INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Secondary metabolism is indeed essential to the form and function of plants and also to the survival of these sessile members in their diverse and dynamic ecosystems. The products range from relatively simple phenolic esters to the complex, heterogeneous polymers of lignin. These metabolites have a wide range of often disparate functions ranging from providing mechanical strength to pest deterrence to facilitation of symbiosis (Strack, 1997). The general phenylpropanoid pathway is central to secondary metabolism from various perspectives (Dooner et al., 1991; Chapple and Ellis, 1992; Davin and Lewis, 1992; Chapple, 1994; Boudet et al., 1995; Dixon et al., 1996; Douglas, 1996; Mol et al., 1998; Weisshaar and Jenkins, 1998; Whetten et al., 1998; Dixon and Steele, 1999). Secondary metabolism is highly variable not only among plant species but also among different cell types within a plant, and generalizations from specific examples provide a useful framework but not conclusive insights for other systems (Lewis and Yamamoto, 1990; Campbell and Sederoff, 1996). The general phenylpropanoid pathway and the associated pathways are thus essentially a composite.

The phenylpropanoid pathway produces from L-Phe or L-Tyr a number of hydroxycinnamate derivatives by a series of hydroxylations and methylations culminating at one end in sinapic acid in angiosperms (Fig. 1 in Campbell and Sederoff, 1996). Sinapic acid and a number of preceding intermediates are in turn used as the precursors for various compounds (Dixon et al., 1996). Sinapine is one such product derived from sinapic acid (for the pathway, see Chapple et al., 1992). It is found only in crucifer seeds (Regenbrecht and Strack, 1985; Bouchereau et al., 1991), but its role is unknown. Strack (1981) speculated that it is a storage reserve for seedling growth in Raphanus sativus, but it does not have this function in Arabidopsis (Chapple et al., 1992). The presence of sinapine in oilseed Brassicas (notably rapeseed/canola) is undesirable because it is an antinutritional component of the seed meal (Bell, 1993). It imparts a bitter flavor, and its metabolite is responsible for the fishy odor in the eggs of certain brown-shelled egg laying hens that are fed on rapeseed/canola meal (Pearson et al., 1980; Ismail et al., 1981). Sinapine is present in all Brassica species (Velasco and Möllers, 1998; Wang et al., 1998;) and in nearly all of the crucifers that have been examined (Regenbrecht and Strack, 1985; Bouchereau et al., 1991).

We have been investigating metabolic engineering as a means to reduce sinapine in canola seeds. Here we present the cloning, characterization, and antisense suppression of the CYP84 family of cytochrome P450-dependent mono-oxygenase genes from Brassica napus that resulted in up to 40% reduction in the sinapine content. This work was completed while the available genetic evidence implicated the Arabidopsis ferulate-5-hydroxylase (F5H) in the production of 5-OH ferulate from ferulate, a precursor in the sinapate synthesis pathway (Chapple et al., 1992; Meyer et al., 1996). Very recently, a sweetgum (Liquidamber styraciflua) CYP84 member (Osakabe et al., 1999) and the Arabidopsis F5H (Humphreys et al., 1999) have been shown to catalyze the hydroxylation of coniferaldehyde far more efficiently than ferulate when expressed in yeast. The two enzymes, however, show a difference in their substrate spectrum. We discuss our results in light of these new observations.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Isolation of Three Unique F5H cDNA Clones from B. napus

Five cDNA clones (BNF5H) hybridizing to an Arabidopsis F5H probe were isolated from a stem cDNA library of B. napus. The cDNA inserts in two clones appeared truncated but otherwise identical in sequence to the insert in another clone (BNF5H1). Thus only BNF5H1 and the remaining two clones, BNF5H2 and BNF5H3 were characterized further. These three showed nucleotide sequence polymorphism in the open reading frame (ORF) and a significant difference in the 3'-untranslated region (UTR) sequences. The nucleotide sequences of the inserts in these clones have been deposited in GenBank (accession nos. AF214007, AF214008, and AF214009).

The insert in BNF5H1 (1,880 bp) has a 38-nucleotide 5'-UTR, a 1,560-nucleotide ORF, and a 282-nucleotide 3'-UTR, whereas BNF5H2 (1,884 bp) has a 51-nucleotide 5'-UTR, a 1,560-nucleotide ORF, and a 273-nucleotide 3'-UTR. The first ATG in the putative ORF of BNF5H2 is preceded by an in-frame TAG sequence at 36 position relative to the ATG. The flanking sequences in both of the BNF5H1 and BNF5H2 (AAT ATG GAG) have the consensus nucleotide present (underlined) in the plant translational initiation context sequence (Joshi et al., 1997). Thus, the ORF of 1,560 nucleotide present in BNF5H1 and BNF5H2 is predicted to encode a 520-amino acid polypeptide of 58.5 kD. The third cDNA clone (BNF5H3) with an insert of 1,835 nucleotide appeared to be truncated at the 5' region up to the 18th nucleotide of the ORF. This clone has a 292-nucleotide 3'-UTR.

