Characterization of recombinant prolidase from Lactococcus lactis – changes in substrate specificity by metal cations, and allosteric behavior of the peptidase Soo I. Yang and Takuji Tanaka Department of Food and Bioproduct Sciences, College of Agriculture and Bioresources, University of Saskatchewan, Saskatoon, Canada Fermented foods have significant nutritional value and are receiving growing attention from health-conscious consumers. During fermentation, microbial activity changes the chemical, physical and nutritional attri- butes of the food materials. One of the main changes during fermentation is the production of peptides and amino acids via proteolysis. These compounds are major factors contributing to the flavor of fermented foods. Of these amino acids and peptides, hydrophobic peptides exhibit undesirable bitterness in the fermented foods [1]. For example, in Cheddar cheese, hydropho- bic peptides ranging from 2–23 residues were found to be responsible for bitterness [2]. Hydrophobic peptides produced during fermenta- tion undergo further hydrolysis through general pepti- dase reactions, which result in reduced bitterness. However, peptides that contain proline behave differ- ently from other peptides during general proteolysis. As proline is structurally and chemically unique among the 20 naturally occurring amino acids due to its imine structure, proline-containing peptides are much less susceptible to further enzymatic hydrolysis [3,4]. Thus, hydrolysis of peptides during fermentation can ultimately produce proline-containing dipeptides, such as Xaa-Pro and Pro-Xaa. Ishibashi et al. [5] reported two interesting observations regarding pro- line-containing dipeptides: most of these dipeptides are bitter, and Xaa-Pro is generally more bitter than Pro-Xaa. Proline-containing dipeptides tend to be accumulated as a result of the low susceptibility to enzymatic hydrolysis, and these dipeptides are bitter; therefore, peptidases specific for proline-containing dipeptides could control the bitterness of fermented foods. Keywords bitterness; metallopeptidase; overexpression; PepQ; proline Correspondence T. Tanaka, Department of Food and Bioproduct Sciences, College of Agriculture and Bioresources, University of Saskatchewan, 51 Campus Drive, Saskatoon, Saskatchewan S7N 5A8, Canada Fax: +1 306 966 8898 Tel: +1 306 966 1697 E-mail: takuji.tanaka@usask.ca (Received 1 August 2007, revised 15 October 2007, accepted 16 November 2007) doi:10.1111/j.1742-4658.2007.06197.x The Lactococcus lactis NRRL B-1821 prolidase gene was cloned and over- expressed in Escherichia coli. Under suboptimum growth conditions, recombinant soluble and active prolidase was produced; in contrast, inclu- sion bodies were formed under conditions preferred for cell growth. Recombinant prolidase retained more than half its full activity between 30 and 60 °C, and was completely inactivated after 30 min at 70 °C. CD anal- ysis confirmed that prolidase was inactivated at 67 °C. The enzyme was active under weak alkali to weak acidic conditions, and showed maximum activity at pH 7.0. Although these characteristics are similar to those for other reported prolidases, this prolidase was distinctive for two kinetic characteristics. Firstly, different substrate specificity was observed for its two preferred metal cations, zinc and manganese: Leu-Pro was preferred with zinc, whereas Arg-Pro was preferred with manganese. Secondly, the enzyme showed an allosteric response to changes in substrate concentra- tions, with Hill constants of 1.53 for Leu-Pro and 1.57 for Arg-Pro. Mole- cular modeling of this prolidase suggests that these unique characteristics may be attributed to a loop structure near the active site. Abbreviations IPTG, isopropyl thio-b- D-galactoside; LAB, lactic acid bacteria. FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS 271 Lactic acid bacteria (LAB) are widely used to pro- duce fermented foods. LAB are nutritionally fastidious and require amino acids as exogenous nutrients [6,7]. Required amino acids are assimilated in the form of peptides that are produced from proteins by LAB extracellular proteinase [8]. The assimilated peptides are further hydrolyzed by peptidases in LAB in order to supply free amino acids for metabolism. LAB may have as many as 18 peptidases for efficient hydrolysis of the imported peptides [9]. Of these peptidases, four are proline-specific: proline iminopeptidase, prolinase, X-prolyl dipeptidyl aminopeptidase and prolidase [9]. Prolidase (EC 3.4.13.9) is specific to Xaa-Pro dipep- tides, which can only be hydrolyzed by this peptidase [8]. As mentioned above, Xaa-Pro dipeptides are bitter. Therefore, reduction of the dipeptides via prolidase may lead to the reduction of bitterness in fermented food products. Prolidases have been reported from some microbial sources, such as Lactobacillus delbrueckii subsp. bulgar- icus CNRZ 397 [4,10], Lactobacillus casei subsp. casei IFPL 731 [11], Pyrococcus furiosus [12] and Lb. helveti- cus [13]. These prolidases have preferential activity on Xaa-Pro dipeptides that have a hydrophobic amino acid as the N-terminal residue. The prolidases do not hydrolyze Pro-Pro or Gly-Pro, and have little activity on hydrophilic Xaa-Pro