The AC-rich motif, Pal-box, is an important cis-acting element for gene expression involved in phenylpropanoid biosynthesis. A cDNA clone (Ntlim1) encoding a Pal-box binding protein was isolated by Southwestern screening. The deduced amino acid sequence is highly similar to the members of the LIM protein family that contain a zinc finger motif. Moreover, Ntlim1 had a specific DNA binding ability and transiently activated the transcription of a β-glucuronidase reporter gene driven by the Pal-box sequence in tobacco protoplasts. The transgenic tobacco plants with antisense Ntlim1 showed low levels of transcripts from some key phenylpropanoid pathway genes such as phenylalanine ammonia-lyase, hydroxycinnamate CoA ligase and cinnamyl alcohol dehydrogenase. Furthermore, a 27% reduction of lignin content was observed in the transgenic tobacco with antisense Ntlim1.
Lignin is a complex phenolic polymer that reinforces the walls of certain cells in the vascular tissues of higher plants. Lignin plays an important role in mechanical support, water transport and pathogen resistance. In the pulp and paper industry, lignin must be removed by harsh chemical treatments, which is a costly process both to the mill and the environment. Genetic modification of the lignin content and composition of woody plants is receiving considerable attention in current forest biological studies and wood formation. Lignin is derived from dehydrogenative polymerization of monolignols, p-coumaryl alcohol, coniferyl alcohol and sinapyl alcohol. Based on tracer experiments and subsequent enzymatic analysis of the intermediate steps, it has been shown that monolignols are synthesized through the phenylpropanoid biosynthetic pathway ( Higuchi 1985). The lignin biosynthetic pathway is relatively well studied, and most genes encoding the enzymes in this pathway have been identified in several plant species ( Whetten et al. 1998 ). The targeted manipulation of specific genes and enzymes has complemented the chemical characterization of lignin mutants in a variety of plant species ( Whetten et al. 1998 ). These studies have exploited a range of different genes encoding enzymes involved in the control of lignification at three levels. The first involves the enzymes of the common phenylpropanoid pathway, phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H) and hydroxycinnamate CoA ligase (4 CL). Second is the methylation step of monolignols, caffeate/5-hydroxyferulate-O-methyltransferase (COMT) and ferulate-5-hydroxylase (F5H). The third level involves the last steps of monolignol biosynthesis, hydroxycinnamoyl-CoA:NADPH oxidoreductase (CCR) and cinnamyl alcohol dehydrogenase (CAD). Several attempts to produce transgenic tobacco with antisense or sense-suppression methods have reported the modification of lignin content and monomeric composition ( Boudet 1998).
To understand the regulation of lignin biosynthesis, we focused on a transcription factor that controls expression of genes encoding enzymes in this pathway. The horseradish peroxidase (HRP) encoded by the prxC2 gene is involved in the lignification process ( Kawaoka et al. 1992 ). We have reported previously on the cloning of a cDNA encoding TFHP1, which specifically bound to the G-box motif in the prxC2 promoter ( Kawaoka et al. 1994 ). Later analysis of the regulatory properties of the prxC2 promoter revealed another positive cis-acting element POP (Pal-box related cis-acting element of prxC2), which contained a Pal-box like motif at the position between −178 and −136 bp from the translation start site (P. Kaothien et al. unpublished data). As shown in Fig. 1, the Pal-box motif, CCA (C/A) (A/T) A (A/C) C (C/T) CC, is a highly conserved sequence involved in the gene expression of phenylpropanoid biosynthesis, such as PAL ( Cramer et al. 1989 ; Lois et al. 1989 ; Ohl et al. 1990 ), 4 CL ( Douglas et al. 1987 ), CHS ( Dron et al. 1988 ) and CAD ( Feuillet et al. 1995 ).
We describe here the isolation of a cDNA encoding a tobacco protein, termed Ntlim1, which specifically bound to the Pal-box sequence. The Ntlim1 protein has two LIM domains which are cysteine-rich polypeptides composed of two special zinc fingers separated by a two-amino acid spacer. The only other report about plant LIM protein is the sunflower PLIM-1 which is expressed specifically in pollen ( Baltz et al. 1992 ). Although the LIM domain is comprised of a double semi-conservative zinc finger motif, there is still not enough proof for its DNA binding activity. Furthermore, the LIM domain may function as an interface for protein–protein interaction ( Crawford et al. 1992 ; Sadler et al. 1992 ). In this paper, we show that the Ntlim1 protein is able to bind to the Pal-box sequence and regulates transcription of the phenylpropanoid biosynthesis genes. The results of gel retardation assays strongly suggest that the LIM motifs of Ntlim1 function as DNA binding domains. Interestingly, suppression of the Ntlim1 caused simultaneous reduction in the transcript levels of some phenylpropanoid pathway genes and low lignin content in transgenic plants.
