Characterization of cis-elements required for vascular expression of the Cinnamoyl CoA Reductase gene and for protein–DNA complex formation

Authors

  • Eric Lacombe,

    1. Signaux et Messages Cellulaires chez les Végétaux, UMR CNRS-UPS 5546, Pôle de Biotechnologie Végétale, BP 17, Auzeville, 31 320 Castanet Tolosan, France, and
    Search for more papers by this author
      The first two authors contributed equally to this work.
  • Jan Van Doorsselaere,

    1. Signaux et Messages Cellulaires chez les Végétaux, UMR CNRS-UPS 5546, Pôle de Biotechnologie Végétale, BP 17, Auzeville, 31 320 Castanet Tolosan, France, and
    2. Laboratorium voor Genetica, Department of Plant Genetics, Flanders Interuniversity Institute for Biotechnology, Universiteit Gent, K.L. Ledeganckstraat 35, B-9000, Belgium
    Search for more papers by this author
      The first two authors contributed equally to this work.
  • Wout Boerjan,

    1. Laboratorium voor Genetica, Department of Plant Genetics, Flanders Interuniversity Institute for Biotechnology, Universiteit Gent, K.L. Ledeganckstraat 35, B-9000, Belgium
    Search for more papers by this author
  • Alain M. Boudet,

    1. Signaux et Messages Cellulaires chez les Végétaux, UMR CNRS-UPS 5546, Pôle de Biotechnologie Végétale, BP 17, Auzeville, 31 320 Castanet Tolosan, France, and
    Search for more papers by this author
  • Jacqueline Grima-Pettenati

    Corresponding author
    1. Signaux et Messages Cellulaires chez les Végétaux, UMR CNRS-UPS 5546, Pôle de Biotechnologie Végétale, BP 17, Auzeville, 31 320 Castanet Tolosan, France, and
      For correspondence (fax +33 5 62 19 35 02; e-mail
    Search for more papers by this author

  • The first two authors contributed equally to this work.

    The EMBL accession number for the sequence of the EgCCR promoter is AJ132750.

For correspondence (fax +33 5 62 19 35 02; e-mail grima@smcv.ups-tlse.fr).

Summary

Cinnamoyl-CoA reductase (CCR) catalyses the first specific step in the biosynthesis of monolignols, the monomeric units of lignins. We examined the developmental regulation of the Eucalyptus gunnii EgCCR promoter by analysing the expression of EgCCRGUS fusions in tobacco. EgCCR promoter activity was strongest in lignified organs (stems and roots) consistent with the EgCCR mRNA level in these organs. Histochemical analysis showed expression in vascular tissues (cambium, young differentiating xylem, ray cells, internal and external phloem) of stems and roots in agreement with in situ hybridization data. Promoter deletion analysis and gain-of-function experiments identified the sequences between positions −119 and −77 as necessary and sufficient for expression in vascular tissues of stems. Electrophoretic mobility-shift assays showed that this region is specifically recognized by nuclear proteins present in tobacco stems, giving rise to two retarded complexes, LMC1 and LMC2. Using overlapping EgCCR fragments and mutated oligonucleotides as competitors in gel-shift assays, two DNA–protein interaction sites were mapped. Finally, the role of protein–protein interactions in the formation of the LMC1 and LMC2 complexes was investigated using the detergent sodium deoxycholate, and protein fractionation onto a heparin Sepharose column.

Introduction

Lignins, the second most abundant natural plant compounds on earth after cellulose, are complex three-dimensional aromatic polymers resulting from the oxidative polymerization of predominantly three hydroxycinnamyl alcohols (p-coumaryl, coniferyl and sinapyl alcohols), commonly referred to as monolignols. These polymers are deposited in the secondary cell wall of specialized cells found mainly in the vascular tissues, providing additional reinforcement of the cell wall and conferring properties such as imperviousness and resistance to biodegradation. As a consequence, lignified tissues play important roles in structural support and water/nutrient conduction, and in response to environmental stimuli such as mechanical stresses or pathogen attacks. These capabilities are essential for terrestrial life, and lignification has probably played a critical role in the successful colonization of land by plants ( Lewis & Davin, 1994).

The biosynthesis of lignins begins with the common phenylpropanoid pathway starting from phenylalanine and leading to the hydroxycinnamoyl-CoA esters. (The hydroxycinnamoyl-CoA esters are the common precursors of a wide range of end-products involved in various aspects of plant development and defence such as flavonoids, lignins, coumarins, phlobaphenes and many low molecular-weight phenolics; Hahlbrock & Scheel, 1989.) The hydroxycinnamoyl-CoA esters are then channelled into the lignin branch pathway to produce monolignols via two successive reductive steps catalysed by cinnamoyl-CoA reductase (CCR) and cinnamyl alcohol dehydrogenase (CAD) ( Boudet et al. 1995 ). CCR catalyses the conversion of hydroxycinnamoyl-CoA esters to their corresponding cinnamaldehydes, and these are subsequently converted to the corresponding monolignols by CAD.

The organization and transcriptional regulation of the genes of the general phenylpropanoid encoding l-phenylalanine ammonia-lyase (PAL) and 4-coumarate CoA ligase (4CL), are areas of active research. Depending on the species, PAL and 4CL are encoded by several genes which are largely controlled at the transcriptional level in response to both developmental and environmental cues ( Chapell & Hahlbrock, 1984; Dixon & Paiva, 1995; Hahlbrock & Scheel, 1989; Lawton & Lamb, 1987). In species examined in detail, PAL and 4CL expressions were shown to be co-ordinately regulated ( Bell-Lelong et al. 1997 ; Dixon & Paiva, 1995; Hahlbrock & Scheel, 1989; Logemann et al. 1995 ; Lois et al. 1989 ).

