PAL (L-phenylalanine ammonia-lyase), the first enzyme of phenylpropanoid biosynthesis, is often encoded by multigene families in plants. A PCR-based approach was used to isolate cDNA clones corresponding to the four PAL genes of tobacco (Nicotiana tabacum). By careful comparison of cDNA and genomic clones, a new PAL gene (PAL4) was defined. PCR amplification of PAL sequences from cDNA led to the generation of chimaeric clones between PAL1 and PAL4, and incorrect annotation of PAL4 ESTs (expressed sequence tags) as PAL1 in the EST database has given rise to a randomly shuffled tentative consensus sequence. The PAL2 previously described in the literature was shown, by domain swapping experiments with PAL1, to possess a single nucleotide substitution leading to an inactive enzyme. The altered amino acid resulting from this substitution maps to the base of the active site pocket in the three-dimensional structure of PAL. The inactive PAL2 allele could not be recovered from 13 different tobacco cultivars examined. PALs 1–4 were co-expressed in multiple plant organs, and were also co-induced following exposure of cell cultures to yeast elicitor or methyl jasmonate. All four tobacco PAL proteins expressed in Escherichia coli displayed normal Michaelis–Menten kinetics, with Km values between 36 and 60 μM. Co-expression of different PAL proteins in E. coli resulted in formation of heterotetramers, which possessed kinetic properties within the same range as those of the individual homotetramers. The potential physiological function of heterotetrameric PAL forms is discussed.
- domain swapping
- gene family
- L-phenylalanine ammonia-lyase (PAL)
- phenylpropanoid metabolism
- tobacco (Nicotiana tabacum)
The phenylpropanoid pathway leads to the biosynthesis of a wide range of plant natural products including flavonoids, hydroxycinnamic acids, coumarins, stilbenes, lignin and condensed tannins, which collectively have diverse biological functions as phytoalexins, signal molecules, structural components, flower pigments or UV protectants . PAL (L-phenylalanine ammonia-lyase; EC 22.214.171.124) plays a crucial role at the interface between plant primary and secondary metabolism by catalysing the deamination of L-phenylalanine to form trans-cinnamic acid.
The properties, regulation, expression and cellular distributions of PAL have been extensively studied , and the crystal structures of PAL from parsley  and the yeast Rhodosporidium toruloides  have been determined. The number of PAL genes ranges from one to five in most plant species studied [5–7]. The various proteins encoded are assumed to be involved in different secondary metabolic pathways, for example for formation of structural components or plant defence, although in many cases this remains to be experimentally determined. In the French bean (Phaseolus vulgaris), PAL1 is expressed in roots, leaves and shoots, PAL2 is expressed in roots, shoots and petals, and PAL3 is expressed only in roots. All three bean PAL genes are activated by wounding, whereas only PAL1 and PAL3 are induced by fungal infection . In quaking aspen, one PAL form may be associated with formation of condensed tannins, the other with lignin production .
Although the phenylpropanoid pathway in tobacco (Nicotiana tabacum) has been the subject of several studies, and PAL has been partially purified from tobacco leaves and suspension cultures [10,11], information about tobacco PAL genes/enzymes is limited as well as misleading. Fukasawa-Akada et al.  identified four PAL genes in N. tabacum by genomic Southern blot hybridization. They could be divided into two subfamilies, with a member of each family originating from each of the two progenitor species Nicotiana tomentosiformis and Nicotiana sylvestris (Table 1). The existence of another tobacco PAL gene (named PAL2) with high sequence similarity to PAL1 [TC (Tentative Consensus) 7464 in The Institute for Genomic Research Gene Index (http://compbio.dfci.harvard.edu/tgi/)] and belonging to the same subfamily, was predicted . However, different designations for the PAL family in tobacco had previously been established, and the PAL gene identified by Nagai et al.  had also been named PAL2 (TC3936) (designated PAL3 by Fukasawa-Akada et al. ). A PAL3 was characterized by Pellegrini et al.  (TC4054), but designated PAL4 by Fukasawa-Akada et al. . To date, none of the corresponding PAL enzymes has been shown to be catalytically active in vitro, and the fourth tobacco PAL gene has yet to be described.
PAL is highly regulated at both the transcriptional and post-transcriptional levels, and metabolic channelling involving enzyme co-localization is proposed to play a major role in flux control through this enzyme . In this respect, tobacco PAL1 was found to co-localize with the membrane-bound C4H (cinnamate 4-hydroxylase; EC 126.96.36.199) cytochrome P450, whereas PAL2 was localized in the cytosol [16,17]. Characterization of all four PAL enzymes in tobacco will help determine the structural requirements for metabolic channelling.
In view of the discrepancies in the nomenclature used in different studies, and the incomplete description of the tobacco PAL gene family, a re-evaluation of the molecular biology and biochemistry of tobacco PALs is warranted. In the present paper, we report the molecular characterization of the full complement of PAL enzymes from tobacco, including PAL4. All four PAL enzymes were biochemically characterized and their expression profiles studied. Surprisingly, PAL2 sequences previously reported in the literature encode an inactive PAL protein as the result of a single nucleotide polymorphism which changes an amino acid at the base of the active site pocket. The origins of this and other confusing features of the tobacco PAL sequences represented in current gene databases are discussed. We also provide direct evidence that PAL is able to form heterotetramers when co-expressed in Escherichia coli, and discuss the physiological implications of this finding.
