A tryptophan residue is identified in the substrate binding of penicillin G acylase from Kluyvera citrophila

A tryptophan residue is identified in the substrate binding of penicillin G acylase from Kluyvera citrophila

Enzyme and Microbial Technology 40 (2007) 1389–1397 A tryptophan residue is identified in the substrate binding of penicillin G acylase from Kluyvera...

985KB Sizes 3 Downloads 12 Views

Enzyme and Microbial Technology 40 (2007) 1389–1397

A tryptophan residue is identified in the substrate binding of penicillin G acylase from Kluyvera citrophila R. Suresh Kumar a , A.A. Prabhune a , A.V. Pundle a , M. Karthikeyan b , C.G. Suresh a,∗ a

Division of Biochemical Sciences, National Chemical Laboratory, Dr. Homi Bhabha Road, Pune 411008, India b Information Division, National Chemical Laboratory, Dr. Homi Bhabha Road, Pune 411008, India Received 31 July 2006; received in revised form 22 September 2006; accepted 11 October 2006

Abstract Penicillin acylases are important enzymes in pharmaceutical industry for the production of semi-synthetic ␤-lactam antibiotics via the key intermediate 6-aminopenicillanic acid. The penicillin G acylase purified from Kluyvera citrophila (KcPGA) on modification with tryptophanspecific reagents such as N-bromo succinamide (NBS) and 2-hydroxy 5-nitrobenzylbromide (HNBB) showed partial loss of activity and substrate protection. Various solute quenchers and substrate were used to probe the microenvironment of the putative reactive tryptophan through fluorescence quenching. Homology modeling of KcPGA structure has been carried out. Docking substrate on this modeled KcPGA structure identifies the tryptophan residue that is directly influenced by substrate binding. To confirm the biological significance of this particular tryptophan, we did a sequence comparison of PGAs from various organisms. The sequence alignment clustered the matches into two sets, those closer to (>40% identical) KcPGA and had the tryptophan of interest present in them formed the first set, while those less identical (<30%) to KcPGA and the particular tryptophan absent in them formed the second set. It is clear from the reported kinetic parameters of representative members of these two sets that the affinity for penicillin G (penG) of the former class is several times better. Thus, based on our studies we suggest that the tryptophan residue in the identified position is important for binding substrate penG by the acylases. © 2006 Elsevier Inc. All rights reserved. Keywords: K. citrophila; Penicillin G acylase; Tryptophan modification; Fluorescence measurement; Substrate-docking; Sequence alignment

1. Introduction The microbial enzymes such as penicillin G acylases (penicillin amidohydrolases, PGAs, EC 3.5.1.11) have a high impact on the pharmaceutical industry by their application in the production of antibiotics. They are employed in the deacylation of benzyl penicillin to 6-aminopenicillianic acid (6-APA), the precursor molecule for production of semi-synthetic penicillins [1–3]. Their high efficiency has resulted in the replacement of conventional chemical processes in favor of enzymatic ones by the industry [4]. The penicillin G acylases (PGAs) have been purified and characterized from various sources [5–11]. The PGA from Kluyvera citrophila (KcPGA) has attracted attention due to its better and more suitable features for industrial applications as compared to PGA from Escherichia coli (EcPGA). It is comparatively easier to immobilize KcPGA and shows more stability



Corresponding author. Tel.: +91 20 25902236; fax: +91 20 25902648. E-mail address: [email protected] (C.G. Suresh).

0141-0229/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.enzmictec.2006.10.022

towards extreme conditions of temperature, pH, and presence of organic solvents [12–15]. The stabilization of the enzyme by immobilization is also reported [16]. KcPGA is translated as a single-chain precursor consisting of 844 amino acid residues in the cytoplasm. Subsequent processing by removal of a 26 residue signal peptide and a 54 residue spacer peptide produces in periplasm the mature enzyme in the form of a heterodimer consisting of an ␣- and a ␤-chain of 209 and 555 amino acid residues, respectively [17]. It has been reported that altering amino acid residue Gly ␤21 affects protein maturation [18]. The serine residue with a newly generated free ␣-amino group at the N-terminus of ␤-subunit acts both as nucleophile and as base in catalysis. Thus, KcPGA can be placed in the Ntn hydrolase family, since having Ser at the N-terminus of ␤-chain and the modeled three-dimensional structure showing characteristic Ntn-hydrolase ‘␣–␤–␤–␣’ tertiary fold [19,20]. There have been reports of this enzyme being cloned and overexpressed in E. coli [21,22]. The active site of the enzyme has drawn special attention due to its importance in understanding the catalytic mechanism and substrate specificity for application in protein engineering

1390

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

[23–25]. Thus, it is important to identify the residues that influence the specificity, activity and contribute to the stability of the enzyme. The residues Arg ␣145, Phe ␣146 and Tyr ␤31 have been previously identified to participate in substrate binding [26,27]. Here, we show by using active site chemical modification studies, kinetic analysis, fluorescence spectroscopy, homology modeling, molecular dynamics and bioinformatics techniques the importance of a particular tryptophan residue in substrate binding of PGA enzymes. 2. Materials and methods Benzylpenicillin (penicillin G or penG) and 6-APA were kindly provided by Hindustan Antibiotics Ltd., India. Tryptophan specific reagent N-bromo succinamide (NBS) and 2-hydroxy 5-nitrobenzylbromide (HNBB) were purchased from ICN Biochemicals (Ohio, USA). Other analytical grade reagents were procured from Sigma (USA).

