*For correspondence (fax +1 765 494 7897; e-mail email@example.com). †Co-first authors: R.F. and C.M.M. contributed equally to this research. ‡Present address: DuPont Central Research and Development, Biochemical Sciences and Engineering, Experimental Station PO Box 80328, Wilmington, DE 19880-0328, USA.
Ferulate 5-hydroxylase (F5H) is a cytochrome P450-dependent monooxygenase that catalyses the hydroxylation of ferulic acid, coniferaldehyde and coniferyl alcohol in the pathways leading to sinapic acid and syringyl lignin biosynthesis. Earlier studies in Arabidopsis have demonstrated that F5H over-expression increases lignin syringyl monomer content and abolishes the tissue-specificity of its deposition. To determine whether this enzyme has a similar regulatory role in plants that undergo secondary growth, we over-expressed the F5H gene in tobacco and poplar. In tobacco, over-expression of F5H under the control of the cauliflower mosaic virus 35S promoter increased lignin syringyl monomer content in petioles, but had no detectable effect on lignification in stems. By contrast, when the cinnamate 4-hydroxylase (C4H) promoter was used to drive F5H expression, there was a significant increase in stem lignin syringyl monomer content. Yields of thioglycolic acid and Klason lignin in C4H–F5H lines were lower than in the wild-type, suggesting that F5H over-expression leads to a reduced deposition or an altered extractability of lignin in the transgenic plants. Histochemical analysis suggested that the novel lignin in C4H–F5H transgenic lines was altered in its content of hydroxycinnamyl aldehydes. Transgenic poplar trees carrying the C4H–F5H transgene also displayed enhanced lignin syringyl monomer content. Taken together, these data show that hydroxylation of guaiacyl-substituted lignin precursors controls lignin monomer composition in woody plants, and that F5H over-expression is a viable metabolic engineering strategy for modifying lignin biosynthesis in forest species.
Lignin is thought to be essential for the existence of plants in a terrestrial environment because it provides mechanical support to tracheary element cell walls enabling them to resist the tension generated during transpiration. Unfortunately, the polymer has a number of properties that interfere with the use of lignified plant materials. For example, the removal of lignin during the pulping process is expensive and potentially environmentally hazardous. As a result, metabolic engineering of the lignin biosynthetic pathway has been suggested as a method for producing genetically modified trees with superior pulping characteristics that would be of value to the forestry industry ( Campbell & Sederoff 1996; Dean & Eriksson 1992; Whetten & Sederoff 1991). In addition, the lignin polymer is intimately associated with the cell wall polysaccharides of forages. Lignin interferes with the digestion of these carbohydrates by limiting their availability to hydrolytic enzymes and thereby decreases the nutritional value of plant tissues ( Jung & Deetz 1993). Consequently, the biotechnological manipulation of lignin content and monomer composition is of significant commercial interest.