Tobacco plants were of cultivar N. tabacum cv Xanthi nc for all experiments reported in this paper, except the screening for inactive PAL2 alleles. This was performed with leaf material from cultivars Xanthi nc, Kt19, KDH-960, TN 90, TND 94, MD 40, MD 402, MD A30, KY 908, Oxford 207, Metacomet, Poquonock and SCANTIC. The last 11 cultivars were obtained from the USDA National Plant Germplasm System (http://www.ars-grin.gov/npgs/). Xanthi and Kt19 were from our own seed collection. Plants were grown in soil or Turface MVP™ (Profile Products) under greenhouse conditions (16 h day length). Plant tissues were collected, frozen in liquid nitrogen and pulverized in a tissue grinder (6770-Freezer/Mill, SPEX CertiPrep). N. tabacum NT-1 cell suspension cultures were maintained by regular subculture into fresh culture medium (Murashige and Skoog Basal Medium) and incubated at 24 °C with shaking at 130 rev./min. At 3 days after subculture, cultures (40 ml batches) were treated with YE (yeast elicitor; 50 μg/ml glucose equivalents) or 25 μM MJ (methyl jasmonate) in ethanol [16,18]. Control cells received the same amounts of distilled water or ethanol.
Unless specified, chemicals were obtained from Sigma–Aldrich.
Isolation of PAL genes
Total RNA was extracted from frozen tissue using either TRI Reagent (Molecular Research Center) according to the manufacturer's protocol, or the RNeasy Plant Mini Kit (Qiagen). DNA was extracted with a DNeasy Plant Kit (Qiagen).
RNA and DNA concentrations were quantified using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies), and RNA quality was verified with an Agilent 2100 Bioanalyzer (Agilent Technologies). First strand cDNA synthesis was performed with Superscript III reverse transcriptase (Invitrogen).
PCR primers were designed using either the Lasergene sequence analysis software DNASTAR or Primer3 software (http://frodo.wi.mit.edu/primer3/). Specific primers for the 5′- and 3′-UTRs (untranslated regions) were employed to obtain near-full-length cDNA clones for the four PAL genes and, in the case of PAL4, also a near-full-length genomic clone. For a list of primers see Supplementary Table S1 (at http://www.BiochemJ.org/bj/424/bj4240233add.htm).
PAL4 was amplified from genomic DNA with TaKaRa LA Taq DNA polymerase (TaKaRa Bio), and from cDNA with KOD Hot Start DNA Polymerase (Novagen), according to the manufacturers' protocols. PCR products were purified, cloned into the pGEM-T Easy vector system (Promega) and used for transformation of E. coli (MAX Efficiency DH5α competent cells, Invitrogen). The clones were sequenced, and, in a second round of PCR, NdeI and XhoI restriction sites were introduced at the ends of the coding regions using Cloned Pfu DNA polymerase (Stratagene).
Semi-quantitative RT–PCR (reverse transcription–PCR) was performed with specific primers for the four PAL genes (Supplementary Table S1) using GoTaq DNA polymerase (Promega). PCR products were visualized with SybrSafe DNA Gel Stain (Invitrogen). Analysis of agarose gels was performed with ImageQuantTL. The 1 kb band of the DNA ladder (Promega) was used for calibration (=100).
Plasmid construction and generation of PAL variants
For protein expression, the PAL sequences were cloned into the expression vector pET15b with an N-terminal His-tag and a thrombin cleavage site or, with the stop codon removed, into pET29a(+) with a C-terminal His-tag (Novagen).
Hybrid PALs were generated using the gene splicing overlap extension approach as described previously . Gene fragments were recombined by PCR using Cloned Pfu DNA polymerase (Stratagene). Single-exchange site-directed mutagenesis was performed with the QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's protocol.
The N-terminal HA-epitope (amino acid residues 98–106 of human influenza virus haemagglutinin: YPYDVPDYA) was introduced into the PAL2 and PAL4 sequences by PCR with primers coding for the HA nucleotide sequence (see Supplementary Table S1).
For testing of the formation of PAL heterotetramers, two different PAL genes were co-expressed in the pETDuet-1 vector system (Novagen). PAL1 with SacI/NotI restricition sites was cloned into MCS1 for expression with an N-terminal His-tag. HA-tagged PAL2 or PAL4 with NdeI/XhoI restriction sites were cloned into MCS2. The control vectors for these experiments contained only one PAL sequence.
Expression and purification of recombinant PAL
Transformed E. coli Rosetta 2 (DE3) pLysS cells (Novagen) were grown in LB (Luria–Bertani) broth containing 0.1 mg/ml carbenicillin (pET15b, pETDuet-1) or 0.05 mg/ml kanamycin (pET29a) and 34 μg/ml chloramphenicol at 37 °C. Expression of recombinant protein was induced by addition of 0.25 mM IPTG (isopropyl β-D-thiogalactoside). Cells were harvested after 6 h of incubation at 28 °C. Recombinant PAL proteins were purified using the MagneHis Protein Purification System (Promega). Purified protein was stored in 50 mM Tris/HCl, pH 8.0, 30% glycerol and 1 mM TCEP-HCl [tris-(2-carboxyethyl)-phosphine hydrochloride; Hampton Research] at −80 °C.
Analyses were carried out with tagged PAL preparations unless otherwise stated. The N-terminal His-tag was removed where indicated with thrombin protease according to the manufacturer's recommendations (Novagen). Protein concentrations were determined with the Bio-Rad protein assay (Bio-Rad Laboratories) using BSA as standard.