In bean, PAL is encoded by a small gene family of differentially regulated genes where one, PAL2, appears to be mainly associated with the synthesis of lignin monomers. In transgenic tobacco plants, the PAL2 promoter specifies GUS expression in differentiating xylem tissue, cortical and epidermal cells of stems, root tips, pigmented sectors of petals and pollen ( Bevan et al. 1989 ; Liang et al. 1989 ; Shufflebottom et al. 1993 ). Vascular expression is part of the complex programme of developmental expression of PAL2 regulating the synthesis of functionally diverse phenylpropanoid products. In-depth studies of PAL2 promoter activity during development have led to the delineation of cis-elements involved in vascular expression ( Hatton et al. 1995 ; Leyva et al. 1992 ). Similar studies have also been undertaken for the promoter of 4CL ( Hauffe et al. 1993 ; 1995; Neustaedter et al. 1999 ).

In contrast with the considerable insights into the regulation of the expression of these phenylpropanoid genes, provided mainly by promoter-driven reporter-gene expression, very little information is available about the genes encoding enzymes specific for monolignol biosynthesis such as CAD and, especially, CCR. Functional analysis of the eucalyptus CAD promoter in transgenic poplar revealed gene expression primarily in regions undergoing active lignification: phloem fibres, differentiating xylem, xylem ray cells and vascular cambium ( Feuillet et al. 1995 ; Hawkins et al. 1997 ). This pattern of expression was confirmed by in situ hybridization ( Hawkins et al. 1997 ) and immunolocalization techniques ( Samaj et al. 1998 ). A similar vascular pattern of expression was found in transgenic tobacco plants containing either the tobacco ( Walter et al. 1994 ) or the eucalyptus CAD promoter (V. Lauvergeat, personal communication), suggesting that annual herbaceous plants such as tobacco can be used to study the expression of genes from a perennial woody angiosperm.

We have now focused our attention on the EgCCR promoter because, as the first dedicated enzyme in monolignol biosynthesis, CCR occupies a key position between the general phenylpropanoid metabolism and the lignin-specific branch. As such, CCR may be considered as a potential control point regulating the carbon flux towards lignins. This hypothesis has been recently confirmed by gene transfer experiments ( Piquemal et al. 1998 ), taking advantage of the characterization of the first cDNA encoding CCR in Eucalyptus gunnii ( Lacombe et al. 1997 ). Through ectopic expression of a homologous antisense gene, significant downregulation of CCR has been obtained in transgenic tobacco plants leading to both qualitative and quantitative changes in lignin profiles.

In order to gain an insight into the transcriptional regulation of this gene, we investigated the temporal and spatial expression of the E. gunnii CCR gene using promoter–reporter gene (GUS) fusions in transgenic tobacco. We have shown that EgCCR promoter activity is correlated with lignification throughout plant development. Functional analysis of 5′ deletions identified the sequences from −119 to −77 (relative to the transcription start) as necessary for vascular expression in stems and root tips, thereby defining a minimal EgCCR promoter which directs expression in tissues in which the full-length promoter is active. Electrophoretic mobility-shift assays were carried out to reveal interactions between stem nuclear proteins and potential cis sequences within this minimal promoter region. The data obtained suggest that the shifted bands observed are the result of the binding of a multimeric complex at two different sites, one of them corresponding to an MYB consensus site.

Results

Expression of EgCCR–GUS fusion in transgenic tobacco

In E. gunnii, CCR is encoded by a single gene which shows high similarity in its coding region with dihydroflavonol-4-reductase (DFR) genes ( Lacombe et al. 1997 ). The sequence of the promoter region (EgCCR) is shown in Fig. 1. The position of the transcription start ( Fig. 1) was determined using 5′ RACE, at a C numbered +1. Multiple putative TATA boxes were found in an AT-rich region (position −16/−53) upstream of the initiation transcription site ( Fig. 1). It is worth noting the presence of one /CT/ (n = 15) track located at position −135. Computer-aided nucleotide comparison identified an AC-rich region (position −90/−81) which is conserved among promoters of several phenylpropanoid genes ( Bell-Lelong et al. 1997 ; Douglas, 1996; Feuillet et al. 1995 ; Hatton et al. 1995 ).

Figure 1.

Nucleotide sequence of the 5′-flanking region of the Eucalyptus gunnii CCR gene.

The translation start codon is underlined. A long (TA) stretch (−16/−53) which could serve as a TATA box is indicated in italics. The transcription start (+1) is shown in bold. Three direct repeats are underlined. The ACI element (position −88) is shown in bold. Deletion-end points corresponding to positions −1079, −898, −775, −491, −324, −218, −176, −119 and −77 are shown by arrowheads. The EMBL accession number of the CCR promoter sequence is AJ132750.

In order to assess the developmental regulation of the EgCCR gene, a 1668 bp fragment of the promoter of the E. gunnii EgCCR gene (position −1448 to +220) was translationally fused to the β-glucuronidase (GUS) reporter gene. This fragment, referred to as full-length promoter ( Fig. 1), contains a sequence of 1448 bp upstream of the putative transcriptional site, the 5′ untranslated region and the first nine codons corresponding to the CCR protein. The resulting construct was introduced into tobacco via Agrobacterium-mediated transformation. Segregation of kanamycin resistance in the progeny of self-pollinated tobacco transformants, together with Southern blot analysis, indicated that the number of transgene loci ranged from one to three or more (data not shown). The temporal and spatial distribution of EgCCR promoter-driven gene expression was investigated both in in vitro- and greenhouse-grown tobacco primary transformants.

The levels of GUS activity were assayed quantitatively in stems, leaves and roots of 3-week-old, in vitro-grown, independent transformants. Although the amount of GUS activity varied among individual plants, the pattern of GUS expression driven by the EgCCR promoter was conserved, with a preference for roots followed by stems and leaves ( Fig. 2a). On average, GUS activity was 1.7-fold higher in roots than in stems, and 9.5-fold higher than in leaves. The observed variation in transgene expression between individual transformants is probably due to ‘position effect’ and/or transgene copy number.

Figure 2.

GUS activity levels in different organs from tobacco plants harbouring the full-length EgCCR-GUS construct.

Different tissues were harvested from seven independent transformants grown in vitro (a) or in greenhouse conditions (b). GUS activity was measured and the mean value calculated for each tissue. 1 U = 1 nmole MU (4-methyl-umbelliferone) produced per hour per μg protein.