A number of genes and their encoded enzymes have been targeted in efforts aimed at altering lignin quantity or quality in transgenic plants (reviewed in Baucher et al. 1998 ; Whetten et al. 1998 ). Many of these experiments have focused on sense and antisense suppression of caffeic acid/5-hydroxyferulic acid O-methyltransferase (COMT) expression. This approach has met with variable results, probably owing to the degree of COMT suppression achieved in each study. The most dramatic effects were seen in experiments using homologous COMT constructs in tobacco ( Atanassova et al. 1995 ) and poplar ( Lapierre et al. 1999 ; Van Doorsselaere et al. 1995 ). These studies revealed that down-regulation of COMT expression led to a decrease in lignin syringyl monomer content and a concomitant appearance of 5-hydroxyguaiacyl residues in the transgenic lignin. These manipulations also resulted in the generation of a brown midrib phenotype reminiscent of maize bm3 mutants in which the endogenous COMT locus is defective ( Vignols et al. 1995 ). Antisense suppression of cinnamyl alcohol dehydrogenase (CAD) activity in transgenic tobacco and poplar also generated a pigmented lignin which exhibited an increased aldehyde content, and was more easily extracted than the lignin in control plants ( Baucher et al. 1996 ; Halpin et al. 1994 ). An orange–brown lignin, distinct from that in CAD and COMT antisense transgenic plants, was observed in tobacco plants carrying cinnamoyl CoA reductase (CCR) antisense constructs ( Piquemal et al. 1998 ). The CCR antisense plants also deposited less lignin than wild-type, a phenotype correlated with the collapse of tracheary elements. The reduction in lignin in these plants preferentially affected guaiacyl monomers; consequently, the lignin in these plants also had a higher syringyl to guaiacyl monomer ratio. Suppression of 4-hydroxycinnamoyl CoA ligase (4CL) expression has resulted in phenotypes ranging from modest changes in lignin monomer composition to substantial decreases in total lignin deposition ( Hu et al. 1999 ; Kajita et al. 1997 ; Lee et al. 1997 ). These studies all demonstrated the feasiblity of using metabolic engineering strategies to modify lignin quantity and quality, and showed that lignin biosynthesis is much more plastic than previously had been thought.
For many years, the phenylpropanoid pathway was considered a branched but linear pathway ( Lewis & Yamamoto 1990). According to this model, the phenylpropane skeleton of phenylalanine is converted to hydroxycinnamic acids, which serve as precursors for flavonoids, lignin and other hydroxycinnamic acid derivatives. During the last decade, a different route for the biosynthesis of lignin monomers has received attention ( Kneusel et al. 1989 ; Kühnl et al. 1989 ; Pakusch et al. 1989 ; Pakusch et al. 1991 ; Schmitt et al. 1991 ; Ye 1997; Ye & Varner 1995; Ye et al. 1994 ; Zhong et al. 1998 ). This so-called ‘alternative pathway’ involves the activation of p-coumaric acid to its coenzyme A thioester, followed by hydroxylation and methylation reactions that generate feruloyl CoA. Considering that ferulic acid can also be synthesized by the free acid pathway and can be activated to its CoA thioester by 4CL, lignin monomer biosynthesis probably occurs via a cross-linked network of pathways. Indeed, the continued accumulation of guaiacyl lignin in COMT-suppressed plants ( Atanassova et al. 1995 ; Van Doorsselaere et al. 1995 ) suggested that the alternative pathway is a major contributor to lignin biosynthesis in woody plants. Support for this hypothesis has been provided by the finding that tobacco plants down- regulated in caffeoyl CoA O-methyltransferase (CCoAOMT) activity have a lower total lignin content than wild-type ( Zhong et al. 1998 ).
None of the lignin metabolic engineering strategies described above led to a substantial increase in the syringyl content of lignin, a trait that is correlated with improved chemical degradability of lignocellulosic material ( Chiang & Funaoka 1990; Chiang et al. 1988 ). We have recently demonstrated that ferulate 5-hydroxylase (F5H) over-expression in Arabidopsis abolishes the tissue specificity of syringyl lignin accumulation and results in the deposition of a lignin that is derived almost entirely from syringyl monomers ( Meyer et al. 1998 ). Although these data demonstrate that F5H is a regulatory site for lignin biosynthesis in Arabidopsis, woody plants may have alternative mechanisms for the regulation of phenylpropanoid pathway flux. For F5H over-expression to be a useful approach for the modification of lignin in tree species, it would be necessary to demonstrate the efficacy of this metabolic engineering strategy in plants that undergo secondary growth. Because recent studies have shown the utility of tobacco and poplar as model woody plants for experiments in the genetic engineering of lignin, we chose to investigate the impact of F5H over-expression in these species.