The ORFs in BNF5H1 and 2 are very similar to each other (98% nucleotide sequence identity; 99% amino acid sequence identity; Fig. 1). They are similar to the Arabidopsis F5H ORF (520 amino acids; Meyer et al., 1996) at the nucleotide (90% identity) and amino acid (93% identity) levels. The near-complete ORF in BNF5H3 (514 amino acids) also shows the same level of identity to the corresponding portion, but lacked a codon for Pro at position 39, which is present in the other F5Hs. The Arabidopsis F5H is the first member of CYP84, a new family of cytochrome P450-dependent mono-oxygenases (Meyer et al., 1996). Recently, a coniferaldehyde 5-hydroxylase clone from sweetgum has been isolated (Osakabe et al., 1999). This CYP84 member is 72% identical to Arabidopsis and BNF5H protein. The B. napus F5Hs also contain a high degree of amino acid sequence conservation in the Pro-rich region immediately following the N-terminal hydrophobic region and in the C-terminal heme-binding region (residues P₄₅₀ FGXGRRXCXG₄₆₀ of BNF5H1), which is characteristic to P450 mono-oxygenases. Thus, the three BNF5Hs were deemed to belong to CYP84. Further, as alluded to above, the three cDNA clones were considered to represent three CYP84 genes. Additional evidence for this is provided below.

View larger version (41K): [in this window] [in a new window]   Figure 1.   Comparison of the deduced amino acid sequence of the B. napus F5H (BNF5H) with that of Arabidopsis (Meyer et al., 1996; ATHF5H) and sweetgum (Osakabe et al., 1999; CAld5H). Hyphens indicate identical amino acid, and the variants are as shown. The parenthesis in BNF5H3 denotes incomplete amino-terminal portion,  indicates a gap, and the asterisks mark the end.

B. napus and Its Progenitors Have at Least Two Groups of F5H Genes

The molecular relationship of BNF5H1, -2, and -3 was studied further by examining the 3'-UTR and the ORFs in the three cDNA clones and the introns in the corresponding genomic segments. The nucleotide sequence of the 3'-UTR of BNF5H1 differs from that of BNF5H2 by 13% and from that of BNF5H3 by 37%. Similarly, the 3'-UTR of BNF5H2 differs from that of BNF5H3 by 37%. Aside the absence of a Pro codon in BNF5H3, the ORFs of the three clones showed nucleotide polymorphism as originally detected by restriction digestion patterns (data not shown) and subsequently by nucleotide sequence analysis.

A genomic clone was identified by probing a B. napus cv Westar library with a BNF5H1 probe. A preliminary analysis showed that the region corresponding to the central one-third of the ORF contained two introns (data not shown). Two sets of PCR primers (see "Materials and Methods") were used as individual pairs to amplify this region from a B. napus genomic DNA sample, and the independently derived amplicons (0.95 or 1.1 kb) were cloned into pBluescript SK⁺ (Stratagene, La Jolla, CA). The inserts in nine of these were sequenced completely. Both of the 0.95- and 1.1-kb categories contained two introns and the exon regions. The exon sequence allowed assignment of the amplicons to the three cDNA clones. Three types of amplicons corresponding to BNF5H1, -2, and -3 were found collectively. The congeneric introns thus identified were compared among themselves. The two introns are situated at identical positions in all three genes with respect to the stop codon (Fig. 2). However, the length and nucleotide sequence show significant differences. The introns of BNF5H1 and -2 are more similar to each other than to those of BNF5H3. Further, the nucleotide sequence dissimilarity scores for the introns shown in Figure 2 support the relationship gleaned from the comparison of the UTRs. Thus these results collectively establish two organizational groups of BNF5H genes within the B. napus CYP84 family, one comprising BNF5H1 and -2 and the other BNF5H3.

View larger version (25K): [in this window] [in a new window]   Figure 2.   Organizational characteristics of the introns, pertaining to the grouping of the three BNF5H genes. The two introns found in the shaded region corresponding to the cDNA are located at identical points with reference to the stop codon (taa; 1,050 and 619 nucleotides upstream) in the predicted ORF. The length of the congeneric introns vary as shown. The nucleotide sequence dissimilarity from pairwise comparisons is shown on the left for intron 1 and on the right for intron 2. The regions corresponding to the unshaded portions of the ORF have not been examined for the presence of introns.

The cultivated B. napus used here is a descendent of a natural amphidiploid (n = 19) containing the genomes of both Brassica rapa (n = 10) and Brassica oleracea (n = 9). It was of interest to find the association, if any, of the above gene groups to the diploid species. Southern-blot analysis of genomic DNA detected four bands in B. napus and two in each of the two cultivated species of B. rapa and in B. oleracea (Fig. 3), suggesting the presence of at least four genes in B. napus and at least two in the other species. The third band in the HindIII digest of B. oleracea might be due to an additional gene or due to an internal cleavage site in one of the genes. The attribution of the individual genes to the B. rapa or B. oleracea lineage was done by amplification of the genomic DNA with gene-specific primers from the coding region. The specificity of these primer pairs was ascertained with the respective cDNA controls. As shown in Figure 4, BNF5H3 is present in both of the B. rapa and B. oleracea, BNF5H1 only in B. rapa, and BNF5H2 only in B. oleracea. As expected, B. napus yielded the amplicons for all three genes. The above-noted molecular relationship among the three BNF5H genes vis-a-vis the genealogical relationship was confirmed from the conservation of similarity among the amplicons at their nucleotide sequence level. For example, the BNF5H1 amplicon was more similar to its counterpart from the other two species than to the amplicons of BNF5H2 or BNF5H3 from any of the three species (data not shown). This is noteworthy in view of the closer relationship of BNF5H1 to BNF5H2 than to BNF5H3.

