An active site homology model of phenylalanine ammonia-lyase from Petroselinum crispum Dagmar Ro¨ ther 1 ,La ´ szlo ´ Poppe 2 , Gaby Morlock 1 , Sandra Viergutz 1 and Ja ´ nos Re ´ tey 1 1 Institute for Organic Chemistry, University of Karlsruhe, Germany; 2 Institute for Organic Chemistry, Budapest University of Technology and Economics, Hungary The plant enzyme phenylalanine ammonia-lyase (PAL, EC 4.3.1.5) shows homology to histidine ammonia-lyase (HAL) whose structure has been solved by X-ray crystal- lography. Based on amino-acid sequence alignment of the two enzymes, mutagenesis was performed on amino-acid residues that were identical or similar to the active site resi- dues in HAL to gain insight into the importance of this residues in PAL for substrate binding or catalysis. We mutated the following amino-acid residues: S203, R354, Y110, Y351, N260, Q348, F400, Q488 and L138. Deter- mination of the kinetic constants of the overexpressed and purified enzymes revealed that mutagenesis led in each case to diminished activity. Mutants S203A, R354A and Y351F showed a decrease in k cat by factors of 435, 130 and 235, respectively. Mutants F400A, Q488A and L138H showed a 345-, 615- and 14-fold lower k cat , respectively. The greatest loss of activity occurred in the PAL mutants N260A, Q348A and Y110F, which were 2700, 2370 and 75 000 times less active than wild-type PAL. To elucidate the possible func- tion of the mutated amino-acid residues in PAL we built a homology model of PAL based on structural data of HAL and mutagenesis experiments with PAL. The homology model of PAL showed that the active site of PAL resembles the active site of HAL. This allowed us to propose possible roles for the corresponding residues in PAL catalysis. Keywords: phenylalanine ammonia-lyase; PAL; MIO; site- directed mutagenesis; homology model. Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5) is a very important plant enzyme that catalyses the conversion of L -phenylalanine into E-cinnamic acid which is the precursor of a great variety of phenylpropanoids, such as lignins, flavonoids and coumarins [1,2]. Because of its central role in plant metabolism, PAL is a potential target for herbicides [2]. The related enzyme histidine ammonia-lyase (HAL; EC 4.3.1.3) catalyses a very similar reaction, converting L -histidine into E-urocanic acid. Amino-acid sequence comparison of histidine and phenylalanine ammonia-lyases from different organisms revealed that there are several homologous regions indicating that their active sites are very similar [3]. For about 30 years it has been believed that a dehydroalanine acts as electrophilic prosthetic group at theactivesiteofbothHALandPAL[4–6].Recentlythe three-dimensional structure of HAL was solved by X-ray crystallography revealing that the electrophilic prosthetic group 3,5-dihydro-5-methylidene-4H-imidazol-4-one (MIO) is the catalytically essential moiety rather than the dehydro- alanine (Fig. 1) [7]. It has been proposed that this MIO group is generated by autocatalytic cyclization of the A142- S143-G144 moiety of HAL. This process resembles the fluorophore formation of the green fluorescent protein [8]. More recently, we provided spectroscopic evidence for the presence of a prosthetic MIO group at the active site of PAL [9]. Here we report the exchange of several amino-acid residues in PAL that are identical or similar to active site residues of HAL and evaluation of their importance in substrate binding and catalysis by enzyme kinetic behaviour of the mutants and by a homology model of PAL. MATERIALS AND METHODS Bacterial strains and plasmids Wild-type PAL and PAL mutants were overexpressed in E. coli BL21(DE3) cells. The gene coding for phenylalanine ammonia-lyase from P. crispum was changed to the codon- usage of E. coli and cloned in vector pT7-7 followed by a transformation in E. coli BL21(DE3) cells containing vector pREP4-GroESL [10]. Site-directed mutagenesis Phenylalanine ammonia-lyase mutants were produced by following the instruction manual of the QuickChange(tm) Site-Directed mutagenesis kit (Stratagene) [11]. The oligo- nucleotides used in the mutagenesis reactions were: S203A(+): 5¢-catcactgctgccggcgacctgg-3¢, S203A(–): 5¢-cca ggtcgccggcagcagtgatg-3¢; Q488A(+): 5¢-cagcacaacg ctgacgt taac-3¢, Q488A(–): 5¢-gttaacgtcagcgttgtgctg-3¢; Q488E(+): Correspondence to J. Re ´ tey, Institute of Organic Chemistry, University of Karlsruhe, Richard-Willsta ¨ tter-Allee, D-76128 Karlsruhe, Germany. Fax: + 49 721 6084823, Tel.: + 49 721 6083222, E-mail: biochem@ochhades.chemie.uni-karlsruhe.de Abbreviations: PAL, phenylalanine ammonia-lyase; HAL, histidine ammonia-lyase; MIO, 3,5-dihydro-5-methylidene-4H-imidazol-4-one. Enzymes: histidine ammonia-lyase (EC 4.3.1.3); phenylalanine ammonia-lyase (EC 4.3.1.5). Note: in this paper, the numbering of amino acids for PAL from P. crispum is consistent with the SWISS-PROT database (P24481) record but not with the numbering used in previous PAL papers. (Received 15 February 2002, accepted 8 May 2002) Eur. J. Biochem. 