Portal de Periódicos da CAPES

Identification of the Substrate Specificity-conferring Amino Acid Residues of 4-Coumarate:Coenzyme A Ligase Allows the Rational Design of Mutant Enzymes with New Catalytic Properties

2001; Elsevier BV; Volume: 276; Issue: 29 Linguagem: Inglês

10.1074/jbc.m100355200

1083-351X

Hans‐Peter Stuible, Erich Kombrink,

Microbial Natural Products and Biosynthesis

4-Coumarate:coenzyme A ligases (4CLs) generally use, in addition to coumarate, caffeate and ferulate as their main substrates. However, the recently cloned Arabidopsis thaliana isoform At4CL2 is exceptional because it has no appreciable activity with ferulate. On the basis of information obtained from the crystal structure of the phenylalanine-activating domain of gramicidin S-synthetase, 10 amino acid residues were identified that may form the substrate binding pocket of 4CL. Among these amino acids, representing the putative “substrate specificity motif,” only one residue, Met293, was not conserved in At4CL2, compared with At4CL1 and At4CL3, two isoforms using ferulate. Substitution of Met293 or Lys320, another residue of the putative substrate specificity motif, which in the predicted three-dimensional structure is located in close proximity to Met293, by smaller amino acids converted At4CL2 to an enzyme capable of using ferulate. The activity with caffeate was not or only moderately affected. Conversely, substitution of Met293 by bulky aromatic amino acids increased the apparent affinity (K m) for caffeate up to 10-fold, whereas single substitutions of Val294 did not affect substrate use. The results support our structural assumptions and suggest that the amino acid residues 293 and 320 of At4CL2 directly interact with the 3-methoxy group of the phenolic substrate and therefore allow a first insight into the structural principles determining substrate specificity of 4CL. 4-Coumarate:coenzyme A ligases (4CLs) generally use, in addition to coumarate, caffeate and ferulate as their main substrates. However, the recently cloned Arabidopsis thaliana isoform At4CL2 is exceptional because it has no appreciable activity with ferulate. On the basis of information obtained from the crystal structure of the phenylalanine-activating domain of gramicidin S-synthetase, 10 amino acid residues were identified that may form the substrate binding pocket of 4CL. Among these amino acids, representing the putative “substrate specificity motif,” only one residue, Met293, was not conserved in At4CL2, compared with At4CL1 and At4CL3, two isoforms using ferulate. Substitution of Met293 or Lys320, another residue of the putative substrate specificity motif, which in the predicted three-dimensional structure is located in close proximity to Met293, by smaller amino acids converted At4CL2 to an enzyme capable of using ferulate. The activity with caffeate was not or only moderately affected. Conversely, substitution of Met293 by bulky aromatic amino acids increased the apparent affinity (K m) for caffeate up to 10-fold, whereas single substitutions of Val294 did not affect substrate use. The results support our structural assumptions and suggest that the amino acid residues 293 and 320 of At4CL2 directly interact with the 3-methoxy group of the phenolic substrate and therefore allow a first insight into the structural principles determining substrate specificity of 4CL. 4-Coumarate:coenzyme A ligase phenylalanine-activating domain Arabidopsis thaliana 4CL Populus tremuloides 4CL 4-Coumarate:coenzyme A ligase (4CL,1 EC 6.2.1.12) is an essential enzyme of the phenylpropanoid biosynthetic pathway, catalyzing the activation of various hydroxylated and methoxylated cinnamic acid derivatives to the corresponding thiol esters in a two-step reaction. During the first step, coumaric acid and ATP form a coumaroyl-adenylate intermediate with the simultaneous release of pyrophosphate. In the second step, the coumaroyl group is transferred to the sulfhydryl group of CoA, and AMP is released (1Knobloch K.-H. Hahlbrock K. Eur. J. Biochem. 1975; 52: 311-320Crossref PubMed Scopus (184) Google Scholar, 2Becker-André M. Schulze-Lefert P. Hahlbrock K. J. Biol. Chem. 1991; 266: 8551-8559Abstract Full Text PDF PubMed Google Scholar). Despite their low overall sequence identity, one highly conserved peptide motif is common to 4CLs, luciferases, fatty acyl-CoA synthetases, acetyl-CoA synthetases, and specialized domains within peptide synthetase multienzymes. This conserved, putative AMP binding domain has been used as the most important criterion to group the above-listed proteins in one superfamily of adenylate-forming enzymes (3Fulda M. Heinz E. Wolter F.P. Mol. Gen. Genet. 