peptides. Previous research has shown that, in the absence of prolidase, LAB growth is retarded by 13% [14]. Most previous research has concentrated on kinetic characterization of LAB physiological activity. As a consequence, there is little information on the expression of recombinant prolidases, the functionality of each residue or protein engineering of this enzyme. In the present study, the prolidase-coding gene, pepQ, was isolated from Lactococcus lactis NRRL B-1821 and cloned. Characterization of the recombinant protein revealed some interesting characteristics of this proli- dase. Moreover, this research provides the means to investigate the structure–function relationships of proli- dase, hence providing a greater understanding of the characteristics of this peptidase, which would be of industrial use in the debittering of fermented foods. Results and Discussion Cloning and expression of Lc. lactis prolidase The prolidase gene of Lc. lactis NRRL B-1821 was successfully isolated using PCR. The DNA sequence of the isolated gene (GenBank accession number EU216565) was virtually identical to that reported for Lc. lactis Il1403 (GenBank accession numbers NC_002662 and AE006395). The sequence of the iso- lated gene has a base difference compared with Lc. lac- tis Il1403, and this substitution results in an amino acid change from Tyr67 to His67 in the putative amino acid sequence. Attempts were made to express the gene using the tac promoter on pKK223-3, and the expres- sion system produced a large amount of recombinant protein under the preferred growth conditions (37 °C in LB broth) of the Escherichia coli TOP 10 F’ hosts. The amount of recombinant protein reached about 50% of total cell extracts, as determined from SDS– PAGE gel densitometry (data not shown). However, the growth under these conditions yielded recombinant prolidase as inclusion bodies. In theory, inclusion bodies can be refolded into an active form; however, it is uncertain whether the refolded proteins have an identical fold to that of the native proteins. Therefore, the above expression system was refined in order to optimize the conditions to produce soluble protein without inclusion body formation. Optimization of expression and purification of recombinant Lc. lactis prolidase The overexpression of recombinant proteins can dis- turb host cell metabolism through their activities. To avoid these negative effects, host cells often produce recombinant proteins as inclusion bodies [15]. As pro- lidases are highly specific for Xaa-Pro, which does not have any specific roles in metabolism of the host cells, activity of prolidases would not have harmful effects on the host cells. Therefore, the rapid and ample expression itself would have caused sufficient stress to the host metabolism. Growth conditions were exam- ined in order to decrease host stress by altering the conditions for expression. Two possibilities for decreasing the stress were postulated: (a) that unfavor- able conditions for E. coli growth (resulting in slower growth) would retard expression of the recombinant prolidase, and (b) that conditions allowing the host more resources for themselves, i.e. rich media or weak induction of expression, could compensate for the con- sumption of energy and substances diverted to produce the recombinant prolidase. Various culture conditions were examined using a pKK223-3–prolidase clone as described in Experimental procedures, and it was found that low temperatures with vigorous aeration yielded recombinant prolidase as a soluble protein. Optimum results were achieved using a low concentra- tion of chloramphenicol (1 lgÆmL )1 ), induction with 1mm isopropyl thio-b-d-galactoside (IPTG) at A 600 = 0.5, and vigorous aeration (200 r.p.m. with a low volume of medium in a large vessel) at 16 °C. Recombinant Lactococcus lactis prolidase S. I. Yang and T. Tanaka 272 FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS Under these conditions, 40 h cultures produced 20– 40% of soluble proteins as recombinant prolidase, as determined by densitometry of the SDS–PAGE gel. Extension of the culture beyond 40 h did not increase prolidase production. Purification of recombinant prolidase was achieved using a two-step process: ammonium sulfate precipita- tion and anion-exchange column chromatography (Fig. 1). Crude extracts were found to contain a sub- stantial amount of recombinant prolidase (40 kDa; lane C), compared with non-induced cell extracts (lane B). Ammonium sulfate precipitation removed most of the contaminating proteins from the crude extracts (lane D). A final purification step using a DEAE– Sephacel column resulted in a single prolidase band, as evidenced by SDS–PAGE (lane E). A 900 mL culture yielded 18.2 mg of purified prolidase, with a purifica- tion factor of 11.8 and 89% recovery of activity from the crude extracts (Table 1). The purified prolidase had 197.2 unitsÆmg )1 (where one unit is as defined in Experimental procedures) of specific activity using 2mm Leu-Pro dipeptide as the substrate in 20 mm sodium citrate buffer (pH 6.5) ⁄ 1mm ZnCl 2 at 50 °C. Characterization of recombinant Lc. lactis prolidase The molecular mass of recombinant prolidase was esti- mated using mass spectrometry and gel filtration. Based