269, 3065–3075 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02984.x 5¢-cagcacaacgaagacgttaac-3¢, Q488E(–): 5¢-gttaacgtctcgttg tgctg-3¢; Y351F(+): 5¢-caggaccgttttgctctgcg-3¢, Y351F(–): 5¢-cgcagagcaaaacggtcctg-3¢; Y110F(+): 5¢-ccgactcctttggcg ttacc-3¢, Y110F(–): 5¢-ggtaacgccaaaggagtcgg-3¢; R354A(+): 5¢-cgttatgctctggctacctctcc-3¢, R354A(–): 5¢-ggag aggtagccag agcataacg-3¢; N260A(+): 5¢-gcactggttgctggtaccgctg-3¢, N260A(–): 5¢-cagcggtaccagcaaccagtgc-3¢;Q348A(+):5¢-aa acgaaagcggaccgttat-3¢, Q348A(–): 5¢-ataacggtccgctttcgg ttt-3¢; F400A(+): 5¢-ggtggtaacgcccaggggac-3¢, F400A(–): 5¢-gtc ccctgggcgttaccacc-3¢; L138H(+): 5¢-gatccgcttccacaacg ctg-3¢, L138H(–): 5¢-cagcgttgtggaagcggatc-3. The mutations were verified by sequence analysis using the dideoxynucleotide chain-termination method [12]. Protein expression and purification E. coli BL21 (DE3) cells carrying the plasmids with the genes for wild-type PAL and PAL mutants were cultured and PAL was purified as described previously [10]. SDS/PAGE and Western blot analysis SDS/PAGE was carried out according to Laemmli [13] using 10% polyacrylamide gels. The gels were stained with Coomassie Brillant Blue R250. Western Blot analyses were performed following a previously described method using nitrocellulose blotting filters [14,15]. Wild-type PAL and mutants were detected with rabbit polyclonal antibodies raised against PAL from P. crispum (the antibody was a generous gift of N. Amrhein, Eidgeno ¨ ssische Technische Hochschule, Zu ¨ rich). Enzyme assay and determination of protein concentration PAL activity was measured spectrophotometrically at 30 °C following the formation of E-cinnamate at 290 nm. The assay was performed in 1-cm quartz cuvettes by modifica- tion of the method described in [16] with enzyme concen- trations varying between 10 and 20 lg for active enzymes and between 0.3 and 0.4 mg for less active mutants. The enzyme was preincubated at 30 °C for 5 min in 750 lLof 0.1 M Tris/HCl pH 8.8. The reaction was started by adding 250 lLofa20-m ML -phenylalanine solution. Wild-type enzyme and moderately active mutants were measured in intervals of 1 min for 5 min, less active enzyme mutants were measured in intervals of 5 min for 20 min. For determination of K m and V max , L -phenylalanine concentra- tions were varied from 0.01 to 5 m M . Kinetic constants (K m , V max ) were determined using a double reciprocal plot [17]. The isolated enzymes were electrophoretically pure as verified by staining with Coomassie Brillant Blue R250 and therefore it was possible to measure the turnover numbers (k cat ) with the relative molecular mass of 311.313 for the tetrameric PAL. Determination of protein concen- tration was carried out according to Warburg & Christian [18,19], Murphy & Kies [20] and Groves et al. [21]. BSA was used as reference protein for the measurements. Sequence comparison Amino-acid sequences of HAL from P. putida [22], HAL from Homo sapiens [23] and PAL from P. crispum [24] were extracted from the SWISS-PROT database. Sequence alignment was carried out using the computer program MALIGN (HUSAR, DKFZ Heidelberg). MALIGN is a HUSAR adaptation of the program MAP [25]. Homology modelling Model of PAL 64–531 monomer. The sequence of PAL (Swiss-Prot: P24481) was submitted to SWISS-MODEL (Automated Protein Modelling Server) [26,27]. The PAL structure homology model resulting from this first approach, which was folded over the HAL structure (PDB: 1B8F), contained amino acids 64–531 [28]. This partial model was optimized by different molecular me- chanics force fields using CHARMM [29], Amber-95 [30,31] and MM3Prot [32] force field implementations in the TINKER [33] package; Amber3 and MM+ force field implementations in HYPERCHEM [34] package; and the GROMOS 96 43B1 [35] implementation in SWISS - PDBVIEW- ER . Calculations were performed on 300–850 MHz Pen- tium III computers running under WINDOWS 95, WIN- DOWS 98 or LINUX ( REDHAT 6.2). A switched smoothing function, which gradually reduced nonbonding interactions to zero from a 10-A ˚ inner radius to a 14-A ˚ outer radius, was generally applied. Otherwise, all the calculations here and later were performed by using default settings of the program packages. These optimized structures were compared by GROMOS single point energy calculations and Fig. 1. The MIO moiety in HAL from P. putida. The mechanism of the PAL reaction through Friedel–Crafts type attack of MIO on the phenyl ring of L -phenylalanine. 3066 D. Ro ¨ ther et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Ramachandran plot analyses using the SWISS - PDBVIEWER 3.6 package [27]. For further modelling, the PAL 64–531 monomer optimized by MM+ force-field of HYPERCHEM [34] was used. During the MM+ optimization, the A202- S203-G204 triad was replaced by MIO in this structure. Model of PAL 64–531 homotetramer. The raw PAL 64–531 fragment homotetramer was built by the SWISS - PDBVIEWER 3.6 [27] package using the MM+ optimized PAL 64–531 fragment model and the cell parameters of the HAL structure (space group: I222) [7]. The MIO structures were rebuilt by MM+ calculation and kept frozen in the homotetramer during Amber3 optimization performed by HYPERCHEM [34]. Model of PAL 1–63 fragment: The 1–63 fragment of PAL was built using secondary structure prediction data (obtained by using PHDsec, version 5.94–317 [36,37], and 3D-PSS Psi-Pred [38,39]). Raw three-dimensional models corresponding to the predicted secondary structure were built and optimized in the HYPERCHEM [34] and SWISS - PDBVIEWER 3.6 [27] packages. Evaluation of the raw three- dimensional models as described for the PAL 64–531 fragment resulted in the best fold of the 1–63 fragment. Model of PAL 532–716 fragment. Swiss-MODEL [26,27] has not found a template eligible for modelling the terminal fragment (532–716) of the PAL sequence (Swiss- Prot: P24481). Although several reasonable hits were found when the 532–716 fragment was submitted to 3D-PSS search [38,39] (e.g. 15% identity over 184 amino acids with 1FPW chain A; 19% identity over 184 amino acids with 1JBA chain A; 22% identity over 172 amino acids with 2FHA; 18% identity over 153 amino acids with 1BG7; 17% identity over 179 amino acids with 1AKE chain A; 22% identity over 162 amino acids for 1GGQ chain A), the first approach models built from these templates by SWISS - PDBVIEWER 7.02 [27] showed no proper contacts with the core 64–531 fragment. The secondary structure prediction data for the whole PAL sequence (by using PHDSEC [36,37]) showed 14% identity over 184 residues between the PAL 532–716 sequence fragment and the 91–275 fragment of chain B in the aspartase structure (PDB accession no. 