Engineering and Directed Evolution of a Ca2+ Binding Site A-Deficient AprE Mutant Reveal an Essential Contribution of the Loop Leu75–Leu82 to Enzyme Activity

An aprE mutant from B. subtilis 168 lacking the connecting loop Leu75–Leu82 which is predicted to encode a Ca2+ binding site was constructed. Expression of the mutant gene (aprEΔLeu75–Leu82) produced B. subtilis colonies lacking protease activity. Intrinsic fluorescence analysis revealed spectral differences between wild-type AprE and AprEΔL75–L82. An AprEΔL75–L82 variant with reestablished enzyme activity was selected by directed evolution. The novel mutations Thr66Met/Gly102Asp located in positions which are predicted to be important for catalytic activity were identified in this variant. Although these mutations restored hydrolysis, they had no effect with respect to thermal inactivation of AprEΔL75–L82 T66M G102D. These results support the proposal that in addition to function as a calcium binding site, the loop that connects β-sheet e3 with α-helix c plays a structural role on enzyme activity of AprE from B. subtilis 168.


Introduction
Currently, there is a high level of commercial interest for subtilisins that work under extreme biochemical conditions [1,2]. Therefore, understanding the structure and function of subtilisins is fundamental to employing rational and directed evolution strategies in order to enhance activity and/or change substrate specificity for these proteins [3,4]. However, there are structural motifs in subtilisin E (AprE) which have proven to affect enzyme activity and still remain uncharacterized. For instance, crystallographic analysis revealed that residues Leu 75 , Asn 77 , Ile 79 , and Val 81 located in the connecting loop Leu 75 -Leu 82 together with Gln 2 and Asp 41 form a calcium binding site (CBS) in subtilisin BPN' [5]. Furthermore, it is known that residues Gly 83 -Ser 85 , conserved among several members of the subtilisin family [6], form a stretch bend which lies at the C-terminal edge of the loop connecting β-sheet e3 to α-helix c. These residues are located 1.5 nm away from and on the opposite side of the catalytic residues Asp 32 , His 64 , and Ser 221 [7]. Despite their far location from the catalytic residues, mutations in this region induce changes on both substrate specificity and enzyme activity of subtilisins. For instance, a single Ser 85 Ala mutation increased twice the kcat of B. subtilis 168 AprE [6], and a Val 84 Ile mutation not only increased the Km of subtilisin BPN' but also adapted the enzyme to work at a lower than normal temperature [7].
Members of the subtilisin family usually possess two calcium binding sites (CBSs), named CBSA and CBSB [8]. Each CBS displays different affinity for the calcium ion [8]. In this report, evidence is presented supporting the idea that in addition to the role as a calcium binding site, the loop connecting β-sheet e3 with α-helix c (residues Leu 75 -Leu 82 ) also plays an important role in the enzyme activity of subtilisin E from B. subtilis 168.

Bacterial Strains, Plasmids, and Growth Conditions.
Bacterial strains used in this work are listed in Table 1. The growth medium used routinely was Luria-Bertani (LB) [9]. Preparation of competent E. coli and B. subtilis cells and their transformations were performed as previously described [10,11].

Site-Directed Mutagenesis of AprE.
Codons 75 through 82 from wild-type aprE [13,14] were eliminated with the Altered Sites II Site-Directed Mutagenesis System Kit (Promega, Madison, WI) using the oligonucleotide 5 -GCTTGGGCTAACGCC * AGCGGCAATCGTACC-3 (asterisk denotes the location of the in-frame deletion).

Random
Mutagenesis of AprEΔLeu 75 -Leu 82 . Random mutagenesis was carried out as follows. Strain B. subtilis PERM570 (Table 1) was grown to an O.D. 600 nm of 0.5; the cell culture was supplemented with 2 mM H 2 O 2 and incubated for 48 hours at 37 • C. Cells were serially diluted, and aliquots of 100 μL were inoculated on LB agar plates supplemented with skimmed milk. The plates were incubated 12 hours at 37 • C, and colonies exhibiting caseinolytic activity were selected and transferred to a fresh plate. The plasmids of selected colonies were isolated and used to retransform B. subtilis 1A751 and E. coli DH5α. The aprEΔL 75 -L 82 variant generated through this protocol was fully sequenced on both strands.

