGT and Tyr433 in HpGGT) at the apex of the bacterial lid loop gates the active site by hydrogen bonding to a conserved asparagine residue (Asn411 in EcGGT and Asn400 in HpGGT) adjacent to the catalytic threonine [18,45]. The corresponding residue for gating the active site of HsGGT is Phe433 [21], while its side chain is not hydrogen bonded with the analogous residue, Asn401. In our case, the absence of lip loop in BlGGT apparently gives the enzymatic pocket more freedom for substrate entrance and product exit ( Figure 3D). As shown in Table 2, the apparent K m value of S460A was only 27% higher that of the wild-type enzyme and that for S461A was a little bit lower. These observations indicated that the enzyme-substrate affinity has not changed much after the replacements of Ser460 and Ser461 by alanine. The crystal structure of EcGGT complexed with L-glutamate has further revealed that the carboxyl group is hydrogen bonded with Arg114 Nη, Ser462 Oγ, Ser463 N, and Ser463 Oγ via water, and the α-amino group has interactions with Asn411 Oδ, Gln430 Oε, and Asp433 Oδ ( Figure 4A). The γ-glutamyl carbonyl oxygen of enzyme-bound L-glutamate is also hydrogen bonded with the main-chain amino atoms of Gly483 and Gly484. Besides, there are some hydrophobic interactions between hydrophobic parts in the ligand and the enzyme. It can be seen that a number of hydrogen-bonding interactions in the active site of EcGGT are disrupted after the replacements of Ser462 and Ser463 by alanine ( Figure 4B,C and Table S2). Interestingly, the enzyme-ligand interactions of EcGGT are not perfectly conserved in the active sites of HsGGT and BlGGT ( Figure 4D,G and Tables S3 and S4). As shown in Figure 4E,F and Table S3, Ala-replacement at Ser451 of HsGGT has a certain degree of impact on the enzyme-ligand interactions, while the replacement of Ser452 by Ala does not make any changes on the original interactions. It is important to note that the hydrogen-bonding interactions within the enzymatic pocket of BlGGT are almost entirely conserved in the predictive S460A and S461A models ( Figure 4H,I and Table S4). Conclusively, a slight difference in the active-site architecture between BlGGT and the compared enzymes together with the high conservation of the enzyme-ligand interactions may potentially explain why S460A and S461A can fully reserve their transpeptidation activity, whereas the respective mutants in EcGGT and HsGGT do not.
It is also worthy of mention that, under the standard assay conditions, there was no detectable transpeptidase activity in ∆M462, ∆S460-M462, ∆S461-M462, and ∆P464 (Table 1). This was further confirmed by performing the enzyme assay at 25 • C, indicating that the loss of their transpeptidation activity is unrelated to the heat-induced denaturation. Together with the above-mentioned results, it might be suggested that this conserved sequence is essential for the proper activity of BlGGT and may indirectly involve in the reaction cycle by leaving the catalytic threonine (Thr399) and substrate in the spatial positions that efficiently facilitate nucleophilic attack.

Spectroscopic Characterization of the Structural Properties of Enzyme Preparations
Giving the fact that the structural impact caused by the studied mutations may contribute to changes in the enzymatic activity of BlGGT, we here sought to investigate the structural alterations occurring in the deletion and Ala-replacement mutants by fluorescence and CD spectroscopy. As compared to the intrinsic tryptophan emission spectrum of BlGGT, the fluorescence intensity of the freshly prepared deletion mutants was slightly enhanced by 6.5% to 9.7% ( Figure 5A). It can be seen that the fluorescence spectra for BlGGT and ∆Met462 were all maximized at a wavelength of 341.6 nm. Apparently, the fluorescence peak maxima of ∆S460-Met462, ∆S461-Met462, and ∆P464 were slightly shifted to either red or blue by less than 0.4 nm ( Figure 5A). However, the freshly prepared Ala-replacement mutants exhibited up to 24% decrease in the fluorescence intensity, with no profound shift in the peak maxima ( Figure S2A). These results indicate that only minor alterations in the molecular structure have occurred upon the introduction of deletion and Ala-replacement mutations into the conserved 458 PLSSMXP 464 sequence. Furthermore, there were no major differences in the CD spectra of wild-type enzyme and its relevant mutants ( Figure 5B and Figure S2B). These results suggest that both deletion and Ala-replacement mutations have no detrimental consequences on the secondary structural content of BlGGT. Conclusively, based on the results of fluorescence and CD measurements, we may exclude any gross structural changes for a conceivable explanation on the impaired activities of the mutant enzymes. It is worthy of note that the extent of autocatalytic processing of the wild-type and mutant enzymes was not closely linked to the overall spectra of fluorescence and CD spectroscopy significantly ( Figure 5A,B and Figure S2). Very recently, the crystal structure of T399A-BlGGT has been solved at 2.89 Å resolution [24]. A Cα-carbon trace superposition of T399A-BlGGT and the processed enzyme by Pica and his coworkers has shown that the root mean square error of the Cα atomic coordinates is 0.57 Å [24], which suggests a high degree of structural similarity between them. Therefore, the content of secondary and tertiary structural elements of each mutant appears to most closely resemble that of BlGGT. This might be one possible explanation for why their CD and fluorescence spectra are highly similar to each other.