Isolation of a cDNA encoding Pal-box binding protein
To investigate the existence of the DNA binding protein to the Pal-box like motif in tobacco nucleus, we prepared a double-stranded oligonucleotide probe P that corresponded tandemly to three copies of the Pal-box like motif (–CCACTTGAGTAC-) in the horseradish peroxidase prxC2 promoter region. The binding of nuclear protein to probe P was tested by gel retardation assay using the nuclear extracts from tobacco leaves and stems. As a result, the retarded band appeared in the presence of nuclear extracts from tobacco leaves and stems at the same position and was maintained even in the presence of poly (dI-dC)·poly (dI-dC) as a non-specific competitor. Whereas in the presence of excess amounts of the homologous non-labeled DNA fragment as a specific competitor, the retarded complex was not observed (data not shown), thus confirming the presence of nuclear factors that specifically bind to the Pal-box like motif in tobacco leaves and stems.
To isolate cDNAs which encode the Pal-box binding protein, a λgt11 expression library constructed with mRNA from tobacco stems was screened by using the probe P. One positive clone, no. 101, was obtained from the screening of 1 000 000 recombinant phages. The 1 kb cDNA insert of clone 101 was amplified by PCR and subcloned into the pNoTA/T7 plasmid. The recombinant plasmid p101 was used for further experiments. The cDNA insert of the p101 contained an open reading frame that encodes a protein of 200 amino acid residues ( Fig. 2a). Since this cDNA did not have a poly(A) + sequence and Northern blot analysis showed that the size of the mRNA was about 1.2 kb ( Fig. 2d,e), the inserted cDNA of p101 was probably lacking the 3′ untranslated region. From a GenBank database search, the deduced amino acid sequence of p101 proved to be highly homologous to a plant LIM protein, the PLIM-1 of late pollen genes in the sunflower ( Baltz et al. 1992 ) and Arabidopsis L2 (GenBank accession number X91398). We designated the protein encoded by this cDNA, Ntlim1.
The LIM motif defines one class of the zinc-binding domain and was originally recognized in, and named after, the protein products of the lin-11, isl-1 and mec-3 genes ( Taira et al. 1995 ). The gene products of lin-11 and mec-3 transcriptionally regulate genes involved in cell fate determination and differentiation in Caenorhabditis elegans, and the isl-1 gene encodes a rat insulin I gene enhancer-binding protein. The LIM proteins have diverse functional domains, such as homeodomains or protein kinase domains ( Taira et al. 1995 ). Furthermore, results from the motif search showed that the deduced amino acid sequence of the Ntlim1 protein contains two repeats of conserved amino acid sequences of the LIM domain (LIM1 and LIM2; Fig. 2b). The alignment of amino acid sequences of Ntlim1 and other LIM proteins, e.g. sunflower PLIM-1 ( Baltz et al. 1992 ), chicken CYSR ( Wang et al. 1992 ), rat CRP2 ( Okano et al. 1993 ), human RHM1 ( McGuire et al. 1989 ) and mouse RHM2 ( Boehm et al. 1991 ) are shown in Fig. 2(c). To determine the number of genes in the tobacco genome that are related to the gene for Ntlim1, Southern hybridization was carried out. Three hybridizing bands were obtained with genomic DNA digested with either EcoRV or DraI, indicating that there are at least three copies of Ntlim1 in the tobacco genome (data not shown).
Mode of expression of Ntlim1 mRNA
Total RNA was extracted from tobacco roots, stems and leaves. Ten μg of the total RNA was subjected to electrophoresis and blotted, and the accumulation of Ntlim1 mRNA in each organ was determined by Northern hybridization. The Ntlim1 mRNA was abundant in stems and also detected in leaves and roots ( Fig. 2d). We extracted total RNA from tobacco leaves that had been incubated in phosphate buffer for several hours after wounding to determine whether the transcription of the Ntlim1 gene was induced by wounding. The mRNA transcribed from the Ntlim1 was observed at 0 h (before wounding) and the level remained almost constant for 24 h, although it was slightly increased at 2 h after wounding ( Fig. 2e). The Ntlim1 gene is not induced by wounding.
LIM domain has DNA-binding activity
In order to determine whether the Ntlim1 protein does indeed bind to the Pal-box like sequence, we prepared protein extracts from E. coli carrying the expression vector pGEX-2TK that produces glutathion-S-transferase (GST) fusion protein by isopropyl-β- d-thiogalactopyranoside (IPTG) induction. The fusion protein, GST-Ntlim1 was purified through Glutathione-Sepharose 4B resin. After digestion with thrombin protease, the band of purified Ntlim1 protein (24 kDa) was observed by SDS-PAGE ( Fig. 3a). The DNA binding ability of the purified Ntlim1 was tested by gel retardation assay using radiolabeled probe A, which is the double-stranded oligonucleotide fragment of the POP sequence ( Fig. 3b). The purified GST protein was used as a negative control (lane 5). A retarded band showing DNA binding activity of Ntlim1, which disappeared in the presence of 100 molar excess of non-labeled probe A, was observed (lane 1 and 2). When non-labeled probe A with disrupted Pal-box (two nucleotides substituted) was added as a competitor, the signal of the retarded band was weakened (probe mA, lane 3). The Pal-box binding specificity of Ntlim1 was also confirmed by competition assay using excess amounts of non-labeled Pal-box (probe B, lane 4).