View larger version (73K): [in this window] [in a new window]   Figure 3.   Southern-blot analysis of BNF5H-like sequences in the genomic DNA of the amphidiploid B. napus and its two diploid progenitor representatives. HindIII (H) and EcoRI (E)-digested genomic DNA were probed with a BNF5H1 cDNA fragment (nucleotide 256-nucleotide 702). HindIII-digested bacteriophage -DNA size markers in kb are as shown.

View larger version (60K): [in this window] [in a new window]   Figure 4.   Gene-specific amplification of BNF5H sequences from B. napus, B. rapa, and B. oleracea genomic DNA. The positive controls with the BNF5H cDNAs and negative control with no template DNA are as shown. The BNF5H1-specific primers flank the introns, hence the difference in size between the genomic and cDNA amplicons. The primer sets used here were as follows: set A for BNF5H1; set C for BNF5H2; and set E for BNF5H3 (see "Materials and Methods").

Expression Profile of the BNF5H Genes in B. napus

Northern-blot analysis of total RNA from various parts of B. napus with an BNF5H cDNA probe showed a very abundant transcript of 1.9 kb in the stem and much less in the root, leaf, pod, bud, and the least in the seed tissue (Fig. 5). The expression of other phenylpropanoid pathway genes, Phe ammonia-lyase and cinnamate 4-hydroxylase also followed this general pattern, but the cinnamate 4-hydroxylase gene expression was almost equally abundant in both of the stem and root tissue (data not shown). A 25-cycle reverse transcriptase (RT)-PCR of poly(A⁺) RNA with gene-specific primers showed an amplicon of predicted size for each primer pair in all of these tissues, indicating expression of the three genes in these tissues (data not shown). The RT-PCR analysis was repeated with total RNA and with fewer cycles (15-cycle reaction) to provide a better distinction of the relative transcript abundance. A Southern-blot hybridization analysis of the resulting amplicons was done with a common probe as described in "Materials and Methods" (Fig. 6).

View larger version (34K): [in this window] [in a new window]   Figure 5.   Northern-blot analysis of BNF5H gene expression in B. napus tissue. Twenty micrograms of total RNA for each sample was electrophoresed, blotted, and probed with a BNF5H1 (nucleotides 256-702) cDNA probe. rRNA, Ethidium bromide-stained ribosomal RNA. The relative abundance of BNF5H transcript was measured in a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

View larger version (37K): [in this window] [in a new window]   Figure 6.   RT-PCR analysis of 10 µg of total RNA with gene-specific primers. The controls with the BNF5H cDNAs or no template DNA are as shown. The primer sets used here (see "Materials and Methods") were as follows: set A for BNF5H1; set D for BNF5H2; and set E for BNF5H3. The BNF5H1-specific primers used here did not yield an intron-containing amplicon, indicating absence of genomic DNA contamination in the template. Note that the primer set A flanking an intron gave a larger amplicon with genomic DNA as shown in Figure 4. Amplicon size is in kb.

In general the stem tissue showed an abundant expression of all three genes, and the pods and seeds showed the least transcript abundance for all three genes. BNF5H3 was expressed nearly uniformly in young and old leaf, root, bud, and flower but much less in pods and seeds. It is interesting that BNF5H2 expression was greater in the seed tissue than in the root, bud, flower, or pod tissue. Also, it showed a high level of expression in mature leaves. BNF5H1 also showed a variation in the expression levels. Thus although all of the tissues that were examined here expressed all three genes, there were differences in the level of expression. This experiment was repeated with different primer pairs that confirmed the expression profile.

F5H Transgenic Lines of B. napus and Their Molecular Genetic and Biochemical Attributes

A total of 15 transgenics with pJOY43 (napin-sense BNF5H1) and six with pJOY44 (napin-antisense BNF5H1) were obtained. HindIII-restricted genomic DNA of these were probed for T-DNA copy number as noted in Figure 7 to find the lines for generating homozygotes (discussed below). The napin promoter is seed specific, and its activity spans the mid-phase of seed development (Joseffson et al., 1987), during which sinapine is synthesized (Vogt et al., 1993).

View larger version (16K): [in this window] [in a new window]   Figure 7.   Diagrammatic representation of the T-DNA region of the A. tumefaciens plasmids used for the genetic transformation of B. napus. E and H, EcoRI and HindIII restriction endonuclease sites, respectively; RB and LB, The right and left borders, respectively, of the T-DNA. 35S, CaMV35S-AMV leader promoter module from Datla et al. (1993); napin, the B. napus napin promoter from Kohno-Murase et al. (1994). NOS-T, Nopaline synthase terminator region; GUS::NPTII, the bifunctional fusion gene containing -glucuronidase and neomycin phosphotransferase II (Datla et al., 1991). These plasmids are derivatives of the pHS723 vector that can replicate in A. tumefaciens (Hirji et al., 1996). Southern hybridization of HindIII-digested genomic DNA with a GUS probe would estimate the number of T-DNA insertions in the transgenics.

The sinapine content of the mature seeds from the primary transgenic lines (T₀ seeds) was assayed on the assumption that perturbation of BNF5H gene expression would show an alteration. Not all of the primary lines could be assayed because of the poor seed yield in some. None of the 10 napin-sense BNF5H1 transgenics showed an enhanced sinapine content, but two of them had much less sinapine than the rest or the control (Fig. 8, see "Primary lines" lot). Four of the five napin-antisense BNF5H1 lines also showed a reduction (Fig. 8, see "Primary lines" lot). The T₀ seeds from those primary transgenic lines with a single-T DNA insertion were retained for deriving homozygotes: T₀ seeds to give T₁ plants and the seeds from these (T₁ seeds) to give T₂ plants and T₂ seeds therefrom. The homozygotes were identified by their non-segregating -glucuronidase phenotype from a screening of approximately 40 seedlings from each plant. The average concentration of extractable sinapine in the two homozygotes of the pHS723 transgenic lines (vector control) was 9.0 ± 0.3 mg g¹ seed. This value was taken as 100% for all of the comparisons. It was surprising that the T₂ homozygous seeds from napin-sense lines also contained less sinapine (Fig. 8, the "Homozygous lines" lot), but the reduction was not as strong as the apparent reduction in the T₀ samples from the tissue culture-derived plants (Fig. 8, the "Primary lines" lot). The single homozygous napin-antisense line also showed a modest reduction.