1JSW) [40,41]. Because aspartase is also an ammonia- lyase and its structure shows substantial structural similarity to that of the template HAL (long parallel helices and a quasi tetrameric structure), the PAL 532–716 fragment was modelled using chain B of the aspartase structure as template by the SWISS - PDBVIEWER 7.02 [37] package providing the 532–716 model fitting best to the core 64–531 fragment. Model of the full PAL 1–716 monomer. The full PAL 1–716 monomer was obtained from the optimized PAL 1–63, 64–531 and 532–716 fragments by manual adjust- ments and rigid-body approach optimizations using by the TINKER package [33]. The rms fit of four full PAL 1–716 monomers over the Amber3-optimized PAL 64–531 frag- ment homotetramer indicated the necessity of modifications in the 621–640 loop of the full PAL model avoiding spatial overlap. The necessary loop modifications in the full PAL monomer were made by the SWISS - PDBVIEWER 3.6 [27] package, followed by Amber3 optimization of the outside area (fragments 1–70 and 525–716 in each monomer chain) of the reconstituted full PAL homotetramer by the HYPER- CHEM [34] program. Quality assessment of the model Analysis of the full PAL homotetramer was performed by the SWISS - PDBVIEWER 7.02 [27] package. No significant clashes were found in the contact regions between the unique chains. The aspherical nature of the active homo- tetramer form of PAL is supported by the sedimentation constant and the Stokes’ radius data found by sucrose density gradient centrifugation of PAL from potato [42]. Of the 716 amino acids in the chain A of the model, 43 were outside of the likely Phi/Psi combinations in the Ramachandran plot (including Gly and Pro). Most of the deviations were found in the 64–531 and the 532–716 fragments (29 and 14, respectively), whereas only one difference was found in the 1–63 fragment. Further quality assessments were made by WHAT IF v4.99 [43,44] [e.g. bond lengths Z score 0.528 ± 0.012; 27 bumps between atomic pairs over 0.1 A ˚ ; four Gly residues have unusual backbone oxygen position; the rms Z score for all improper dihedrals (1.253) was within normal ranges; packing Z scores less than )2.50 were found for I141: )3.68, H309: )2.93, R52: )2.80, F 621 : )2.73, E291: )2.51, P699: )2.50] and PROCHECK V3.5 [45,46] through EMBL the Biotech Validation Suite, Heidelberg [47]. Although some deviations in bond lengths, angles, side chain planarities and hydrogen bonding environments were found by these checks, mainly in the terminal 532–716 fragment, the PROCHECK overall average G factor for the protein ()0.35) was acceptable. Because the model gave reliable assembly around the active site, for which a good match of residues from three chains is necessary, it seems to be trustworthy, at least in this region. Optimization and substrate fit within the active site area Analysis of the full PAL homotetramer showed that S203 is fully covered by residues of three monomer subunits within a global area of 25 A ˚ radii. This part was cut from the full PAL homotetramer model, and used for modelling the substrate-free and substrate binding states of the active site by MM+ calculations [34] within 15-A ˚ radii around S203. The outside sphere between 15 and 25 A ˚ of the whole 25-A ˚ radii globe was kept ÔfrozenÕ during the calculations, which were performed on 1949 atoms within the 15-A ˚ inside area. Conformational analysis of phenylalanine in its zwitter- ionic state by PM3 calculations of PC SPARTAN PRO [48] package was performed [28], and the lowest energy confor- mation was used as starting structure of the substrate. The zwitterionic L -Phe structure was docked to the substrate- free active site model by applying the following consider- ations: (a) the C 2 position of the phenyl ring of L -Phe should be close enough to the methylene of the MIO to perform the nucleophilic addition to the C ¼ C double bond; and (b) the NH 3 + and the pro-S b-H should be antiperiplanar [28]. Several, slightly different arrangements satisfied these requirements. These starting structures containing the zwitterionic L - PHE substrate were optimized by MM+ Ó FEBS 2002 Active-site model of phenylalanine ammonia-lyase (Eur. J. Biochem. 269) 3067 method of the HYPERCHEM [34] program, and the one of the lowest energy was considered as the best fit (Fig. 5A). The r-complex-like intermediate state was obtained by constructing a single bond between the L -Phe C 2 and MIO methylidene C atoms, correcting the atom and bond types and orders, and relaxing the structure by MM+ optimiza- tion (Fig. 5B). The E-cinnamate/ammonia binding model was obtained from the r-complex model by breaking the appropriate bonds, correcting the atom and bond types and orders, and optimizing the structure by MM+ method (Fig. 5C). Overlay of the models on the HAL structure The substrate-free active site model (25 A ˚ radius around the MIO) for PAL was overlaid onto the similar portion of the experimental structure of HAL containing a sulfate ion [7] by the SWISS - PDBVIEWER 3.7 [27] package (Fig. 4A). The models were visualized using the WEBLABVIEWER [49] program. The two models for the active sites containing the cationic intermediate state of the substrate for PAL and HAL [50] were aligned similarly (Fig. 4B). Calculation of the charge distribution in the L -Phe – MIO r-complex intermediate The [ L -Phe-MIO] r-complex model was cut off from the whole r-complex containing the active site model. Semiempirical (CNDO, MNDO, AM1, PM3, ZINDO/1, ZINDO/S) and ab initio (STO-3G) calculations for atomic charge distributions were performed on this truncated r-complex model by using the built-in methods of the HYPERCHEM [34] program. No change/reversal in polariza- tion order of hydrogen atoms (see PM3 results in Fig. 6B) calculated by the different methods was found. RESULTS AND DISCUSSION Amino-acid sequence