To verify the DNA binding domain of Ntlim1, two truncated constructs, containing either LIM1 domain (amino acids 10–61) or LIM2 domain (amino acids 110–161) of Ntlim1, were made and inserted into a pGEX-4T-1 expression vector. The resulting GST-LIM1 and GST-LIM2 fusion proteins were produced in E. coli and purified through Glutathione-Sepharose 4B resin in the same way as the GST-Ntlim1 fusion protein. Following thrombin treatment, the bands of expected size, i.e. 11.6 kDa for LIM1 and 8 kDa for LIM2, were observed by SDS-PAGE ( Fig. 3a). The purified LIM1 and LIM2 proteins were tested for their DNA binding ability by gel retardation assay using a radiolabeled probe B. In Fig. 3(c), we observed the retarded bands showing the DNA binding ability of both LIM1 and LIM2 proteins. These bands disappeared when competitors, 100 molar excess of non-labeled probe B and probe C (the shorter version of probe B), were added to the binding reactions. These results indicate that the LIM domain has DNA-binding ability.
Ntlim1 recognizes AC-rich motif
To determine the consensus DNA sequence for Ntlim1 binding, we performed polymerase chain reaction (PCR) based binding site selection, which involved cycles of Ntlim1 binding to digested tobacco genomic DNAs, purification of DNA-fusion protein complexes by glutathion-sepharose 4B, and PCR amplification of the selected sequences. These steps were repeated for five cycles. The Ntlim1 protein selected 20 diverse sequences from the digested tobacco genome (50–200 bp length). Maximum matching analysis with the Pal-box motif was carried out using the Genetyx-win program (Software Development, Tokyo, Japan). As shown in Fig. 4, the first six bases (base nos 1–6) were restricted as compared with the latter six bases. The consensus binding sequence was found to be CCAC(A/C)AN(A/C)N(C/T)(A/C) from the fifth-round binding experiment, and this sequence is similar to the Pal-box motif.
Ntlim1 is a transcription factor
There are three notable domains of Ntlim1, the LIM1 domain, the LIM2 domain and the acidic domain (amino acids 164–200). Although the isoelectric point (pI) of the entire Ntlim1 protein is 8.39, the pI of the acidic domain at the C-terminal region is 3.92 determined by computer analysis with the Genetyx-win program (Software Development, Tokyo, Japan). To determine whether Ntlim1 functions as a transcription activator in vivo, we constructed 11 kinds of effector plasmids under the control of the cauliflower mosaic virus (CaMV) 35S core promoter, and three kinds of chimeric GUS genes as reporter plasmids ( Fig. 5a). The P reporter plasmid was constructed using double-stranded synthetic oligonucleotides. A three copy sequence of the Pal-box like motif was ligated to the −90 bp of CaMV 35S promoter. The GP reporter plasmid contained a G-box like motif which has an enhancer effect ( Menkens et al. 1995 ), fused into the P reporter. The reporter C2 is a 307 bp of DNA fragment containing horseradish peroxidase prxC2 promoter that has both the G-box and the Pal-box like sequences fused to the GUS structural gene. Transient expression of GUS in tobacco protoplasts was measured 24 h after the effector, reporter and a CaMV 35S::luciferase (LUC) construct as an internal control were co-introduced by electroporation. As shown in Fig. 5(b), co-introduction of each reporter, P, GP or C2, with the full-length Ntlim1 construct (LIM) increased GUS activity approximately three- to fourfold compared to the negative control. However, when other effectors were introduced, increased GUS activity was not obtained. Actual GUS/LUC value for co-introduction of the reporter GP with the negative control (non) was 1.5-fold higher than that for of the reporter P. These results indicate that all three domains of Ntlim1 are definitely required for its transactivation activity in vivo, and that the Ntlim1 protein is a transcription factor for the expression of prxC2 gene.
Since we have verified that the Ntlim1 protein, as well as its LIM1 and LIM2 domains, were capable of binding to the Pal-box like motif in vitro ( Fig. 3c), we wanted to know whether the LIM domains of Ntlim1 are able to function as DNA binding domains in vivo. Therefore, we investigated the transcription activation effects of constructs, AD-L1, AD-L2 and AD-L1L2, containing LIM1 and/or LIM2 domains fused to the budding yeast (Saccharomyces cerevisiae) Gal4 trans-activation domain ( Ma et al. 1988 ), on the GUS activity from reporter constructs, P or GP in tobacco protoplasts ( Fig. 5a). Cells containing AD-L1, AD-L2, AD-L1L2 and AD-LIM effector constructs gave four to five times higher GUS activities than those of the negative controls, non and AD ( Fig. 5c). Each LIM domain is sufficient for DNA binding both in vivo and in vitro.