View larger version (37K): [in this window] [in a new window]   Figure 8.   HPLC analysis of the sinapine content of the seeds from transgenic lines with BNF5H1 (shown in the illustration as F5H) under the control of a napin promoter. Three replications were done for each transgenic line. The alphanumeric designation in the bars of the "Primary lines" lot identify the parents of the corresponding homozygotes in the "Homozygous lines" lot. There are additional homozygotes for which no corresponding primary line data were generated here because of the sample size in the parents. The sinapine content in mg g1 mature seeds for the homozygous samples were as follows: vector control (VC) line 1, 9.1 ± 0.1; line 2, 8.8 ± 0.3; C1, 8.9 ± 0.2; C2, 7.6 ± 0.5; C3, 7.3 ± 0.8; C4, 6.9 ± 0.6; and D1, 7.5 ± 1.1. The average of the homozygous vector control, 9.0 ± 0.3 mg g1 mature seeds, was taken as 100% for the relative values shown.

We examined the impact of constitutive antisense suppression of BNF5H by generating transgenic lines with inverted BNF5H1 under the control of a CaMV35S promoter (pJOY 42). From 22 transgenic lines, 18 were assayed, and these showed various levels of sinapine in the T₀ seeds (Fig. 9, see "Primary lines" lot). As with the other transgenic lines above, homozygotes were derived, and the seeds from five lines were assayed for their sinapine content. It was reduced in all of them by up to 40% in two lines (Fig. 9, see "Homozygous lines" lot).

View larger version (36K): [in this window] [in a new window]   Figure 9.   HPLC analysis of the sinapine content of the seeds from CaMV35S-antisense BNF5H1 (shown in the illustration as 35S anti-F5H) transgenic lines. Three replications were done for each transgenic line. The vector control (VC) is the same as in Figure 8. The sinapine content in mg g1 mature seeds for the homozygous samples was as follows: vector control (VC); line 1, 9.1 ± 0.1; line 2, 8.8 ± 0.3; B1, 7.1 ± 0.8; B2, 6.4 ± 0.6; B3, 5.4 ± 0.5; B4, 5.3 ± 0.3; and B5, 6.5 ± 0.8. The average of the homozygous vector control, 9.0 ± 0.3 mg g1 mature seeds, was taken as 100% for the relative values shown.

The BNF5H expression in the homozygotes was analyzed in northern and western blots. The older stem (bottom 10-15 cm) tissue had been found to have the most abundant expression of the native BNF5H gene (Fig. 5), and this tissue was examined in the 35S antisense-BNF5H1 transgenics (Fig. 10). With the exception of line B3, the transgenic lines did not show any significant reduction in the BNF5H transcript level.

View larger version (71K): [in this window] [in a new window]   Figure 10.   Northern-blot analysis of homozygous (T2) transgenics with a BNF5H probe. Fifteen micrograms of total RNA from the bottom 10-cm portion of stem tissue from each of the 35S antisense-BNF5H1 transgenics or from the seeds of each of the napin sense-BNF5H1 transgenic lines of B. napus was electrophoresed, blotted, and probed; the seeds from the latter were collected over a 20- to 30-d period after flowering in each line. VC, Empty vector transgenic line control; rRNA, ethidium bromide-stained ribosomal RNA. Here and in subsequent figures, the alphanumeric designations B1 through B5 and C1 through C4 denote homozygous transgenic lines (T2), whose lineage has been referred to in Figures 8 and 9.

A similar BNF5H gene expression pattern was also observed when the top 15-cm stem was probed (data not shown). Napin-sense BNF5H1 transgenic seeds showed a very abundant BNF5H transcript level as compared with the vector-alone transformed plants (Fig. 10), as expected for the expression characteristic of this promoter (Joseffson et al., 1987). Thus, one of the 35S antisense lines had a slightly diminished BNF5H transcript level, whereas all of the napin-sense lines had a very high level in the seeds. We then examined the level of BNF5H polypeptide in the 35S antisense BNF5H1 transgenics by immunoblot analysis of crude extracts. The stem tissue of the 35S antisense BNF5H1 transgenic plants (except line B1) showed lower levels of F5H polypeptide than the vector-transformed plants (Fig. 11).

View larger version (34K): [in this window] [in a new window]   Figure 11.   Immunoblot analysis of 35S antisense-BNF5H1 stem extracts. Fifteen micrograms of total protein extracts was boiled, separated on SDS-PAGE, blotted, and probed with rabbit antisera raised against a truncated recombinant BNF5H produced in Escherichia coli. VC, Vector-alone transgenic line. Mature and young stem refer to bottom 10 to 15 cm and top 15 cm of stem, respectively. Equal loading of protein was verified by amido black staining of the membranes.