comparison and site-directed mutagenesis Amino-acid sequence comparison of HAL and PAL from different organisms showed a sequence identity of about 40 and 20% when comparing different sequences of HAL among one another and comparing sequences of HAL and PAL, respectively [3]. Recently, the X-ray structure of HAL from P. putida was solved by Schwede et al., who discovered a new electrophilic prosthetic group at the active site, namely MIO (Fig. 1) [7]. The four active sites of HAL are constructed by the assembly of two subunits of the tetrameric enzyme. Active site amino-acid residues in the case of HAL from P. putida are S143, the predominant precursor of MIO group, as well as R283, Y53, E414, Y280, Q277, F329, N195 and H83. These active site residues may be important for substrate binding, catalysis or MIO formation [7,50]. All active site residues in HAL have an analogous residue in the PAL protein sequence except residues H83 and E414. Figure 2A shows the amino-acid comparison of HAL from P. putida [22], HAL from H. sapiens [23] and PAL from P. crispum [24]. Active site residues of HAL and the respective residues in the PAL sequence are shown in red. Figure 2B shows the predictions for the secondary structure of PAL using four different methods. The more reliable predictions are marked with blue colour, the less reliable ones with grey. To examine the importance of this residues for PAL activity, we constructed enzyme mutants at the correspond- ing sites using the QuickChange TM Site-Directed mutagen- esis system [11]. We carried out mutagenesis reactions in residues S203, R354, Y110, Y351, N260, Q348, F400 of the PAL amino-acid sequence and also in residues L138 and Q488 that replace residues H83 and E414 in HAL, respectively. The result of the mutagenesis experiments were verified by sequence analysis [12]. Protein expression and purification The plasmids containing the mutated genes were trans- formed in cells of E. coli BL21(DE3) that were previously transformed with the chaperonin-expressing plasmid pREP4-groESL [10]. After expression and purification of wild-type PAL and the various PAL mutants SDS/PAGE and Western-Blot analysis of crude extracts were performed to check expression levels and monomeric size of the enzyme variants. In all cases similar quantities of recombinant enzymes were produced showing the same monomeric size. Western-Blot analysis revealed that all enzyme variants were detected with anti-PAL antibodies. After purification of wild-type PAL and the different PAL mutants yields between 5 and 30 mg pure enzyme per litre cell culture were obtained. Kinetic characterization of the enzyme mutants Steady state kinetic parameters of wild-type PAL and the PAL mutants were measured at substrate concentrations varying from 0.01 to 5 m ML -phenylalanine. Table 1 shows the kinetic constants of wild-type PAL and the constructed and measured PAL mutants. The factor k cat wtPAL /k cat mutPAL is the k cat or turnover number of wild-type PAL divided by k cat of the respective PAL mutant. It is a measure of how many times the mutant is less active compared to the wild-type enzyme. Pure wild-type PAL showed a k cat of 13.5 s )1 and a K m of 0.12 m M in agreement with previously reported values [51]. Comparison of the K m values of wild- type PAL and the various mutants revealed, that except mutants L138H and L138H/Q488E all other mutants showed a slightly higher affinity for their substrates with K m values varying between 0.019 and 0.057 m M .Mutation in L138 in the PAL sequence, which is the counterpart of H83 in the HAL sequence resulted in a 14-fold decrease in activity and a decreased affinity for L -phenylalanine. The measured K m value is about 13 m M and is therefore about 100 times higher than that of wild-type PAL. In the double mutant L138H/Q488E, the K m value is increased further to 55 m M and the enzyme is about 145 times less active than the wild-type enzyme. The PAL residues L138 and Q488 and the unsimilar counterparts H83 and E414 in the HAL amino-acid sequence may be important for the substrate specificity of the homologous enzymes PAL and HAL. Our expectation that the L138H/Q488E mutant of PAL shows activity with L -histidine was not fulfilled. Although this mutant showed a similar K m value for L -histidine as wild- 3068 D. Ro ¨ ther et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Fig. 2. Amino-acid sequence alignment of HAL from P. putida [22], HAL from Homo sapiens [23] and PAL from P. crispum [24] performed with MALIGN (HUSAR). Active site amino-acid residues of HAL and respective residues in PAL are marked with colour (A). Secondary structure of the PAL model (line: Mod) aligned with the secondary structure predictions of PHDSEC [36,37] (lines: PHD and SUB) and 3 D - PSS PSI - PRED [38,39] (line: sspPP). The less reliable 1–63 and 532–716 fragments are in grey, the more reliable 64–531 fragment is in blue (B). Table 1. Kinetic constants of wild-type PAL and PAL mutants. PAL activity was measured by monitoring the formation of E-cinnamate at 290 nm in the presence of purified enzyme. The enzyme was preincubated at 30 °Cin0.1 M Tris/HCl (pH 8.8). The reaction was started by addition of a 20- m ML -phenylalanine solution. The L -phenylalanine concentrations were varied from 0.01 to 5 m M . The kinetic constants K m (m M )andV max (UÆmg )1 or lmolÆmin )1 Æmg )1 ) were determined using a double reciprocal plot [17]. Turnover numbers or k cat values were determined with the relative molecular mass 311 313 for the tetrameric PAL. Determination of protein concentration was carried out according to Warburg & Christian [18], Murphy & Kies [20] and Groves et al. [21]. K m (m M ) k cat (s )1 ) k cat /K m (m M )1 Æs )1 ) Factor (k cat wtPAL /k cat mutPAL ) wt PAL 0.12 ± 0.004 13.5 ± 0.1 112.5 S203A PAL 0.019 ± 0.001 0.031 ± 0.0001 1.63 435 R354A PAL 0.057 ± 0.003 0.104 ± 0.005 1.82 130 Y110F PAL – 0.00018 – 75,000 Y351F PAL 0.024 ± 0.004 0.057 ± 0.001 2.38 235 N260A PAL 0.033 ± 0.003 0.005 ± 0.001 0.152 2,700 Q348A PAL 0.03 ± 0.01 0.0057 ± 0.0004 0.19 2,370 F400A PAL 0.027 ± 0.005 0.039 ± 0.001 1.44 345 Q488A PAL 0.033 ± 0.002 0.022 ± 0.002 0.667 615 Q488E PAL 0.057 ± 0.006 2.1 ± 0.04 36.8 6 L138H PAL 13.5 ± 0.6 0.99 ± 0.02 0.073 14 L138H/Q488E PAL 55 ± 4.9 0.093 ± 0.004 0.0017 145 Ó FEBS 2002 Active-site model of phenylalanine ammonia-lyase (Eur. J. Biochem. 