2.1. Enzyme assay KcPGA was prepared and purified using reported procedures [27]. The PGA activity was assayed by incubating 32 ␮g of KcPGA (0.37 ␮mol) at 40 ◦ C for 10 min. The volume of reaction mixture was made up to 1 ml in 50 mM potassium phosphate buffer (pH 7.5) containing 1 mM penG. 6-APA formed was estimated using p-dimethyl aminobenzaldehyde [28]. One unit of enzyme activity was defined as the amount of enzyme required to produce 1 ␮mol of 6-APA per min under standard assay conditions. Kinetic constants Km and kcat were determined by incubating the enzyme at various concentrations of penG (0.1–20 mM) under standard assay conditions, then fitting a linear regression curve to data points using Lineweaver–Burk plot.

2.2. Protein estimation Protein concentration was determined according to the method of Lowry et al. [29] using BSA as standard. 2.2.1. Modification of tryptophan with N-bromo succinimide To measure the effect of Trp modification on enzyme activity, 320 ␮g (3.72 nM) KcPGA in 10 mM sodium acetate buffer, pH 5.5 was incubated with different concentrations of NBS (10–100 ␮M). Ten microlitre aliquots were withdrawn every 2 min intervals for 15 min and the residual activity measured under standard assay conditions. The number of tryptophan residues that reacted with NBS was calculated by measuring the decrease in absorbance at 280 nm as described by Spande and Witkop [30]. Under same conditions of enzyme as described above NBS (5–200 ␮M) was added in aliquots of 10 ␮l in 10 installments at every 2 min intervals and absorbance recorded. After each addition, an aliquot of 10 ␮l was removed and the reaction arrested by the addition of 90 ␮l of 50 mM l-tryptophan. The residual activity was determined under standard assay conditions. NBS mediated inactivation was monitored by measuring the decrease in absorbance at 280 nm. The same procedure was followed in the presence of enzyme inhibitor phenylacetate. The number of tryptophan residues modified was determined using an estimated value of 5500 M−1 cm−1 for extinction coefficient. Enzyme samples incubated in the absence of NBS served as control. 2.2.2. Modification of tryptophan with 2-hydroxy 5-nitrobenzylbromide KcPGA in same conditions as used for reactions with NBS was incubated with 10–40 mM HNBB. The solution of HNBB was freshly prepared in dry acetone. Aliquots of 10 ␮l were withdrawn every 2 min after an initial lapse of 10 min till the end of 20 min and the residual activity measured under standard assay conditions. The number of tryptophan residues modified was determined based on estimated OD at 410 nm and assuming a molar absorption coefficient of 18,000 M−1 cm−1 .

2.3. Substrate protection studies The protection provided by substrate and a competitive inhibitor during modification reactions of NBS and HNBB was determined by incubating the enzyme with varying concentrations of benzylpenicillin (substrate) or phenylacetate (competitive inhibitor) to final concentration of 50 mM, prior to treatment with modifying reagents under the reaction conditions. Thereafter, the reacted mixture as well as control was passed through sephadex PD 10 desalting column and the residual activity determined under standard assay conditions. Deactivation in the column was estimated using control.

2.4. Circular dichroism analysis Both the NBS and HNBB modified KcPGA samples were passed through sephadex G25 column to remove excess reagent and then the CD spectra were recorded. Untreated enzyme passed through the same column is used to record native spectra.

2.5. Fluorometric studies Fluorescence measurements were performed on a Perkin-Elmer spectroflourimeter LS 5B, using an excitation and emission slit width of 5 nm, 3.81 ␮M of KcPGA in 10 mM sodium acetate buffer, pH 5.5 was excited at 280 nm and the emission spectra was recorded in the range of wavelength 300–400 nm. Fluorescence quenching due to modification with NBS and HNBB was measured by adding 2 ␮l aliquots of the modifier from 500 ␮m and 20 mM stocks, respectively, till the relative fluorescence intensity lowered substantially. Suitable controls were included to correct for changes in enzyme dilution. Fluorescence quenching due to substrate and competitive inhibitor was measured by the progressive addition of 10 ␮l aliquots of either penG or phenylacetate from a 500 mM stock. Correction was applied to compensate for changes in enzyme concentration due to addition of substrate and inhibitor solutions. Substrate protection against chemical modification was determined by incubating the enzyme sample with 0.5 mM penG prior to the addition of 200 ␮M NBS, the fluorescence quenching was then measured. To assess the extent of tryptophan participation in substrate binding, fluorescence measurement was carried out with the addition of penG in aliquots of 0.5 mM (starting with 0.5 mM and final concentration of 2 mM) to NBS modified KcPGA. The intensity of PGA saturated with penG (F∞ ) was obtained from experimental data by plotting 1/(F0 − F) versus 1/[S] and extrapolating the curve to Y-axis. Here, F0 is defined as the fluorescence intensity of the enzyme alone and F is the fluorescence intensity of the enzyme at particular penG concentration [S]. log [(F0 − F)/(F − F∞ )] was plotted against log [S] to estimate the association constant (Ka ) of the complex which was same as the value of [S] when log [(F0 − F)/(F − F∞ )] = 0. The change in free energy of association was determined using the equation: G = −RT ln Ka

(1)

The thermodynamic parameters viz. enthalpy change (H) and entropy change (S) were calculated from vant Hoff’s analysis of temperature dependence of Ka by using the equation: ln Ka =

S −H + . RT R

(2)

Enthalpy change was calculated from the slope of the curve of ln Ka versus 1/T which equals (−H/R). The entropy change was then obtained from the equation: G = H − TS.