on the gene sequence, an estimated molecular mass for the prolidase monomer of 40 kDa (39 970 Da) was determined. The estimated molecular mass (40 164 Da) determined by mass spectroscopy of purified prolidase confirmed this value. This molecular mass may not reflect the native state as mass spectrometers dissociate the protein molecules during the analysis process. The molecular mass of prolidase in the native state (active form) was roughly estimated using four size-exclusion columns. Prolidase appeared in the void volume fraction of the Bio-Gel P-60 column (exclusion limit of 60 kDa) and in later fractions using other columns: Sephadex G-100 (100 kDa) and G-150 (150 kDa), and Bio-Gel P-200 (200 kDa). These results indicate that prolidase is larger than 60 kDa, but smaller than 100 kDa. Only a dimeric structure can have a molecular mass within this range, as the monomer of this prolidase is 40 kDa as shown by mass spectrometry, SDS–PAGE and gene sequence. We therefore propose that Lc. lactis prolidase forms a homodimeric structure with a molecular mass of approximately 80 kDa. This proposed dimeric struc- ture is in agreement with the X-ray crystal structures of P. furiosus (Protein Data Bank accession number 1PV9 [16]) and Pyrococcus horikoshii OT3 (Protein Data Bank accession number 1WY2) prolidases. Recombinant Lc. lactis prolidase exhibited activity over a broad range of temperatures, showing similar activities between 35 and 55 °C (Fig. 2). The reaction rate dropped to 67% at 60 °C, and no activity was observed above 70 °C. This range is broader than that for Lb. delbrueckii prolidase, which has its highest activities between 40 and 50 °C [4]. Figure 2 also shows the temperature stability. The enzyme retained more than 60% of its activity after 30 min incubation below 50 °C; however, incubation at 60 °C decreased Fig. 1. SDS–PAGE gel showing the final purification of recombinant Lactococcus lactis prolidase. Samples from each step of purification process were compared by SDS–PAGE. Lane B, whole cell extracts of non-induced culture; lane C, crude extracts of induced culture; lane D, after 60% ammonium sulfate precipitation; lane E, after DEAE–Sephacel chromatography purification. The arrow indicates a molecular mass of 40 kDa. Lane A shows the molecular mass markers. Table 1. Purification of Lactococcus lactis recombinant prolidase. Purification process Total protein (mg) Total activity (units a ) Specific activity (unitsÆmg )1 ) Yield (% activity) Purification (fold) Cell extract 272 4542 16.7 100 1 Ammonium sulfate precipitation 41 3665 89.4 81 5.4 DEAE–Sephacel 18.2 3589 197.2 79 11.8 a One unit of prolidase activity is defined as hydrolysis of 1 lmol of peptide in 1 min. S. I. Yang and T. Tanaka Recombinant Lactococcus lactis prolidase FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS 273 the residual activity to 22%. Little activity was observed after incubation at 70 °C. The thermal stability is comparable to that of Lb. casei prolidase [11] and higher than that of Lb. delbrueckii proli- dase [4]. This loss of activity in the recombinant Lc. lactic prolidase was most likely due to denaturation between 60 and 70 °C. In order to confirm this speculation, CD analysis was employed. The CD signal started to decline at 60 °C and reached a minimum at 71 °C (Fig. 3), with the denaturation temperature estimated as 67 °C. This observation indicated that the enzyme began to lose structure at 60 °C and was completely denatured at around 70 °C, thereby supporting the speculation that the loss of prolidase activity results from denaturation of the enzyme. Enzyme activity was measured between pH 4 and 10 (Fig. 4). Activity was detected between pH 6.0 and 8.0 and reached a maximum at pH 7 for both Leu-Pro and Arg-Pro. The optimum pH was consistent with values reported for Lb. delbrueckii (pH 6.0) [4], P. furiosus (pH 7) [12], Lb. casei (pH 6.5–7.5) [11] and partially purified Lc. lactis subsp. cremoris AM2 pro- lidases (pH 7.35 and 8.25) [17], although these enzymes worked in narrower pH ranges for Leu-Pro than that for Lc. lactis prolidase. The reported prolidases vary in their metal require- ments, e.g. Lb. delbrueckii prolidase requires zinc [4], P. furiosus prolidase prefers cobalt and manganese [12], and Lb. casei enzyme can utilize magnesium, manganese and cobalt [11]. The recombinant Lc. lactis prolidase showed its highest activity for Leu-Pro with zinc, but the activity with manganese was 21.5% of that with zinc (Table 2). Activity was not detected with other divalent cations, i.e. cobalt, magnesium, nickel, copper and calcium. Substrate specificity of recombinant Lc. lactis prolidase To date, all known prolidases are dipeptide-specific, and cannot hydrolyze larger peptides. Similar results were observed for the prolidase examined in this study (Table 2). Recombinant Lc. lactis prolidase exhibited activity for Leu-Pro, Val-Pro, Phe-Pro, Arg-Pro and Lys-Pro. Based on the peptide hydrolysis assay employed in this study, i.e., quantification of free 0 25 50 75 100 20 30 40 50 60 70 Tem p erature (°C) Relative activity ( : 40°C = 100%) Residual activity ( : 20°C = 100%) Fig. 2. Effect of temperature on recombinant Lactococcus lactis prolidase activity. Open