Immunoblot analyses of the seed extracts from the wild-type or vector-alone transgenic lines did not identify a clearly visible polypeptide corresponding to the one found in the stem tissue, indicating that its presence in the seeds is at the detection limit of this method (data not shown). Although the seeds from the napin-sense lines contained a very high level of BNF5H transcript, there was no detectable increase in the BNF5H polypeptide.

All of the T₂ plants were normal for their vegetative and reproductive phases of growth in a controlled cabinet (data not shown). B. napus leaves contain a sinapoylmalate synthase activity, which is implicated in the synthesis of sinapoyl malate from sinapate-derived precursors (Strack et al., 1990). Thus the 35S antisense F5H lines should show a reduction in the sinapoyl malate content if the BNF5H gene product(s) were involved in the biosynthesis of sinapate. We found a reduction in the sinapoyl malate content of the T₂ plants by up to 25% relative to the vector-alone transgenics (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Multiplicity of the CYP84 Family Members in B. napus and Its Progenitors

We have identified two groups comprising BNF5H1 and -2 in one and -3 in the other. Another very closely related gene in the BNF5H3 group might have escaped detection here. The BNF5H genes identified here add three more members to the CYP84 family, which previously contained the Arabidopsis F5H (Meyer et al., 1996) and subsequently a sweetgum coniferaldehyde hydroxylase (Osakabe et al., 1999). We have further shown that the two groups of the B. napus CYP84 genes are unlikely to have converged from the hybridization of the two diploid parents. Instead, the two groups must have formed prior to the divergence of B. oleracea and B. rapa. The greater divergence between BNF5H3 and either of the other two genes suggests that this duplication might have occurred very early in the lineage of the Brassica spp., whose ancestor remains unknown (Truco et al., 1996).

A large number of genes exist as families in Brassica spp. In some cases the gene multiplicity has been attributed in part to the amphidiploid nature of B. napus (Slocombe et al., 1994). However, the diploids or their ancestor may have undergone a polyploidy event (Kianian and Quiros, 1992; Truco et al., 1996; Anderson and Warwick, 1999). The presence of two groups of BNF5H genes in the diploid Brassica spp. might also be due to such polyploidy. The presumptive genetic basis aside, the closer similarity of BNF5H1 to BNF5H2, despite their genealogical separation, suggests that the genes got "fixed" for whatever functional requirements. That the gene-specific primers designed for B. napus sequences indeed amplified DNA from the cultivars of B. rapa and B. oleracea chosen at random further illustrates this conservation. It will be interesting to study the biochemical attributes of the BNF5H gene products, especially in view of the observation that P450 mono-oxygenases differing very little in their primary structure can exhibit different activities (Wachenfeldt and Johnson, 1995).

Expression Pattern of the CYP84 Members

The observation that the stem tissue had the most abundant steady-state level of BNF5H mRNA resembles the expression profile of F5H in Arabidopsis (Ruegger et al., 1999) and other phenylpropanoid metabolism-related genes (Lee and Douglas, 1996; Bell-Lelong et al., 1997; Mizutani et al., 1997; Hu et al., 1998; Inoue et al., 1998). It is consistent with a role for F5H in lignification. Despite the production of sinapoyl esters in the seeds, the expression of BNF5H here was no greater than that in the roots, pods, buds, or flowers. Thus, there was no evidence for the ester accumulation being supported by an enhanced BNF5H transcription level in the seeds. In contrast we observed a higher level of transcripts in the roots than in the seeds. Canola-associated microbes have been found in the rhizosphere (Siciliano et al., 1998), and hitherto unidentified phenylpropanoid compound(s) in the roots may have a role in these interactions. Whereas similar gene expression pattern with Phe ammonia-lyase (Wanner et al., 1995), cinnamate 4-hydroxylase (Bell-Lelong et al., 1997), 4-coumarate:coenzyme A ligase (Lee et al., 1995), and chalcone synthase (Schmid et al., 1990) might be linked to defense-related synthesis of flavonoids and phytoalexins, perhaps BNF5H gene expression is also pertinent to this.

Against the modest difference in the spatial expression profile of the three BNF5H, it is interesting to note the ubiquitous expression of BNF5H3, a gene that occurs in all three Brassica spp. The presence of two BNF5H3 genes in B. napus, one from each of the two fusion parents, may be a contributing factor. Given the minor differences in the gene sequences, it will be necessary to determine the expression profile of the three promoters transgenically to discern the differences, if any, in the temporal pattern, and to gain further insights into the spatial expression pattern.

Reduction in the Sinapate Ester Content of the Transgenics

In the 35S-antisense BNF5H plants there was a noticeable reduction in the BNF5H polypeptide level and a reduction of the sinapine content by up to 40%. The BNF5H transcript level was affected in only one of the transgenics. This suggests that the antisense BNF5H transcripts did not affect the transcription of the endogenous BNF5H gene or the stability of the transcript as much as the translation of BNF5H protein was affected. Antisense suppression results generally in a concurrent reduction of both of the transcript and the polypeptide (Bourque, 1995). However, there is at least one report where the transgenic plants suffered a substantial reduction in the amount of polypeptide without encountering a similar effect on transcription (Temple et al., 1993). The mechanism of antisense suppression is not completely elucidated; antisense transcripts may hybridize with the endogenous sense transcript, destabilize the formation of the ribosomal complex, and prevent translation (Mol et al., 1994).