(3)

To study the nature of the microenvironment of tryptophan residue 8 M acrylamide, 5 M potassium iodide (KI) and 5 M cesium chloride (CsCl), each separately, has been added to the enzyme sample and tested. 10 mM sodium thiosulfate was included in case of KI to prevent formation of I3 −

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

1391

2.7. Sequence comparison

ions. Corrections were made for contributions from buffer and quenchers to fluorescence and to compensate for the dilution caused by the addition of quenchers. At the highest concentrations of the quenchers, the volume changes accounted was only less than 10% of the initial volume of enzyme solution. The Stern–Volmer equation was used to analyze the quenching of tryptophan fluorescence:

The program BLAST [35] was used to search for homologous sequences of KcPGA in the database. The program CLUSTALW was used for multiple sequence alignment [36]. The postscript outputs from aligned sequences were generated using Espript 2.2 [37].

F0 = 1 + Ksv [Q] F

3. Results

(4)

where F0 and F are the fluorescence intensities in the absence and in the presence of the quencher [Q] the concentration of the quencher, and Ksv is the effective quenching or the Stern–Volmer constant [31]. The fraction of the fluorophore accessible to the quencher was determined using modified Stern–Volmer plot [32]:

 F  0 F0 − F

=



1 1 + (fa Ksv ) (1/[Q]) fa

 (5)

where fa is the effective fractional accessible fluorescence. The parameters F, F0 , and [Q] are same as those defined in Eq. (4).

2.6. Homology modeling and docking A multiple sequence alignment between KcPGA and other penicillin G acylases was carried out using the programs MOE (Chemical Computing Group Inc., http://www.chemcomp.com/) and ClustalW (http://www.ebi.ac.uk/clustalw/). These sequence alignments were subsequently used for modeling KcPGA based on a homologous PDB structure (code, 1GK9) and generated 10 models in the MOE program. These models were subjected to coarse energy minimization procedure in order to remove any possible short-contacts between atoms. The best model predicted by MOE had a score of ±2.9. The predicted models were further evaluated for correct geometry, stereochemistry, and minimum energy distributions. The programs SFCHECK and PROCHECK [33] were used to establish reliability of the modeled structure and to assess the quality of predicted models of KcPGA. In addition, to compare the variability of models, C␣ atom traces and backbone atoms were superposed onto the template crystal structure and RMSD between the positional parameters of equivalent atoms calculated. The protein structures were visualized and analyzed by using programs PyMOL0.97 (http://www.pymol.org) and QUANTA (Accelrys, USA). The differences in active site characteristics of the reference crystal structure and the predicted models were studied by generating the surfaces of active site cavities. The composition of residues, shape, and binding pattern of the active site was also investigated. Molecular dynamics (MD) calculations for enzyme structure were carried out using the MOE program. The CHARMm22 force field distribution provided in MOE was employed. Energy minimization after adding hydrogen atoms was performed to remove bad contacts. Substrates were built using fragments from CHEMDRAW [Cambridgesoft, UK]. The resultant structures were energy minimized using 1000 steps of steepest descents minimization followed by conjugate gradient minimization until convergence reached 0.01 kcal mol−1 A−1 with a distance dependent dielectric constant of 80. These molecular structures served as inputs for docking experiments. The substrate was docked using pharmacophore search algorithm Ph4Dock [34]. Ten possible conformations of the protein-bound substrate were ranked according to the ligand’s internal energy. The van der Waals and electrostatic interaction energy terms of the potential energy function were used and the conformation of side chains kept flexible. The first binding mode had the lowest van der Waals energy term and its orientation was found conducive for deacylation. The two most favorable binding modes of ligands were selected for evaluating protein–substrate complexes. The energy minimization using MMFF94s force field in MOE was carried out to reach a final energy gradient of 0.01 kcal mol−1 A−1 . In the ‘protein + ligand’ minimizations, all the protein side chains were kept fully flexible. The sum of the internal energies of protein and ligand in the complex was subtracted from the total potential energy of the energy-minimized complex to obtain the actual interaction energy between protein and ligands.

3.1. Tryptophan modification by NBS and HNBB Treatment of KcPGA with 100 ␮M NBS and 40 mM HNBB resulted in 72 and 60% loss of activity, respectively, that tentatively indicated the presence of tryptophan residues in the active site. Under the conditions of treatment, the HNBB can simultaneously affect both tryptophan and cysteine residues. However, absence of cysteine residues in KcPGA eliminates the ambiguity about the type of residue modified. Spectrophotometric titration of KcPGA using 5–100 ␮M NBS at pH 5.5 resulted in a progressive decrease in absorption at 280 nm. The number of tryptophan residues oxidized per mole of the enzyme calculated by the extrapolation of tangent from the inflexion point of progression curve was one. Complete inactivation of the enzyme occurred upon modification of a single tryptophan residue. Fig. 1 shows the time dependent inactivation of KcPGA. As can be seen, all the tryptophans that are accessible to modification are oxidized within the first 10 min of the reaction. The plots of the logarithm of residual activity versus time of incubation with various NBS concentrations were linear up to the end of test periods, indicating pseudo-first-order

Fig. 1. Determination of the order of modification reaction of KcPGA with respect to tryptophan-specific modifying reagent NBS at pH 5.5 and 25 ◦ C. NBS concentrations were 10 ␮M (), 20 ␮M (), 40 ␮M () and 100 ␮M (×). Inset: Kinetics of inhibition of KcPGA by NBS. The pseudo-first-order rate constants (Kiapp) were plotted against various concentrations of NBS.