In experiments investigating PAL heterotetramer formation, PAL proteins were expressed in E. coli and the cells were then lysed in BugBuster Master Mix (Novagen) in the presence of protease inhibitor cocktail (EDTA-free, Roche Applied Science, Mannheim, Germany) and 1 mM TCEP-HCl. The supernatant was incubated for 30 min at 4 °C in 20 mM Tris/HCl, pH 7.5. In control experiments, PAL proteins with His-tags or HA-tags were expressed separately, and then incubated together at comparable concentrations. Purification was performed with the MagneHis Protein Purification System.
Size-exclusion chromatography was performed with Superdex 200 (10/30) on an Äkta purifier (GE Healthcare) in 50 mM Tris/HCl, pH 8.5, and 150 mM NaCl, at a flow rate of 0.5 ml/min. Gel Filtration Standard (Bio-Rad Laboratories) was used for calculating the molecular mass of the PAL tetramer. Homogeneity of the purified protein was verified by analysis on a denaturing NuPAGE 4–12% Bis-Tris gel (Invitrogen). Gels were stained with SimplyBlue Safe Stain (Invitrogen). IEF (isoelectric focusing) was performed with vertical Novex IEF gels pH 3–10 (Invitrogen) using non-denaturating Novex IEF sample buffer (Invitrogen). IEF Marker (SERVA Electrophoresis, Heidelberg, Germany) was used for determining pI values. Gels were stained with Coomassie Blue.
PAL proteins (homo- and hetero-tetramers) were identified by protein gel blot analysis. Proteins were transferred on to nitrocellulose using an XCell II Blot Module (Invitrogen). His-tagged PAL was identified with His-tag monoclonal antibody from mouse (Novagen), and PAL with HA-epitope was identified with anti-HA antibody produced in rabbit. The secondary antibody used was either goat anti-mouse or goat anti-rabbit IgG–alkaline phosphatase conjugate. Colour reaction was performed with Western Blue stabilized substrate for alkaline phosphatase (Promega).
Assay of PAL activity
PAL activity was determined spectrophotometrically as described by Edwards and Kessmann . Formation of the product trans-cinnamic acid was monitored at 290 nm with a UV-2401 PC spectrophotometer (Shimadzu, Kyoto, Japan). The reaction was routinely carried out at 30 °C with 0.1 μg/ml protein in 50 mM Tris/HCl, pH 8.5, containing 12 mM L-phenylalanine; the control contained the same amount of D-phenylalanine. Kinetic data analysis was performed using the computer program HYPER32.exe, Version 1.0.0.
PAL4 is a new PAL gene closely related to PAL1
Specific primers for the 5′- and 3′-UTRs of PAL1 , PAL2  and PAL3  (Supplementary Table S1) were initially used to clone PAL1–PAL3 from cDNA generated using RNA from yeast elicited tobacco NT-1 cell suspension cultures.
The sequences of the PCR-amplified products of PAL1 were mostly identical with the published sequence, but some of the clones contained a PAL1 sequence with different 5′-UTR and N-terminal sequences. The considerable differences in the 5′-UTR suggested that the two PAL1 sequences we had amplified might actually be from different genes rather than representing different alleles of the same gene.
In view of the equivocal results with amplification of PAL1, we set about acquiring the complete sequence of the published PAL1 gene and the new PAL gene (which we named PAL4). New specific primers (Supplementary Table S1) were designed for the 5′-UTRs of PAL1 and PAL4 and the 3′-UTR of PAL1, and both cDNA and genomic DNA were used as templates. These primers amplified two different PAL sequences from genomic DNA extracted from tobacco leaves. Of six clones sequenced, two were identical with the published PAL1 sequence , and four represented the new PAL4 sequence.
These results allowed the design of a specific 3′-UTR primer for PAL4 (Supplementary Table S1), which was used to amplify PAL4 from cDNA derived from elicited tobacco cell culture, confirming our assumption that PAL4 is a separate gene and is expressed in tobacco. PAL4 sequences derived from cDNA and genomic DNA were identical, except for the intron contained in the genomic sequence. Both nucleotide sequences were deposited in the GenBank® database (EU883669 and EU883670).
For comparison, PAL1 was amplified in parallel from cDNA with specific 5′- and 3′-UTR primers (Supplementary Table S1). In addition to amplicons identical with the published PAL1 sequence , this also resulted in amplification of chimaeric PALs, with the 5′-UTR and N-terminus matching the newly identified PAL4 and the C-terminus matching PAL1, although the sequence switch occurred in different places in different clones. Furthermore, comparison of the PAL4 sequence with the TC sequence of PAL1 derived from EST (expressed sequence tag) sequencing [TC7464, DFCI NtGI (N. tabacum Gene Index)] showed that the assembled TC is in fact a mixed PAL1/PAL4 sequence. Alignment of PAL1 and PAL4 with seven of the ESTs annotated as PAL1 and used for establishing the PAL1 TC-sequence revealed that several of the ESTs are, in fact, PAL4 sequences (EB443260, EB447219 and DW004402). Thus the C-terminus and 3′-UTR of TC7464 is a random mixture of PAL1 and PAL4 (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/424/bj4240233add.htm).