Ntlim1 affects gene expression of phenylpropanoid biosynthesis
To assess the molecular effects of Ntlim1 on phenylpropanoid metabolism, a 1.0 kb fragment of Ntlim1 was inserted in sense and antisense orientations between the CaMV 35S promoter and the 3′ terminator sequence of the nopaline synthase gene. These constructs were introduced into tobacco plants by the Agrobacterium-mediated transformation. Twenty independent primary transgenic plants (T1) of each construct were produced and all lines contained the respective T-DNA by PCR analysis. Subsequently, five sense lines and seven antisense lines were screened by measuring the Ntlim1 transcript levels in the stems and were grown to maturity in the greenhouse. Their self-pollinated seeds (T2) were collected and used for further investigation. Ten T2 kanamycin-resistant transgenic plants of each line were grown in the greenhouse. All of them showed normal growth rate and flowered at similar times. No remarkable abnormal growth or flower color changes were found in the transgenic lines except in several antisense plant lines that showed slightly increased stem height. We isolated cDNA clones encoding some of the structural genes, PAL, 4CL and CAD, involved in phenylpropanoid biosynthesis from tobacco by PCR. The expression levels of the transgene and endogenous Ntlim1, and these structural genes were investigated by Northern blotting using random-primed probes. Total RNAs were extracted from stems in the kanamycin-resistant T2-transgenic plants of each line. As shown in Fig. 6, the sense tobacco plants (S2, S5, S7, S8 and S10) exhibited strong expression of the transgene Ntlim1 at the predicted position of 1.0 kb and the expression levels of PAL, 4CL and CAD were relatively the same or slightly increased as compared with those in the control plant. In the tobacco plants carrying the antisense Ntlim1 gene (A1, A2, A3, A6, A7, A9 and A19), the signals of antisense Ntlim1 were not detected except in A1. Probably, the antisense Ntlim1 transcript might be unstable in tobacco cells. However, in the A2, A9 and A19 plants, the endogenous Ntlim1 (1.2 kb) expression was completely suppressed and the mRNA levels of PAL, 4CL and CAD were not detectable. The A1, A3, A6 and A7 plants exhibited lower expression levels of the endogenous Ntlim1, and corresponding reduced transcript levels of PAL, 4 CL and CAD genes.
We also measured the enzymatic activities of PAL, 4CL and CAD in stems of transgenic and control plants. As shown in Fig. 7, no increased enzyme activities were observed in the sense transgenic plants (S2 and S5) as compared with the control plant, suggesting the possibility of post-transcriptional regulation. The antisense plants exhibited reduced enzyme activity levels, but activity could still be measured in A2 and A9 plants that had very low transcript levels. These results imply that Ntlim1 is able to simultaneously regulate the expression of many genes involved in phenylpropanoid biosynthesis.
Lignin content in transgenic plants
To examine the effect of simultaneous reduction of the transcript levels in many lignin biosynthetic genes, the analysis of the content and monomeric composition of lignin in the transgenic plants was performed. The lignin content of the cell wall residue (CWR) of stem xylem tissues was measured by the gravimetric Klason procedure. Lignin determinations were carried out on T2 kanamycin-resistant transgenic lines after 10 weeks of growth in the greenhouse. The lignin contents of lines S2, A6, A9 and the control plant were measured. No differences could be detected between the sense line S2 and the control plant. In contrast, 10% and 27% reduction was observed in A6 and A9, respectively ( Table 1).
Lignin content is expressed as a percentage (w/w) of cell wall residues (CWR) by Klason lignin.
Syringaldehyde and vanillin were obtained when the CWR was treated by an alkaline nitrobenzene oxidation procedure. Values are expressed as μmol per gram of CWR.
All values represent the mean ± SD of results from the analysis of three to six plants.
21.2 ± 0.6 (100)
124.4 ± 3.1
157.2 ± 4.8
22.3 ± 0.5 (105)
129.5 ± 3.3
153.6 ± 2.9
19.1 ± 0.3 (90)
89.5 ± 3.3
103.6 ± 4.9
15.5 ± 0.3 (73)
31.5 ± 0.9
67.3 ± 3.7
To determine the monomeric composition of lignin in the CWR, we examined the CWR from each plant by alkaline nitrobenzene oxidation analysis. Vanillin and syringaldehyde are generated from the non-condensed fraction of lignin by this procedure ( Chiang & Funaoka 1988). These benzaldehyde derivatives are considered to be derived from guaiacyl and syringyl units in lignin, respectively. The relative abundance of guaiacyl and syringyl units in lignin can be summarized as the S/V ratio (the ratio of the amounts of syringaldehyde and vanillin). No remarkable change of the S/V ratio was detected in the S2. The A9 plant showed a lower value of the S/V ratio and a decreased amount of total aldehydes. These data indicate that lignin in the transgenic plants with decreased expression level of the Ntlim1 contained more condensed units that could not be degraded by oxidation procedures.