In order to study promoter activity throughout plant development, 3-month-old, greenhouse-grown primary transformants were examined for GUS activity at different stages of development. Highest GUS activity was found in old stems, followed by young stems, roots and petioles. Very low levels of GUS activity were found in leaves, flowers and fruits ( Fig. 2b). In contrast to in vitro-grown plantlets, young stems and roots exhibited similar levels of activity. Moreover, GUS activity was threefold higher in older parts of the stem compared to the younger parts. In contrast, no significant difference was observed between leaves at different stages of development. The EgCCR promoter showed a strong activity, mainly in stems and roots which is consistent with the steady-state CCR mRNA levels in these organs, as assayed by Northern blot analysis ( Fig. 3).

Figure 3.

CCR mRNA accumulation in tobacco.

RNA was isolated from stems, roots and leaves from 2-month-old tobacco plants (2 weeks post-acclimation). RNA gel blots were performed using the tobacco CCR cDNA ( Piquemal et al. 1998 ) as a probe.

The EgCCR promoter directs GUS expression preferentially in vascular tissues

In order to determine the cell- and tissue-specific expression pattern directed by the EgCCR promoter, GUS activity was assayed histochemically in in vitro- and greenhouse-grown primary transformants, as well as in seedlings from F1 progeny. Histochemical analysis showed that GUS activity was consistently localized in the vascular tissues of all organs examined ( Fig. 4). Blue staining was observed in roots and was restricted to the vascular bundles of the hypocotyl in young seedlings from the F1 generation ( Fig. 4a). In primary transformants, GUS activity was highest in stem tissues undergoing active lignification ( Fig. 4b). GUS activity was also high in the vascular cylinder of the root ( Fig. 4c,d), and in vascular tissues of fruits, petioles, leaves and flowers (data not shown).

Figure 4.

Histochemical localization of GUS activity in different tissues of transgenic tobacco plants containing the full-length CCR-GUS construct (a–e) and the Δ-119/+220 deletion (f–g).

(a) GUS-stained 7-day-old seedlings; (b) cross-section of stems from in vitro-grown plants showing GUS expression in vascular tissues (cambium, xylem and phloem fibres); (c) cross-section of roots from in vitro-grown plants showing GUS expression in the vascular cylinder (xylem, phloem, endodermis, cortex, epidermis and exodermis); (d) cross-section of roots from greenhouse-grown plants showing GUS expression in the vascular cylinder (xylem, phloem, endodermis, epidermis and exodermis); (e) longitudinal section of roots from in vitro-grown plants showing GUS expression in the root tip and vascular cylinder; (f) cross-section of stems from in vitro-grown plants showing GUS expression in vascular tissues (cambium, xylem and phloem fibres); (g) whole roots from in vitro-grown plants showing GUS expression in root tip.

The pattern of expression conferred by the EgCCR promoter in stems of greenhouse-grown tobacco plants (data not shown) was similar to that observed in stems of in vitro-grown plants ( Fig. 4b), i.e. vascular cambium, xylem ray cells and phloem fibres. No blue staining was detectable in cortex (not shown) or in stem pith cells. In contrast, intense blue staining was observed in virtually all tissues of roots of in vitro-grown plants ( Fig. 4c), whereas in roots of greenhouse-grown plants, GUS activity was mainly associated with the vascular cylinder and limited to a few cells in the cortex ( Fig. 4d). GUS activity in cortical cells of roots of in vitro-grown plants could account, at least in part, for the high level of GUS activity in these organs ( Fig. 2a). Similar observations have already been reported for tobacco transformed with homologous or heterologous CAD promoter–GUS fusions, and were thought to be brought about by the tissue culture conditions ( Feuillet et al. 1995 ; Walter et al. 1994 ). Indeed, we observed that when in vitro-grown plants containing the EgCCRGUS construct were grown on medium protected from light, very low expression was detected in the cortex, suggesting that exposure of roots to light might be responsible, at least in part, for promoter activity in cortical cells (data not shown).

The vascular-specific pattern of EgCCR promoter activity in stems and roots of greenhouse-grown tobacco plants is in good agreement with the in situ hybridization data obtained previously on poplar stem ( Lacombe et al. 1997 ) and root sections (S. Hawkins, unpublished results), showing that CCR mRNA accumulation was restricted to vascular tissues.

Apart from the vascular-preferential expression driven by the EgCCR promoter and supporting the role of CCR in lignification, GUS activity was also observed in non-lignified tissues. Blue staining was consistently found in root tips, independently of culture conditions. A longitudinal section of the root tip revealed GUS activity to be concentrated in the epidermal and cortical cells ( Fig. 4e) in the elongation zone of the root. This finding underlines the fact that, although the major metabolic role for monolignols is as lignins precursors, they can also be involved in the synthesis of other derivatives such as lignans or dihydroconiferylglucosides which, interestingly, have been shown to promote cell division ( Lynn & Chang, 1990).

Delineation of cis-elements required for vascular-preferential expression of the EgCCR promoter

In order to define the position and function of cis-sequences that determine the vascular-specific pattern of expression, a series of 5′ deletions of the EgCCR promoter ( Fig. 1) were translationally fused to the GUS gene and transferred into tobacco. Depending on the construct, four to 21 independent transformants were obtained and examined for GUS activity by both quantitative assays and histochemical staining.

Despite a considerable variation in GUS activity between the individual transformants containing a specific construct, fluorimetric measurements of GUS activity in stems and roots revealed strong activities in transformants with deletions up to base pair −119 ( Table 1). The majority of transformants harbouring the Δ-77/+220 construct displayed a very low or even undetectable GUS activity in all organs, indicating the presence of critical positive regulatory elements in the region between −119 and −77.

Table 1.  GUS activity levels in stems and roots from in vitro-grown tobacco plants containing CCR–GUS fusions
ConstructGUS activity (U) in stemsGUS activity (U) in roots
0–0.010.1–11–100–0.010.01–11–10
  1. One unit of GUS activity (U) corresponds to one nmole of methyl-umbelliferone produced per hour and per μg protein.