PAL4 has high nucleotide sequence identity with PAL1 in the coding region (97.3%, giving 98.3% identity at the protein level) (Supplementary Table S2 at http://www.BiochemJ.org/bj/424/bj4240233add.htm). Similarity of the nucleotide sequence of PAL4 to those of PAL2 and PAL3, which both belong to subfamily II , is slightly lower, at 83.3% and 83.5% respectively. The above results suggest that the very close sequence identity between the open reading frames of PAL1 and PAL4 can lead to the generation of chimaeric PAL1:PAL4 amplicons when using PCR, as has been described previously for target sequences of other multigene family members [21,22], and to incorrect TC assembly from ESTs.
With a calculated molecular mass of 78 kDa (DNASTAR), PAL4 encodes the largest of the four tobacco PAL proteins. The gPAL4 genomic clone consists of 4656 nucleotides with an open reading frame encoding a polypeptide of 717 amino acids. The intron separates exon I (406 bp) and exon II (1748 bp) at the same position as in gPAL1; this position for the intron appears to be characteristic of all PAL genes . The intron in PAL4 is 2214 nucleotides compared with 1932 nucleotides in PAL1, and the two introns share 52.9% sequence identity. As predicted by Fukasawa-Akada et al. , PAL4 (designated PAL2 in their work) has an EcoRI restriction site (which is absent from PAL1) in its intron close to the 5′ end of exon 2.
The published tobacco PAL2 sequence encodes an inactive enzyme
A PAL2 clone matching the exact sequence described previously [13,23] was obtained from the authors and expressed in E. coli. Surprisingly, the purified protein showed no catalytic activity, although it was soluble and formed a tetramer as verified by gel filtration (Supplementary Figure S2 at http://www.BiochemJ.org/bj/424/bj4240233add.htm). The cause of the inactivity was investigated by domain-swapping experiments between PAL1 and PAL2.
The PAL sequence was divided into four segments approximately conforming to the previously described domains ; domain 1 (the N-terminus; residues 1–78, numbering according to PAL4), domain 2 (residues 79–191), domain 3 [the core domain including the MIO (methylidene imidazolone) group; residues 192–534)] and domain 4 (the shielding domain; residues 535–717) (Figure 1A). Single domains in the nucleotide sequence were exchanged by gene splicing overlap extension, and the specific activity of the purified recombinant proteins was determined. The hybrid PALs were designated according to the origin of the four domains (PAL1=P1111, PAL2=P2222). All wild-type and hybrid PALs gave similar SDS/PAGE protein profiles when expressed in E. coli, with the enzyme migrating as a major band of approx. 70 kDa, somewhat smaller than the calculated molecular mass of approx. 80 kDa (Supplementary Figure S3 at http://www.BiochemJ.org/bj/424/bj4240233add.htm). All hybrid PALs containing the core domain 3 of PAL2 were inactive, but all were active when they contained the core domain of PAL1 with a similar specific activity (Figure 1B). P1121 and P2212 differ in ten residues, one of which (an aspartate residue at position 495 which is a glycine in all other PALs) seemed most likely to cause a difference in enzyme activity. This residue was therefore introduced into the sequences of PAL1 and P2212 by single-exchange site-directed mutagenesis. The exchange of G495D in PAL1 and P2212 resulted in an inactive enzyme (Figure 2). Thus the aspartate residue at position 495, which arises due to a single nucleotide polymorphism at nucleotide position 1484 (1469 in the PAL2 sequence), is responsible for the lack of activity of the PAL2 protein as defined by Nagai et al. .
PCR of genomic DNA (from tobacco leaves from 13 different varieties, see the Experimental section) and cDNA (derived from RNA from tobacco cell suspension cultures) with specific primers for the 5′- and 3′-UTRs of PAL2 (P2UTR5, P2/3UTR3) (Supplementary Table S1) exclusively yielded (active) PAL2 enzyme with a glycine residue in position 495; both this and the G459D mutant were biochemically characterized as described below.
The sequence of the PCR-amplified product of tobacco PAL3 differed from the sequence published by Pellegrini et al.  in only one amino acid, an alanine residue in place of an arginine at position 578 (PAL4 numbering). The TC4054 sequence from the NtGI Gene Index contains an additional nucleotide in this position, causing a frame shift. Since the three other tobacco PALs all contain an alanine in this position, our PAL3 sequence is probably the correct one.
A new classification of the tobacco PAL family
Supplementary Figure S4 (at http://www.BiochemJ.org/bj/424/bj4240233add.htm) shows an alignment of the four PAL protein sequences characterized in this work, and the sequence similarity between their coding regions is summarized in Supplementary Table S2. On the basis of these studies, we propose a new classification of the four tobacco PAL genes (Table 1).
Expression and purification of tobacco PAL proteins
The four PAL proteins were expressed in E. coli with either C-terminal or N-terminal His-tags. Protein expression levels and yields were comparable for all four PAL proteins independent of the tag used for purification, with approx. 0.4 mg of purified protein obtained per 10 ml of culture. However, PAL3 and PAL4 with C-terminal His-tags had a significantly lower specific activity (approx. 6 nkat·mg−1) than the corresponding PAL1 and PAL2 C-terminal His-tag fusions (17 and 15 nkat·mg−1 respectively) (Figure 3). Preparations of all PAL proteins with C-terminal His-tags contained more degradation products that were retained during the purification process than did preparations of N-His-tagged proteins (Supplementary Figure S3), suggesting that the presence of a tag at the C-terminus can destabilize the protein structure leading to degradation and the loss of activity seen in Figure 3. PAL4 without a His-tag exhibited the highest specific activity (29 nkat·mg−1).