269) 3069 type HAL, its k cat value was about 8000 times lower. Construction of the PAL mutants Q488A and Q488E leads to a 615-fold and sixfold lower activity compared to that of the wild-type enzyme, respectively. Q488 is the PAL counterpart of E414 in the HAL amino-acid sequence, but it seems to play a different role in the catalysis by PAL. Distances obtained in models (Fig. 5A,C: Q488 NH1 to N1 of the MIO is  3.3 A ˚ ) suggest that the Q488 may interact with the MIO ring. Its replacement by an acidic residue in mutant Q488E has no dramatic effect on the catalytic activity of this mutant, whereas replacement by a nonpolar residue in PAL mutant Q488A leads to a more severe decrease. The effects are not as dramatic as in the mutagenesis of residue E414 of HAL. HAL mutant E414A showed a more than 20 900-fold lower activity compared to the wild-type enzyme [50]. Therefore it was assumed that E414 acts as a base in HAL catalysis. Mutagenesis in residue S203 of the PAL gene led to an enzyme variant with a k cat of 0.031 s )1 . The mutant S203A is therefore 435 times less active than the wild-type enzyme. This is an  10 times higher activity that was previously reported for this active site mutant [52]. The counterpart of residue Y280 at the active site of HAL is Y351 in the amino- acid sequence of PAL. PAL mutant Y351F showed a by factor 235 reduced activity compared to that of the wild- type enzyme. PAL mutants N260A and Q348A showed a 2700- and 2370-fold diminished activity, respectively. This indicates that both residues may play important roles in substrate binding or catalysis. PAL mutant F400A shows a less dramatic effect; this mutagenesis resulted in an enzyme Fig. 3. Comparison of X-ray structure of HAL [7] and homology model structure of PAL. Compar- ison of the schematic representation of (A) tetrameric and (B) monomeric structures of HAL (left) and PAL (right). Catalytically important residues are shown as stick models: MIO moiet- ies are shown in green, the other residues are coloured by elements: C, grey; N, blue; O, red). The less reliable 1–63 and 532–716 fragments of the PAL model are coloured in grey. 3070 D. Ro ¨ ther et al. (Eur. J. Biochem. 269) Ó FEBS 2002 with 345 times lower activity than the wild-type enzyme. This residue seems to interact with the r-complex interme- diate by p-stacking and thereby contributing to its stabil- ization [28]. Amino-acid residue R354 in the PAL sequence is the counterpart of residue R283 in the HAL sequence. The R354A mutant showed 130 times lower activity than the wild-type enzyme, whereas the Y110F mutant of PAL resulted in an almost inactive enzyme. PAL mutant Y110F is > 75 000 times less active than the wild-type enzyme and is therefore the least active PAL mutant examined so far. Y110 probably plays a very important role that can be explained by enzyme modelling. Homology modelling of PAL To gain further insight into the possible role of the active site residues a theoretical model for PAL was built by homology modelling [28] (Figs 3 and 4). According to the model, the active structure is a homotetramer as in the case of HAL (Fig. 3A). The catalytically important residues are located on two distinct regions in both HAL and PAL (Fig. 3B) at highly isosteric positions. The active site region of the PAL homotetramer (Fig. 4) closely resembles that of the X-ray structure of HAL [7,50,53]. All the residues in the PAL model occupy the expected positions postulated by comparison with the HAL sequence and structure. Modelling of a zwitterionic sub- strate (Fig. 5A), the r-complex forming between the substrate and MIO (Fig. 5B) and the product E-cinna- mate/ammonia (Fig. 5C) into the active site of the PAL model gave further insight into the role of the amino-acid residues in catalysis (Fig. 6A). These ligand-binding models confirmed the hypothesis concerning the p-stacking role of F400. This residue may stabilize the intermediate r-complex and prevent abstraction of the proton from the ortho- position of the aromatic ring by excluding any basic group [28]. The close vicinity of Y351 to the pro-S b-H of the substrate in the r-complex model (3.61 A ˚ , Table 2) indi- Fig. 4. The substrate-free active site model of PAL overlaid on the experimental structure of HAL. (A) The protein chains are coloured differently. The lighter colours and the labels in yellow are related to the PAL model. (B) Com- parison of the active sites in the r-complex intermediate containing model of PAL with the cationic intermediate binding model of HAL [50]. The thick bonds and labels in yellow are related to the PAL model. The thin lines and white labels are related to the HAL model. Ó FEBS 2002 Active-site model of phenylalanine ammonia-lyase (Eur. J. Biochem. 