1994; 242: 241-249Crossref PubMed Scopus (72) Google Scholar). In good agreement with this theoretical classification and prediction of functional similarity, the mutational analysis of At4CL2 from Arabidopsis thaliana experimentally corroborated such a close functional relationship between 4CL and the adenylation and substrate recognition domains of nonribosomal peptide synthetases (4Stuible H.-P. Büttner D. Ehlting J. Hahlbrock K. Kombrink E. FEBS Lett. 2000; 467: 117-122Crossref PubMed Scopus (99) Google Scholar). Considerable progress has been made in determining the structural basis of substrate specificity in the peptide synthetase system (5Conti E. Stachelhaus T. Marahiel M.A. Brick P. EMBO J. 1997; 16: 4174-4183Crossref PubMed Scopus (596) Google Scholar, 6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar, 7Stachelhaus T. Marahiel M.A. J. Biol. Chem. 1995; 270: 6163-6169Abstract Full Text Full Text PDF PubMed Scopus (141) Google Scholar). The crystal structure of the N-terminal adenylation subunit (phenylalanine-activating domain (PheA)) of gramicidinS-synthetase 1 obtained in the presence of AMP andl-phenylalanine, a substrate structurally related to coumaric acid, allowed both the localization of catalytic important positions and the identification of 10 amino acids lining the phenylalanine binding pocket (5Conti E. Stachelhaus T. Marahiel M.A. Brick P. EMBO J. 1997; 16: 4174-4183Crossref PubMed Scopus (596) Google Scholar). All substrate binding pocket constituents of PheA (except Lys517, which functions as the active site in adenylate formation and is therefore obviously not involved in substrate discrimination) could be identified within a 100-amino acid residue-comprising region, which is flanked by the conserved core motifs A3 and A6 of peptide synthetases (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar) exhibiting high sequence identity with the Box I(LPFSSGTTGLPKG) andBox II(GEICIRG) sequences of 4CLs, respectively. Using different sequence alignment strategies, the substrate specificity conferring amino acids corresponding to the residues identified in PheA have also been localized in other peptide synthetases (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar, 8Challis G.L. Ravel J. Townsend C.A. Chem. Biol. 2000; 7: 211-224Abstract Full Text Full Text PDF PubMed Scopus (671) Google Scholar). The importance of several of these positions for substrate recognition and specificity has been verified by site-directed mutagenesis (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar). On the basis of these data, the concept of a substrate binding pocket with a limited number of contact residues being involved in substrate specificity determination has been established for several peptide synthetases (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar, 8Challis G.L. Ravel J. Townsend C.A. Chem. Biol. 2000; 7: 211-224Abstract Full Text Full Text PDF PubMed Scopus (671) Google Scholar, 9von Döhren H. Dieckmann R. Pavela-Vrancic M. Chem. Biol. 1999; 6: R273-R279Abstract Full Text PDF PubMed Scopus (73) Google Scholar). The position of 4CL at a metabolic branch point connecting general phenylpropanoid metabolism with different end product-specific pathways makes 4CL a promising target for biotechnological manipulation. The suitability of 4CL for product or pathway engineering is not only suggested by its position within the plant secondary metabolism but is additionally supported by recent studies that indicate that 4CL isoforms by virtue of their different substrate specificities and expression patterns channel phenolic precursors either to the lignin or the flavonoid biosynthetic pathway (10Hu W.-J. Kawaoka A. Tsai C.-J. Lung J. Osakabe K. Ebinuma H. Chiang V.L. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 5407-5412Crossref PubMed Scopus (215) Google Scholar, 11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). We recently cloned and biochemically characterized three 4CL isoforms from A. thaliana (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). Although showing differences in detail, all three 4CLs were able to activate their typical substrates, coumarate and caffeate. Unexpected, however, was the observation that only the isoforms At4CL1 and At4CL3 were able to convert ferulate to the corresponding thiol ester, whereas At4CL2 was not (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). Both the minor structural differences existing between caffeate (3,4-dihydroxy cinnamic acid) and ferulate (3-methoxy-4-hydroxy cinnamic acid) and the high sequence identity found between the three At4CL isoforms suggested that only a limited number of amino acid residues are responsible for the inability of At4CL2 to convert ferulate to its thiol ester. Therefore, we decided to study the molecular mechanism preventing ferulate activation by At4CL2 in detail to unravel the principle rules of substrate specificity determination in the 4CL system. Standard DNA techniques were performed as described elsewhere (12Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual. 2nd ed. Cold Spring Harbor Laboratory Press, Plainview, NY1989Google Scholar). For plasmid amplification, theEscherichia coli strain DH5α (Life Technologies, Inc.) was used. Protein expression was performed using the E. colistrain M15(pRP4). The pQE30-based At4CL2 expression plasmid used for mutagenesis experiments has previously been described (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). Point mutations were introduced into At4CL2 by polymerase chain reaction amplification of the entire At4CL2 expression plasmid by using two mutated oligonucleotide primers, each complementary to opposite strands of the vector. Double mutants were created by polymerase chain reaction amplification of already existing At4CL2 mutant plasmids using a new pair of mutated oligonucleotide primers. Conditions for polymerase chain reaction-based mutagenesis were as follows: 1 cycle at 94 °C (1 min), 16 cycles at 94 °C (1 min), 59 °C (1 min), and 68 °C (14 min), and 1 final polymerization step at 68 °C (14 min). All components necessary for this mutagenesis procedure were included in the commercial QuikChange kit (Stratagene). One primer sequence of each pair of complementary mutagenesis primers is listed in TableI.Table ISequences of PCR primers used for the construction of At4CL2 mutant proteinsPrimer sequence1-aModified codons are underlined.MutationGGTGTAAAGTCACGGTGGCTCCAGTCGTGCCACCGATCGTTTTAGCM293PGGTGTAAAGTCACGGTGGCTGCCGTCGTGCCACCGATCGTTTTAGCM293AGGTGTAAAGTCACGGTGGCTATGATGGTGCCACCGATCGTTTTAGCV294MGGTGTAAAGTCACGGTGGCTATGCTCGTGCCACCGATCGTTTTAGCV294LGGTGTAAAGTCACGGTGGCTTTCGTCGTGCCACCGATCGTTTTAGCM293FGGTGTAAAGTCACGGTGGCTTACGTCGTGCCACCGATCGTTTTAGCM293YGGTGTAAAGTCACGGTGGCTTGGGTCGTGCCACCGATCGTTTTAGCM293WCTGAGCTCGGTTAGGATGGTTCTGTCTGGAGCAGCTCCTCTTGGK320LCTGAGCTCGGTTAGGATGGTTGCGTCTGGAGCAGCTCCTCTTGGK320A1-a Modified codons are underlined. Open table in a new tab The entire DNA sequences of all mutated reading frames were determined on ABI Prism 377 DNA sequencers (Applied Biosystems, Forster City, CA). To identify the putative substrate pocket-lining amino acids of 4CL, sequence alignments were generated with the PILEUP program of the Genetics Computer Group program package, version 10.0, with the gap weight and gap length weight set to 2 and 8, respectively (13Devereux J. Haeberli P. Smithies O. Nucleic Acids Res. 1984; 12: 387-395Crossref PubMed Scopus (12062) Google Scholar), using the following sequences (with GenBank accession numbers given in parentheses brackets): A. thaliana At4CL2 (AAD47193), At4CL1 (AAD47191), At4CL3 (AAD47194), Brevibacillus brevisgramicidin S-synthetase 1 (PheA; CAA33603), and other 4CL sequences as listed previously (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). Expression and purification of At4CL2 proteins were performed as recently described (4Stuible H.-P. Büttner D. Ehlting J. Hahlbrock K. Kombrink E. FEBS Lett. 2000; 467: 117-122Crossref PubMed Scopus (99) Google Scholar). 4CL activity was determined with the spectrophotometric assay previously described, using caffeic acid and ferulic acid as phenolic substrates (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar, 14Knobloch K.-H. Hahlbrock K. Arch. Biochem. Biophys. 1977; 184: 237-248Crossref PubMed Scopus (120) Google Scholar). The change of absorbance during CoA ester formation was monitored at 363 nm for caffeoyl-CoA and at 345 nm for feruloyl-CoA (15Stöckigt J. Zenk M.H. Z. Naturforsch. 1975; 30c: 352-358Crossref Scopus (281) Google Scholar). TheK m and V max values for caffeate and ferulate were estimated at fixed concentrations of ATP (5.5 mm) and CoA (0.3 mm) by linear regression of v/s against s (Hanes plot). Since K m and V max values could not be determined for all mutant enzymes because of their low activity, specific activity was determined in addition at standard conditions (0.2 mm phenolic substrate, 5.5 mmATP, 0.3 mm CoA) for all 4CL variants. Protein concentrations were determined according to Bradford (16Bradford M.M. Anal. Biochem. 