Δ−1448+2204334
Δ−1079+2208282
Δ−898+2203232
Δ−775+220431
Δ−491+220211
Δ−324+22033
Δ−218+22066
Δ−176+220312
Δ−119+2201394
Δ−77+220972

To investigate the effects of 5′ promoter deletions on the tissue-specific pattern conferred by the EgCCR promoter, in vitro-grown transformants were histochemically stained for GUS activity. Deletions up to −119 in the EgCCR promoter produced similar patterns of GUS expression as compared to the full-length promoter ( Table 2). In stems, GUS activity was confined to the vascular cambium, the xylem ray cells and the internal and external phloem fibres ( Fig. 4f; Table 2). In roots, staining was observed in the vascular cylinder and in root tips ( Fig. 4; Table 2). Interestingly, deletion of sequences upstream of position −324 abolished EgCCR promoter activity in cortical cells of roots, suggesting that the sequences responsible for light-mediated EgCCR gene expression are located upstream of this position. Consistent with the observation that markedly reduced overall GUS activity was detected in extracts from plants harbouring the construct Δ-77/+220, no visible GUS staining was detected in stems and roots of these transformants, even after prolonged incubation of tissues ( Table 2). Taken together, these data suggest that the minimal promoter region containing all cis-sequences necessary for vascular expression is located within −119 to −77 bp of the putative transcription start.

Table 2.  Histochemically localised GUS expression in stems and roots from in vitro-grown tobacco plants containing CCR-GUS deletions and −55/35S fusions
ConstructionGUS expression in stemsGUS expression in roots
EpCoPhlXylPiNo.CoVCRTNo.
  1. No, number of plants with detectable GUS activity.

  2. The histochemically-assessed levels of GUS expression in the tissues are summarized as: +++, expresson in >75% of plants; ++, expression in 50–75% of plants; +, expression in 25–50% of plants; –, expression in 0% of plants; nd, not determined.

  3. Ep, epidermis; Co, cortex; Phl, phloem; Xyl, xylem; Pi, pith; VC, vascular cylinder; RT, root tip.

Δ−1448+220+++++++7+++++++++7
Δ−1079+220+++++++6ndndndnd
Δ−898+220+++++++4ndndndnd
Δ−324+220++++++11++++++7
Δ−218+220++++++15++++++11
Δ−176+220++++++2++++2
Δ−119+220++++++18++++++9
Δ−77+2201815
(A)−117–68:-55/35S1010
(B)−68–117:-55/35S88
(C)−218–68:-55/35S1313
(D)−68–218:-55/35S++++++9+++++++++9
(E)-55/35S33

To assess whether an EgCCR promoter fragment can activate a minimal heterologous promoter, a 150 bp (−218 to −68) and a 49 bp (−117 to −68) fragment were cloned in both sense and antisense orientations upstream of the CaMV 35S minimal promoter −55 (−55/35S, Keller & Heierli, 1994) ( Fig. 5). This minimal promoter (E) did not give rise to histochemically detectable GUS activity, either in stems or in roots of transgenic tobacco ( Keller & Heierli, 1994).

Figure 5.

Constructs used for gain-of-function experiments.

EgCCR promoter fragments −117/−68 and −218/−68 were cloned in front of the −55/35S minimal promoter. (a) −117/−68 fragment in sense orientation; (b) −117/−68 fragment in reverse orientation; (c) −218/−68 fragment in sense orientation; (d) −218/−68 fragment in reverse orientation; (e) constuction −55/35S used as negative control. The black box indicates the localization of potential DNA-binding sites ( Fig. 7) within the different fragments. The transcription initiation site is indicated by +1.

Histochemical analysis of plants transformed with these constructs revealed no detectable GUS activity in plants containing constructs A and B (containing fragment −117/−68 neither in sense (A) or antisense (B) orientation, nor in transformants harbouring construct C (−218/−68 fragment in sense orientation) ( Table 2). Interestingly, only construct D (−218/−68 fragment in antisense orientation) could direct strong GUS activity in both stems and roots ( Table 2). In stems of transformants containing construct D, blue staining was observed in the vascular cambium and in xylem ray cells ( Fig. 6a) in a similar fashion to that observed for the full-length EgCCR promoter and for deletions up to −119 ( Fig. 4b,f). This result indicates that EgCCR promoter sequences located between positions −218 and −68 are necessary and sufficient to direct vascular-specific expression of GUS activity in stems. Since we have shown that deletion −119/+220 was able to confer a vascular pattern of expression in stem, it is likely that the cis-regulatory elements responsible for vascular expression are located between −119 and −68 bp. The lack of expression obtained using constructs A and B suggests the importance of the position of the potential regulatory sequences contained in the −117/−68 EgCCR promoter fragment relative to the TATA box. Constructs A, B and C led to a similar position of potential cis-regulatory sequences relative to the TATA box, which is different to construct D containing the −218/−68 promoter fragment in antisense orientation ( Fig. 5). In this latter construct, the distance of putative regulatory elements with respect to the TATA box would be similar to that in the −119 deletion.

Figure 6.

Histochemical localization of GUS activity in different tissues of transgenic tobacco plants containing −68–218:55/35S construct.

(a) Cross-section of stem from in vitro-grown plant showing GUS activity in vascular tissues (Ca, cambium; R, ray cells; X, xylem); (b) root from in vitro-grown plant showing GUS activity in virtually all cell types (RT, root tip).

In contrast, in roots of plants harbouring the D construct, GUS staining was observed in virtually all tissues as well as in root tips ( Fig. 6b). This pattern of expression is reminiscent of that of the full-length promoter, but is different from the pattern conferred by deletion −218/+220, since no GUS activity was detected in the cortex with this construct. This result indicates that the region −68/+220 (not present in construct D) may contain important regulatory sequences restricting EgCCR promoter activity to roots. Since sequences from position −68 to +1 are mainly composed of TATA-box sequences, the differences observed may result from the deletion of regulatory sequences located within the 5′ leader sequence of the EgCCR gene.