The Pal-box like motif is a cis-acting element for expression of the basic horseradish peroxidase prxC2 gene as shown from the results of transient expression analysis using tobacco protoplasts (P. Kaothien, Y. Shimokawatoko, A. Kawaoka, K. Yoshida, and A. Shinmyo, unpublished data). The Pal-box is a common cis-acting element for several phenylpropanoid synthetic genes, such as PAL ( Cramer et al. 1989 ; Lois et al. 1989 ; Ohl et al. 1990 ), 4CL ( Douglas et al. 1987 ), CHS ( Dron et al. 1988 ) and CAD ( Feuillet et al. 1995 ). Antirrhinum MYB genes are known to have a Pal-box binding function and to regulate phenylpropanoid biosynthesis ( Tamagnone et al. 1998 ). Here, we demonstrate that a novel Pal-box binding factor was isolated by Southwestern screening from the expression library of tobacco stems. The predicted amino acid alignment of the isolated clone exhibited a high similarity to LIM proteins. The cloned Pal-box motif binding factor encodes a protein consisting of 200 amino acid residues and shares a 55% identity with the sunflower PLIM-1 protein, which is a transcription factor required for expression of late pollen genes ( Baltz et al. 1992 ). This protein contains a consensus amino acid sequence defined as the LIM domain which is a unique cysteine-rich motif, with two zinc fingers connected by a short spacer of invariably two amino acid residues ( Dawid et al. 1995 ). The LIM proteins form a diverse group which includes LIM-homeodomain proteins, LIM-only protein and LIM-kinase ( Taira et al. 1995 ). Very little is known about the function of the LIM proteins in higher plants.
The LIM domain may provide an interface for protein–protein interaction ( Crawford et al. 1992 ; Sadler et al. 1992 ) and the LIM-only protein Lmo2 is part of a sequence-specific DNA-binding complex ( Wadman et al. 1997 ). The sunflower pollen protein PLIM-1 has been reported to bind to nucleic acid in vitro, but no specific DNA binding sequence was described ( Baltz et al. 1996 ). Thus, there is still not enough evidence to suggest that the LIM domain is able to bind to DNA. We have now shown that the Ntlim1 protein expressed in E. coli bound to a Pal-box like motif in vitro. Furthermore, either the LIM1 or the LIM2 domain was sufficient for its DNA-binding ( Fig. 3). The results presented in Fig. 5 indicate that Ntlim1 also bound to the Pal-box like motif in vivo (tobacco protoplasts). The Ntlim1 protein contains two LIM domains, LIM1 and LIM2, and an acidic domain without any other putative domain, therefore it was classified as a LIM-only protein. The data we present here demonstrate for the first time that a LIM-only protein can be a sequence-specific DNA binding transcription factor. The DNA-binding function of the zinc finger motif is similar between the LIM-only proteins and the GATA transcription factors ( Perez-Alvarado et al. 1994 ). Erythroid transcription factor GATA-1 was the first member of the GATA family to be cloned, and these proteins all have two zinc fingers and bind the DNA sequence, GATA ( Weiss & Orkin 1995). Therefore, the LIM domain may function in a similar way to GATA-1 zinc fingers that mediate both DNA-binding and protein–protein interaction ( Osada et al. 1995 ; Perez-Alvarado et al. 1994 ). From NMR structure analysis of the LIM2 domain of the CRP2 protein, the carboxy-terminal CCCC module is structurally related to the DNA-binding domain of GATA-1 ( Konrat et al. 1997 ). The Ntlim1 protein, however, did not recognize the GATA sequence from the result of PCR-based binding sequence analysis ( Fig. 4). Further structural analysis is needed to determine the recognition-sequence of the Ntlim1 protein.
Transient transactivation experiments indicate that Ntlim1 has a domain serving as a transcription activator ( Fig. 5). The most likely candidate for this domain is the C-terminal acidic polypeptide which contains only 37 amino acids long. No homology to known transcription activation domains has been observed and there are no distinctive features in its amino acid sequence in the C-terminal region of Ntlim1. Random acidic amino acid peptides from E. coli have been reported to substitute for the acidic activator domain of Gal4 in yeast ( Ma & Ptashne 1987). The Gal4 activator can activate transcription in tobacco ( Ma et al. 1988 ). Plant transcription factors, such as maize VP1 and Dof1, have acidic domains that function in transcription activation in plants ( McCarty et al. 1991 ; Yanagisawa & Sheen 1998). All of the evidence supports the hypothesis that an acidic domain of Ntlim1 may function as a transcription activator.