Expression of the Arabidopsis F5H gene in transgenic tobacco
To over-express F5H in tobacco we employed two chimeric constructs previously used to evaluate the impact of F5H over-expression on syringyl lignin and sinapate ester biosynthesis in Arabidopsis ( Meyer et al. 1998 ; Ruegger et al. 1999 ). In these constructs, the regulatory sequences of the Arabidopsis F5H gene have been replaced with either the constitutive cauliflower mosaic virus 35S promoter (35S–F5H) or the Arabidopsis cinnamate 4-hydroxylase (C4H) promoter (C4H–F5H). At least 40 independent transformants were generated for each of the above constructs by Agrobacterium-mediated transformation of tobacco leaf disks. To identify a series of plants with a range of transgene expression levels, F5H mRNA levels in each of the kanamycin-resistant lines were analysed by RNA gel blot hybridization. Figure 1 shows the steady-state levels of F5H mRNA in the leaves of 11 selected representative independent primary transformants. Under the hybridization conditions employed, mRNA complementary to the F5H cDNA from Arabidopsis was undetectable in wild-type tobacco leaves, indicating a low level of F5H expression in the leaves of untransformed plants and/or limited nucleotide homology between the tobacco and Arabidopsis F5H genes. By contrast, high levels of F5H transcript were detected in 35S–F5H transgenic tobacco plants (A–D), consistent with the expected constitutive expression provided by this promoter. F5H mRNA was also detected in leaves of most of the C4H–F5H transformants (G–K); however, the levels were substantially less than in the 35S–F5H lines (note the difference in exposure time for the two blots in Fig. 1). In some transformants, levels of F5H mRNA were relatively low (F) or even undetectable (E). Line E was included for characterization at the biochemical level as a transformation control. Despite the large differences in F5H expression in the various lines, no obvious differences were observed in the transformed plants when compared to wild-type (data not shown).
Impact of 35S promoter-driven F5H expression on lignin in petioles and stems
Nitrobenzene oxidation (NBO) releases substituted benzaldehyde and benzoic acid derivatives from lignin, and the relative content of these monomers is indicative of the presence of guaiacyl (vanillin and vanillic acid) and syringyl (syringaldehyde and syringic acid) units in the polymer. To survey the effects of F5H over-expression in primary transformants, initial NBO analyses were conducted on xylem from petioles of T0 plants. These experiments revealed that wild-type petiole lignin has a syringyl monomer content of 14 mol%, consistent with the low levels of syringyl units often found in primary xylem ( Venverloo 1971) ( Table 1). In 35S–F5H transgenics, lignin syringyl monomer content was increased to between 22 and 40 mol%. These data are consistent with those obtained using the 35S–F5H construct in Arabidopsis ( Meyer et al. 1998 ). They indicate that the 35S–F5H construct can increase lignin syringyl monomer content, at least in tissues that normally deposit lignin dominated by guaiacyl units.
Table 1. . Impact of 35S promoter-driven F5H expression on lignin monomer composition in tobacco petioles and stems as determined by nitrobenzene oxidation
Line designations refer to those shown in Fig. 1. Data represent the mean of six replicates ± SD.
Sum of vanillin + vanillic acid;
sum of syringaldehyde + syringic acid.
70.0 ± 13
11.5 ± 2.0
81.5 ± 15
14.3 ± 1.09
64.0 ± 16
35.5 ± 9.5
99.5 ± 22
35.7 ± 6.06
56.5 ± 5.5
26.0 ± 2.5
82.5 ± 8.0
31.3 ± 0.50
43.0 ± 8.0
12.0 ± 1.5
55.5 ± 10
22.4 ± 1.53
32.5 ± 8.5
21.5 ± 5.5
54.5 ± 14
40.0 ± 1.86
277 ± 32
270 ± 30
545 ± 54
49.3 ± 2.80
289 ± 19
314 ± 33
605 ± 50
52.0 ± 1.67
203 ± 16
213 ± 18
416 ± 30
51.2 ± 1.76
214 ± 18
258 ± 18
473 ± 29
54.7 ± 2.20
290 ± 20
229 ± 15
520 ± 35
44.2 ± 0.15
To evaluate the effect of the 35S–F5H transgene on lignin monomer composition in secondary xylem, stem sections from 4-month-old T0 plants were subjected to NBO analysis. The syringyl monomer content of wild-type tobacco stem lignin was found to be 50 mol% ( Table 1). Surprisingly, despite high levels of F5H mRNA in 35S–F5H plants, syringyl monomer content was unaffected. These data indicated that expression of F5H under the control of the 35S promoter is not sufficient to increase syringyl monomer content in secondary xylem.