The irreversible inactivation of BlGGT by heat has already been reported to follow the one-state process [49]. As shown in Figure 5C, the wild-type enzyme started to denature at around 48.1 • C and was transformed from its native state into a completely unfolded polypeptide chain at 70.2 • C. However, the unfolding process of the deletion mutants occurred at lower temperatures and ended up at 60.2 • C. The unfolding curves of BlGGT, ∆M462, ∆S460-M462, ∆S461-M462, and ∆P464 all indicated one phase transition with an apparent T m value of 61.8 ± 0.5, 42.1 ± 0.2, 41.2 ± 0.2, 41.9 ± 0.3, and 42.5 ± 1.2 • C, respectively. A review article has stated that there are numerous factors, including hydrogen-bonding networks, specific side-chain hydrophobic and packing interactions, and charge clusters, related to the thermal stability of proteins [50]. The results of heat-induced unfolding transition may give clues to the abolishment of some BlGGT-stabilized forces by the deletion-associated mutations. In contrast, the apparent T m values for S460A and S461A were relatively invariant at around 61 • C ( Figure S2C). These results indicate that there is no significant change in the overall thermal stability of the enzyme as the consequences of Ser460 →Ala and Ser461 → Ala mutations.
Fluorescence spectroscopy is a sensitive, rapid and noninvasive analytical technique that can provide information on the conformational changes that underlie protein function [34]. The AEW that reports the changes of emission maximum and fluorescence intensity was used to assess the conformational stability of GdnHCl-treated enzymes. As shown in Figure 5D, the AEW value of BlGGT in the absence and presence of 4.0 M GdnHCl was 351.2 and 357.3 nm, respectively. It has previously been stated that a fully exposed tryptophan residue in the completely unfolded protein has an emission maximum of approximately 356 nm [34]. In this regard, BlGGT is completely unfolded under the GdnHCl concentrations of above 4 M. Alongside with that, the enzyme gave a midpoint transition of 2.9 M GdnHCl. However, the value of ∆M462, ∆S460-M462, ∆S461-M462, and ∆P464 was significantly reduced to 1.2, 1.2, 1.1, and 0.9 M, respectively. These observations clearly indicate that the structural integrity of BlGGT are markedly affected by the deletion-associated mutations.

Impact of Mutations on the Spatial Positions of Self-Activating and Catalytic Residues
Detailed knowledge about the three-dimensional structure of enzymes is crucial for studying their structure-function relationship. X-ray and NMR are the most frequently used experimental methods for solving enzyme structures, but this would be a much more difficult and time-consuming task during protein structure determination [51]. Thus, computer-based modeling provides an attractive option to generate predictive models of three-dimensional enzyme structures [52]. Computer modeling of the local environment surrounding the conserved 458 PLSSMXP 464 sequence was, therefore, performed to elucidate the impact of the deletion-associated mutations on the self-activation of BlGGT. As it was mentioned above, amino acid residues Glu398, Thr415, Thr417, Arg571, and Met462 may play a practical role during the autocatalytic processing of BlGGT ( Figure S1A and Figure 6A). The likely spatial positions of these residues in the predictive model of P458A basically resemble those of the wild-type enzyme ( Figure 6B). The conservation of the spatial locations of these residues allows this mutant enzyme to ensure an efficient self-activation (Table 1). It is worthy of note that the spatial arrangements of Glu398, Thr415, Thr417, and Arg571 are slightly altered in the predictive models of ∆S460-M462, ∆S461-M462, and ∆M462 ( Figure 6C-E). Except for the above changes, a more profound shift in the spatial arrangement of Met462 can be seen in the model of ∆P464 ( Figure 6F). Such changes in the spatial arrangements appear to interfere with the hydrogen-bonding networks and hydrophobic interactions that primarily held these critical residues in their relative spatial positions of BlGGT ( Figure S1A). The crystal structure of BlGGT complexed with an l-glutamate molecule has recently been determined and refined to an atomic resolution of 3.022 Å [20]. This study suggests that some important amino-acid residues, including Arg109, Glu438, Ser460, Ser461, Gly481, and Gly482, might be involved in l-glutamate binding. As