Northern blot analysis showed that at least two genes encoding enzymes of general phenylpropanoid metabolism (PAL and 4CL) and one gene in the lignin branch (CAD) were affected by Ntlim1 at the transcription level ( Fig. 6). Transgenic plants with the sense Ntlim1 showed the same or slightly increased expression levels of PAL, 4CL and CAD as compared with the control plants, while the antisense Ntlim1 transgenic plants showed decreased expression of these genes. Thus, suppression of Ntlim1 expression could block many steps in phenylpropanoid biosynthesis and Ntlim1 might also regulate the expression of some other genes in this pathway that contain the Pal-box sequence in their promoter regions. The Pal-box is reported as a common recognition site of MYB proteins ( Douglas 1996; Grotewold et al. 1994 ). Overexpression of AmMYB308 repressed the expression of C4H, 4CL and CAD in tobacco, but did not affect the steady state level of PAL transcript ( Tamagnone et al. 1998 ). Different MYB factors could co-ordinately regulate the expression of PAL ( Tamagnone et al. 1998 ). There are many copies of the MYB gene family in a plant genome and the DNA-binding specificity of plant MYB proteins varies considerably ( Martin & Paz-Ares 1997). The binding site performance and affinity of MYB proteins is also likely to be strongly influenced by other protein factors that interact with them ( Jin & Martin 1999). From our results, the sequence recognized by the Ntlim1 protein is similar to that recognized by MYB proteins, although the last six bases in the consensus of Ntlim1 recognition sequence were not rigid ( Fig. 4). The discrepancy of the effect on the PAL gene expression between Ntlim1 and AmMYB308 protein might result from the smaller difference in their DNA-binding ability. It is also possible that both Ntlim1 and AmMYB308 proteins are members of the Pal-box binding complex although they compete in binding to the Pal-box. Therefore, the Ntlim1 could bind to the PAL promoter while the AmMYB308 could not. To clarify this matter, further analysis of the Pal-box binding complex by an immunological experiment is needed.
Transgenic plant S2 with a high expression level of Ntlim1 showed no increase of lignin content ( Table 1). One possible explanation is the post-transcriptional regulations of PAL, 4CL and CAD because we could not detect any increase of these enzyme activities ( Fig. 7). As expected, the low expression level of PAL, 4CL and CAD in the A9 plant with the antisense Ntlim1 resulted in a 27% reduction in its lignin content compared with the control plants. Several attempts to reduce lignin content by inhibiting the activity of enzymes in the phenylpropanoid biosynthesis pathway have been reported ( Douglas 1996). Transgenic plants with repressed COMT or CAD activities resulted in a modified lignin structure instead of reduced lignin content, suggesting that neither enzyme limits the lignin accumulation ( Dwivedi et al. 1994 ; Halpin et al. 1994 ). However, a reduction in the expression of PAL or 4CL has previously been shown to cause lower lignin content and altered lignin composition in transgenic plants ( Elkind et al. 1990 ; Lee et al. 1997 ). Lignin is an essential component of the plant cell wall and severe reduction in lignin content (> 40%) can cause an abnormal growth phenotype in herbaceous plants. For example, transgenic tobacco with suppression of PAL or overexpression of MYB-related transcription factors showed an abnormal growth phenotype ( Elkind et al. 1990 ; Tamagnone et al. 1998 ). Transgenic tobacco plants with antisense CCR showed a 47% reduction of Klason lignin, but the normal growth of the plant was severely affected ( Piquemal et al. 1998 ). Caffeoyl-CoA O-methyl transferase (CCoAOMT) antisense tobacco plants with up to 66% reduction in lignin content grew as normal as the wild-type plants but their vessel shape was significantly collapsed ( Zhong et al. 1998 ). Recently, transgenic aspen trees with downregulated 4CL exhibited normal phenotype when the lignin content was reduced by up to 45% because the reduction in the lignin content was compensated by an increase in cellulose content ( Hu et al. 1999 ). In this experiment we detected no visible abnormal growth in the transgenic tobacco with sense and antisense Ntlim1 and also found no sign of vessel collapse in the A9 plant with up to 27% reduction in the lignin content. Thus, Ntlim1 may function as a weak activator that helps maintain basal expression of the genes encoding enzymes in the phenylpropanoid biosynthesis pathway and resulted in the retained basal enzyme activities in lignifying tissues. Our results indicate that the genetic engineering of transcription factors involved in lignin biosynthesis can be an efficient strategy for producing plants with improved pulping proceeding or digestibility.
Nicotiana tabacum cv Petit Havana SR-1 was used. Tobacco plants were grown in a greenhouse at 25°C under a 16 h light and 8 h dark photoperiod.
Screening of the library and sequencing
Total RNA was isolated from tobacco stem tissues that were grown for 6 weeks in a greenhouse. Using oligo(dT) and random primers, a cDNA library was constructed in λgt11 vector (Stratagene, La Jolla, CA, USA). The amplified library was screened with a digoxigenin-labeled double-stranded oligonucleotide fragment that corresponded to a triple repeat of a Pal-box like motif, -CCACTTGAGTAC-. We used essentially the same screening protocol as Singh et al. (1988) . An inserted DNA fragment of a positive clone was amplified by polymerase chain reaction using forward and reverse primers for λgt11 vector. The amplified DNA fragment was recloned into pNoTA/T7 vector by PCR Cloner (5 Prime, 3 Prime Inc., Boulder, CO, USA). Sequence analysis was performed on double-stranded DNA templates, using DyeDeoxyTM Terminator Cycle Sequencing Kit (Perkin Elmer, Foster City, CA, USA).