Size-exclusion chromatography of PAL1 and PAL2-D495 (both with a C-terminal His-tag) indicated molecular masses of 262 and 270 kDa respectively, consistent with the enzymes existing as tetramers (Supplementary Figures S2 and S5A at http://www.BiochemJ.org/bj/424/bj4240233add.htm). Increasing the protein loading resulted in the presence of an additional peak of approx. 1552 kDa, representing an aggregated hexameric version of the PAL1 tetramer (Supplementary Figure S5B). However, PAL enzymatic activity was primarily associated with the peak of protein representing the PAL tetramer (Supplementary Figure S5C).
All purified recombinant proteins were stable without loss of activity for several months in storage buffer at −80 °C.
Effects of reducing agents on PAL activity
Sulfhydryl-group-specific reagents can strongly affect PAL activity. Thus tobacco PAL in a relatively crude preparation was competitively inhibited by 2-ME (2-mercaptoethanol) , and parsley PAL was irreversibly inhibited by DTT (dithiothreitol) after incubation for several hours . On the other hand, increased enzymatic activity of tobacco PAL was observed in the presence of some reducing agents . This apparent contradiction prompted us to re-investigate the effects of reducing agents on individual recombinant tobacco PAL proteins.
All of the recombinant tobacco PAL proteins were competitively inhibited by 2-ME, with a Ki of 3 mM determined for PAL1. However, purification of recombinant PAL1 from E. coli in the absence of reducing agents resulted in lower specific activity (16 nkat·mg−1), which could be increased to 22 nkat·mg−1 with inclusion of 1 mM TCEP-HCl, a reducing agent which lacks sulfhydryl groups. It is likely that reducing agents are required for stabilizing the enzyme in an active conformation , or alternatively preventing the formation of disulfides . Because PALs from several sources appear to be sensitive to sulfhydryl-group-specific reagents , TCEP-HCl was added to the storage solution of all PAL proteins at a final concentration of 1 mM. It had no effect on the catalytic properties of the enzyme.
Kinetic properties of tobacco PALs
To determine whether the four different tobacco PAL genes might have different functions in plants, we first determined the kinetic properties of the encoded enzymes. Removal of the N-terminal His-tag by cleavage with thrombin did not result in significant changes in the specific activities compared with those of the N-terminal His-tagged protein (Figure 3). However, as the presence of a C-terminal His-tag appeared to affect the stability of the enzyme, particularly for PAL3 and PAL4 (see above), the kcat/Km ratios for these proteins (9090 and 10290 s−1·M−1 respectively) were reduced compared with those of PAL1 and PAL2 His-tag fusions (18650 and 21324 s−1·M−1 respectively). Thus for the kinetic studies reported below, all enzymes were expressed as N-terminal His-tag fusions, with the tag subsequently removed by thrombin cleavage.
All four PAL proteins exhibited Michaelis–Menten kinetics with comparable apparent Km values, ranging from 36.4 μM for PAL3 to 59.8 μM for PAL1 (Table 2). The Km reported for PAL from tobacco cell suspension cultures (presumably a mixture of the PAL proteins) was 30 μM . In contrast, a Km value of 220 μM was reported for partially purified tobacco PAL in an assay containing 4 mM 2-ME, a competitive inhibitor  (see above). In the present study, comparable values of between 150 μM and 200 μM were measured with recombinant PAL proteins in the presence of 8.6 mM 2-ME.
Near identical turnover numbers of 1.09 s−1 and 1.14 s−1 were determined for PAL1 and PAL2 respectively, whereas PAL3 had the lowest kcat value of 0.78 s−1 and PAL4 the highest (1.53 s−1), assuming four active sites per holoenzyme (Table 2). PAL2 and PAL4 possess similar kcat/Km ratios, which are slightly higher than those of PAL1 and PAL3 (Table 2). These values are within the same range as the kcat/Km ratios of between 25500 and 51200 s−1·M−1 determined for recombinant PAL from Arabidopsis thaliana . Overall, our studies indicate that the kinetic properties of the four PAL forms are relatively similar.
Formation of PAL heterotetramers
PAL is generally assumed to form homotetramers, but the appearance of multiple forms of native PAL with similar molecular masses but different pI values has been observed on several occasions [26,27] and the possibility that these may represent heterotetramers has been discussed . We therefore developed an approach to address the ability of tobacco PAL forms to assemble into heterotetramers in vitro following co-expression in E. coli using the pETDuet-1 vector system.
PAL1 was cloned with an N-terminal His-tag (to be used for purification), whereas a second PAL protein, either PAL2 or PAL4, was co-expressed (in the same E. coli cells) with PAL1 as a protein fusion with an N-terminal HA-epitope  to allow identification by immunoblotting. Co-expressed PAL forms were isolated from the soluble protein fraction of the bacterial cell lysate with MagneHis Ni-Particles that bind the polyhistidine-tag but not the HA-tag (as checked with individual His–PAL1 and HA–PAL2 prior to analysis of co-expressed proteins). SDS/PAGE analysis of the resulting proteins revealed major PAL subunits of approx. 70 kDa, but this technique is not able to resolve the small differences in molecular mass between the different PAL subunit forms (Supplementary Figure S6 at http://www.BiochemJ.org/bj/424/bj4240233add.htm). However, gel blot analysis of the purified protein with specific antibodies against His-tag and HA-epitope detected the presence of equal signals for both His-tagged PAL1 and HA-tagged PAL2 or PAL4 (Figure 4). In the critical control experiment, the two different PAL proteins were expressed separately in E. coli, and the extracts were then mixed, pre-incubated, and purified under the same conditions as above. In this case, gel blot analysis identified only His-tagged PAL1 as being bound to the MagneHis particles; HA-tagged PAL2 was eluted from the beads because it does not contain a His-tag, and was therefore not detected (Figure 4). Thus only co-expression in vivo leads to a PAL complex in which the HA-tagged PAL2 can be retained on the MagneHis beads through association with His-tagged PAL1. These data indicate the formation of heteromers (presumably heterotetramers) between PAL1 and PAL2 or PAL4.