1976; 72: 248-254Crossref PubMed Scopus (222164) Google Scholar) with bovine serum albumin as standard. To identify the substrate binding pocket-lining amino acids within the At4CL2 primary sequence, different multiple sequence alignments of the three 4CL isoforms of A. thaliana were performed with several adenylation domains of peptide synthetases, including PheA, luciferases, as well as all other available 4CL sequences (for details, see “Experimental Procedures”). We assumed that the functional relatedness between these three groups of enzymes is also reflected at the structural level, as recently described for PheA and firefly luciferase (5Conti E. Stachelhaus T. Marahiel M.A. Brick P. EMBO J. 1997; 16: 4174-4183Crossref PubMed Scopus (596) Google Scholar, 17Conti E. Franks N.P. Brick P. Structure. 1996; 4: 287-298Abstract Full Text Full Text PDF PubMed Scopus (552) Google Scholar). In good agreement with this presumption, the reduced sequence alignment shown in Fig.1 reveals such structural similarity by the colinear organization of the two highly conserved sequence motifs common to 4CLs (Box I and Box II) and PheA (core motifs A3 and A6) and the length and sequence of the intervening region. Because in PheA all substrate binding pocket constituents are known to be located between the peptide motifs A3 and A6, we predicted that the substrate specificity conferring amino acids of 4CL are located between the putative AMP binding domain (Box I) and the so-called GEICIRG domain (Box II). Furthermore, the unusual features of At4CL2, which activates 3,4-dihydroxy cinnamic acid (caffeate) as its preferred substrate but is incapable of converting 3-methoxy-4-hydroxy cinnamic acid (ferulate) to the corresponding thiol ester, should be reflected in the constituents of the substrate binding pocket in comparison with the ferulate-using enzymes At4CL1 and At4CL3. By extraction of the nine putative substrate binding pocket-forming residues of At4CL1, At4CL2, and At4CL3 from the alignment with PheA, we identified two differences among the three Arabidopsis 4CL “substrate specificity motifs” (Fig. 1). First, the bulky methionine residue at position 293 in At4CL2 is substituted by proline and alanine residues in At4CL1 and At4CL3, respectively (Fig. 1). Because At4CL1 and At4CL3, in contrast to At4CL2, are able to activate ferulate, the observed difference suggested that ferulate use could be regulated by a size exclusion mechanism mediated by the bulkiness of the amino acid at position 293. Second, in At4CL3, position 320 (At4CL2 numbering) is occupied by a leucine, whereas At4CL1 and At4CL2 both carry a lysine in this position (Fig. 1). Although the different amino acid residues at position 320 allowed no obvious explanation of the available biochemical data on substrate use, in particular the missing ferulate activation by At4CL2, we assumed that this position could also be important for determining substrate specificity, because it should be located in close proximity to the other variable amino acid (Met293) if the enzyme has a three-dimensional structure similar to that of PheA. To test whether the bulky methionine residue at position 293 in At4CL2 is responsible for the lack of capacity of the enzyme to activate ferulate, Met293 was replaced by proline or alanine as suggested from the primary sequences of At4CL1 and At4CL3, respectively (Fig. 1). Both mutant enzymes, M293P and M293A, retained their activity with caffeate, and the K mand V max values for this substrate were comparable with those of the wild-type enzyme (TableII). However, as a novel catalytic activity, both mutant enzymes were able to activate ferulate, exhibiting K m values fairly similar to those reported for At4CL1 (199 µm) and At4CL3 (166 µm; Ref. 11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar). Although the specific activities of these new At4CL2 variants with ferulate were ∼50-fold higher than the residual activity displayed by the wild-type enzyme (2.3 picokatals/mg), caffeate was still the preferred substrate because of the significantly lower K m value (Table II). This is clearly illustrated by calculation of theV max:K m ratios, which for both mutants varied from 8.2 to 8.6 for caffeate and 0.5 to 1.1 for ferulate.Table IIKinetic properties of Arabidopsis 4CL2 mutant enzymesEnzymeCaffeateFerulateK mV maxSA2-aSpecific activity (based on protein), determined under standard conditions with 0.2 mmphenolic substrate.K mV maxSA2-aSpecific activity (based on protein), determined under standard conditions with 0.2 mmphenolic substrate.