The EgCCR fragment required for vascular expression contains two sites for DNA–protein interactions

Based on the 5′ unilateral deletion analysis and the gain-of-function experiments, we predicted that the minimal region of the EgCCR promoter responsible for vascular expression in stems and root tips (−119/−68) would interact specifically with transcription factors that can mediate transcriptional activation. In order to test this hypothesis, we performed electrophoretic mobility-shift assays (EMSA) after incubating tobacco stem nuclear extracts with a labelled −117/−68 EgCCR fragment. As shown in Fig. 7, two shifted bands (low mobility complex), named LMC1 and LMC2, respectively, were observed. Competition with a 100-fold molar excess of unlabelled specific competitor (−117/−68) greatly reduced complex formation, whereas a similar amount of a non-specific DNA competitor (unrelated DNA) had no effect ( Fig. 7). These results suggest that one or more DNA-binding proteins in tobacco stem nuclear extracts interact specifically with cis-acting element(s) within the −117/−68 promoter fragment.

Figure 7.

Gel-shift assay with tobacco stem nuclear extracts using the −117/−68 CCR probe and overlapping CCR promoter fragments (wild-type or mutated) as competitor.

Tobacco stem nuclear extracts were used with the −117/−68 EgCCR fragment as a probe. The sequence of fragments A, B and C and the mutated fragments (Am1–Am4 and Bm5–Bm8) are shown. The mutated bases are underlined. 500 ng of each fragment were used as competitor in the binding assay. Specific competition was observed with fragments A and B, and with several mutated fragments (see text). The interaction sites within the −117/−68 EgCCR fragment are indicated in bold. Bases shown to be important for the binding of nuclear proteins were mapped (bold), defining two interaction sites. The binding site contained in fragment A was named BS1 (binding site 1), whereas the binding site in fragment B was called ACI as its sequence partially overlaps ACI.

In order to localize the site(s) of DNA–protein interaction within the −117/−68 fragment, three overlapping promoter fragments (A, B and C; Fig. 7) were used as unlabelled competitors of complex formation. Effective competition using 100-fold competitor excess was observed with fragment A and to a lesser extent with fragment B, but not with fragment C ( Fig. 7). Competition was observed for both complexes, although LCM2 never disappeared completely under the experimental conditions used. Since fragments A and B compete independently, we concluded that they both contain sequences that are important for the LMC formation. This was further confirmed by EMSA performed using either A or B as labelled probes. Both fragments were able to form LMC1 and LMC2, although the intensity of the shifted bands with fragment B was far less intense than with fragment A, confirming that the affinity of the nuclear proteins was greater for fragment A than for fragment B ( Fig. 8).

Figure 8.

EMSA using fragments −117/−68 (lane 1), B (lane 2) and A (lane 3), respectively, as probes, and nuclear protein fraction from tobacco stems.

In order to map the sequences critical for the formation of the complexes more precisely, we used mutated versions of fragments A and B (Am1–Am4 and Bm5–Bm8; Fig. 7) as competitors in EMSA experiments. The mutated versions Am1 and Am4 were still able to compete as efficiently as fragment A, whereas Am2 and Am3 turned out to be ineffective competitors. These results suggest that in the AGCGGG motif, contained in the core of fragment A, at least two bases are critical for the formation of the complex.

Among the mutated versions of fragment B, Bm7, and to a lesser extent Bm8, were less effective than fragment B in inhibiting the formation of the complexes. Unexpectedly, fragment Bm6 appeared to be a more effective competitor than fragment B. Changing the G residues at positions −86 and −85 to T residues seemed to increase the competitiveness of fragment B. No change in the effectiveness of competition was observed with Bm5. These experiments suggested that one or more of the G residues (positions −75, −76, −81, −82 and −83) are the most important in the formation of both complexes with fragment B. The two other G residues (positions −86 and −85) also appear to be involved in the interaction, although they do not seem to be as important as the first ones.

Taken together, the EMSA experiments suggest that the formation of protein–DNA complexes LMC1 and LMC2 is probably due to interactions occurring at two distinct sites within the −117/−68 fragment.

A complex of proteins binds to the −117/−68 EgCCR fragment

As suggested by the EMSA experiments, two different cis-sequences are necessary for LMC1 and LMC2 formation. These complexes are probably the result of binding of a multiprotein complex to the −117/−68 CCR fragment. In order to test this hypothesis, we treated the nuclear extracts with the non-denaturating detergent sodium deoxycholate (DOC), which is known to disrupt protein–protein interactions ( Després et al. 1995 and references therein). Figure 9 shows that DNA-binding activity was lost when extracts were pretreated with 0.1% DOC, but not when extracts were pretreated with 1% of the non-denaturating detergents CHAPS or Triton X-100. This suggests that the complexes formed in the absence of the detergent consist of multiple proteins that interact with each other and are unable to bind DNA alone in such experimental conditions.

Figure 9.

Binding of tobacco stem nuclear proteins to the −117/−68 probe after treatment of the extracts with different non-ionic detergents (sodium deoxycholate (DOC), CHAPS and Triton X-100).

Gel-shift assays were performed with tobacco stem extracts pretreated with three detergents at different concentrations. The LMC1 and LMC2 complexes were not formed when stem proteins were pretreated with DOC (0.1 or 1%).

As a first step toward the biochemical characterization of the transacting factors involved in the formation of LMC1 and LMC2, proteins contained in stem nuclear extracts were fractionated according to their affinity for DNA on heparin Sepharose. The binding activity of proteins contained in the flow-through and protein fractions eluted with increasing concentrations of NaCl (0.5, 1, 2 and 3 m) to the −117/−68 EgCCR fragment was tested using EMSA. Figure 10 shows that DNA-binding activity is recovered in the protein fraction eluted with 1 m NaCl, and consists of a single retarded complex that presents identical mobility properties as complex LMC2 (data not shown). The absence of LMC1 in either of the fractions suggests that this complex could result from the interaction of an additional protein, which does not interact directly with DNA, with the proteins responsible for LMC2 formation.

Figure 10.