IEF and subsequent gel blot analysis of PAL purified from E. coli expressing both PAL1 and PAL2 also indicated the formation of heteromers, as PAL1–His mono(tetra)mers were not detected (Figure 5). The pI of PAL1–His was determined to be 6.77 (calculated 6.67), and PAL2–HA determined to be 6.17 (calculated 6.23). The main band for the protein obtained after co-expression of PAL1 and PAL2 in E. coli had an estimated pI of 6.38, in good agreement with the calculated pI of 6.44 (DNASTAR) for a PAL1–PAL2 heterotetramer with His- and HA-tags.
Finally, size-exclusion chromatography of the PAL1–PAL2 heteromer revealed exactly the same molecular mass as seen previously when individual PAL1 and PAL2 preparations were examined, namely a major peak of active PAL protein corresponding to the tetramer, with a catalytically inactive higher-molecular-mass aggregate also being present at high protein loadings (compare Supplementary Figure S7 with Supplementary Figures S5 and S2 at http://www.BiochemJ.org/bj/424/bj4240233add.htm).
We then determined the kinetic properties of the purified PAL1–PAL2 heterotetramer, and compared them with those of a PAL1–PAL2 heterotetramer formed with inactive PAL2 (PAL1–PAL2D495) (Table 3). PAL1–PAL2 exhibited kinetic properties between those of PAL1 and PAL2 homotetramers with N-terminal His-tags. In comparison, PAL1–PAL2D495 heterotetramers had a Vmax of almost exactly half that of PAL1 with an N-terminal His-tag, although the Km value was similar to that of PAL2–His. This provides additional evidence for the formation of heterotetrameric forms of PAL, in this case containing both active and inactive subunits.
Negative co-operativity, as has been described for PALs from various plants [26,29,30], was not observed on kinetic analysis of either PAL heterotetramers or mixtures of individual PAL proteins (results not shown).
Tissue-specific expression of PAL genes
For formation of PAL heterotetramers to have physiological significance in planta, the different PAL genes must be expressed in the same tissues at the same times. Transcript levels of the different PAL proteins in different plant tissues were therefore investigated by semi-quantitative RT–PCR with PAL gene-specific primers for each of the different PAL transcripts. Transcripts of all PALs were found in every tissue investigated (Figure 6) at approximately comparable levels, with some exceptions. PAL4 was readily detected in stamen, xylem and roots, but was barely detectable in stem pith tissue, suggesting a specific role in lignifying cells. PAL2 transcript levels in all tissues were lower than those of PAL1 and PAL3, except in xylem and leaves. Transcript levels of most PALs were low in mature leaves and pith.
Elicitation of PAL transcripts
Overall PAL expression is induced following exposure of tobacco cell cultures to YE or MJ [23,31]. To determine how this response operates at the level of induction of individual PAL forms, tobacco NT-1 cell suspension cultures were induced with YE or MJ and harvested at different time points. Semi-quantitative RT–PCR was performed with specific primers for the four PAL genes.
PAL1 and PAL4 were expressed constitutively at higher levels than PAL2 and PAL3 in cell suspension cultures; however, statistical analysis indicated that their apparent induction by YE and MJ was in most cases not significant, due to the high values in unelicited controls at 4 h (Figure 7). This suggests that PAL1 and PAL 4 transcription is very sensitive to handling of the cultures, i.e. during addition of water or dilute ethanol to the controls. PAL3 was strongly induced within 1 h of exposure to YE, but PAL2 induction was weaker and delayed by approx. 2 h (results not shown). A previous study had suggested that only PAL2 (designated PAL A) was elicitor-inducible . Expression of PAL transcripts peaked 4 h after exposure to YE, whereas MJ induction of PAL transcripts was slower (results not shown), in agreement with earlier results .
A complete description of the tobacco PAL gene family
Fukasawa-Akada et al.  proposed the presence of four single-copy PAL genes in the tobacco genome belonging to a small multigene family consisting of two subfamilies. N. tabacum is amphidiploid and assumed to have arisen by interspecies hybridization between N. tomentosiformis and N. sylvestris [12,32].
The previously elusive PAL4 sequence had been wrongly annotated as PAL1 due to the formation of chimaeric PAL sequences. However, the identification of the PAL4 gene in the present study now makes it possible to complete the characterization of the tobacco PAL gene family. Here we show that all four PAL genes encode functional isoenzymes and that all are expressed in multiple tobacco tissues. PAL4 corresponds to a gene previously designated as PAL2 by Fukasawa-Akada et al. . As predicted by this group, PAL4 has a high similarity to PAL1 and both belong to subfamily I; however, they derive from different progenitor species.