Expression and Purification of Wild-Type and AprE
Mutants. Wild-type aprE and aprEΔL 75 -L 82 BamHI/BamHI fragments encoding the preproenzymes were cloned in plasmid pUSH2 [12] to introduce an in-frame six histidinecoding sequence at the 3 end of both aprE sequences. This strategy generated the strains, E. coli PERM223 harboring pPERM222 (pUSH2-aprE) and E. coli PERM494 harboring pPERM494 (pUSH2-aprEΔL 75 -L 82 ), respectively. Wild-type and AprE variants were expressed and purified from the culture media of B. subtilis 1A751 by metal affinity chromatography on a Ni-NTA-agarose column (Quiagen; Valencia, CA) as previously described [6]. Protein concentrations were determined by using the Coomassie (Bradford) Protein Assay Kit (Pierce; Rockford, IL).

Subtilisin Intrinsic Fluorescence (IF) Assays.
Fluorescence spectra data were obtained after equilibration of a mixture containing 4 μM of either wild-type or mutant AprEΔL 75 -L 82 in 10 mM Pipes pH 7.5 at 25 • C in the presence or absence of 0.5 mM EGTA in a spectrofluorophotometer RF-5301PC (Shimadzu, Japan) equipped with both a thermostated cell and constant stirring. Fluorescence spectra were recorded between 280-450 nm upon exciting the protein at 280 nm.

Thermal Unfolding Followed by Intrinsic Fluorescence.
Subtilisin E samples were placed into a 2 mL quartz cuvette; changes in intrinsic fluorescence were measured at 340 nm using an excitation wavelength of 280 nm (4 nm bandwidth) and emission wavelength from 300 to 400 nm (4 nm bandwidth). Temperature was ramped from 25 to 90 • C with a 1 • C min −1 . Thermal unfolding data were normalized to where x is the temperature in Kelvin where the enzyme was completely unfolded. Thermodynamic parameters were calculated by nonlinear least-squares fitting to following scheme. Two-state model between native (N) and unfolded (U) states N → U. Data were analyzed using the thermal following equation: where T is temperature in K, T m is the temperature at midpoint, and ΔH m is the enthalpy at the T m , respectively.

Enzyme Kinetics.
The synthetic peptide Succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (s-AAPF-p-Na, Sigma Chemical Co. St. Louis, MO) was used as substrate; assays were performed in 100 mM Tris-HCl (pH 8.0) and 5 mM CaCl 2 at 37 • C. The amount of p-nitroanilide released was measured by recording the absorbance increase at 410 nm. Enzyme activity was expressed as units/mg protein. Velocity data were fitted to the Michaelis-Menten equation by nonlinear regression.

Thermal Stability of the Enzymes.
Purified wild-type or variant AprE enzymes (0.7 mg/mL) were incubated in 100 mM Tris-HCl (pH 8.0) and varying concentrations of both CaCl 2 (100 μM-5 mM) and NaCl (0 or 100 mM). The wild-type and variant AprE were either previously treated or not with 100 μM EDTA and then incubated on ice for 15 minutes before testing for thermal stability. The activity remaining after a given time of incubation was determined at 37 • C using s-AAPF-pNa as the substrate. The temperatures tested for enzyme stability were between 50-65 • C.

Data Analysis. Thermal inactivation kinetics for both
WT and AprE variant were studied fitting the inactivation data to (3) by nonlinear regression and using the iterative program Microcal Origin, as described in studies of thermal enzyme inactivation [15]. The equation used was the following: where A R represents the (%) of residual activity at a given time (t), A 0 is the initial relative activity, considered as 100%, and k is the rate constant for enzyme inactivation in min −1 . Equation (3) describes a one-step process (4) for enzyme inactivation; from the native (N) to the inactive state (I):