Southern and Northern analyses
Genomic DNA was isolated from tobacco leaves and Southern blot analysis was performed as described by Maniatis et al. (1982) . Total RNA was extracted from the tobacco tissue ( Chomczynski & Sacchi 1987), fractionated on a formaldehyde-agarose gel and blotted onto a nylon membrane, Hybond N (Amersham Pharmacia Biotech, Bucks, UK). The membrane was hybridized with a digoxigenin-dUTP labeled probe using a DIG labeling kit (Boehringer Mannheim GmbH, Mannheim, Germany). Hybridization was performed according to the instructions from Boehringer Mannheim GmbH. Arabidopsis actin gene (CD3–118) was obtained from the Arabidopsis Biological Resources Center (Columbus, OH, USA). Wheat ribosomal DNA was used as an internal control ( Apples & Dvorak 1982).
Expression of Ntlim1 in Escherichia coli
The cells of E. coli JM109, carrying the expression vector pGEX-2 TK (Amersham Pharmacia Biotech) containing the cDNA sequence encoding Ntlim1 were grown to an absorbance of 660 nm of 0.8 in LB broth. IPTG was then added to the culture at a final concentration of 1 m m and the cells were cultured at 25°C for an additional 2 h. Cells were harvested by centrifugation and the crude extract was used for purification of Ntlim1 protein by glutathion-sepharose 4B according to the manufacturer's instructions (Amersham Pharmacia Biotech).
Gel retardation assay
The binding reaction was carried out in a 20 μl reaction mixture, containing protein (500–600 ng in the case of Ntlim1, LIM1, LIM2 or GST protein), 2 ng of radiolabeled DNA probe, 10 m m Tris–HCl pH 8.0, 80 m m KCl, 0.016% NP40, 7.5% glycerol, 1 m m DTT and 1 μg poly (dI-dC)·poly (dI-dC) (Amersham Pharmacia Biotech). The reaction mixtures were pre-incubated at 23°C for 15 min, mixed with radiolabeled DNA probe, and allowed to proceed for a further 30 min at 23°C. Each sample was applied, without dye, onto a 5% non-denaturing polyacrylamide gel in 0.5 × TBE buffer. Electrophoresis was carried out at 15 V cm−1 in 0.5 × TBE buffer. Following the electrophoresis, the gel was dried and exposed overnight to X-ray film at −80°C. For the competitive binding assay, excess amounts of non-labeled DNA fragments (competitors) were added to the reaction mixture at the pre-incubation step.
Binding site analysis
Binding site analysis basically followed the methods used by Kinzler & Vogelstein (1989). The pool of DNA used was prepared tobacco genomic DNA digested with restriction enzyme HaeIII and HincII that produced a blunted end. The digested genomic DNAs were attached with the EcoRI–NotI–BamHI adapter (Takara Syuzo, Otsu, Japan) by T4 DNA ligase. The binding reaction was performed as it was for the gel retardation assay using the E. coli extract of overexpressed GST-Ntlim1 fusion protein. The DNA-fusion protein complexes were purified by glutathion-sepharose 4B (Amersham Pharmacia Biotech) and were digested by thrombin protease. Using primer corresponding to the adapter sequence, PCR was carried out as follows: 5 min at 94°C, then 1 min at 55°C, 1 min at 72°C and 1 min at 94°C for 25 cycles, followed by a final 10 min extension at 72°C. After five rounds of the binding and purifying reactions, PCR products were recloned in pNoTA/T7 using PCR Cloner II (5 Prime, 3 Prime Inc) and 28 clones were sequenced.
Effector plasmids construction
Construction of chimeric genes was performed basically using the method of Maniatis et al. (1982) . Effector plasmids were constructed under the control of CaMV 35S promoter in a modified plasmid pBI221-mcs that placed a multi-cloning site into the GUS structural gene. Double-stranded synthetic oligonucleotide based on the synthetic linker III for the multi-cloning site ( Malik & Wahab 1993) was inserted at the BamHI–SacI site in pBI221 ( Jefferson et al. 1987 ). To generate LIM, which has a full-length cDNA of Ntlim1, the BamHI fragment of Ntlim1 was ligated into the same site of pBI221-mcs. L1 and L1L2 were then produced using reverse primer for pUC19, and L80 primer 5′-GCTTCTCGAGTTTTGGTGTACCTTCAAGC-3′ or L164 primer 5′-TCTCCTCGAGAAGTTGAATATGGATGATGT-3′, respectively. The amplified DNA fragments of L1 and L1L2 were digested by restriction enzymes BamHI and XhoI, then inserted at the same site of pBI221-mcs. The L2 was amplified by L2 primer 5′-CACAGGGAGCCAAAGTGACAAGCAAGCATGTT-3′ and the L164 primer. The L1A was created by digested XhoI and SacI sites of the effector plasmid L1L2, and then inserted translationally to a XhoI–SacI digested PCR fragment that covered the acidic domain of Ntlim1 using L164c primer 5′-ACATCTCGAGATTCAACTTATC-
AAGGA-3′ and forward primer for pUC19. The L2A was also constructed by the ligation of a BamHI–SacI digested DNA fragment amplified by PCR using L2 and forward primers.