The homozygous napin-antisense BNF5H line showed a reduction of only 17%. Obviously, more lines are required to address the impact of seed-specific antisense suppression, but the additional T₀ lines from this study (Fig. 8) are not suitable owing to their multiple transgene copies and genetic segregation (data not shown). The modest reduction of the sinapine content in the napin transgenic lines might be due to variability or suppression of the native BNF5H genes. The transgene transcript was abundant in the sense lines but there was no increase in the polypeptide level. We have not identified the basis for this result.

The generalized pathway for phenylpropanoid metabolism depicts F5H catalyzing the formation of 5-hydoxyferulate, a precursor of sinapate, and sinapate in turn as the precursor for sinapine and for sinapoyl CoA in two bifurcated pathways (Chapple et al., 1992; Meyer et al., 1996, 1998; Ruegger et al., 1999; Fig. 1 in Campbell and Sederoff, 1996). Sinapoyl CoA has been considered as the precursor for sinapyl alcohol, which is then polymerized into syringyl (S) lignin. This inference has, however, been challenged very recently (Chen et al., 1999; Humphreys et al., 1999; Osakabe et al., 1999). Osakabe et al. (1999) demonstrated that the sweetgum CYP84 product carries out the hydroxylation of coniferaldehyde (ConAld) to 5-OH ConAld approximately 140 times more efficiently than that of ferulate to 5-OH ferulate, thus prompting the term ConAld hydroxylase to refer to the enzyme. Further, ConAld non-competitively inhibits ferulate hydroxylation. Humphreys et al. (1999) have simultaneously shown that ConAld (K_m = 1 µM), but not ferulate (K_m = 1 mM), is the preferred substrate for the Arabidopsis F5H, which is 72% identical in its primary structure to the sweetgum ConAld hydroxylase.

What, then, is the basis for our observation that antisense suppression of the CYP84 members in B. napus diminished sinapine synthesis in the seeds? Sinapine is accumulated mostly in the cotyledons in B. napus (Fenwick, 1979). If the enzymatic properties of B. napus F5H are similar to that of the sweetgum ConAld hydroxylase, a supposed absence of ConAld in the non-lignified cotyledons would explain the participation of the B. napus enzyme in ferulate hydroxylation in vivo. Alternatively, as Humphreys et al. (1999) have speculated for Arabidopsis, an aldehyde oxidase might yield sinapate from 5-OH ConAld-derived sinapoyl aldehyde, and thus any constraint on the flux to 5-OH ConAld would also diminish sinapine synthesis. Clearly, a biochemical characterization of the BNF5H gene products is necessary to address the potential functional differences among them.

As Lindberg and Negishi (1989) have shown, a single amino acid substitution in the coumarin 7 hydroxylase, a P450 mono-oxygenase, can expand its substrate specificity to include a steroid. It is interesting that the Arabidopsis F5H but not the sweetgum gene product (both from a yeast expression system) can hydroxylate coniferyl alcohol (Humphreys et al., 1999; Osakabe et al., 1999), suggesting potential differences in the catalytic repertoire of CYP84 enzymes as well. It will also be informative to compare the activity of the CYP84 gene products from various species, and thus the identification of three independent BNF5H genes will be useful in this regard. The apparent minor differences among the BNF5H gene products may have some enzymological implications.

Implications of BNF5H Down-Regulation for Canola Meal Improvement

Canola meal is a significant source of protein supplement in animal feed. Our results show that it is possible to achieve a measure of reduction in the sinapine content by down-regulating BNF5H, thus offering an avenue for improving the meal quality. Besides sinapine reduction, this approach may offer other collateral benefits. Cherney et al. (1991) had found that the brown midrib mutants of maize and sorghum, which are lower in the relative content of syringyl lignins, are more digestible as forage. The digestibility of forage is inversely related to its lignin-derived methoxyl content (Dixon et al., 1996). The syringyl lignin contains two methoxyl groups on each aromatic ring, in contrast to the guaiacyl lignin that contains one. Thus, the feed would be more digestible if it contained less syringyl lignin even if the total lignin content of a feed material, typically guaiacyl + syringyl lignins, remained unaltered. The F5H null mutant of Arabidopsis has very little syringyl lignin. However, it is interesting that cauliflower mosaic virus promoter-directed production of F5H in this mutant restores the syringyl lignin content, and the use of a cinnamate 4-hydroxylase promoter instead renders the lignin almost completely syringyl type (Meyer et al., 1998; Marita et al., 1999). Thus, the down-regulation of the F5H genes in B. napus may have a favorable impact on the lignin composition and meal digestibility, which are yet to be determined.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material and Growth Conditions

Brassica napus cv Westar, Brassica oleracea cv Horizon, and Brassica rapa cv Green Valiant plants were used in this study. The plants were planted in Rediearth (Grace Horticultural Products, Ajax, Canada) potting medium, watered with a fertilizer solution (0.2 g/L N:P:K, 20:20:20; Plant Products Company, Brampton, Canada), and grown in a chamber under 16-h light/25°C and 8-h dark/20°C cycle. The light intensity was 380 µmol m² s¹ photosynthetic photon flux density. The samples were collected, frozen in liquid nitrogen, and stored at 80°C until needed.

Chemicals

Purified sinapate esters used as the reference standards were initially the kind gifts of Drs. D.I. McGregor (Agriculture and Agri-Food, Saskatoon, Canada) and B.E. Ellis and S.X. Wang (University of British Columbia, Vancouver), and subsequently were isolated and purified from canola by Ms. K. Gossen (National Research Council-Plant Biotechnology Institute, Saskatoon, Canada). HPLC grade solvents were purchased from J.T. Baker (Phillipsburg, NJ), and all of the other reagents were of reagent grade from Sigma (St. Louis).