Impact of C4H promoter-driven F5H expression on lignin in stems
Previous work in Arabidopsis demonstrated that over-expression of F5H under the control of the 35S promoter could increase lignin syringyl monomer content to no more than 30 mol%, in comparison to approximately 20 mol% in wild-type. By contrast, the C4H–F5H transgene led to the deposition of a more highly syringyl-enriched lignin ( Meyer et al. 1998 ). Considering that wild-type tobacco stem lignin contains relatively high levels of syringyl monomers (approximately 50 mol%), modification of lignin monomer composition in this tissue may require the more efficacious C4H–F5H construct. NBO analysis of stem tissue from the C4H–F5H transgenic lines G–K revealed that their lignin syringyl monomer content was as high as 84 mol%, which exceeded that found in both wild-type plants and 35S–F5H transgenics ( Table 2). In the C4H–F5H lines, although lignin content as measured by total NBO yield varied according to the degree of secondary growth in the samples tested, lignin syringyl monomer content was positively correlated with the level of transgene F5H mRNA abundance measured in leaves. C4H–F5H lines expressing little or none of the F5H transcript (lines F and E, respectively) showed either no change or only a small increase in lignin syringyl monomer content compared to wild-type. No consistent effect on total lignin content was observed by NBO.
Table 2. . Impact of C4H promoter-driven F5H expression on lignin monomer composition in tobacco stems as determined by nitrobenzene oxidation
Line designations refer to those shown in Fig. 1. Data represent the mean of six replicates ± SD.
Sum of vanillin and vanillic acid;
b sum of syringaldehyde and syringic acid.
310 ± 26
321 ± 22
630 ± 45
50.1 ± 1.40
171 ± 58
152 ± 60
314 ± 117
48.1 ± 1.67
219 ± 39
384 ± 73
605 ± 109
63.7 ± 1.99
112 ± 19
289 ± 58
401 ± 36
71.9 ± 1.35
114 ± 20
442 ± 89
555 ± 109
79.4 ± 0.57
123 ± 8.5
484 ± 91
605 ± 99
79.6 ± 1.91
154 ± 17
560 ± 81
715 ± 94
78.4 ± 1.64
76 ± 8.5
408 ± 61
485 ± 69
84.2 ± 0.76
Analysis of lignin monomer composition using the DFRC method
Because NBO analysis of lignin does not distinguish between monomers derived from lignin depolymerization and those arising from other wall-bound phenolics, a second method for lignin monomer analysis was employed. The recently developed DFRC (derivatization followed by reductive cleavage) method for lignin analysis specifically targets β-O-4 alkyl-aryl-ether linkages, the most frequent inter-unit linkage found in native lignin ( Lu & Ralph 1997). By selectively cleaving these ether bonds and simultaneously maintaining the C6–C3 phenylpropane structure, this analytical procedure gives rise to characteristic coniferyl and sinapyl alcohol diacetates in quantities that accurately reflect the amount of guaiacyl and syringyl units in the polymer. When applied to wild-type tobacco stem samples, DFRC analysis revealed a lignin syringyl monomer content of 20–40 mol% ( Fig. 2). Similar levels of syringyl residues were found in lignin from stems of the 35S–F5H transgenic plants (lines C and D) and the C4H–F5H transgenic line that does not express the Arabidopsis F5H gene (line E). Consistent with the previous NBO data, DFRC analysis of the C4H–F5H lines G, J and K demonstrates that the lignin in the stems of these plants is comprised of approximately 80 mol% syringyl residues.