269) 3071 cates that this residue might act as the b-H abstracting base. The residues Y110, Q348 and R354 might play a role in binding the carboxylate moiety of the substrate, whereas residues Y110 and N260 can interact with amine/ammonia in the substrate and product, respectively. In the case of HAL it was shown that in the substrate- free state of the enzyme the MIO prosthetic group is not substantially polarized [50]. In HAL, presumably, activation/polarization of the MIO happens when the substrate itself is approaching it and the MIO carbonyl oxygen can be stabilized by partial protonation from N1-H of the substrate. Analysis of the acidity/charge distribution in the r-complex intermediate in the PAL reaction by semiempirical and ab initio calculations (Fig. 6B) revealed that the C3-H hydrogen atom is almost as positively charged as the pro-S b-H, and therefore it can stabilize the negatively charged MIO carbonyl oxygen in a similar way. Interaction of the C6-H of the L -phenylalanine with Y453- O – can have a ring preorientation effect and it can also increase the nucleophilic character of the phenyl moiety of the substrate. It was observed that the b-H abstraction in the PAL reaction does not seem to be rate determining for good substrates [54]. The attack by MIO is partially rate determining; another rate-limiting step seems to be product release. The kinetic significance of these two steps is substrate-dependent [28]. Inspection of the whole PAL tetrameric model reveals that residue Y110, whose mutation to F showed the most dramatic decreasing effect on the reaction rate, is located at the edge of a channel (see also Fig. 4A) through which the substrate can enter into or the product can be released from the active site. Residue Y110 in PAL (and also in HAL its counterpart Y53) seems to play twofold role (Fig. 7). The first role of it may be a protonation state conversion: ammonium ions (outside from active site in aqueous solution) , ammonia/amine (Fig. 7A,F). Secondly, it may serve as anchoring/orienting Fig. 6. Model for the ammonia elimination from the r-complex inter- mediate of the PAL reaction (A) and the charge distribution in the L -Phe–MIO r-complex intermediate calculated by PM3 method (B). Fig. 5. Calculated models for the zwitterionic L -phenylalanine binding (A), for the r-complex intermediate (B), and for the E-cinnamate/ ammonia binding (C) state of the PAL active site. 3072 D. Ro ¨ ther et al. (Eur. J. Biochem. 269) Ó FEBS 2002 group for the carboxylate of the substrate/product during the elimination/addition step in its protonated form through hydrogen bonding (Fig. 7D). Presumably, the enzymic base exists in the substrate-free state of the enzyme as its conjugated acid (Fig. 7A,F), from which the amino group of the substrate can abstract a proton during entering into the active site (Fig. 7C). Interaction of the aromatic ring of the substrate with the MIO group forms the cationic intermediate in which the pro-S b-H is acidified and thus the elimination step can take place (Fig. 7D). After elimination the unsaturated product can be released first (Fig. 7E), followed by leaving the ammonia as ammonium ion (Fig. 7F). Conclusions for the mechanism of action of phenylalanine ammonia-lyase About 30 years ago Hanson & Havir [1,6] proposed a mechanism for the PAL reaction starting with the addition of the a-amino group of the substrate to the prosthetic electrophile at that time believed to be dehydroalanine. Because this mechanism cannot explain how the nonacidic b-H of the substrate can be abstracted by the enzymic base [54], about 6 years ago another mechanism was suggested involving an electrophilic attack at the phenyl ring by the prosthetic group [52]. The new mechanism was supported by the easy reaction of alternative substrates, especially 3-hydroxyphenylalanine (m-tyrosine), which facilitated the electrophilic attack. Furthermore, the S203A mutant of PAL lacking the electrophilic MIO group reacted easily with 4-nitrophenylalanine, in which the b-H is more acidic [52]. In the older literature there are also data that seem to support the Hanson mechanism. Peterkofsky found that HAL catalyses the incorporation of [2- 14 C]urocanate into L -histidine but not of 15 NH 4 + [55]. Unfortunately, in his experiments the concentration of NH 4 + was too low (10–60 m M ); on the basis of these and other previous experiments [56,57], he assumed that the HAL (as well as the PAL) reaction was irreversible. Later, it was shown that at higher NH 4 + concentrations (up to 6 M ) both ammonia- lyase reactions can be completely reversed [28,58]. Never- theless, Peterkofsky’s experiments support the existence of an enzyme–NH 4 + intermediate and the idea that ammonia is released after the other product (urocanate or cinnamate). This does not confirm, however, that ammonia is covalently bonded to the prosthetic electrophile (MIO) as the Hanson mechanism requires. Because the mutant ammonia-lyases lacking MIO (S143A for HAL and S203A for PAL) still catalyze the reaction (about 10 3 times more slowly with the natural substrates, but much faster for their nitro-deriva- tives [52,59]), it is clear that in these cases covalent binding of NH 4 + cannot occur. Modelling the active sites contain- ing the product (cinnamate: Fig. 5C and urocanate: Fig. 2C in [50]) suggest that the product may be released first, and before leaving thus blocks the narrow channel for the release of ammonia/NH 4 + (Fig. 7). Therefore, an ionic bonding of ammonia/NH 4 + is a reasonable assumption. In another paper purportedly supporting the Hanson mechanism, 15 N and 2 H-isotope effects on the PAL reaction are described [54]. The ammonia produced in the PAL reaction was converted in two steps into N 2 whose 15 N-content was determined by mass spectrometry. The latter was 1% less than the one expected from the natural abundance of 15 N (0.369%). Even if these small effects are real, it is not clear in which step they occurred. Because the PAL reaction is not irreversible and the Peterkofsky experiment showed that Fig. 7. Proposed mode of entering the substrate (A, B and C), reaction (D) and the release of the products (E,F) for the PAL and HAL reac- tions. Y denotes Y110 and Y53, N denotes N260 and N195, B denotes Y251 and E414 for PAL and HAL, respectively. Table 2. Distances in models of PAL’s active site. Selected distances (A ˚ ) in models for the zwitterionic L -phenylalanine binding (A), for the r-complex intermediate containing (B), and for the E-cinnamate/ammonia binding (C) state of the PAL active site are listed. Atomic pairs Model 5A Model 5B Model 5C S203 C3 –Phe C2¢ 4.18 1.54 4.65 S203 O1 –Phe C3¢ 3.78 3.05 3.73 Y435 O4¢ –Phe C6¢ 3.35 3.93 3.30 Y351 OH –Phe C3 4.62 3.61 4.56 Q488 O1 –Phe C3 6.97 6.04 6.88 Q488 NH1 –MIO N1 3.34 4.20 3.24 N260 O1 –Phe N2 4.68 4.58 4.88 Y110 OH –Phe N2 3.14 3.80 3.38 Y110 OH –Phe O1 3.01 2.90 3.10 Q348 N4 –Phe O1 3.39 3.93 3.97 R354 NH1 –Phe O¢1 3.36 3.26 4.54 R354 NH2 –Phe O¢1 3.93 3.36 5.03 Ó FEBS 2002 Active-site model of phenylalanine ammonia-lyase (Eur. J. Biochem. 