shown in Figure S3, the predictive models of P458A, L459A, and M462A display small differences in the spatial positioning of L-glutamate-binding residues. This could be the reason why these three mutants exhibited the specific transpeptidation activity comparable to that of the wild-type enzyme. In the catalytic cycle of BlGGT, the spatial orientation of the side-chain hydroxyl group of the catalytic Thr399 is critical for the nucleophilic attack at the carbonyl carbon of the γ-glutamyl substrates ( Figure 7A and Scheme S1). As shown in Figure 7B-E, the spatial orientation of this side-chain hydroxyl group is dramatically altered upon different combinations of sequence deletion. In the most widely accepted reaction mechanism (Scheme S1), the side-chain hydroxyl group of the highly conserved Thr (Thr399 in BlGGT), which is also responsible for the autocatalytic processing, attacks the C = O group of the γ-glutamyl-compound to form a γ-glutamyl-enzyme intermediate. The intermediate then reacts with water, to release glutamate in a hydrolysis reaction, or with an acceptor, to give a transpeptidation reaction forming new γ-glutamyl compounds. Therefore, alterations to the spatial orientation of this side-chain hydroxyl group will definitely affect the nucleophilic attack and eventually lead to the full abolishment of the enzymatic activity (Table 1). It is also worthy of mention that the spatial position of Met462 in BlGGT is notably changed after the site-directed replacement of Ser460 and Ser461 with Ala ( Figure 7F,G). These spatial rearrangements will in turn change the hydrogen-bonding networks and open up the catalytic site of the enzyme, which may be more favorable to the entrance of the substrates and finally improve the catalytic activity of S460A and S461A. Although the spatial position of Met462 is notably shifted in the predictive models of S460A and S461A, the spatial orientation of the side-chain hydroxyl group of the catalytic Thr399 remains unchanged ( Figure 7F,G). The spatial reservation of the catalytic Thr399 is probably the key that allows these two enzymes to work properly. As noted earlier, P464A only retained about 47% of the specific transpeptidation activity as compared to the wild-type enzyme. The profound decrease in the specific transpeptidation activity is probably caused by a slight change in the spatial positioning of Met462 after the Ala-replacement ( Figure 7H).

Conclusions
In summary, the results of the combined effort of mutational and functional investigations rationalize the significance of the conserved 458 PLSSMXP 464 sequence in the autocatalytic processing and enzymatic activity of BlGGT. Clearly, the secondary and tertiary structures of BlGGT have not been changed profoundly by the deletion and Ala-replacement mutations, but the spatial alterations in amino acid residues involved in autocatalytic processing and L-glutamate-binding have really happened to some of the mutants. In the deletion-associated mutants, the disruption of hydrogen-bonding networks in the local surrounding environment of the active site may freeze the nucleophilic residue (Thr399) in a catalytically incompetent state that in turn leads to a complete loss of the enzymatic activity. However, further work is needed to more precisely elucidate the functional influence of Ala-replacements, especially the structural determinations of S460A and S461A.
Supplementary Materials: The following are available online at http://www.mdpi.com/2218-273X/9/9/508/s1, Figure S1: Local environments surrounding the highly conserved PLSSMXP region of BlGGT, Figure S2: Intrinsic fluorescence (A) and Far-UV (B) spectra, and thermal unfolding curves (C) of BlGGT, S460A, and S461A, Figure S3: The catalytic environments of P458A (A), L459A (B), and M462A (C), Table S1: Overlapping complementary primers used in the site-directed mutagenesis, Table S2: Comparison of hydrogen-bonding interactions in the enzymatic pockets of EcGGT and the predictive S451A-EcGGT and S452A-EcGGT models, Table S3: Comparison of the hydrogen-bonding interactions in the enzymatic pockets of HsGGT and the predictive S451A-HsGGT and S452A-HsGGT models, Table S4: Comparison of the hydrogen-bonding interactions in the enzymatic pockets of BlGGT and the predictive S460A and S461A models, Scheme S1: The proposed catalytic mechanism of GGT enzymes, Scheme S2: The proposed mechanism for intramolecular autocatalytic processing of BlGGT.