µmpicokatals/mgpicokatals/mgµmpicokatals/mgpicokatals/mgWild type24 ± 1.5158 ± 9144 ± 8ND2-bND, not determined, activity too low.ND2.3 ± 0.9M293P25 ± 8206 ± 38183 ± 25224 ± 46243 ± 9115 ± 7M293A24 ± 7.5206 ± 46188 ± 41558 ± 45284 ± 3678 ± 14M293P/V294M21 ± 4.6187 ± 44168 ± 42105 ± 27210 ± 37140 ± 40M293A/V294L24 ± 7190 ± 40170 ± 32162 ± 61290 ± 34165 ± 22V294M22 ± 4134 ± 23120 ± 18NDND4.9 ± 1V294L23 ± 4.6114 ± 11104 ± 10NDND3.2 ± 0.6M293F63 ± 17177 ± 63136 ± 41NDND<0.3M293Y224 ± 53176 ± 3088 ± 5NDND<0.1M293W180 ± 4244 ± 1023 ± 6NDND8.6 ± 0.6K320L74 ± 13147 ± 6107 ± 4151 ± 50234 ± 34140 ± 12K320A55 ± 6130 ± 12105 ± 840 ± 2135 ± 15108 ± 7K320L/M293P48 ± 6119 ± 3397 ± 2946 ± 7270 ± 50263 ± 62K320A/M293P84 ± 1963 ± 645 ± 658 ± 592 ± 871 ± 7All values are the mean ± S.D. of at least three independent determinations with affinity-purified protein.2-a Specific activity (based on protein), determined under standard conditions with 0.2 mmphenolic substrate.2-b ND, not determined, activity too low. Open table in a new tab All values are the mean ± S.D. of at least three independent determinations with affinity-purified protein. To test the specificity of the observed mutant phenotype after alteration of amino acid residue 293, we generated a second set of mutants at the neighboring position 294 that is also variable among theArabidopsis 4CLs (Fig. 1). Again, the newly introduced amino acids were selected according to the sequences of At4CL1 and At4CL3. The hereby generated At4CL2 variants, V294M and V294L, exhibited slightly reduced V max values and specific activities with caffeate, whereas no significant activity was observed with ferulate. Thus, despite its close proximity to Met293, the valine at position 294 is apparently not involved in determining substrate specificity in 4CL. However, the results obtained with the double mutants M293P/V294M and M293A/V294L demonstrate that substitutions of Val294 have a modulating influence on ferulate activation when combined with a second mutation, leading to a 2–3-fold decrease of the K m values in comparison with the single mutants M293P and M293A (Table II). Because a size reduction of the amino acid residue at position 293 seemed to be essential for the acceptance of ferulate as a substrate by At4CL2, we postulated that the introduction of bulky aromatic amino acids at this position should not only prevent ferulate (4-hydroxy-3-methoxy cinnamate) activation but should also reduce the efficiency of caffeate (3,4-dihydroxy cinnamate) activation. Therefore, it was no surprise that none of the three At4CL2 mutant enzymes, M293F, M293Y, and M293W, exhibited any significant enzymatic activity with ferulate (Table II). Substitution of Met293 by the smallest aromatic amino acid, phenylalanine, also did not appreciably affect theK m (only 2-fold) and V maxvalues for caffeate, whereas the bulkier amino acids tyrosine and tryptophan had significant effects. The M293Y substitution resulted in a 10-fold increase in the K m value without affectingV max, whereas by the M293W substitution, both parameters were affected, leading to a drastically reduced enzymatic capacity of the mutant enzyme (Table II). Collectively, these results support the model that the bulkiness of the amino acids lining the substrate binding pocket is an important factor determining substrate specificity in the 4CL system. To study a possible contribution of Lys320 on the catalytic properties of At4CL2, we substituted this residue with leucine, as suggested from the sequence comparison (Fig. 1), and with alanine. Compared with the wild-type enzyme, both substitutions, K320L and K320A, resulted in a 2–3-fold increase in the K mfor caffeate use, whereas V max was not affected, although a large, positively charged amino acid was replaced by smaller, aliphatic amino acids (Table II). In addition, both new mutant enzymes exhibited remarkable activity with ferulate, similar to the 4CL variants M293A and M293P (Table II). The properties of the mutant enzyme with the K320L substitution resembled those of the ferulate-activating At4CL2 variants carrying single point mutations at position 293 (Table II). In contrast, the K320A substitution led to a 4-fold decrease in the K m value for ferulate when compared with the K320L mutation, although no alanine was found at this position in any of the available 4CL sequences. The lowK m value (40 µm) in combination with the moderate V max value (135 picokatals/mg) resulted