Biochemical properties of tobacco PAL proteins
PAL1 and PAL4 have comparable enzymatic activities, but vary in the catalytic impacts of their C-termini, as seen by the significant difference in specific activity between PAL1 and PAL4 with C-terminal His-tags. Although the two proteins share 98.3% similarity at the protein level, PAL4 without a tag has a higher Vmax and turnover rate than PAL1. The same effect of the C-terminus can be observed among the members of subfamily II. PAL3 with a C-terminal His-tag, similar to the PAL4 derived from N. sylvestris, has a 50% lower specific activity than PAL2. Pilbák et al.  proposed that the C-terminal multi-helix extension in eukaryotic PALs plays an important role in regulation of the enzyme activity by destabilizing the active conformation of the Tyr110-loop. The highly flexible N-terminus, which shows the highest divergence among the tobacco PAL proteins as well as among PAL sequences from other sources , seems to have less impact than the C-terminus on enzymatic properties, as seen from the similar kinetic values for PAL with or without an N-terminal His-tag; the N-terminus may be involved instead in interactions with other cellular components .
PAL4 exhibited the highest specific activity of the tobacco PAL proteins (29 nkat·mg−1). This value compares with 40 nkat·mg−1 for recombinant PAL from parsley , and is slightly higher than that for PALs purified from plant tissues: e.g. rice, 4.6 nkat·mg−1 ; bean, 5.7 nkat·mg−1 ; and tomato, 0.5 nkat·mg−1 . The kcat data indicate that tobacco PALs have slightly lower turnover rates than PALs from other sources such as A. thaliana (1.8–3.2 s−1) ; Nostoc punctiforme (1.96 s−1); Anabaena variabilis (4.3 s−1) ; maize (10.6 s−1) ; and parsley (13.5 s−1) .
Overall, our data are in agreement with published Km values for recombinant PALs from other plant sources such as parsley (15–24.5 μM) , A. thaliana (68–71 μM)  and French bean (52 μM) . Purified PAL from tomato exhibited slightly higher Km values with higher deviation between different gene products (121 μM for PAL2 and 840 μM for PAL1) , recombinant maize PAL had a Km of 270 μM  and rice PAL had a Km of 500 μM .
As verified in the present study, 2-ME is an inhibitor of PAL. However, reducing agents stabilize the enzyme activity. Because DTT has been found to irreversibly inhibit PAL , it is advisable that PAL is purified and stored in the presence of reducing agents without sulfhydryl groups, such as TCEP.
The number of active sites per PAL tetramer has been the subject of debate . Results of early labelling experiments suggested two active sites per homotetramer , and Bolwell et al.  assumed two active sites per tetrameric enzyme for their calculations with bean PAL. However, the crystal structures of PAL from parsley and R. toruloides indicate four functional active sites [3,4]. Regulatory processes have been described for PAL that may render some active sites inaccessible; these include phosphorylation  or conformational changes of two highly mobile loops [3,33]. It is possible that in PAL3 and PAL4 with C-terminal His-tags only two active sites are accessible; this could explain why their kcat/Km ratios are only approx. 50% those of PAL1 and PAL2 with C-terminal His-tags.
Negative co-operativity of PAL preparations has been observed when analysing the kinetics of PAL during early stages of purification [6,26], and the formation of heterotetramers has been suggested as an explanation for this phenomenon. However, we show in the present study that tobacco PAL heterotetramers do not exhibit negative co-operativity, and neither do mixtures of different recombinant tobacco PAL proteins. Altered kinetic properties could conceivably arise as a result of differential post-translational modifications or interactions with specific proteins only present in plant extracts.
Origin and lack of activity of PAL2 G459D
The present study shows that the previously published PAL2 sequence encodes an inactive enzyme. Although the physiological significance of this observation cannot be explained at present, a regulatory function for an apparently inactive PAL2 allele is possible if the encoded protein can associate into heterotetrameric forms of the enzyme. However, screening of multiple tobacco cultivars failed to recover the inactive PAL2 allele, suggesting that it is either a rare event in tobacco germplasm, or else an artefact. It is also interesting to note that the sequence of a bean PAL1 has been published with an aspartate residue in the corresponding position to Asp459 in inactive tobacco PAL2, whereas PAL2 and PAL3 both contain a glycine . However, the catalytic activity of this PAL1 enzyme was not determined.
Alignment of tobacco PAL sequences with the PAL sequences from parsley and the yeast R. toruloides, whose crystal structures have been elucidated [3,4], revealed that Gly495 (Gly493 in parsley) is the first residue of the α-helix α17, one of the three central α-helices forming an electropositive platform for the catalytic MIO group, which is anchored to the three helices through non-covalent bonding . In R. toruloides PAL, this residue corresponds to Ala506, whose amide group shares a hydrogen bond with the hydroxy group of Ser210 (Thr202 in all tobacco PALs), and the hydroxy group of Ser275 (most likely Ser267 of helix α11 in tobacco PAL). Exchange of Gly495 with the negatively charged aspartic acid apparently disrupts the coherence of the MIO environment in a way that renders the enzyme inactive.
Cell- and organ-specific expression of tobacco PAL genes
In plants possessing PAL multigene families, attempts have been made to assign different metabolic functions to the individual PAL proteins. In quaking aspen (Populus tremuloides), differential expression of two PAL genes was observed, one associated with formation of condensed tannins, the other with lignin production . In French bean, PAL3 is expressed only in roots , and this may reflect a function in the formation of root-specific secondary metabolites.