Results and Discussion
Mutations in the stretch bend Gly 83 -Ser 85 lying at the Cterminal edge of the loop connecting the β-sheet e3 with the α-helix c of AprE led to changes on both substrate specificity and enzyme activity of subtilisins [6,7]. These findings strongly suggest that this region has an important structural role for enzyme activity in AprE. Therefore, this loop was eliminated by site-directed mutagenesis, and the resulting aprEΔL 75 -L 82 mutant gene ( Figure 1) was expressed in B. subtilis 1A751, a strain lacking protease activity as determined on casein plates ( Figure 1). In fact, the cell free culture medium of this strain possessed no activity against hide powder azure and only 3% of the activity showed by the strain expressing the wild-type aprE gene against azocasein (Results not shown). A version of subtilisin BPN' lacking the CBSA and containing stabilizing mutations has been previously produced [16,17]. Refolding of this protein was greatly facilitated by the absence of the Ca-loop while retaining high levels of activity [17]. However, as described here in the absence of stabilizing mutations deletion of the CBSA on AprE resulted in a dramatic loss of enzyme activity. Therefore, the loop L 75 -L 82 may be important for structural integrity of not only the binding site but also the active site. Changes in intrinsic fluorescence are excellent for monitoring the polarity of Trp environment and hence are sensitive to protein conformation [18,19]. Therefore, the emission fluorescence spectra of AprE and AprEΔL 75 -L 82 were recorded with excitation at 280 nm. The AprEΔL 75 -L 82 spectrum showed an emission maximum of ∼358 nm which was red-shifted by 14 nm relative to the peak of the wildtype AprE spectrum (Figure 2(a)). These data suggest that the side chains of the aromatic residues are more exposed and subtilisin BPN' share 86% similarity; in fact the three tryptophan residues existing in mature subtilisin BPN' (i.e., Trp 106 , Trp 113 , and Trp 241 ) are present in equivalent positions in AprE (i.e., Trp residues 105, 112, and 240, resp.) [5,20,21]. On the other hand, a previous study suggested that in subtilisin BPN' , Trp 113 , is virtually nonfluorescent; the largely exposed Trp 241 contributes 20% of the fluorescence, whereas the partially exposed Trp 106 accounts for the majority of the emission [22]. Therefore, the increased fluorescence intensity observed in AprEΔL 75 -L 82 could be attributed to perturbations in the local environment of residues Trp 105 and/or Trp 112 which are located near to the deleted loop L 75 -L 82 . As noted above, deletion of the loop 75-82 abolished the calcium binding potential at site A while leaving intact the calcium binding site B. To further investigate this notion, wild-type and mutant AprE proteins were incubated in the presence of 0.5 mM EGTA, a concentration enough to chelate Ca 2+ only from the CBSB [23]. As shown in Figure 2(b), elimination of Ca 2+ from CBSB induced in both enzymes a small decrease in their fluorescence intensity with respect to the nontreated native enzymes (Figure 2(b)). These results are in agreement with the presence of an intact CBSB in both the wild-type and the AprEΔL 75 -L 82 enzymes.
The structural consequences of loop L 75 -L 82 removal from AprE resulted in the lost not only of the CBSA but also of enzyme activity. Therefore, a directed evolution strategy was used to search for amino acid substitutions in the mutant enzyme that could restore enzyme activity. A plasmid containing aprEΔL 75 -L 82 was expressed in a hypermutagenic strain of B. subtilis deficient on the mutM mutY and sodA genes that also lacked protease activity as described above. After several rounds of mutagenesis for aprEΔL 75 -L 82 , three colonies exhibiting extracellular protease activity against casein were recovered. The colony with the highest protease activity was selected to further characterize its phenotype; the clone was called aprEΔL 75 -L 82 Var1. Interestingly, the cell free culture medium of this strain recovered 27% and 65% of the activity exhibited by the strain expressing the wildtype aprE gene against hide powder azure and azocasein, respectively (Results not shown).
Analysis of the nucleotide sequence of aprEΔL 75 -L 82 Var1 revealed the existence of two nonsense mutations that resulted in amino acid substitutions, Thr 66 Met and Gly 102 Asp. The mutant gene named aprEΔL 75 -L 82 T 66 MG 102 D was cloned in pUSH2, and the resulting construction was expressed in the protease deficient strain B. subtilis IA751 (Figure 1). Calculation of kinetic constants kcat and Km from initial rate measurements of hydrolysis of s-AAPF-pNa revealed that the relative catalytic efficiency of AprEΔL 75 -L 82-T 66 M G 102 D was of around 7.4% as compared with the wild-type AprE enzyme (Table 2).
In order to understand the effect of these mutations in the structure of the AprEΔL 75 -L 82 T 66 M G 102 D enzyme, the medium temperature of denaturation (T m ) was calculated for the three enzymes. Results showed that the T m value of AprEΔL 75 -L 82 was around four degrees higher than that of the wild-type protein (Figure 3), indicative of a more stable enzyme. Interestingly, the T m value of the AprEΔL 75 -L 82 T 66 M G 102 D mutant was between the T m values of the AprEΔL 75 -L 82 and wild-type enzymes (Figure 3). These results suggest that the stabilities of the two variants are essentially the same.
The CBSA absence and compensatory mutations on the activity of AprEΔL 75 -L 82 T 66 M G 102 D were determined.   To this end, AprE and AprEΔL 75 -L 82 T 66 M G 102 D were incubated with 0.1 mM Ca 2+ and 100 mM Na + , respectively. Under these incubation conditions, binding sites A and B of subtilisin BPN' were saturated 95% with Ca 2+ and Na + , respectively [24]. The kinetic parameters for thermal inactivation were calculated using (3) to better correlate the effect of amino acid residues substitutions (Thre 66 Met and Gly 102 Asp) on the calcium dependent stability of AprEΔL 75 -L 82 .  (Table 3). On the other hand, in the presence of 0.1 mM Ca 2+ , the t 1/2 for the wild-type enzyme was around 17 times higher than that of the mutant enzyme. However, in the presence of 0.1 mM EDTA, that is, in the absence of calcium, both enzymes showed a similar inactivation rate (Table 3). Therefore amino acid residues substitutions (Thre 66 Met and Gly 102 Asp) led to recover of enzyme activity but had no effect with respect to thermal inactivation of AprEΔL 75 -L 82 .