in a relatively highV max:K m ratio of 3.4 for ferulate, thereby slightly exceeding theV max:K m ratio of 2.4 for caffeate. The stronger mutant phenotype obtained with the K320A substitution in comparison with the K320L exchange again indicates that an increase of the available space for the substrate binding pocket allows a higher activity with bulky substrates. To analyze potential additive effects, two enzyme double mutants were created, each carrying ferulate specificity-conferring amino acids at both positions identified to be important (293 and 320). When the K320L/M293P variant and the K320L single mutant were compared with respect to caffeate use, no additive effect of the second mutation was apparent (Table II). However, for ferulate activation, theK m value of the K320L/M293P variant was 3–5-fold lower than the K m values of the corresponding enzymes with single mutations, whereas the V maxvalue was not affected (Table II). The combination of lowK m (46 µm) and highV max (270 picokatals/mg) made ferulate with aV max:K m ratio of 5.9, instead of caffeate with a V max:K mratio of 2.4, the preferred substrate of this new At4CL2 variant. In this case, the introduction of a second mutation had an additive effect and changed the substrate preference of the enzyme from caffeate to ferulate. A more complex situation was observed with the K320A/M293P double mutant, which exhibited K m values for ferulate and caffeate fairly similar to those of the K320A single mutant. In addition, however, the V max values and the specific activities were also clearly reduced for both substrates, and a positive influence of the second mutation on ferulate use was not apparent (Table II). In conclusion, the introduction of small amino acids at either position, 293 or 320, of the At4CL2 primary sequence promoted ferulate activation by this enzyme and generated a new substrate use profile. Together with the occurrence of additive effects of both mutations, as observed with the K320L/M293P double mutant, these results indicate that both substituted amino acid residues represent substrate binding pocket constituents, which are located in close proximity to each other in the three-dimensional structure of the protein and at the same time contact the 3-methoxy or 3-hydroxy groups of ferulate and caffeate, respectively. The results presented in this study demonstrate the possibility of rationally designing 4CL activity by site-directed mutagenesis. The generation of new enzymatic specificities using At4CL2 as a model system represents a first step toward modification of biosynthetic processes located downstream of the general phenylpropanoid metabolism. Starting from the crystal structure of PheA (5Conti E. Stachelhaus T. Marahiel M.A. Brick P. EMBO J. 1997; 16: 4174-4183Crossref PubMed Scopus (596) Google Scholar), we identified nine amino acid residues in the At4CL2 primary sequence that may line the binding pocket for the phenolic substrate. Because the amino acid sequences of At4CL2 and PheA are only ∼15% identical in the region of ∼100 amino acids that harbors all the substrate binding and specificity-conferring constituents, it has to be emphasized that the assignment of functionally important amino acids in At4CL2 on the basis of sequence alignments is not necessarily correct for each predicted position. The functional relevance has to be verified experimentally for each position, and it may well be that other or additional positions than those deduced from the alignment with PheA are necessary and important for the determination of substrate specificity in the 4CL system. Nevertheless, the strong influence on ferulate and caffeate use exerted by variation of the amino acid residues at positions 293 and 320 in At4CL2 argue strongly in favor of a correct assignment of at least these two positions. The present results strongly suggest that both residues interact directly with the 3-hydroxy or 3-methoxy groups of the phenolic substrates and thus represent two of the amino acids lining the substrate binding pocket. The At4CL2 residues Met293 and Lys320 are likely to correspond to the PheA amino acids Thr278 and Ile299, respectively. In PheA, Thr278 and Ile299 are neighboring residues in the three-dimensional structure, both of which are located at the same side of the phenylalanine binding pocket (5Conti E. Stachelhaus T. Marahiel M.A. Brick P. EMBO J. 1997; 16: 4174-4183Crossref PubMed Scopus (596) Google Scholar). By comparing the nine substrate specificity-conferring amino acids of several peptide