circles represent the observed activity of fresh prolidase at each temperature. The activities are expressed as activities relative to the activity at 40 °C. Closed triangles repre- sent the residual activity of Lc. lactis prolidase after 30 min treat- ment at each temperature. The residual activities are expressed as activities relative to the activity after 20 °C incubation. Tem p eratue (°C) Normalized value (%) 0 20 40 60 80 100 20 30 40 50 60 70 80 90 Fig. 3. Thermal denaturation observed by CD. The observed CD signal at 222 nm is plotted against temperature. The signal intensity is expressed relative to the value at 20 °C. The determined dena- turing temperature, 67 °C, is indicated by a vertical dashed line. 0 20 40 60 80 100 45678910 pH Relative activity (pH 7 = 100%) Fig. 4. pH dependency of recombinant Lactococcus lactis proli- dase. The activity of Lc. lactis prolidase was measured using two dipeptides over a range of pH values. The observed activities are expressed as the activity relative to that at pH 7.0. Open squares and closed circles represent Leu-Pro and Arg-Pro, respectively. Recombinant Lactococcus lactis prolidase S. I. Yang and T. Tanaka 274 FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS proline, no hydrolysis was observed for Gly-Pro, Glu- Pro, Asp-Pro or the two tripeptides Leu-Leu-Pro and Leu-Val-Pro. Interestingly, substrate specificity was dependent on the catalytic metal cation. No activity towards Pro-Pro was seen in the presence of zinc, but low activity was seen in the presence of manganese. Moreover, the preference for dipeptide changed from Leu-Pro to Arg-Pro in the presence of manganese. A comparison of the crystal structure of P. furiosus prolidase and the sequence-based model of Lc. lactis prolidase (Fig. 5) indicated that the S 1 sites are composed mainly of hydrophobic residues (Phe190, Leu193 and Ile308 of Lc. lactis prolidase), suggesting a preference towards hydrophobic residues at the N-terminus of the dipeptides. In fact, the Xaa-Pro dipeptides preferred by Lc. lactis prolidase were mostly hydrophobic peptides, as shown in Table 2. However, the preference for Arg-Pro, and the metal-dependent substrate specificity cannot be explained by the nature of the S 1 site residues described. A notable difference between P. furiosus and Lc. lactis prolidases in the active site area is the length of the loop structure that is contributed by the other subunit and covers the S 1 site (yellow ribbon for P. furiosus and cyan ribbon for Lc. lactis in Fig. 5). The crystal structure of P. furiosus suggests that this loop forms part of the S 1 site. The loop is longer in Lc. lactis by four residues, and the middle of the loop is composed of charged residues (Asp36, His38, Glu39 and Arg40), whereas the Arg295 Ser307 Arg295 Ser307 S 1 ′ S 1 ′ S 1 S 1 Fig. 5. Active site superposition of Lactococcus lactis and Pyrococcus furiosus prolidases. The residues in the active sites of a Lc. lactis prolidase subunit are indicated by thick lines. The S 1 site (Phe190, Leu193 and Val302; blue), S 1 ‘ site (His292, Tyr329 and Arg337; green), and substrate size-limiting residues (Pro306, Ser307 and Ile308; orange) and metal-chelating residues (Asp221, Asp232, His296, Glu325 and Glu339; cyan) are shown. Corresponding residues in P. furiosus are indicated by thin lines. The size-limiting arginine, Arg295, in P. furiosus prolidase [16], and the corresponding residue, Ser307, in Lc. lactis prolidase are labeled. Ribbon models show the loop contributed from the other subunit. The Lc. lactis and P. furiosus prolidase loop structures are in cyan and yellow, respectively. Leu37B of P. furiosus prolidase and Asp36B of Lc. lactis prolidase are shown in the line model on the ribbons. The illustration was generated using the VMD molecular modelling program [27]. Table 2. The relative activities of recombinant Lactococcus lactis prolidase in the presence of zinc or manganese for various peptide substrates. Activity was measured with 2 m M peptides in 20 mM sodium citrate (pH 6.5) and 1 mM metal (zinc or manganese) chlo- ride; activities are expressed relative to the activity for Leu-Pro in the zinc reaction mixture. Substrates Zinc Manganese Leu-Pro 100.0 ± 0.4 21.5 ± 0.6 Phe-Pro 23.8 ± 0.5 15.2 ± 2.7 Val-Pro 14.4 ± 0.4 14.7 ± 3.8 Arg-Pro 12.0 ± 0.9 42.5 ± 1.2 Lys-Pro 6.6 ± 0.5 2.5 ± 0.4 Pro-Pro < 0.1 0.8 ± 0.1 Glu-Pro < 0.1 < 0.1 Gly-Pro < 0.1 < 0.1 Asp-Pro < 0.1 < 0.1 Leu-Leu-Pro < 0.1 < 0.1 Leu-Val-Pro < 0.1 < 0.1 S. I. Yang and T. Tanaka Recombinant Lactococcus lactis prolidase FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS 275 P. furiosus loop has two hydrophilic residues (Thr34 and Ser35). It is speculated that this loop structure contributes to the preference for the Arg-Pro dipep- tide, i.e. methylene groups (b, c, d–carbons) are accommodated in the S 1 site, and the amino group of the side chain is associated with the negatively charged residues (Asp 36 and Glu39) on the loop structure. Allosteric behavior of recombinant Lc. lactis prolidase The relationships between substrate concentrations and reaction rates were examined in order to determine kinetic parameters. Plots of substrate concentration against observed catalytic rate showed sigmoidal curves