269) 3073 there is a fast reverse reaction from the enzyme-NH 4 + intermediate [55], discrimination of 15 N can occur in the later steps. We consider the results presented here and in our previous papers [50–52] and conclude that the mechanism involving the attack of MIO at the aromatic portion of the substrates is more consistent with the experimental data and modelling studies than the alternative mechanism proposed in some older studies [1,54,55]. ACKNOWLEDGEMENTS We thank Prof. G. E. Schulz and M. Baedeker (University of Freiburg) for providing us with the new PAL expression system and also Dr M. Stieger (Hoffmann-La Roche, Basel, Switzerland) for vector pREP4-GroESL carrying the HSP-60 system. D. R. thanks the Land Baden-Wu ¨ rttemberg for a scholarship for graduate students. This work was supported by the Deutsche Forschungsg- emeinschaft and the Fonds der Chemischen Industrie. Financial support by the Hungarian OTKA (T-033112) is also gratefully acknowledged. REFERENCES 1. Hanson, K.R. & Havir, E.A. (1978) An introduction to the enzymology of phenylpropanoid biosynthesis. Rec. Adv. Phy- tochem. 12, 91–137. 2. Hahlbrock, K. & Scheel, D. (1989) Physiology and molecular biology of phenylpropanoid metabolism. Annu.Rev.Plant.Phys. Plant Mol. Biol. 40, 347–369. 3. Taylor, R.G., Lambert, M.A., Sexsmith, E., Sadler, S.J., Ray, P.N., Mahuran, D.J. & McInnes, R.R. (1990) Cloning and expression of rat histidase. Homology to two bacterial histidases and four phenylalanine ammonia-lyases. J. Biol. Chem. 265, 18192–18199. 4. Wickner, R.B. (1969) Dehydroalanine in histidine ammonia-lyase. J. Biol. Chem. 244, 6550–6552. 5. Givot, I.L., Smith, T.A. & Abeles, R.H. (1969) Studies on the mechanism of action and the structure of the electrophilic center of histidine ammonia-lyase. J. Biol. Chem. 244, 6341–6353. 6. Hanson, K.R. & Havir, E.A. (1970) L -Phenylalanine ammonia- lyase. IV. Evidence that the prosthetic group contains a dehydroalanyl residue and mechanism of action. Arch. Biochem. Biophys. 141, 1–17. 7. Schwede, T.F., Re ´ tey, J. & Schulz, G.E. (1999) Crystal structure of histidine ammonia-lyase revealing a novel polypeptide modifica- tion as the catalytic electrophile. Biochemistry 38, 5355–5361. 8. Ormo ¨ ,M.,Cubitt,A.B.,Kallio,K.,Gross,L.A.,Tsien,R.Y.& Remington, S.J. (1996) Crystal structure of the Aequorea victoria green fluorescent protein. Science 273, 1392–1395. 9. Ro ¨ ther, D., Merkel, D. & Re ´ tey, J. (2000) Spectroscopic evidence for a 4-methylidene imidazol-5-one in histidine and phenylalanine ammonia-lyases. Angew. Chem. 112, 2592–2594. Angew. Chem. Int. Eds 39, 2462–2464. 10. Baedeker, M. & Schulz, G.E. (1999) Overexpression of a designed 2.2 kb gene of eukaryotic phenylalanine ammonia-lyase in Escherichia coli. FEBS Lett. 457, 57–60. 11. Braman, J., Papworth, C. & Greener, A. (1996) Site-directed mutagenesis using double-stranded plasmid DNA templates. Methods Mol. Biol. 57, 31–44. 12. Sanger, F., Nicklen, S. & Coulson, A. (1977) DNA sequencing with chain-termination inhibitors. Proc. Natl Acad. Sci. USA 78, 177–181. 13. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 277, 680–685. 14.Towbin,H.,Staehlin,T.&Gordon,A.(1979)Electrophoretic transfer of proteins from acrylamide gels to nitrocellulose sheets: Procedures and applications. Proc.NatlAcad.Sci.USA76, 4350– 4354. 15. Burnette, W.N. (1981) ‘Western-blotting’: electrophoretic trans- fer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibodies and radionated protein A. Anal. Biochem. 112, 195–203. 16. Zimmermann, A. & Hahlbrock, K. (1975) Light-induced changes of enzyme activities in parsley cell suspension cultures. Arch. Biochem. Biophys. 166, 54–62. 17. Lineweaver, H. & Burk, D. (1934) The determination of enzyme dissociation constants. J. Am. Chem. Soc. 56, 658–666. 18. Warburg, O. & Christian, W. (1942) Isolierung und Kristallisation des Ga ¨ rungsfermentes Enolase. Biochem. Z. 310, 384–421. 19. Layne, E. (1957) Spectrophotometric and turbidimetric methods for measuring proteins. Methods Enzymol. 3, 447–454. 20. Murphy,J.B.&Kies,M.W.(1960)Noteonspectrophotometric determination of proteins in dilute solutions. Biochim. Biophys. Acta 45, 382–384. 21. Groves, W.E., Davis, F.C. Jr & Sells, B.H. (1968) Spectro- photometric determination of microgram quantities of protein without nucleic acid interference. Anal. Biochem. 22, 195–210. 22. Consevage, M.W. & Phillips, A.T. (1990) Sequence analysis of the hutH gene encoding histidine ammonia-lyase in Pseudomonas putida. J. Bacteriol. 172, 2224–2229. 23. Suchi, M., Harada, N., Wada, Y. & Takagi, Y. (1993) Molecular cloning of a cDNA encoding human histidase. Biochim. Biophys. Acta 1216, 293–295. 24. Lois, R., Dietrich, A., Hahlbrock, K. & Schulz, W. (1989) A phenylalanine ammonia-lyase gene from parsley: structure, reg- ulation and identification of elicitor and light responsive cis-acting elements. EMBO J. 8, 1641–1648. 25. Huang, X. (1994) On global sequence alignment. Comput. Appl. Biosci. 10, 227–235. 26. Peitsch, M.C. (1996) ProMod and Swiss-Model: Internet-based tools for automated comparative protein modelling. Biochem. Soc. Trans. 24, 274–279. 27. Peitsch, M.C. & Guex, N. (1997) Swiss-Model and the Swiss- PDBviewer: an environment for comparative protein modelling. Electrophoresis 18, 2714–2723. 