1392

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397 Table 1 Protection of KcPGA against inactivation by tryptophan-specific reagent

Fig. 2. The titration of KcPGA with NBS is plotted. Tryptophan residues were quantified by the stepwise addition of NBS as described in the text.

kinetics of inactivation. The individual slopes of the plots were calculated to determine the respective first-order rate constant Kapp. The order of the reaction (n) in the case of NBS determined from plots of log [Kapp] versus log (NBS concentration [M]) gave value n = 1 indicating that the modification of a single tryptophan residue resulted in inactivation of a mole of the enzyme (inset of Fig. 1). Modification of KcPGA with 40 mM HNBB showed time dependent decrease in residual activity and the maximum absorbance observed at 410 nm. Extrapolation of the tangent from inflexion point, in the plot of residual activity versus number of tryptophan residues modified, indicated modification of a single tryptophan residue per mole of the enzyme (Fig. 2). No variation was detected in case of modification under denaturing conditions.

Treatment

Enzyme activity (% of the initial activity)

None N-Bromosuccinamide (100 ␮M) Benzylpenicllin (50 mM) + N-bromosuccinamide (100 ␮M) Phenylacetate (50 mM) + N-bromosuccinamide (100 ␮M) HNBBr (20 mM) Benzylpenicllin (50 mM) + HNBBr (20 mM) Phenylacetate (50 mM) + HNBBr (20 mM)

100 30 74

68

42 82 85

3.4. Circular dichorism analysis The shape of the CD spectra of modified KcPGA was similar to that of untreated enzyme indicating that no serious structural changes occurred on treatment with NBS and HNBB (Fig. 3). 3.5. Fluorometric studies Fluorescence spectra of untreated KcPGA showed λmax at 348 nm upon excitation at 280 nm. Addition of small increments

3.2. Substrate protection studies The protective action of penG (substrate) and phenylacetate (competitive inhibitor) on the inactivation of KcPGA by NBS is shown in Table 1. In the presence of 50 mM (highest concentration tried) each of penG and phenylacetate the percentage of original activity retained by KcPGA was 74 and 68, respectively. Thus it is clear that both the substrate penG and the inhibitor phenylacetate affect the tryptophan modification to the same extent. 3.3. Michaelis–Menten kinetics The Km values of the native and NBS modified KcPGA estimated were 16 and 53 ␮M, whereas the values of kcat remained same (63 s−1 ) (Table 2).

Fig. 3. The CD spectra of untreated KcPGA () and that chemically modified with NBS (—) and HNBB (). The spectra were recorded on a JASCO-710 spectropolarimeter from 190 to 260 nm using 1 cm path legth at 25 ◦ C and an enzyme concentration of 0.5 mg/ml in 10 mM sodium acetate buffer, pH 5.5. X-axis: wavelength, Y-axis: ellipticity.

Table 2 The values of Km and kcat for unmodified and NBS or HNBS treated enzyme Reagent concentration

Km (␮M penicillin G)

kcat (s−1 )

kcat /Km (␮M−1 s−1 )

Unmodified KcPGA PGA + N-bromosuccinamide (100 nM) PGA + 2-hydroxy 5-nitrobenzylbromide (10 mM)

16 54 51

63 63 63

3.9 1.2 1.2

The Vmax and Km values of native and modified penicillin G acylases were estimated from the Lineweaver–Burk plots by measuring the initial reaction rate at standard assay condition. kcat values were calculated from equation Vmax = kcat [E]t.

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

of NBS and HNBB resulted in progressive lowering (55.5% decrease) of relative fluorescence from 366 to 160 AU (arbitrary units). A blue shift of λmax from 332 to 327 nm was also observed at higher concentrations of tryptophan modifying reagents. Incremental addition of PGA substrate (penG) and competitive inhibitor (phenylacetate) resulted in progressive decrease of relative fluorescence intensity from 208 to 178 AU (85.6% of original). The absorption spectra of these compounds do not produce any peak in this region implying that the quenching is purely due to interaction with tryptophan and not due to secondary absorption by the ligands themselves. Red shift of λmax from 348 to 357 nm was observed when tested with both the ligands. No fluorescence quenching was observed upon addition of penG to NBS modified KcPGA (Fig. 4). However, a red shift of λmax from 348–352 nm was observed. The association constants for the binding of penG to untreated and NBS modified KcPGA are 7.94 × 102 and 2.43 × 102 M−1 , respectively. The change in value of standard free energy of binding (G◦ ) for penG is −16.2 kJ mol−1 . The substrate protection against fluorescence quenching by NBS was observed when KcPGA was pre-incubated with penG before the addition of NBS. The drop in fluorescence upon addition of NBS at the highest concentration was only 10.2% (178 to 160 AU) and λmax remained unaltered; the blue shift observed previously on treatment with NBS was absent in this case (Fig. 5). The fluorescence of native enzyme was quenched by the addition of ionic and neutral quenchers. The percentage of quenching

1393

Fig. 5. Fluorescence spectra of NBS treated and substrate-protected KcPGA (3.81 ␮M, 2 ml), excited at 295 nm and λmax recorded at 348 nm. Unmodified (), modified using NBS concentrations 100 ␮M (), 200 ␮M (䊉). Spectra of enzyme bound with penicillin G 1 mM (), NBS modified enzyme protected with substrate, at NBS concentrations 100 ␮M (), 200 ␮M ().