Electrophoretic mobility shift assay using the −117/−68 EgCCR fragment and different protein fractions eluted from the heparin-Sepharose column.

Lane 1, flow-through; lanes 2–5 protein-eluted, respectively, with 0.5, 1, 2 and 3 m NaCl.

These results suggest that at least three proteins are responsible for both LMC1 and LMC2 formation. Two of these proteins can interact directly with the two binding sites, respectively, contained within the −117/−68 EgCCR fragment, and they need to interact with each other in order to stabilize the interaction with DNA. LMC1 formation is probably due to the recruitment of a third protein partner by the proteins present in LMC2.

Discussion

The CCR gene is preferentially expressed in vascular tissues

The fact that important functions are performed by lignins in plants suggests that elaborate control mechanisms must exist to ensure the synthesis of lignins at the right place at the right time. Cinnamoyl CoA-reductase catalyses the reduction of the hydroxycinnamoyl-CoA esters to the corresponding cinnamaldehydes, the first specific step in the biosynthesis of the monolignols. In order to gain an insight into the signal transduction pathway that mediates EgCCR gene expression, we have functionally analysed the EgCCR promoter from E. gunnii.

The GUS reporter system was used to characterize the temporal and spatial patterns of EgCCR promoter activity at the cell and tissue levels. Histochemical analysis showed that tissue-specific expression of the EgCCR–GUS gene was correlated with areas undergoing lignification: cambial cells, young xylem, ray parenchyma cells, phloem fibres (internal and external) and endodermis. Therefore the activity of the EgCCR promoter in tobacco is strongly associated with vascular tissues, consistent with the role of CCR in providing phenylpropanoid monomers for polymerization into lignins. The consistency of the GUS data presented here with the Northern blot analysis and in situ hybridization in stems ( Lacombe et al. 1997 ) suggests that expression of the EgCCR gene is controlled, at least in part, by the activity of its promoter.

The vascular expression pattern conferred by the EgCCR promoter is reminiscent of that reported earlier for the CAD promoter ( Feuillet et al. 1995 ; Hawkins et al. 1997 ) suggesting that the two genes, acting consecutively in the lignin branch pathway, may be co-ordinately regulated. Expression in vascular tissues is also part of the more complex developmental patterns conferred by promoters of genes encoding enzymes further upstream in the pathway, such as PAL and 4CL.

Regulatory sequences that mediate vascular-specific expression in stems, roots and root tips

As a first step towards the identification of cis-regulatory elements involved in the control of EgCCR gene expression in vascular tissues, we have analysed truncated versions of the promoter in transgenic plants. The removal of sequences from −1448 to −119 did not result in a dramatic loss of expression in vascular tissues in stems and roots as the Δ-119/+220 truncated promoter produced similar patterns of GUS expression to the ‘full-length’ promoter. In contrast, deletion Δ-77/+220 abolished expression completely, thereby defining the sequence between −119 and −77 as necessary for expression in vascular tissues and in root tips.

Surprisingly, when sequences spanning positions −117 to −68 were fused to the 35S minimal promoter (−55 : 35S), they were unable to direct GUS activity in any tissue tested. Only fragment −218/−68, cloned in reverse orientation in front of this minimal promoter, could promote GUS activity in vascular tissues of stems and in root tips, but not in the vascular cylinder of roots, underlying the importance of the relative position to the TATA box of cis-sequences contained in the −117/−68 EgCCR fragment.

In addition, gain-of-function experiments stressed that the developmental pattern of EgCCR expression in roots probably involve a combination of positive and negative cis-regulatory sequences that, when acting together, give rise to a pattern of expression restricted to the vascular bundle. Whereas sequences contained between −119 and −77 are responsible for constitutive expression in roots, vascular-restricted EgCCR expression involves a putative negative element located in the 5′ leader sequence which abolishes EgCCR promoter activity in cortical cell of the roots.

These data underline the importance of the sequences located between positions −119 and −77 in controlling the developmental regulation of the EgCCR gene. Furthermore, they provide evidence to show that the regulatory mechanisms controlling vascular expression of the EgCCR gene are different in stems and in roots.

Sequence analysis of the −119/−77 bp fragment revealed the presence of an AC element whose sequence is identical to the ACI element (CCCACCTACC) previously identified in the bean PAL2 promoter ( Hatton et al. 1995 ). In the PAL2 promoter, three AC elements (ACI, ACII and ACIII) interact with a G-box to direct a complex pattern of expression. The different AC elements in the PAL2 promoter are responsible for expression in petals (ACI, ACII and ACIII); root tips (only ACI); cortex in the stem (ACI and ACII); and vascular tissues (ACI and ACII). The regulation of PAL2 gene expression is even more complex because the ACI and ACII elements are negative elements for expression in phloem. These data suggest that the AC elements are involved in conferring tissue-specific expression.

AC elements are present in promoters from many genes involved in phenylpropanoid biosynthesis. It has been suggested that those common motifs may provide a mechanism by which different steps of phenylpropanoid metabolism are co-ordinately regulated (for a review see Douglas, 1996).

Trans-acting factors binding with the −117/−68 CCR sequences

EMSA experiments provided evidence for the interaction of nuclear proteins from tobacco stems with DNA sequences within the −117/−68 EgCCR promoter fragment. The close correlation between the binding of nuclear proteins to this fragment and the functional role played by this domain in specifying tissue-specific expression patterns indicates that the transcription factors that interact with this domain are important for EgCCR gene expression.

Binding assays using the non-overlapping double-stranded oligonucleotides A and B as competitors revealed that LMC1 and LMC2 can be independently formed at two independent sites within the −117/−68 EgCCR fragment. Both sites were mapped using mutated versions of A and B in EMSA experiments. The sequence of the binding site located in fragment B corresponds to the ACI element. In contrast, the second binding site (BS1) located in fragment A, which contains several G residues (AGCGGG), does not show any sequence similarity with previously characterized cis-element.