DNA and RNA Methods

The nucleic acid-related enzymes used were from Life Technologies (Burlington, Canada). Oligonucleotide synthesis and DNA nucleotide sequencing were conducted at the National Research Council, DNA Technologies Unit. A B. napus stem cDNA library in Uni-ZAP XR vector (Stratagene) constructed in our laboratory according to the supplier's instruction manual was screened with an Arabidopsis F5H cDNA probe (Meyer et al., 1996; kind gift of Dr. C.C. Chapple, Purdue University, West Lafayette, IN) to isolate the B. napus F5H cDNA clones. The plasmids from the phages were isolated according to the supplier's instruction manual. The clones with insertions >1.5 kb were selected for determining the nucleotide sequence in both directions. LASERGENE Biocomputing Software (DNASTAR, Inc., Madison, WI) was used for sequence analysis.

Genomic DNA was isolated from leaves using a Nucleon PHYTOPURE plant DNA extraction Kit (Amersham, Buckinghamshire, UK). Fifteen micrograms of genomic DNA digested with restriction endonucleases was separated by 1% Tris [tris(hydroxymethyl)aminomethane]-acetate EDTAagarose gel electrophoresis (Sambrook et al., 1989) and transferred onto a GeneScreen Plus membrane (NEN Life Science Products, Boston). The membrane was prehybridized at 42°C for 3 h in 10 mL of prehybridization buffer (Sambrook et al., 1989; 50% [v/v] formamide, 5× Denhardt's reagent, 5× SSC, 0.1% [w/v] SDS, and 100 µg mL¹ denatured salmon sperm DNA). [³²P]dCTP-labeled F5H probe was synthesized from a fragment of BNF5H1 cDNA (nucleotides 256-702) using a rediprime II random primer labeling kit (Amersham). Hybridization with the probe was done overnight in 10 mL of hybridization buffer (prehybridization buffer plus 2% [w/v] dextran sulfate). After hybridization, the membrane was rinsed twice with a solution of 0.2× SSC and 0.1% (w/v) SDS (Sambrook et al., 1989), washed twice with 0.2× SSC and 0.1% (w/v) SDS for 15 min at 50°C, and once with 0.1× SSC and 0.1% (w/v) SDS for 30 min at 65°C. It was then exposed to an x-ray film (Fuji Photo Film, Tokyo) for 1 to 5 d with Quanta III intensifying screens (DuPont, Wilmington, DE).

Total RNA from various tissues was isolated using TRIzol reagent (Life Technologies) according to the supplier's protocol. Northern-blot analysis was done with 15 µg of total RNA that was electrophoretically separated and transferred to a GeneScreen Plus membrane (Sambrook et al., 1989). The BNF5H probe was prepared as above.

PCR Analysis

The introns in the genomic segments were amplified with the following primer pairs: pair 1, for BNF5H1 and BNF5H2, 5'-CGAGTCATGGGCTTCTGTT-3' and 5'-TATC-GCTGACGCTACCGTTCC-3'; pair 2, for BNF5H3, 5'-AGATGAGGAAAGTGTGT-3' and 5'-CCGTAATAACTCCGTT-AAG-3'. The first primer in each of these pairs and in the primer sets outlined below is based on the sense strand, and the second primer is derived from the non-coding strand of the cDNA clone sequences deposited in GenBank (see "Results"). PCR was set up in a 50-µL volume that contained 250 ng of genomic DNA as template, 1× PCR buffer (Life Technologies), 1.5 mM MgCl₂, 200 µM each dNTP, and 50 pmol of each gene-specific primer. The samples were heated at 94°C for 3 min to denature the template, then cooled to 72°C, and immediately 2.5 units of Taq DNA polymerase was added. The amplification was conducted for 30 cycles in a DNA thermal cycler (Perkin-Elmer Applied Biosystems, Foster City, CA) at a setting of 94°C for 45 s, 56°C for 1 min, and 72°C for 2 min for each cycle and a final extension step at 72°C for 10 min. The positive controls contained as the template 20 ng of BNF5H1, -2, or -3 cDNA clones, and the negative control contained only the primers in question.

Poly(A⁺) mRNA was isolated from 5 mg of total RNA from various tissues using an Oligotex mRNA midi kit (Qiagen, Mississauga, Canada). Two micrograms of the eluted mRNA sample was used for the first strand cDNA synthesis by Superscript RNase H Reverse Transcriptase (Life Technologies) following the manufacturer's protocol. The single-stranded cDNA was dissolved in 100 µL of Tris-EDTA buffer, the PCR was set up with 2.5 µL of cDNA template as described above, and amplification was conducted for 25 cycles.

RT-PCR with total RNA isolated from various tissues was done as follows. First strand cDNA was synthesized from 10 µg of total RNA by Superscript RNase H Reverse Transcriptase (Life Technologies). The reaction was stopped by incubating at 68°C for 10 min, and the total volume made up to 50 µL with deionized water. PCR was set up in a 50-µL reaction volume containing 1 µL of the above cDNA preparation as the template, 1× PCR buffer (Life Technologies), 1.4 mM MgCl₂, 200 µM each dNTP, 50 pmol of each of the two gene-specific primers for BNF5H1, -2, or -3, and 2.5 units of platinum Taq DNA polymerase (Life Technologies). After an initial denaturation of DNA for 2 min, amplification was conducted for 15 cycles as described above except that the annealing temperature was at 61°C for 1 min. The positive controls contained 20 pg of relevant BNF5H cDNA as the template, and the negative control contained the primers but no template. After agarose gel electrophoresis, Southern-blot hybridization was done with a probe derived from BNF5H2 (nucleotides 1,310-1,695) that hybridizes to all three BNF5H amplification products.