by acrylamide, CsCl and KI were 80, 50 and 35%, respectively. The effective Stern–Volmer or quenching constant Ksv(eff) and effective fractional accessible fluorescence fa(eff) were obtained from the slope and intercept of F0 /F versus 1/[Q] plots at low [Q] values (Table 3). The modified Stern–Volmer plots were linear for acrylamide, KI and CsCl (Fig. 6). 3.6. Homology modeling of K. citrophila PGA and substrate-docking The multiple sequence alignment showed some conserved tryptophan residues present in all reported PGAs. A high level of sequence identity can guarantee accurate alignment between the target sequence and the template structure. The processed molecule of KcPGA containing subunits ␣ and ␤ appeared similar to EcPGA. The structure of KcPGA modeled in MOE revealed excellent agreement with its predicted secondary structure and with the crystal structure of EcPGA. PROCHECK is used to estimate the percentage of residues whose backbone φ − ψ angles are within the allowed region of Ramachandran map. The result showed that 62.9% of the φ − ψ angles in the KcPGA model lie in the fully allowed region of the Ramachandran plot and the rest in partially or extended partially allowed regions. Both the modeled and the crystal structures on superposition gave an average RMSD (between equivalent C␣ atoms) of ˚ The active site of KcPGA was similar to that of EcPGA 0.54 A. Table 3 The effects of neutral and ionic quenchers on the fluorescence of KcPGA

Fig. 4. Fluorescence spectra of NBS treated KcPGA (3.81 ␮M, 2 ml). Enzyme was excited at 295 nm, and λmax recorded at 348 nm. Unmodified (), modified using NBS concentrations 10 ␮M (), 20 ␮M (䊉), 50 ␮M (×), 100 ␮M (), 200 ␮M (). Inset: fluorescent quenching with different concentrations of substrate penicillin G.

Quencher

15% Quenching

Ksv(eff)

Fa(eff)

Acrylamide (0–0.2 M) KI (0–0.3 M) CsCl (0–1.0 M)

0.008 0.025 0.500

16.25 6.00 2.00

1 0.52 0.22

The protein (3.81 ␮M, 2 ml) was excited at 280 nm and the emission was recorded in the range 300–400 nm. The Ksv values were calculated at low concentration [Q] of acrylamide, KI and CsCl.

1394

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

Fig. 7. (a) Active site environment of KcPGA in the structure modeled using MOE, the docked molecule penicillin G sulphoxide (PGSO) in the active site cavity is also shown. (b) Individual residues of PGA in the active site close to PGSO as obtained from modeled structure is displayed, the residue of interest Trp ␤154 is on the right side.

3.7. Interactions between docked substrate and modeled enzyme

Fig. 6. Modified Stern–Volmer plots: (A) acrylamide, (B) KI, (C) CsCl.

(Fig. 7a). The docking and molecular dynamics studies provided an estimate of the binding energy of substrate penG. There were three best conformations that showed almost the same levels of binding energy and the substrate bound close to residue ␤154 which happened to be the tryptophan. This tryptophan may be providing a favorable hydrophobic binding pocket for the phenyl group to bind when the substrate enters the active site.

Using standard docking procedure, optimally using the available information, the docking of the substrate to penicillin acylase was carried out. The substrate PGSO molecule is shown in Fig. 7b along with the residues of the enzyme in the energyminimized complex; the residues of KcPGA considered as ˚ Majorinteracting with PGSO mostly are closer than 3.5 A. ity of the interactions are with residues of ␤ chain. The carboxyl group of the penam ring interacts with the side chains of Arg ␤262 and Ser ␤384. This serine and main chain carbonyl oxygen of residue ␤23 are interacting with carbonyl oxygen of penam ring. The carbonyl oxygen at the scissile bond interacts with Asn ␤240 side chain, while scissile car˚ away from O␥ atom of Ser ␤1. It may bonyl carbon is ∼4.1 A be noted that the structures of substrate/analog complexes of EcPGA have been already reported [38]. Based on that study a detailed mechanism involving Ser ␤1 as nucleophile, its amino group as base and a host of conserved residues forming oxyanion hole that stabilizes the acylenzyme intermediate has been worked out. We observe that the substrate interacts

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

with analogous residues of KcPGA in our modeling studies as well. The phenyl ring of PGSO, the group that is destined to form acyl-intermediate, is bound in a hydrophobic pocket surrounded by residues Phe ␤24, Phe ␣146, Ile ␤177, Val ␤56 and Trp ␤154 ˚ away from PGSO in (Fig. 7b). Here, the Trp although is ∼6 A the final model, is directly exposed to phenyl ring of PGSO. The phenyl ring of substrate is also closer to main chain carbonyl oxygen of Ser ␤67. In KcPGA ␤154 could be the tryptophan residue whose environment got affected by the binding of substrate or analogs as shown by our spectroscopic experiments. 3.8. Sequence comparison of PGAs In the multiple sequence alignment the sequences were segregated into two groups according to the percentage of similarity and presence of the active site tryptophan. The first group of sequences is closely related with more than 40% sequence identity and a conserved Trp ␤154 present in them. The second group of sequences has only less than 30% identity and no conserved Trp ␤154 present. Substantial variation in PGA activity between the two sets is observed, with the second group having low affinity towards penG analogues (Table 4). 4. Discussion The profound impact of penicillin acylases in the manufacture of antibiotics is the reason for continued interest in studying structural determinants of these enzymes essential for catalysis. Efforts to enhance the catalytic rate by protein engineering have been reported [39]. The available crystal structures of PGA presently are from E. coli [20] and P. retgerri [40]. The studies of these enzymes, in conjunction with their site directed mutants have allowed mapping of some of the residues that form the active site pocket [38]. The PGA employed in semi-synthetic penicillin manufacture is primarily sourced from E. coli. However, KcPGA also can be favorably considered for pharmacological applications for its tolerance to denaturing parameters such as temperature, pH and