for both Leu-Pro and Arg-Pro (Fig. 6). The allosteric behavior indicated by the sigmoidal curves was analyzed using the Hill plot, and Hill coefficients of 1.53 and 1.57, respectively, were obtained for Leu- Pro and Arg-Pro (Fig. 6). Although allosteric behavior is not common among proteinases ⁄ peptidases, it has been reported in several proteinases, e.g. cathepsin C [18] and Helicobacter pylori leucyl aminopeptidase [19]. Interestingly, the latter enzymes share characteristics with Lc. lactis prolidase: they hydrolyze peptides with leucine at the N-terminus, they are metallopeptidases, and their 3D structures share similar domain struc- tures (based on the bovine leucyl aminopeptidase structure, 1LAM [20]). Their similarities in 3D struc- ture include (a) two distinctive domains that fold in a ⁄ b structures, (b) an active site located at the center of the C-terminal domain, and (c) an active site that faces another subunit. The Lineweaver–Burk plot using the Hill coefficient (1 ⁄ s H against 1⁄ v plot) gave a Michaelis constant for Leu-Pro of 3.7 mm and a rate constant of 247.9 s )1 . The constants for known prolidases are: Lb. del- brueckii, 2.2 mm and 225.9 s )1 ; Lb. casei, 0.2 mm and 55.1 s )1 [11]; P. furiosus, 3.0 mm and 271 s )1 [12]. Similar to known prolidases [4,12,17], this Lc. lactis prolidase exhibited substrate inhibition above 5 mm Leu-Pro; at 8 mm, the observed activity was 47% of that at 5 mm. Modelling of Lc. lactis prolidase Molecular modeling provides insight regarding the allosteric nature of this enzyme. A molecular model of Lc. lactis prolidase was successfully constructed and used to evaluate the structure–function relationship of this prolidase. The Lc. lactis model was superposed on the P. furiosus model by comparison of their a-car- bons, yielding a root mean square deviation of 1.59 A ˚ . Figure 5 shows models of P. furiosus (1PV9) and Lc. lactis prolidases around the active site zinc ions. The P. furiosus enzyme, which did not exhibit alloste- ric behavior, had a smaller loop structure over the active site (shown as a yellow ribbon in Fig. 5). Maher et al. [16] discussed the contribution of this loop struc- ture as part of the substrate binding site, and it was suggested that Leu37B (B indicates the contribution from the other subunit) formed the hydrophobic S 1 site in cooperation with Phe178, Ile181 and Ile 290. The comparable residues in the Lc. lactis prolidase are Asp36B, Phe190, Leu193 and Ile308, respectively. The positions and characteristics of the latter three residues are comparable to those of P. furiosus prolidase. How- ever, unlike Leu37B of P. furiosus prolidase, Asp36B is a hydrophilic charged residue, and is located on the longer loop structure (cyan ribbon in Fig. 5) that was discussed in the substrate specificity section. In some enzymes, the loop structures have been shown to con- tribute to the activity of enzymes by changing their shape [21,22]. This suggests that the structure of this loop could take a different shape in the event of sub- strate binding, thus the residue comparable to Leu37B of P. furiosus might not be Asp36B but instead another residue on the longer loop. This flexibility may mediate changes in the overall structure of the enzyme via subunit–subunit interaction. Such changes may trigger the allosteric behavior of Lc. lactis prolidase. Another possibility is that Ser307 works as a key resi- due in the allosteric behaviour of this enzyme. This residue is located close to the substrate-binding site, and is replaced by Arg295 in P. furiosus (Fig. 5). Maher et al. [16] suggested that this residue limited the substrate to dipeptides. We suggest that this residue can cooperate with the loop and contribute allosteric behavior to the peptidase. These suggestions, i.e. the 0 0.001 0.002 0.003 0.004 0.005 0.006 01234 Substrate concentration (mM) Observed rate (µmole·min –1 ) –4 –2 0 2 –3 –2 –1 0 1 2 Ln (substrate) Ln [v/(V max –v )] Fig. 6. Allosteric behavior of Lactococcus lactis prolidase. The plots show the relationship between the observed activity and the con- centration of Leu-Pro (open squares) and Arg-Pro (closed circles). The inset shows the Hill plot of the assay with Leu-Pro. Recombinant Lactococcus lactis prolidase S. I. Yang and T. Tanaka 276 FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS contributions of the loop and Ser307, are being exam- ined by our group. Conclusion In this study, we have produced recombinant prolidase in a soluble, active form. The techniques use to achieve solubilization could be used for other difficult- to-express proteins. Recombinant Lc. lactis prolidase exhibited characteristics similar to other prolidases, but possessed distinctive properties of allosteric behav- ior and metal-dependent substrate specificity. Further structure–function relationship studies will provide insights into the behaviour of prolidase, thus contrib- uting to applications of prolidase in fermented food processing. Experimental procedures Enzymes and chemicals Enzymes for genetic engineering were purchased from Fer- mentas (Burlington, Canada) and Invitrogen (Burlington, Canada). All chemicals used in this study were commer- cially available ACS grade, and were purchased from VWR