synthetases, five highly variable residues were identified, including Thr278and Ile299 in PheA. The high variability of these substrate binding pocket constituents was suggested to reflect their importance for determining substrate specificity (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar). In accordance with this predicted importance, substitution of Thr278 by the larger methionine residue resulted in a 20-fold decrease in the use of the miscognate and bulkier substrate tryptophan compared with the cognate substrate phenylalanine. This observation indicated that the reduction of the available space within the substrate binding pocket was responsible for the increase in substrate specificity (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar). The deduced colocalization of At4CL2 Met293 and Lys320 in the corresponding region of the substrate binding pocket is in good agreement with our experimental results. Substitution of either residue by a smaller amino acid renders the At4CL2 substrate binding pocket accessible to ferulate (Table II). Correspondingly, introduction of bulky aromatic amino acids at position 293 of At4CL2 resulted in a 10-fold increase of the K m value for caffeate (TableII). The data obtained from the peptide synthetase system indicate that the knowledge of specificity-conferring amino acids allows more precise predictions of substrate use by newly discovered adenylation domains before their biochemical characterization than would be possible by overall sequence comparison with functionally characterized enzymes (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar,8Challis G.L. Ravel J. Townsend C.A. Chem. Biol. 2000; 7: 211-224Abstract Full Text Full Text PDF PubMed Scopus (671) Google Scholar). Predictions of substrate use and specificity are also desirable for the 4CL system, particularly in view of the extensive genome sequences that are presently becoming available. For example, although experimental evidence suggests that Arabidopsis contains only three bona fide 4CLs (11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar), several new 4CL-like sequences were recently identified (accession numbers AC011000,AL132967, AL161502, and AL161549) for which a function has not yet been assigned. Despite the observed overall sequence similarity, 4CL activity could not be detected for one of these proteins (AC011000, gene F16P17.9) after heterologous expression in E. coli. 2H.-P. Stuible, unpublished results. Unfortunately, a systematic analysis of the parameters specifying substrate recognition and use by different 4CL isoforms has not yet been possible; in fact, it is impeded by the limited amount of available data correlating structural information with biochemical properties. The characterization of 4CL activities has been carried out largely with proteins purified from various plant species and tissues and revealed the existence of 4CL isoforms with considerable variation in substrate specificity, including the rare occurrence of activity toward sinapate (3,5-dimethoxy-4-hydroxy cinnamate; Refs. 1Knobloch K.-H. Hahlbrock K. Eur. J. Biochem. 1975; 52: 311-320Crossref PubMed Scopus (184) Google Scholar, 18Ranjeva R. Boudet A.M. Faggion R. Biochimie. 1976; 58: 1255-1262Crossref PubMed Scopus (45) Google Scholar, 19Wallis P.J. Rhodes M.J.C. Phytochemistry. 1977; 16: 1891-1894Crossref Scopus (31) Google Scholar, 20Grand C. Boudet A. Boudet A.M. Planta. 1983; 158: 225-229Crossref PubMed Scopus (60) Google Scholar). However, an unambiguous correlation of particular and unusual biochemical properties with a distinct 4CL coding sequence, e.g. by characterization of proteins expressed in E. coli, has been established only for A. thaliana (present work; Ref. 11Ehlting J. Büttner D. Wang Q. Douglas C.J. Somssich I.E. Kombrink E. Plant J. 1999; 19: 9-20Crossref PubMed Scopus (356) Google Scholar) andPopulus tremuloides (10Hu W.-J. Kawaoka A. Tsai C.-J. Lung J. Osakabe K. Ebinuma H. Chiang V.L. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 5407-5412Crossref PubMed Scopus (215) Google Scholar). The Populus protein Pt4CL1 can activate 5-hydroxy ferulate, whereas the isoform Pt4CL2 cannot. By comparison of the Pt4CL1 and Pt4CL2 sequences according to the same criteria as outlined above for the Arabidopsis 4CL system, a total of three amino acid differences can be identified among the nine putative substrate binding pocket constituents. However, a correlation between the bulkiness of these amino acid residues with the reported substrate preference of the Populus enzymes is not obvious, indicating that other factors such as hydrogen bonding, hydrophobicity, or Van der Waals interactions may also influence substrate recognition and use by 4CLs. Experimental evidence in support of the notion that in addition to size exclusion, such other mechanisms contribute to substrate specificity has again been obtained with the peptide synthetase system by Stachelhaus et al. (6Stachelhaus T. Mootz H.D. Marahiel M.A. Chem. Biol. 1999; 6: 493-505Abstract Full Text PDF PubMed Scopus (1005) Google Scholar), who were able to alter the specificity of an adenylation domain from aspartate to asparagine by a single histidine-to-glutamate substitution. There is considerable interest in manipulation of phenylpropanoid metabolism, because phenolic compounds have a wide range of important functions in plants. They serve as structural components (lignin and suberin), protectants against biotic and abiotic stresses (phytoalexins, antioxidants, and UV-absorbing compounds), flower pigments, and signal molecules (21Weisshaar B. Jenkins G.I. Curr. Opin. Plant Biol. 1998; 1: 251-257Crossref PubMed Scopus (409) Google Scholar). More recently, several plant phenolics received considerable attention as health-promoting “nutraceuticals” with antioxidant and anticancer activity (22Dixon R.A. Steele C.L. Trends Plant Sci. 1999; 4: 394-400Abstract Full Text Full Text PDF PubMed Scopus (609) Google Scholar). One approach to alter phenylpropanoid biosynthesis both quantitatively and qualitatively is to increase or reduce the expression of particular genes in transgenic plants (21Weisshaar B. Jenkins G.I. Curr. Opin. Plant Biol. 1998; 1: 251-257Crossref PubMed Scopus (409) Google Scholar). In Arabidopsis, for example, a reduction in lignin content by 50% was achieved by expression of an At4CL1 antisense construct (23Lee D. Meyer K. Chapple C. Douglas C.J. Plant Cell. 1997; 9: 1985-1998Crossref PubMed Scopus (215) Google Scholar), and in tobacco a similar approach resulted in an 80% reduction in lignin content (24Kajita S. Hishiyama S. Tomimura Y. Katayama Y. Omori S. Plant Physiol. 1997; 114: 871-879Crossref PubMed Scopus (105) Google Scholar). In both cases the lignin subunit composition was also drastically altered, which may reflect an eminent disadvantage of the antisense approach, namely that several members of a gene family may be affected in a differential and unpredictable manner. In addition, even though specific inhibition of one particular 4CL isoform may result in a desirable reduction in lignin content in one particular organ or cell type, it may also affect other biosynthetic pathways, such as flavonoid formation, in others, with all the potential detrimental effects. The partial lack of protective flavonoid derivatives may, for example, enhance susceptibility to pathogen infection or to damage by UV light exposure (25Landry L.G. Chapple C.C.S. Last R.L. Plant Physiol. 1995; 109: 1159-1166Crossref PubMed Scopus (521) Google Scholar). A more promising approach for manipulation of phenylpropanoid metabolism in a positive manner using the branch point position of 4CL may require modified structural genes encoding enzymes with clearly defined substrate specificities in combination with selectively altered promoter elements providing organ-, cell type-, and stimulus-specific expression. Correspondingly, the engineering of 4CL variants with enhanced capacity to activate the lignin precursor sinapate or the putative salicylate precursor cinnamic acid would be of great interest for both scientific as well as practical purposes. In conclusion, the data obtained in this study substantiate our prediction that, for its high-level expression, stability as active enzyme, which retains its structural integrity despite extensive mutational alterations, and simple functional analysis in an optical enzyme assay, At4CL2 represents a well suited model system to study the structural basis of substrate specificity in a subclass of adenylate-forming enzymes (4Stuible H.-P. Büttner D. Ehlting J. Hahlbrock K. Kombrink E. FEBS Lett. 2000; 467: 117-122Crossref PubMed Scopus (99) Google Scholar). To overcome the still-existing limitations, especially the lack of a three-dimensional structural model, both the crystallization of the enzyme and further mutational analyses of At4CL2 have been initiated. We thank Dr. Imre E. Somssich and Prof. K. Hahlbrock for critical comments on the manuscript, Prof. K. Hahlbrock also for continuous support of this work, and Roswitha Lentz for excellent technical assistance.