1395

solvent nature, along with consideration of its ease for immobilization. There is need for engineering KcPGA because its catalytic efficiency does not reach up to that of EcPGA. We have focused on the structure–function correlations of this KcPGA as this knowledge holds key to accomplishing successful protein engineering. Our preliminary studies reveal participation of a tryptophan residue in the activity of KcPGA. This residue, hitherto not identified as important for activity or specificity by crystallographic studies, appears to be a conserved residue across all closely related sequences whose specificity for penG is higher (Fig. 8a). NBS is a potent oxidizing agent. Being a highly reactive source of electrophilic bromonium ions (Br+ ), it is capable of adding bromine to the γ–δ carbon–carbon double bond of tryptophan which subsequently gets cleaved. The treatment of KcPGA with 100 ␮M NBS lowered the activity by 70%. Similarly, 58% drop in activity was observed on treatment with 20 mM HNBB. This compound is known to react with tryptophan to form a substituted derivative. Based on the observation that the enzyme is being protected from inactivation by the presence of substrate or a competitive inhibitor, it can be concluded that the modified residue is at the active site or close to it. The conformity of CD spectra of the modified enzyme to that of untreated PGA ascribes the loss of activity to modification of selective residues only and not to any disruption in protein structure. The rate of inactivation when plotted against the concentration of the modifier indicated modification of a single tryptophan. This is in agreement with the value of Kapp obtained from pseudo-first-order kinetics. The apparent Km of the modified enzyme increased while kcat remained unchanged indicating a decrease in substrate affinity as a consequence of modification. This suggests that the concerned tryptophan is involved in substrate binding. The comparison of the fluorescence spectra of native and NBS modified PGA reveals that while their λmax remains same, the AU of the latter is reduced. This confirms that the loss of activity is exclusively due to tryptophan modification and rules out modification of any cysteine residue. It may be remembered that NBS can react, to a lesser degree, with cysteine also. The result

Table 4 Some chosen kinetic parameters of PGAs from various organisms in the sequence alignment shown in Fig. 8a and b Source of PGA

Kluyvera citrophila Escherichia coli Alcaligenes xylosoxidans Alcaligenes faecalis Achromobacter proteobacterium Pseudomonas denitri Shigella boydii Pseudomonas putida Bacillus megaterium Arthrobacter viscosus Bacillus cereus Streptomyces avermitilis

␤-Chain length

555 557 556 551 568 566 560 448 (Single chain) 523 537 537 503 515

Blank indicates that no value is available.

Number of tryptophans in ␤-chain

22 21 19 18 19 19 18 14 14 13 12 14 14

Kinetic parameters Km (␮M)

kcat (s−1 )

16 46 8.9 2 – – – – 420 4500 – 350 –

63 50 68.8 54 – – – – – – – – –

% Homology against K. citrophila PGA

– 85 51 40 52 52 44 84 18 28 27 19 14

1396

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397

Fig. 8. (a) Comparison of sequences of PGA enzymes closely related to KcPGA, sequence across the conserved Trp ␤154 is shown. (b) Comparison of sequences of designated PGAs distantly related to KcPGA to show that the Trp of interest is not conserved in them. The position of Trp ␤154 is indicated by arrow in both the parts (organisms are listed in Table 4).

obtained here is in agreement with the modification study carried out also using cysteine modifiers like DTNB and iodoacetate which have not affected the activity. Difference absorption spectrum for the modification of KcPGA with NBS has a minimum at 280 nm and maximum at 250 nm indicating the formation of oxindolalanine [32]. Weak reaction of NBS with tyrosine is also known. However, the absence of peak at 263 nm in the observed absorbance spectra of KcPGA rule out possibility of tyrosine residue getting oxidized to dityrosine. The microenvironment of the essential tryptophan was studied by measuring the quenching of PGA fluorescence resulted from the binding of ionic and neutral quenchers. In the quenching experiments, the values of fa and Ksv obtained for iodide and cesium ions were lower than the value obtained for acrylamidequenched enzyme. This indicates that the essential tryptophan is located in a hydrophobic environment. Further, the blue shift of λmax in the case of NBS modified enzyme also confirms that the fluorescence contribution is from a tryptophan residue located in hydrophobic environment. The difference spectra of both the unmodified and NBS modified KcPGA showed red shifts suggesting accessibility of modified residue. Compared to Ksv of KI (6), the Ksv of CsCl (2) is lower, suggesting that some tryptophans are in electropositive environment. It has been reported that the enzymes from P. rettgeri [40] and E. coli [20] are inactivated in the presence of tryptophan modifying reagents [41]; however, no kinetic analysis of the inactivation has been reported till now. In the crystal structures of PGA from E. coli and P. rettgeri the Trp ␤154 is near the active site. This particular tryptophan residue occurs also in the immediate vicinity of the active site in the homology modeled KcPGA (Fig. 7b). The closest distance between the tryptophan ˚ There are no considerable and the nucleophile serine is 4.5 A. conformational changes observed in tryptophan on binding the substrate, although some residues near active site are showing changes in their rotamer conformation in the model. The role of other residues in the active site was well demonstrated in the previous studies [41]. Moreover, the presence of tryptophan in the immediate vicinity of the substrate-binding site is known to cause unusual effects