International (Edmonton, Canada). Cultivation of Lactococcus lactis and genomic DNA isolation Lactococcus lactis NRRL B-1821 (Agricultural Research Service culture collection, Peoria, IL, USA) was cultivated in 100 mL of Lactobacillus MRS medium (BD-Difco, Franklin Lakes, NJ, USA) for 24 h at 37 °C without shak- ing. The culture was harvested by centrifugation at 4000 g for 5 min at 4 °C. Harvested cells were treated with pro- teinase K (1 mgÆmL )1 in 50 mm Tris–HCl pH 8.0, 50 mm EDTA, 100 mm NaCl, 0.5% SDS; Roche Diagnostics, Montreal, Canada) at 50 °C for 1 h, then disrupted using phenol. Extracted nucleic acids were collected, and RNA was removed by RNase (Fermentas) treatment. Genomic DNA was purified by ethanol precipitation from the reac- tion mixture. Isolation and cloning of the gene A pair of primers (5¢-GGAGAATTCATGAGCAAAA TTGAACGTATT-3¢;5¢-ATT CTGCAGTTAGAAAATT AATAAGTCATG-3¢) for PCR was designed based on the sequence of the Lc. lactis spp. Il1403 prolidase coding gene (GenBank accession numbers NC_002662 and AE006395) and custom-synthesized (Integrated DNA Technologies Inc., Coralville, IA, USA). The primers possessed EcoRI (N-terminus) and PstI (C-terminus) restriction enzyme sites (indicated by underlining) that flanked the ends of the open reading frame. The PCR reaction mixture contained geno- mic DNA (20 lg), primers (20 pmol each), dNTPs (40 lm each) and Pfu DNA polymerase (0.5 units; Fermentas) in 100 lL of the buffer recommended by the manufacturer. Each PCR reaction cycle consisted of 94 °C for 1 min, 55 °C for 1 min, and 68 °C for 3 min (5 s was added to the 68 °C step for each cycle), and was repeated 30 times. The amplified PCR fragments were hydrolyzed using EcoRI and PstI, and were then introduced into EcoRI–PstI-digested pUC18 plasmids. The recombinant DNA was transformed into E. coli TOP10F’, and positive clones were verified by DNA sequencing. First, the sequenced gene was isolated using EcoRI–PstI restriction enzymes, and subcloned into the same sites of the pKK223-3 vector [23]. Then, the con- structed recombinant DNA was transformed into E. coli TOP10F’. The expression was examined under a variety of conditions in order to obtain recombinant prolidase in a soluble form. The various conditions studied were: (a) cul- ture medium (LB or 2YT medium), (b) the concentration of the inducing agent for the tac promoter (0.1, 1 or 10 mm IPTG), (c) protein synthesis inhibition using sublethal con- centrations of chloramphenicol (0.1 or 1 lgÆmL )1 ), (d) media with higher osmotic pressures (0.5 or 2% w ⁄ v NaCl), (e) schedules of the induction (induced at A 600 =0.4, 0.5, 0.8 or 1.2), (f) the pH of the medium (pH 5.5 or 7.5), (g) the aeration conditions (100 or 200 r.p.m.), (h) the culture temperature (16, 18, 22, 29, 30, 33 or 37 °C), and (i) the duration of the culture (16, 40, 72 or 96 h). These condi- tions were tried individually or concurrently. Purification of the recombinant Lc. lactis prolidase The recombinant E. coli was cultured in LB broth (pH 5.5) at 16 °C. The culture was carried out in 18 500 mL flasks with 50 mL medium in each (total 900 mL). Expression was induced by addition of 1 mm IPTG and chlorampheni- col (1 lgÆmL )1 ) when the A 600 reached 0.5. The culture was vigorously shaken at 200 r.p.m. for 40 h before harvesting. The harvested cells were resuspended in a lysis buffer solu- tion (20 mm sodium citrate buffer, pH 6.0, 1 mm zinc sulfate, 100 mm sodium chloride, 8 lgÆmL )1 RNase and 0.2 mgÆmL )1 lysozyme), and disrupted using ultrasonica- tion. After removal of some proteins from the crude extracts by 40% saturated ammonium sulfate precipitation, the prolidase fraction was recovered using 60% saturated ammonium sulfate precipitation. The recovered prolidase in the precipitate was dissolved in 20 mm sodium citrate (pH 6.0) ⁄ 1mm zinc sulfate and dialyzed against 2 L of the same buffer twice. The dialyzed sample was applied to a DEAE–Sephacel anion exchange column (3 diameter · 15 cm; GE Healthcare, Chalfont St Giles, Buckingham- shire, UK), and prolidase was eluted using a 600 mL linear gradient from 0 to 0.5 m NaCl in 20 mm sodium citrate S. I. Yang and T. Tanaka Recombinant Lactococcus lactis prolidase FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS 277 (pH 6.0) ⁄ 1mm zinc sulfate. The prolidase fractions were concentrated and desalted using an YM30 Amicon Ultracell filtration system (Millipore, Billerica, MA, USA). The pur- ity of the prolidase was densitometrically estimated by SDS–PAGE using Coomassie Brilliant Blue G250 and NIH image software (developed at the US National Institutes of Health and available at http://rsb.info.nih.gov/nih-image/). The purified sample in 20 mm sodium citrate (pH 6.0) ⁄ 1mm zinc sulfate was mixed with the same vol- ume of glycerol and stored at )20 °C until use. Enzyme activity assay The amount of proline liberated from the peptide substrates was determined using the ninhydrin method [24]. Dipep- tides were hydrolyzed in 20 mm sodium citrate buffer (pH 6.5) ⁄ 1mm zinc chloride. The reaction was initiated by the addition of enzyme solution. At 1 min intervals, an aliquot (20 lL) was withdrawn and mixed with 50 lLof glacial acetic acid and 50 lL of ninhydrin reagent (3% w ⁄ v ninhydrin, 60% v ⁄ v glacial acetic acid, 40% v ⁄ v phosphoric acid). The mixture was boiled for 10 min to develop the color, and then cooled on ice. The resulting chromophore was quantified using 515 nm absorption. All measurements were carried out at least in triplicate. One unit of prolidase activity is defined as hydrolysis of 1 lmol of peptide in 1 min. Measurement of substrate specificity The peptide substrates examined were Leu-Pro, Val-Pro, Phe-Pro, Gly-Pro, Arg-Pro, Lys-Pro, Pro-Pro, Asp-Pro, Glu-Pro, Leu-Val-Pro and Leu-Leu-Pro. The peptides (2 mm) were hydrolyzed in 20 mm sodium citrate (pH 6.5) ⁄ 1mm zinc chloride or manganese chloride at 50 °C. pH dependency The pH dependency of prolidase was examined using the following buffer solutions in place of the sodium citrate buffer in the method described above: 20 mm sodium cit- rate (pH 4–5.5), 20 mm MES (pH 6.0–7.0), 20 mm Tris– HCl (pH 7.5–9) and 20 mm sodium borate (pH 10). The activity was analyzed using 2 mm Leu-Pro or Arg-Pro and 1mm manganese chloride at 50 °C. Thermal stability and dependency Recombinant prolidase was incubated in 20 mm sodium cit- rate buffer (pH 6.5) ⁄ 1mm zinc chloride at the designated temperature (20, 30, 40, 50, 60 or 70 °C) for 30 min. The residual activity was determined in order to evaluate the stability of prolidase. The temperature dependency was separately examined in reactions using fresh enzyme at vari- ous temperatures (20, 30, 35, 40, 45, 50, 55, 60 and 70 °C). In both experiments, 2 mm Leu-Pro was used as the substrate. Metallic ion dependency A variety of metal cations were tested for their effects on the prolidase activities. Metal salts of zinc chloride, nickel chloride, cobalt nitrate, copper sulfate, manganese chloride, magnesium chloride and calcium chloride were used. The activities were measured in 20 mm sodium citrate (pH 6.5) ⁄ 1mm solutions of each metal salt with 2 mm Leu- Pro at pH 6.5. Thermal denaturation temperature measurement for recombinant prolidase The CD spectrum of purified prolidase was analyzed in 20 mm sodium phosphate buffer (pH 6.0) using a PiStar- 180 spectroscope (Applied Photophysics Ltd, Leatherhead, Surrey, UK) at the Saskatchewan Structural Sciences Cen- tre (University of Saskatchewan, Saskatoon, Canada). The thermal denaturation temperature was determined from the change in the CD spectrum at 222 nm over a temperature range from 25 to 90 °C. The denaturation temperature was determined as the temperature at which the rate of CD spectrum change reached its maximum [15]. Mass spectrometry The molecular mass of the recombinant prolidase molecule was determined using a mass spectrometer (API Q-star XL hybrid MS system; Applied Biosystems, Foster City, CA, USA) using the electrospray ionization method at the Sas- katchewan Structural Sciences Centre; measurements were carried out on the desalted sample in deionized water. Molecular mass estimation using gel filtration The purified prolidase protein (0.8 lgin10lL) was loaded onto gel filtration columns: Sephadex G-100 and G-150 (GE Healthcare) and Bio-Gel P-60 and P-200 (Bio-Rad Laboratories, Hercules, CA, USA). The size of the columns was 0.5 diameter · 10 cm, and 20 mm sodium citrate buffer (pH 6.5) ⁄ 1mm ZnCl 2 was used as the eluant. The eluate was fractionated, and the activities of the fractions were qualitatively checked using Leu-Pro in order to determine the prolidase fractions. Computational molecular modelling The protein sequence, deduced from the DNA sequence of the prolidase gene, was submitted to the 3d-jigsaw server Recombinant Lactococcus lactis prolidase S. I. Yang and T. Tanaka 278 FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS (http://www.bmm.icnet.uk/servers/3djigsaw/) [25] in order to create an initial molecular model of Lc. lactis prolidase. The initial model was derived based on the crystal structure of P. furiosus (Protein Data Bank accession number 1PV9 [16]) using the default parameters of the server. This model was then energy-minimized using the namd molecular modelling program [26]. The calculation used topology force field data provided with the program, and was carried out in a water- filled box. The cut-off distance was set to 15 A ˚ and the calcu- lation run for 5000 iterations. The minimized model was compared with the P. furiosus model (Protein Data Bank accession number 1PV9 [16]) using the vmd software package [27] on a Macintosh G4 computer. Acknowledgements This research was supported by a grant from the Natu- ral Sciences and Engineering Research Council of Canada. The authors appreciate the assistance of Lili Liu and Guodong Zhang with the laboratory work. George Khachatourians, Rickey Yada (Guelph, Canada), Nicholas Low, Michael Nickerson and Sylvia Yada (Guelph, Canada) are acknowledged for their helpful suggestions in the preparation of the manu- script. References 1 Ishibashi N, Kouge K, Shinoda I, Kanehisa H & Fukui S (1988) A mechanism for bitter taste sensibility in pep- tides. Agric Biol Chem 52, 819–827. 2 Sullivan JJ & Jago GR (1972) The structure of bitter peptides and their formation from caseins. Aust J Dairy Technol 27, 98–104. 