Additional EMSA experiments, together with binding assays using fragments A and B as probes, showed that trans-acting factors responsible for complex formation do not exhibit the same affinity to their respective targets. The binding activity of nuclear extracts to the ACI element is far less important than for the BS1 binding site contained in fragment A. The fact that two interaction sites are present might indicate that two distinct proteins bind to both sites. Gel-shift experiments using DOC-pretreated nuclear extracts support this hypothesis, since the disruption of protein–protein interactions abolishes the binding activity of stem nuclear extracts to the −117/−68 EgCCR fragment. These data further indicate that at least two proteins need to interact prior to binding efficiently to the two cis-elements contained within this fragment. These results are consistent with the fact that Hatton et al. (1995) were unable to demonstrate the formation of DNA–protein complexes using the ACI element contained in the bean PAL2 promoter as probe and tobacco stem proteins. An additional cis-element and its cognate trans-acting factors are likely to be necessary to form a stable protein–DNA complex via protein–protein interaction.

Further evidence in favour of the involvement of a multiprotein complex in LMC1 and LMC2 formation was provided by preliminary purification of DNA-binding proteins from stem nuclear extracts. The absence of LMC1 in heparin-bound proteins indicated that additional proteins that are not developing direct contact with DNA are necessary for the formation of this retarded complex.

These data indicate that complex protein–DNA and protein–protein interactions occur within the −117/−68 EgCCR fragment. A multiprotein component, in which at least two different proteins bind to their respective binding sites, gives rise to LMC2 formation, while a third partner is likely to be responsible for LMC1 formation but is not required for DNA binding.

Since an ACI element has been identified as a binding site for nuclear proteins in the −117/−68 EgCCR fragment, likely candidates to bind to this target sequence are plant MYB-transcription factors. A growing body of evidence shows that MYB proteins bind AC elements ( Grotewold et al. 1994 ; Sablowski et al. 1994 ) and are involved in the coordinate control of transcriptional activity of phenylpropanoid genes (for review see Martin & Paz-Ares, 1997; Weisshaar & Jenkins, 1998). Based on the DNA-binding classification of Romero et al. (1998) , the ACI element present in the EgCCR promoter contains the MBSIIG site [G(G/T)T(A/T)GGT(A/G)] which is bound by a subgroup of MYB factors including AmMYB308. Overexpression of AmMYB308 in transgenic tobacco repressed phenolic acid metabolism and lignin synthesis ( Tamagnone et al. 1998 ). Based on the steady-state transcript levels of PAL, C4H, 4CL and CAD, the authors suggested that AmMYB308 is able to repress the expression of genes encoding all the enzymes between C4H and CAD. These data further suggest that CCR might be transcriptionally regulated by the tobacco AmMYB308 orthologue.

Depending on the set of phenylpropanoid genes considered, MYB factors can activate transcription either alone, as demonstrated for MYB340 and MYB305 from Antirrhinum majus with PAL, CHI and DFR gene expression, or in the presence of an additional factor. In maize the transcriptional activation of genes leading to anthocyanin biosynthesis is the result of combinatorial interactions between a MYB-related regulatory protein Cl (MYB) and a member of the b/HLH (basic helix-loop-helix) family, R. The N-terminal domain of Cl exhibits strong transcriptional activity, but the very low affinity of the Cl protein toward its target site does not allow trans-activation by Cl alone. The binding of Cl, and therefore the activation of the target gene, requires the presence of B which stabilizes the binding of Cl by directly interacting with it ( Goff et al. 1990 ; Goff et al. 1991 ; Goff et al. 1992 ).

Since a growing body of evidence is highlighting the importance of combinatorial control of transcriptional regulation in plants, in some cases involving MYB factors ( Neustaedter et al. 1999 ; Singh, 1998 and references therein), it is conceivable that the proteins(s) binding to the BS1 element could work together with a MYB-related factor (binding the AC1 site) to specify vascular expression of the EgCCR gene. One possibility is that the interaction between these composite proteins could increase the binding affinity of the MYB-related factor to its target site.

Future identification of the protein components of the complex should help to elucidate the mechanism by which the −119/77 nuclear protein complex regulates EgCCR expression in vascular tissues.

Experimental procedures

Plant material

Tobacco plants (Nicotiana tabacum var. Samsun) were grown in vitro and in the greenhouse, as described previously ( Piquemal et al. 1998 ).

RNA gel-blot hybridization

RNA gel-blot analysis was performed according to Piquemal et al. (1998) .

5′ race

5′-RACE was used to map the transcription start using the 5′ RACE System 2.0 (Gibco-BRL, Life Techno Ltd, Paisley, UK). Gene-specific primer GSP1 (5′-ACGCAC-CGAATCTTCC) was used for reverse transcription and a set of two nested gene-specific primers, GSP2 (5′-GGGACTGGTAGGTGGC-CTTTCTC) and GSP3 (5′-TTCCAGCTCTCTCAGATGCC), were used in specific nested PCR reactions. 5′ RACE products were subsequently cloned in pGEM-T Easy (Promega, Madison, USA). Ten positive clones were sequenced in order to map the transcription start.

Promoter–GUS fusions

In order to fuse the promoter with the GUS gene, BamH1 restriction sites were introduced by PCR. Amplification was performed using a 1.8 kb XhoI fragment of the EgCCR genomic clone ( Lacombe et al. 1997 ) as a template; PRGU1 as the 5′ primer containing a sequence located at 1648 bp upstream of the translation initiation codon (5′-GATATCTTCTTTGGGATCCCTCT-CGC, position −1467); and PRGU2 as the 3′ primer complementary to a sequence overlapping the ATG (5′-GGTCTGGCCGGATCCGG-GGAGGGC, position +229). The resulting PCR fragment was cloned in BamH1-cut pOGUS ( Axelos et al. 1989 ), resulting in a translational fusion of the EgCCR promoter with the GUS gene (plamid pCCR6GUS), including nine codons of the EgCCR gene.