in substrate binding and catalysis. For example, the PGA mutant where Phe ␣24, Phe ␣57, Phe ␣146 were mutated to tryptophan showed a 10-fold decrease in kcat and an increase of up to 200% in Km [42]. The involvement of a hydrophobic residue like tryptophan in PGA catalytic activity inferred here is in agreement with the results from other investigations on the role of hydrophobic interactions in the hydrolytic activity of PGA. These include the report on the inhibitory effect of aliphatic alcohol on the activity of the enzyme from E. coli, the magnitude of inhibitory effects vary with the extent of hydrophobicity of the alcohol [43]. On the whole, the chemical modification and fluorescence studies provide strong indication for the presence of an essential tryptophan residue at the active center of KcPGA, which from kinetic analysis appears to be involved in substrate binding. This decisive residue is identified as Trp ␤154 through homology modeling. The same is conserved in closely related PGAs (Fig. 7a) and not in those distantly related ones (Fig. 7b) whose penicillin binding capacities are partly affected. The parameters listed in Table 4 show that the PGAs closely related to KcPGA have better affinity for penG compared to those distantly related and Trp residue discussed here is absent in them. It could be argued that the latter class of enzymes is specific for a different substrate, however, it only reinforces the argument that Trp ␤154 decides the penG specificity. Thus, all the results reported here point to the importance of Trp ␤154 in penG specific activity of PGAs. Acknowledgements R.S.K. is Senior Research Fellow of the Council of Scientific and Industrial Research, New Delhi, India. Authors thank the Department of Biotechnology, New Delhi for financial support. References [1] Shewale JG, Deshpande BS, Sudhakarn V, Amberkar SS. Penicillin acylase: application and potentials. Proc Biochem 1990;25:97–103.

R.S. Kumar et al. / Enzyme and Microbial Technology 40 (2007) 1389–1397 [2] Valle F, Balbus P, Marino P, Bolivar F. The role of penicillin amidase in nature and in industry. TIBS 1991;16:36–40. [3] Bruggink A, Roos R, De Vroom E. Penicillin acylase in the industrial production of ␤-lactam antibiotics. Org Proc Res Dev 1998;2:128–33. [4] Sio CF, Quax WJ. Improved beta-lactam acylases and their use as industrial biocatalysts. Curr Opin Biotechnol 2004;15:349–55. [5] Hunt PD, Tolley SP, Ward RJ, Hill CP, Dodson GG. Expression, purification and crystallization of penicillin G acylase from Escherichia coli ATCC 11105. Protein Eng 1990;7:635–9. [6] Konstantinovic M, Marjanovic N, Ljubijankic G, Glisin V. The penicillin amidase of Arthrobacter viscosus (ATCC 15294). Gene 1994;143:79–83. [7] Das S, Gayen JR, Pal A, Ghosh K, Rosazza JP, Samanta TB. Purification, substrate specificity, and N-terminal amino acid sequence analysis of a betalactamase-free penicillin amidase from Alcaligenes sp. Appl Microbiol Biotechnol 2004;65:281–6. [8] Plhackova K, Becka S, Skrob F, Kyslik P. Isolation and characterization of a new strain of Achromobacter sp. With beta-lactam antibiotic acylase activity. Appl Microbiol Biotechnol 2003;62:507–16. [9] Pinotti LM, Silva RG, Giordano RC, Giordano RL. Inoculum studies in production of penicillin G acylase by Bacillus megaterium ATCC 14945. Appl Biochem Biotechnol 2002;98–100:679–86. [10] Cheng T, Chen M, Zheng H, Wang J, Yang S, Jiang W. Expression and purification of penicillin G acylase enzymes from four different microorganisms, and a comparative evaluation of their synthesis/hydrolysis ratios for cephalexin. Protein Expression Purif 2006;46:107–13. [11] Yang S, Huang H, Zhang R, Huang X, Li S, Yuan Z. Expression and purification of extracellular penicillin G acylase in Bacillus subtilis. Protein Expression Purif 2001;21:60–4. [12] Alvaro G, Fernandez-Lafuente RF, Rosell CM, Blanco RM, Garcia-Lopez JL, Guisan JM. Penicillin G acylase from Kluyvera citrophila. New choice as industrial enzyme. Biotechnol Lett 1992;14:285–90. [13] Fernandez-Lafuente R, Rosell CM, Guisan JM. Enzyme reaction engineering: synthesis of antibiotics catalysed by stabilized penicillin G acylase in the presence of organic cosolvents. Enzyme Microb Technol 1991;13:898–905. [14] Fernandez-Lafuente R, Rosell CM, Guisan JM. Dynamic reaction design of enzymic biotransformations in organic media: equilibrium-controlled synthesis of antibiotics by penicillin G acylase. Biotechnol Appl Biochem 1996;2:139–43. [15] Liu SL, Wei DZ, Song QX, Zhang YW, Wang XD. Effect of organic cosolvent on kinetic resolution of tert-leucine by penicillin G acylase from Kluyvera citrophila. Bioprocess Biosyst Eng 2006;28:285–9. [16] Guisan JM, Alvaro G, Fernandez-Lafuente R, Rosell CM, Garcia-Lopez JL, Tagliani A. Stabilization of heterodimeric enzyme by multipoint covalent immobilization: penicillin G acylase from Kluyvera citrophila. Biotechnol Bioeng 1993;42:455–64. [17] Barbero JL, Buesa JM, Buitrago GG, Mendez E, Perez-Aranda A, GarciaLopez JL. Complete nucleotide sequence of the penicillin acylase gene from Kluyvera citrophila. Gene 1986;49:69–80. [18] Prieto I, Rodriduez MC, Marquez G, Perez-Aranda A, Barbero JL. Changing glycine 21 for glutamic acid in the ␤-subunit of penicillin G acylase from Kluyvera citrophila prevents protein maturation. Appl Microbiol Biotechnol 1992;36:659–62. [19] Brannigan JA, Dodson G, Duggleby HJ, Moody PC, Smith JL, Tomchick DR, et al. A protein catalytic framework with an N-terminal nucleophile is capable of self-activation. Nature 1995;378:416–9. [20] Duggleby HJ, Tolley SP, Hill CP, Dodson EJ, Dodson GG, Moody PCE. Penicillin acylase has a single aminoacid catalytic center. Nature 1995;373:264–8. [21] Wen Y, Feng M, Yuan Z, Zhou P. Expression and overproduction of recombinant penicillin G acylase from Kluyvera citrophila in Escherichia coli. Enzyme Microb Technol 2005;37:233–7.