3 Morel FC, Gilbert C, Geourjon C, Frot-Coutaz J, Port- alier R & Atlan D (1999) The prolyl aminopeptidase from Lactobacillus delbrueckii subsp. bulgaricus belongs to the alpha ⁄ beta hydrolase fold family. Biochim Bio- phys Acta 1429, 501–505. 4 Morel F, Frot-Coutaz J, Aubel D, Portalier R & Atlan D (1999) Characterization of a prolidase from Lactoba- cillus delbrueckii subsp. brugaricus CNRZ 397 with an unusual regulation of biosynthesis. Microbiology 145, 437–446. 5 Ishibashi N, Kubo T, Chino M, Fukui H, Shinoda I, Kikuchi E, Okai H & Fukui S (1988) Taste of proline- containing peptides. Agric Biol Chem 52, 95–98. 6 Juillard V, Le Bars D, Kunji ER, Konings WN, Gripon JC & Richard J (1995) Oligopeptides are the main source of nitrogen for Lactococcus lactis during growth in milk. Appl Environ Microbiol 61, 3024–3030. 7 Mills OE & Thomas TD (1981) Nitrogen sources for growth of lactic acid Streptococci in milk. N Z J Dairy Sci 16, 43–55. 8 Kunji ERS, Mierau I, Hagting A, Poolman B & Konings WN (1996) The proteolytic systems of lactic acid bacteria. Antonie Van Leeuwenhoek 70, 187– 221. 9 Christensen JE, Dudley EG, Pederson JA & Steele JL (1999) Peptidases and amino acid catabolism in lactic acid bacteria. Antonie van Leeuwenhoek 76, 217–246. 10 Rantanen T & Palva A (1997) Lactobacilli carry cryptic genes encoding peptidase-related proteins: characteriza- tion of a prolidase gene (pepQ) and a related cryptic gene (orfZ) from Lactobacillus delbruekii subsp. bulgari- cus. Microbiology 143, 3899–3905. 11 Fernandez-Espla MD, Martin-Hernandez MC & Fox PF (1997) Purification and characterization of a proli- dase from Lactobacillus casei subsp. casei IFPL 731. Appl Environ Microbiol 63, 314–316. 12 Ghosh M, Grunden AM, Dunn DM, Weiss R & Adams MWW (1998) Characterization of native and recombi- nant forms of an unusual cobalt-dependent proline dipep- tidase (prolidase) from the hyperthermophilic archaeon Pyrococcus furiosus. J Bacteriol 180, 4781–4789. 13 Varmanen P, Steele J & Palva A (1996) Characteriza- tion of a prolidase gene and its product and an adjacent ABC transporter gene from Lactobacillus helveticus . Microbiology 142, 809–816. 14 Christensen JE & Steele JL (2003) Impaired growth rates in milk of Lactobacillus helveticus peptidase mutants can be overcome by use of amino acid supple- ments. J Bacteriol 185, 3297–3306. 15 Creighton TE (1993) Conformational properties of polypeptide chains. In Proteins: Structures and Molecu- lar Properties, pp. 171–198. WH Freeman, New York, NY. 16 Maher MJ, Ghosh M, Grunden AM, Menon AL, Adams MWW, Freeman HC & Guss JM (2004) Structure of the prolidase from Pyrococcus furiosus. Biochemistry 43, 2771–2783. 17 Booth M, Jennings V, Fhaolain IN & Ocuinn G (1990) Prolidase activity of Lactococcus lactis subsp. cremoris Am2 – partial purification and characterization. J Dairy Res 57, 245–254. 18 Gorter J & Gruber M (1970) Cathepsin C: an allosteric enzyme. Biochim Biophys Acta 198, 546–555. 19 Dong L, Cheng N, Wang M-W, Zhang J, Shu C & Zhu D-X (2005) The leucyl aminopeptidase from Helicobact- er pylori is an allosteric enzyme. Microbiology 151, 2017–2023. 20 Stra ¨ ter N & Lipcomb WN (1995) Two-metal ion mech- anism of bovine lens leucine aminopeptidase: active site solvent structure and binding mode of l-leucinal, a gem-diolate transition state analogue, by X-ray crystal- lography. Biochemistry 34, 14792–14800. 21 Tanaka T, Teo KSL, Lamb KM, Harris LJ & Yada RY (1998) Effect of replacement of conserved Tyr75 on S. I. Yang and T. Tanaka Recombinant Lactococcus lactis prolidase FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS 279 the catalytic properties of porcine pepsin A. Protein Pept Lett 5, 19–26. 22 Kato H, Tanaka T, Yamaguchi H, Hara T, Nishioka T, Katsube Y & Oda J (1994) Flexible loop that is novel catalytic machinery in a ligase. Atomic structure and function of the loopless glutathione synthetase. Biochemistry 33, 4995–4999. 23 Brosius J & Holy A (1984) Regulation of ribosomal RNA promoters with a synthetic lac operator. Proc Natl Acad Sci USA 81, 6929–6933. 24 Yaron A & Mlynar D (1968) Aminopeptidase P. Biochem Biophys Res Commun 32, 658–663. 25 Bates PA, Kelley LA, MacCallum RM & Sternberg MJE (2001) Enhancement of protein modeling by human intervention in applying the automatic programs 3D-JIGSAW and 3D-PSSM. Proteins 5, 39–46. 26 Phillips JC, Braun R, Wang W, Gumbart J, Tajkhors- hid E, Villa E, Chipot C, Skeel RD, Kale L & Schulten K (2005) Scalable molecular dynamics with NAMD. J Comput Chem 26, 1781–1802. 27 Humphrey W, Dalke A & Schulten K (1996) VMD – visual molecular dynamics. J Mol Graph 14, 33–38. Recombinant Lactococcus lactis prolidase S. I. Yang and T. Tanaka 280 FEBS Journal 275 (2008) 271–280 ª 2007 The Authors Journal compilation ª 2007 FEBS . Characterization of recombinant prolidase from Lactococcus lactis – changes in substrate specificity by metal cations, and allosteric behavior of the peptidase Soo I. Yang and Takuji. protein engineering of this enzyme. In the present study, the prolidase- coding gene, pepQ, was isolated from Lactococcus lactis NRRL B-1821 and cloned. Characterization of the recombinant protein revealed. understanding of the characteristics of this peptidase, which would be of industrial use in the debittering of fermented foods. Results and Discussion Cloning and expression of Lc. lactis prolidase The