Deletion Δ-1079/+220 was constructed by digesting pCCR6GUS with EcoRI and ClaI, end-filling with Klenow DNA polymerase and self-ligation. Deletion Δ-775/+220 was constructed by digesting pCCR6GUS with KpnI, followed by self-ligation. Deletions Δ-898/+220, Δ-491/+220, Δ-176/+220, Δ-119/+220 and Δ-77/+220 were constructed following the same strategy as described for the full-length construct. The promoter fragments were amplified by PCR using the primer PRGU2 in combination with either primer

PRGU4 (5′-CACCGAATTGGATCCACAATAT, position −910, for Δ-900/+220);

PRGU5 (5′-GGCTGCCGGATCCCTCCAGTAA, position −503, for Δ-491/+220);

PRGU11 (5′-AACATGGATCCTAAGGGCAAA, position −187, for Δ-176/+220);

PRGU10 (5′-AAGGAGGGATCCTTATAGGG, position −128, for Δ-119/+220); or

PRGU6 (5′-CGGTAGGTGGATCCTGGTAAAC, position −91, for Δ-77/+220).

Deletions Δ-324/+220 and Δ-218/+220 were obtained using an ExoIII/Mung Bean kit (Stratagene, La Jolla, CA, UA). Fragments −218/−68 and −117/−68 were amplified by PCR using −218 Xho (5′-ATCCTCGAGCCATGGAAAATAAGG), −68 Xho (5′-GACTCGAG- TTACCAAGA), and −117 Xho (5′-GGTCTCGAGAGGGGAGCG), and cloned as XhoI fragments in −55/35S:GUS ( Keller & Heierli, 1994). The orientation of the EgCCR fragment was checked by PCR using appropriate primers.

All EgCCR–GUS fusions were then transferred as EcoRI–HindIII fragments in pBIN19 ( Bevan, 1984), with the exception of deletions Δ-324/+220 and Δ-218/+220. The latter deletions were digested with NdeI, end-filled with Klenow DNA polymerase, digested with HindIII and cloned in SmaI/HindIII-cut pBIN19.

Tobacco transformation and β-glucuronidase assays

Agrobacterium tumefaciens LBA4404 strain was transformed using the freeze–thaw method. Tobacco plants were transformed by the leaf-disc method ( Horsch et al. 1985 ). Multiple independent transformants were obtained and propagated vegetatively in vitro and in the greenhouse. The presence of the transgene was verified by PCR using promoter-specific primers and the GUS primer (Clontech, Palo Alto, CA, USA).

Stems, roots and leaves were harvested from 5-cm-tall, in vitro-grown plants. Old and young stems and leaves (corresponding to the bottom and top of the plants, respectively), roots, petioles, flowers (containing petals, anthers and pistil), and fruits (containing sepals, carpels and ovary) were harvested from greenhouse-grown plants. Tissues were frozen in liquid nitrogen and stored at −80°C.

Quantitative GUS assays were carried out using 4-methylumbelliferyl-β- d-glucuronide as substrate ( Jefferson et al. 1987 ). Protein concentrations were determined using the Bio-Rad protein assay.

Histochemical localization of GUS activity was performed according to Jefferson et al. (1987) with minor modifications. Fragments of stems, roots and leaves were prefixed in 90% acetone under vacuum, then washed twice in 100 m m potassium phosphate buffer (pH 7). Transverse sections (80–150 μm), made by hand or using a vibratome apparatus (Biorad, Ivry sur Seine, France), were incubated for 12–24 h at 37°C in 100 m m NaPi (pH 7), 5 m m EDTA, 0.5 m m ferrocyanide, 0.5 m m ferricyanide, 0.5 mg ml−1 X-Gluc (5-bromo-4-chloro-3-indolyl-β- d-glucuronic acid).

Gel-shift assays

The sequences between −117 and −68 were amplified by PCR with primers 5′-GGTCTCGAGAGGGGAGCG and 5′-GACTCGAGTTAC-CAAGA introducing XhoI sites. The PCR product was cloned in pGEMT (Promega). Complementary oligonucleotides corresponding to fragments A and B were synthesized and annealed. All DNA fragments used as probes in EMSA were 32P-labelled using Klenow DNA polymerase, and purified on polyacrylamide gels ( Maniatis et al. 1982 ).

Nuclear-enriched protein fractions were prepared from stems of 1-m-tall, greenhouse-grown plants according to Curie et al. (1991) . DNA-binding proteins from nuclear extracts were purified on prepacked 1 ml HiTrap heparin Sepharose columns (Pharmacia, Courtaboeuf, France) using fast performance liquid chromatography. Approximately 250 μg of nuclei-enriched protein extracts were loaded, washed with binding buffer (10 m m Tris–HCl pH 8.0, 50 m m NaCl, 7 m mβ-mercaptoethanol, 10% glycerol), and heparin-bound proteins were eluted using binding buffer supplemented with NaCl (0.5, 1, 2 and 3 m). Step-eluted fractions were recovered and dialysed against binding buffer. The efficiency of the dialysis was checked by monitoring conductivity. All procedures were carried out at 4°C using a flow rate of 1 ml min−1.

Binding reactions were performed at room temperature for 30 min in 25 μl binding buffer containing 4 μg nuclear proteins, 1 μg poly(dIdC) : poly(dIdC) and 10 000 cpm 32P probe (0.2–1 ng). Double-stranded oligonucleotides (500 ng) corresponding to fragments A, B and C, and mutated versions of A (Am1–Am4) and B (Bm5–Bm8), were added as competitor in the binding assay. Free and bound DNA were separated on 4% polyacrylamide gels in 0.5× TBE. Gels were then fixed, dried, and exposed to Kodak film at −70°C.

Acknowledgements

We would like to thank B. Keller for the gift of the −55/35S minimal promoter, and Colette Guez for her excellent technical assistance. We are grateful to Pierre Sivadon, Jane Gronow and Simon Hawkins for carefully reading the manuscript. E.L. and J.V.D. were supported by grants from the French Ministère de l'Enseignement Supérieur et de la Recherche, and the French Ministère des Affaires Etrangères, respectively. This work was financially supported by the European Commission, DG XII, FAIR Programme (Contract no. FAIR-CT95-0424) and by the Centre National de la Recherche Scientifique.

Ancillary