Specific primer sets for each of the three BNF5H genes were used: set A and B for BNF5H1; set C and D for BNF5H2, and set E and F for BNF5H3. The nucleotide sequence of these sets are as following: set A [5'-TTCTCGAACCGACCAGCT-3' and 5'-ACAAATAAGGCGCGTGCT-3']; set B [5'-GAGTGGGCATTGACTGAATTAC-3' and 5'-CCATACCAACCACCTTTCC-3']; set C [5'-GGGCATTGACTGAGTTAT-3' and 5'-CCACAGCTCAAGAACCATC-3']; set D [5'-CGCGACAAGAACTCTTGGGTTGAT-3' and 5'-CCACGTT-CAAGAACCATCAACC-3']; set E [5'-ACCCTAAAT-CTTGGCCTGACG-3' and 5'-CCATACCTACTACCAAC-CCTTCG-3']; set F [5'-TTAGAACCGGGAGTAGC-3' and 5'-CCTATTACCAACCCTTCG-3'].

Genetic Transformation of Plants

The plasmids for Agrobacterium tumefaciens-mediated transformation were constructed by inserting a 1.8-kb EcoRI fragment of the BNF5H1 cDNA (this fragment lacks the poly[A] region) in the sense or antisense orientation (Fig. 7). All of the plasmids are based on the binary vector pHS723 constructed in this laboratory (Hirji et al., 1996). A. tumefaciens GV3101 [pMP90] (Koncz and Schell, 1986) derivatives with the pJOY plasmids were obtained by the freeze-thaw method described by An et al. (1988). Genetic transformation of B. napus hypocotyl explants and regeneration of plants were according to Moloney et al. (1989). Putative transgenics were selected by kanamycin resistance and confirmed as transgenics by 5-bromo-4-chloro-3-indolyl--glucuronic acid-based -glucuronidase (GUS) assay (Jefferson et al., 1986) afforded by the bifunctional selection marker (GUS::neomycin phosphotransferase II; Datla et al., 1991) within the T-DNA (Fig. 7). The primary lines with a single-copy of the transgene were identified following Southern blot of HindIII-restricted DNA and probing with a GUS gene fragment. These were selfed to obtain T₀ seeds. Ten T₁ plants were selfed to obtain T₂ seeds. Seedlings from T₂ seeds were screened by the GUS assay to identify homozygous T₂ seed lots.

Immunoblot Analysis

Rabbit antisera were raised against a recombinant BNF5H1 protein, lacking the first 170 amino acids of the deduced nucleotide sequence from the BNF5H1 ORF, produced in Escherichia coli BL21(DE3) pLysS cells carrying a derivative of the plasmid pRSET (Invitrogen, Carlsbad, CA). The polypeptide was purified according to the supplier's instructions prior to immunizing the rabbits. One gram of pulverized B. napus stem tissue was ground with 1.5 mL of buffer (50 mM HEPES [4-(2-hydroxyethyl)-1-piperazine-ethanesulfonic acid], pH 7.5, 1 mM EDTA, pH 7.5, and 10% [v/v] glycerol). The extract was centrifuged at 6,000g for 30 min at 4°C, and the supernatant was assayed for protein concentration by the modified Bradford method (Bio-Rad, Hercules, CA). Fifteen micrograms of protein-equivalent of the extract in the gel loading solution (50 mM Tris, pH 6.8, 2.3% [w/v] SDS, 5% [v/v] -mercaptoethanol, and 10% [v/v] glycerol) was boiled for 10 min and separated in SDS-PAGE (10% acrylamide, w/v) in duplicate gels. The polypeptides were electroblotted onto a polyvinylidene difluoride membrane (Hybond-P, Amersham), and the membrane was probed with the BNF5H antisera, followed by horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham), and chemiluminescent detection with a ECL+Plus western-blotting system (Amersham) according to the supplier's instructions. The duplicate membrane and also the immunoprobed membrane after chemiluminescent detection were stained with 0.1% (w/v) amido black in a 45% (v/v) methanol-5% (v/v) acetic acid solution to verify equal loading and transfer of protein.

HPLC Analysis of Phenolics

Ten seeds, including the seed coat, or 10 mg of freeze-dried leaves from 21-d-old seedlings were ground in 500 µL of 80% (v/v) methanol, incubated at 4°C for 1 h, and frozen at 80°C for 1 h. The extract was thawed and centrifuged at 20,000g for 15 min to obtain the supernatant. The pellet was re-extracted with 500 µL of 80% (v/v) methanol at 4°C for 1 h and centrifuged as above. The two supernatants were pooled, and 20 µL of the pooled extract was run on an HPLC column (Nucleosil C18 AB, Alltech, Deerfield, IL) using an acetonitrile/phosphoric acid (1.5% [v/v]) gradient of 10% to 30% over a 35-min period in an HPLC instrument (Vista 5500, Varian, Palo Alto, CA). UV A₃₃₀ was detected with a Polychrom (model 9065, Varian). The average of sinapine and sinapoyl malate contents per milligram of vector alone transformed transgenic seeds or leaves was calculated as 100%. All of the other data were calculated as percentage sinapine or sinapoyl malate content of vector control plants. The SE was calculated from three replications for each transgenic plant.