Effector plasmids to express Gal4 or its derivative were constructed with plasmids for a yeast two-hybrid system, pGAD424 (Clontech, Palo Alto, CA, USA). An amplified DNA fragment of Ntlim1 using LF primer 5′-CGTATGGATCCATGGCT-
TTTGCA-3′ and LR primer 5′-GATAAGGATCCATCAACTTGGTCG-
GCT-3′ by PCR was digested by BamHI and ligated in frame to pGAD424 at the same restriction site. The AD-LIM, that was observed by PCR using GADF primer 5′-CTCAAGCTTTGCAAAG-
ATGGATAAAG-3′ and LR primers and blunt-ended by T4 DNA polymerase then phosphorylated by polynucleotide kinase, was placed at the SmaI site of pBI221mcs. The AD-L1 and AD-L1L2 were constructed by the same method as AD-LIM using L80 and L164 primer, respectively. The AD-L2 was fused translationally to a DNA fragment of L2 at the BamHI site of pGAD424.
Construction of reporter plasmids and electroporation
The C2 reporter plasmid was a chimeric gene of prxC2(− 307)/GUS and constructed using previously described methods ( Kawaoka et al. 1992 ). The P reporter plasmid was constructed using double-stranded synthetic oligonucleotides. A three copy sequence of the Pal-box like motif (– CCACTTGAGTAC-) was ligated to the EcoRV site at −90 bp of the CaMV 35S promoter. The correctness of orientation of the inserted oligonucleotides was checked by sequencing. The GP reporter plasmid was a G-box like motif (–AACACGTGATA-) fused into the P reporter. Electroporation of the tobacco protoplasts was carried out as described previously ( Kawaoka et al. 1992 ). The GUS activity was assayed in tobacco protoplasts by the method described by Jefferson et al. (1987) , with 4-methyl-umbelliferyl glucuronide as the substrate, and expressed as picomoles of methyl-umbelliferone per minute per mg of protein. The luciferase gene under the control of the CaMV 35S promoter was used as an internal control.
Transformation of tobacco
Chimeric genes of antisense and sense Ntlim1 cDNA under the control of the 35S promoter were ligated into the binary vector pBI121 (Clontech, Palo Alto, USA). Tobacco was transformed with Agrobacterium tumefaciens LBA4404 using the leaf disk transformation method ( Rogers et al. 1986 ). Transformants were selected on MSBN medium containing 100 mg l−1 kanamycin ( Ditta et al. 1980 ). These kanamycin-resistant shoots were placed on hormone free MS agar medium for root regeneration, and finally transferred to soil and grown in a greenhouse.
Isolation of tobacco PAL, 4CL and CAD cDNA clones
For tobacco PAL cDNA (GenBank accession number X78269), a forward primer 5′-CCATCTAATCTGACAGCAGGAAGA, and a reverse primer, 5′-CAGATTGGAAGAGGAGCACCATTC, were synthesized. For tobacco 4CL cDNA ( D43773), a forward primer 5′-TACTCCTCTGGGACGACTGGATTA and a reverse primer 5′-GTCTCCTCTTATGCAAATTTCTCC were used, and for the CAD gene ( A24084), a forward primer 5′-CACCAAGTTAAAAATGATC-
TTGGCATGTCC and a reverse primer 5′-CTGTGAAGTCACACCT-
TTCTCTTTGCAGAA were used. The PCR amplification was carried out using a cDNA (2 ng) synthesized from tobacco stem as a template. The identity of these genes was checked by sequencing.
Enzyme assays were carried out using the crude protein extract from the basal part of stems in 10-week-old tobacco plants. The PAL activity was assayed according to Rhodes & Wooltorton (1971). The 4CL activity was determined with ferulic acid as a substrate ( Grand et al. 1983 ). The activity of CAD was measured with coniferyl alcohol as a substrate by the method of Wyrambik & Grisebach (1975). Protein was quantitated with a protein assay kit (Bio-Rad, Hercules, CA, USA).
Lignin determination was performed on the dried insoluble cell wall residues (CWR) of samples soxhlet extracted with toluene/ethanol, ethanol and water. Klason lignin was measured by the method of Effland (1977). Alkaline nitrobenzene oxidation was analysed according to Kajita et al. (1996) .
We are grateful to Dr Chandrashekhar P. Joshi (Michigan Technological University) for critically reading the manuscript. We thank Dr Yukifumi Uesono (University of Tokyo), Drs Ko Kato and Atsuhiko Shinmyo (Nara Institute of Science and Technology) for their helpful discussion. We also thank Drs Seiichiro Hasezawa and Toshiyuki Nagata (University of Tokyo) for providing tobacco cultured cell BY-2. We are grateful to Masako Obata, Nippon Paper Industries for her technical advice on lignin analysis.
DDBJ/EMBL/GenBank accession number Ntlim1 ( AB023479).