1397

[22] Wen Y, Shi X, Yuan Z, Zhou P. Expression, purification and characterization of His-tagged penicillin G acylase from Kluyvera citrophila in Escherichia coli. Protein Expression Purif 2004;38:24–8. [23] Martin J, Mancheno JM, Arche R. Chemical modification of serine at the active site of penicillin acylase from Kluyvera citrophila. Biochem J 1991;280:659–62. [24] Roa A, Garcia-Lopez JL, Salto F, Cortes E. Changing the substrate specificity of penicillin G acylase from Kluyvera citrophila through selective pressure. Biochem J 1994;303:869–75. [25] Martin J, Mancheno JM, Arche R. Inactivation of penicillin acylase from Kluyvera citrophila by N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline: a case of time-dependent non-covalent enzyme inhibition. Biochem J 1993;291:907–14. [26] Done SH, Brannigan JA, Moody PC, Hubbard RE. Ligand-induced conformational change in penicillin acylase. J Mol Biol 1998;284:463–75. [27] Kumar RS, Suresh CG, PundleA, Prabhune A. Evidence for the involvement of arginyl residue at the active site of penicillin G acylase from Kluyvera citrophila. Biotechnol Lett 2004;26:1601–6. [28] Shewale GJ, Kumar KK, Amberkar GR. Evaluation and determination of 6-aminopenicillinic acid by p-dimethylbenzaldehyde. Biotechnol Technol 1987;1:69–72. [29] Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the folin phenol reagent. J Biol Chem 1951;193:265–75. [30] Spande TF, Witkop B. Determination of the tryptophan content of proteins with N-bromosuccinimide. Meth Enzymol 1967;11:498–506. [31] Eftink MR, Ghiron A. Exposure of tryptophanyl residues in proteins. Quantitative determination by fluorescence quenching studies. Biochemistry 1976;15:672–80. [32] Lehrer SS. Solute perturbation of protein fluorescence. The quenching of the tryptophanyl fluorescence of model compounds and of lysozyme by iodide ion. Biochemistry 1971;10:3254–63. [33] Vagin AA, Richelle J, Wodak SJ. SFCHECK: a unified set of procedure for evaluating the quality of macromolecular structure-factor data and their agreement with atomic model. Acta Crystallogr D 1999;55:191– 205. [34] Goto J, Kataoka R, Hirayama N. Ph4Dock: pharmacophore-based proteinligand docking. J Med Chem 2004;47:6804–11. [35] Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol 1990;215:403–10. [36] Marck C. DNA Strider: a ‘C’ program for the fast analysis of DNA and protein sequences on the Apple Macintosh family of computers. Nucl Acids Res 1988;16:1829–36. [37] Gouet P, Courcelle E, Stuart DI, Metoz F. ESPript: analysis of multiple sequence alignments in PostScript. Bioinformatics 1999;15:305–8. [38] McVey CE, Walsh MA, Dodson GG, Wilson KS, Brannigan JA. Crystal structures of penicillin acylase enzyme–substrate complexes: structural insights into the catalytic mechanism. J Mol Biol 2001;313:139–50. [39] Gabor EM, Janssen DB. Increasing the synthetic performance of penicillin acylase PAS2 by structure-inspired semi-random mutagenesis. Protein Eng Des Sel 2004;17:571–9. [40] McDonough MA, Klei HE, Kelly JA. Crystal structure of Penicillin G acylase from the Bro1 mutant strain of Providencia rettgeri. Protein Sci 1999;8:1971–81. [41] Mahajan PB, Borkar PS. Chemical modification of Escherichia coli penicillin acylase I tryptophan involvement at the catalytic site. Hindustan Antibiotics Bull 1983;25:6–10. [42] Alkema WB, Dijkhuis A-J, de Vries E, Janssen B. The role of hydrophobic active site residues in substrate specificity and acyl transfer activity of penicillin acylase. Eur J Biochem 2002;269:2093–100. [43] Klyosov AA, Svedas VK, Galaev IY. Study of the active center topography of penicillin amidase from E. coli. II. The role of hydrophobicity in the specificity of the enzyme inhibition. Bioorg Khim 1977;3:800–5.