Our initial attempts to determine whether there was differential expression of PAL transcripts in tobacco organs using real-time quantitative RT–PCR were abandoned due to difficulties in designing primers that met the requirements for real-time quantitative RT–PCR while at the same time being specific for each of the four PAL genes which share a relatively low G/C content and high sequence similarity. The results obtained instead with semiquantitative PCR are in agreement with earlier results of Fukasawa-Akada et al.  who found high PAL transcript levels in roots and flowers and lower levels in stem and leaves. All four PAL genes were expressed in every tissue investigated, as also observed in parsley . PAL1 and PAL4 were expressed at similar levels in most tissues, with the greatest differences detected in stems of mature tobacco plants, where PAL4 transcript levels were high in the xylem layer but very low in cortex and pith tissue, a very different pattern from that of PAL1 transcript levels. This suggests a specific role of PAL4 in lignification.
PALs from different sources have been shown to exhibit different expression patterns in response to environmental cues . Of the four PAL genes in Arabidopsis, only PAL1 and PAL2 were induced after a decrease in nitrogen supply and temperature ; these genes, together with PAL4, play a role in lignin synthesis .
In tobacco NT-1 cell suspension cultures, PAL1 and PAL4 were expressed constitutively, and their transcript levels increased within 1 h of elicitation with YE or MJ, whereas PAL2 and PAL3 were induced at later stages post-elicitation. In contrast, Taguchi et al.  found induction of PAL2 (PAL A) with both fungal elicitor and MJ in tobacco cell cultures; PAL1 (PAL B) was not induced. These findings suggest that tobacco PAL2 and PAL3 are linked primarily to inducible stress response reactions.
PAL can exist as a heterotetrameric enzyme
Using three different and partially independent approaches, namely affinity pull-down followed by immunoblot with epitope-specific antibodies, IEF and activity assays, we provide evidence in the present study for the formation of heterotetramers between PAL1 and PAL2 or PAL4 following co-expression in a bacterial system. The fact that no homotetramers were found in the purified fraction suggests preferential formation of heterotetramers. Heterotetramers between PAL1 and inactive PAL2 (PAL2D495) had a specific activity that was 50% that of PAL1–PAL2 heterotetramers, suggesting the presence of only 50% of the active sites. Although a mixed population of heterotetramers with different PAL1–PAL2D495 ratios cannot be excluded, the Vmax data are consistent with the presence of four active sites in fully active PAL. The autocatalytic formation of the prosthetic MIO group has been described for histidine ammonia-lyase, and the mechanism is assumed to be valid for PAL as well . Since residues from three subunits participate in the formation of each active site, it is perhaps remarkable that heterotetramers can form active enzymes. Formation of functional heterodimers in vitro has been shown for O-methyltransferases from Thalictrum tuberosum . Different dimer combinations of the four proteins in vitro resulted in heterodimeric enzymes with substrate specificities that deviate from those of the native homodimeric forms.
Constitutive co-expression of PAL1 and PAL4 in a single cell type (e.g. cell suspension cultures) raises the question of whether formation of PAL heterotetramers occurs in planta, or whether plants possess mechanisms to prevent heterotetramer formation. Metabolic channelling has been proposed to play a role in regulating the phenylpropanoid pathway and controlling flux into different branch pathways . Evidence exists for differential subcellular localizations of distinct PAL proteins in tobacco, and the association of at least one specific PAL protein with C4H, the membrane-associated cytochrome P450 enzyme that catalyses the second step in phenylpropanoid biosynthesis [16,17]. We speculate that, rather than generating variation in catalytic activity, the formation of PAL heterotetramers may direct forms of PAL to alternative subcellular localizations associated with metabolic channelling. This would constitute an additional regulatory mechanism for the formation of phenylpropanoid end-products [44,45]. The next step will be to evaluate whether PAL heterotetramers can be generated in transgenic tobacco plants expressing differentially tagged PAL proteins.
Angelika I. Reichert performed experiments and helped to write the paper, Xian-Zhi He performed experiments, and Richard A. Dixon conceived the work and helped to write the paper.
This work was supported by the Samuel Roberts Noble Foundation.
We thank Kristy Richerson, David McSweeny and Carla Welch for assistance with plant growth and maintenance, Ivone Torrez-Jerez (Plant Biology Division, The Samuel Roberts Noble Foundation, Ardmore, OK, U.S.A.) for the NT-1 cell suspension culture, Dr Mitsuo Okazaki (Department of Applied Biology, Faculty of Textile Science and Technology, Shinshu University, Ueda, Nagano, Japan) for the original tobacco PAL2 clone, and Dr Lenong Li and Dr Maria Monteros for critical reading of the paper.
Abbreviations: C4H, cinnamate 4-hydroxylase; DTT, dithiothreitol; EST, expressed sequence tag; HA, haemagglutinin; IEF, isoelectric focusing; 2-ME, 2-mercaptoethanol; MIO, methylidine imidazolone; MJ, methyl jasmonate; NtGI, Nicotiana tabacum Gene Index; PAL, L-phenylalanine ammonia-lyase; RT–PCR, reverse transcription–PCR; TC, tentative consensus; TCEP, tris-(carboxyethyl)-phosphine; UTR, untranslated region; YE, yeast elicitor
- © The Authors Journal compilation © 2009 Biochemical Society