6
Journal of Biomedicine and Biotechnology The three-dimensional structure of AprE has not been determined but it has been reported for subtilisin BPN' [21]. In fact, as noted above both proteins share 86% identity; therefore their three-dimensional structures are likely to be similar. Thus, the structural analysis using subtilisin BPN' as a model [21] revealed that the mutation Thr 66 Met was found to be in close contact with the active site of the enzyme, in particular interacting with His 64 which acts as a general-base catalyst to activate the γ-OH group of the nucleophile Ser 221 . On the other hand, the mutation Gly 102 Asp was found to occur in the substrate binding subsite S 4 of AprE (Figure 4).
The bulky and nonpolar functional side group of Met suggests that the microenvironment in the active site of the AprEΔL 75 -L 82 mutant was disturbed as a consequence of a polarity change. This alteration may impair the nonpolar residues present in the substrate s-AAPF-pNa (i.e., Phe) that enter in contact with the catalytic residues. Thus, substitution of Thr 66 Met possibly had a positive effect in reestablishing the core environment (polarity) in the active site of AprEΔL 75 -L 82 . Mutations directed to this region might be useful in identifying amino acid substitutions that reestablish the full activity to AprEΔL 75 -L 82 . On the other hand, it has been reported that substitutions of the residues Gly 102 Phe and Ser 128 Phe in savinase, a subtilisin ortholog, blocked the entrance of aromatic residues into the active site pocket, eliminating thus the preference for these residues [25]. Therefore, the introduction of a polar and bulky residue like Asp in position 102 of AprEΔL 75 -L 82 may anticipate an important structural change in the affinity for the substrate.
Overall, the results of the structural and biochemical analysis of the wild-type, AprEΔL 75 -L 82 and AprEΔL 75 -L 82 T 66 M G 102 D proteins, strongly suggest that the local perturbation induced by deletion of the loop L 75 -L 82 were partially compensated by the substitutions T 66 M G 102 D which are located in close vicinity with the catalytic triad of AprE. Therefore, the results described in this work strongly support the idea that in addition to function as a Ca 2+ binding domain, the loopL 75 -L 82 has an important structural role in the enzyme activity of AprE.