28.Gloge,A.,Zon,J.,K } oova ´ ri, A ´ ., Poppe, L. & Re ´ tey, J. (2000) Phenylalanine ammonia-lyase: the use of its broad substrate spe- cificity for mechanistic investigations and biocatalysis. Synthesis of L -arylalanines. Chem. Eur. J. 6, 3386–3390. 29. Brooks, B.R., Bruccoleri, R.E., Olafson, B.D., States, D.J., Swaminathan, S. & Karplus, M. (1983) CHARMM: a program for macromolecular energy, minimization, and dynamics calcula- tions. J. Comput. Chem. 4, 187–217. 30. Cornell,W.D.,Cieplak,P.,Bayly,C.I.,Gould,I.R.,Merz,K.M. Jr, Ferguson, D.M., Spellmeyer, D.C., Fox, T., Caldwell, J.W. & Kollman, P.A. (1995) A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 117, 5179–5197. 31. Kollman, P., Dixon, R., Cornell, W., Fox, T., Chipot, C. & Pohorille, A. (1997) The development/application of a ¢minimalist¢ organic/biochemical molecular mechanic force field using a combination of ab initio calculations and experimental data. In Computer Simulation of Biomolecular Systems (van Gunsteren, W.F.Weiner, P.K.Wilkinson, A.J., eds), Vol. 3, pp. 83–96. Kluwer Academic Publishers, Dordrecht. 32. Lii, J H. & Allinger, N.L. (1991) The MM3 force field for amides, polypeptides and proteins. J. Comput. Chem. 12, 186–199. 33. Ponder, J. (1999) TINKER v3.7, available from http://dasher. wustl.edu/tinker/ 3074 D. Ro ¨ ther et al. (Eur. J. Biochem. 269) Ó FEBS 2002 [...]... Identification of ¨ essential amino acids in phenylalanine ammonia-lyase by sitedirected mutagenesis Biochemistry 36, 10867–10871 ´ 52 Schuster, B & Retey, J (1995) The mechanism of action of phenylalanine ammonia-lyase: the role of prosthetic dehydroalanine Proc Natl Acad Sci USA 92, 8433–8437 53 Baedeker, M & Schulz, G.E (2002) Autocatalytic peptide cyclization during chain folding of histidine ammonia-lyase. .. J.D., Weiss, P.M & Cleland, W.W (1985) Use of nitrogen-15 and deuterium isotope effects to determine the chemical mechanism of phenylalanine ammonia-lyase Biochemistry 24, 2959–2967 55 Peterkofsky, A (1962) Mechanism of action of histidase Aminoenzyme formation and partial reactions J Biol Chem 237, 787–795 56 Mehler, A.H & Tabor, H (1953) Deamination of histidine to form urocanic acid in liver J Biol... 2002 Active- site model of phenylalanine ammonia-lyase (Eur J Biochem 269) 3075 34 Hypercube, Inc (2002) Hyperchem 5.1 and 6.2, available from http://www.hyper.com/products/default.htm 35 Gunsteren, W.F., (1996) Biomolecular simulation: the GROMOS96 manual and user guide Vdf Hochschulverlag, ETH Zurich ¨ 36 Rost, B., Sander, C & Schneider, R (1994) PHD – an Automatic Mail Server for Protein Secondary Structure... 775–784 57 Yoshioka, Y (1955) Histidine deaminase isolated from guinea-pig liver Osaka Daigaku Igaku Zassi 7, 377–387 58 Williams, V.R & Hiroms, J.M (1967) Reversibility of the ÔirreversibleÕ histidine ammonia-lyase reaction Biochim Biophys Acta 139, 214–216 ´ 59 Langer, M., Pauling, A & Retey, J (1995) The Role of Dehydroalanine in the Catalysis by Histidine Ammonia-Lyase Angew Chem 107, 1585–1587 Angew... Giorgianni, F., Schindler, J.F., Viola, R.E & Farber G.K (1993) Crystallization and preliminary X-ray studies of L-aspartase from Escherichia coli J Mol Biol 234, 1248–1256 41 Shi, W., Dunbar, J., Jayasekera, M.M., Viola, R.E & Farber, G.K (1997) The structure of L-aspartate ammonia-lyase from Escherichia coli Biochemistry 36, 9136–9144 42 Havir, E.A & Hanson, K.R (1968) L -Phenylalanine ammonialyase... available from http:// www.wavefun.com/software/pc_spartan_pro/pcpro_main.html 49 Molecular Simulations Inc (2002) Weblabviewer Lite, V4.0, available at http://www.msi.com ´ 50 Rother, D., Poppe, L., Viergutz, S., Langer, B & Retey, J (2001) ¨ Characterization of the active site of histidine ammonia-lyase from Pseudomonas putida Eur J Biochem 268, 6011–6019 ´ 51 Langer, B., Rother, D & Retey, J (1997)... & Hanson, K.R (1968) L -Phenylalanine ammonialyase I Purification and molecular size of the enzyme from potato tubers Biochemistry 7, 1896–1903 43 Vriend, G (1990) WHAT IF: a molecular modelling and drug design program J Mol Graph 8, 52–56 44 Rodriguez, R., Chinea, G., Lopez, N., Pons, T & Vriend, G (1998) Homology modelling, model and software evaluation: three related resources CABIOS 14, 523–528 45... CABIOS, 10, 53–60 37 Rost, B & Sander, C (1994) Combining evolutionary information and neural networks to predict protein secondary structure Proteins 19, 55–72 38 Fischer, D., Barret, C., Bryson, K., Elofsson, A., Godzik, A., Jones, D., Karplus, K.J., Kelley, L.A., Maccallum, R.M., Pawowski, K., Rost, B., Rychlewski, L & Sternberg, M.J (1999) CAFASP-1: Critical assessment of fully automated structure prediction... 39 Kelley, L.A., Maccallum, R & Sternberg, M.J.E (1999) Recognition of remote protein homologies using three-dimensional information to generate a position specific scoring matrix in the program 3D-PSSM, RECOMB 99 In Proceedings of the Third Annual Conference on Computational Molecular Biology (Istrail, S, Pevzner, P., Waterman, M., eds), pp 218–225 The Association for Computing Machinery, New York 40... Stereochemical quality of protein structure coordinates Proteins 12, 345–364 46 Laskowski, R.A., MacArthur, M.W., Moss, D.S & Thornton, J.M (1993) PROCHECK: a program to check the stereochemical quality of protein structures J Appl Cryst 26, 283–291 47 EMBL (2002) Biotech Validation Suite for Protein Structures, available from http://biotech.embl-heidelberg.de:8400/ 48 Wavefunction, Inc (2002) PC Spartan Pro, available . An active site homology model of phenylalanine ammonia-lyase from Petroselinum crispum Dagmar Ro¨ ther 1 ,La ´ szlo ´ Poppe 2 , Gaby Morlock 1 , Sandra. residues in PAL catalysis. Keywords: phenylalanine ammonia-lyase; PAL; MIO; site- directed mutagenesis; homology model. Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5)