Histochemical analysis of lignin in wild-type and transgenic tobacco stems
To investigate potential changes in anatomy and lignin chemistry accompanying F5H over-expression, transverse sections were prepared from petioles and stems of the wild-type and transgenic lines D (35S–F5H) and K (C4H–F5H). In unstained sections, no differences were observed in cell anatomy or xylem pigmentation when the transgenic stem sections were compared to wild-type ( Fig. 3). Examination of unstained sections at higher magnification showed no obvious perturbation of xylem development, wall shape or thickness in F5H over-expressing transgenic tobacco plants (data not shown). To assess changes in lignin chemistry, stem cross-sections were stained with toluidine blue O (TBO), the Wiesner reagent (phloroglucinol/HCl) and the Mäule reagent. TBO is a metachromatic stain that imparts a turquoise colour to lignified cell walls and stains non-lignified cell walls purple. No staining differences were detected using the TBO staining procedure, indicating that F5H over-expression does not lead to large differences in the amount or spatial distribution of lignin ( Fig. 3).
Phloroglucinol stains lignified cells red upon reaction with hydroxycinnamaldehyde groups present in the polymer ( Fig. 3) ( Clifford 1974). The simplicity of this staining procedure has led to its traditional use as a general indicator of cell wall lignification. Using phloroglucinol, the xylem in both the wild-type and the 35S–F5H transgenic stained strongly. In contrast, staining of the C4H–F5H stem sections gave only a weak positive reaction that was restricted to tracheids in the primary xylem and isolated vessel elements in the secondary xylem. Staining is almost undetectable throughout the bulk of the secondary xylem. Treatment of the sections with the reducing agent sodium borohydride prior to phloroglucinol staining ( Fig. 3) eliminates the strong staining in wild-type and 35S–F5H cross-sections and the residual staining seen in the C4H–F5H samples, confirming the aldehyde specificity of phloroglucinol staining. Based upon the results of the NBO and DFRC analyses, as well as spectroscopic and gravimetric measurements of lignin content (see below), the lack of phloroglucinol staining is not due to a significant decrease in lignin content in the C4H–F5H lines. The fact that staining of the C4H–F5H sections before reduction is similar to the staining of the wild-type and 35S–F5H sections after reduction indicates that very few aldehyde groups are present in the lignin of C4H–F5H transgenic tobacco plants.
The Mäule reagent is a histochemical stain that permits the qualitative evaluation of lignin monomer composition in various cell types. In wild-type tobacco petioles, protoxylem elements stain yellow, indicating the presence of guaiacyl lignin in their cell walls. In contrast, metaxylem elements stain red, indicating the presence of at least some syringyl units in their walls ( Fig. 4). This developmental regulation of syringyl unit deposition appears to be partially abolished in the 35S–F5H transgenics because many of the protoxylem elements stain red. The increase in overall staining intensity in the 35S–F5H petioles is consistent with the elevated lignin syringyl monomer content measured by NBO. In wild-type stem cross-sections, the Mäule reagent gives a positive reaction with most xylem cell walls, indicating the presence of syringyl units. Protoxylem and immature secondary xylem stain with TBO and phloroglucinol but give a negative reaction with the Mäule reagent, indicating that the lignin in their cell walls has a low content of syringyl monomers. This developmentally regulated deposition of syringyl residues is perturbed in the F5H transgenics. Syringyl units are detected in the protoxylem of the 35S–F5H transformants despite the fact that this transgene has little or no effect on the overall lignin monomer composition of secondary xylem. In contrast, the Mäule reagent detects syringyl residues in all petiole and stem xylem cells in C4H–F5H transgenics ( Figs 3 and 4). Phloem cap fibres are also more intensely stained than corresponding cells in wild-type and 35S–F5H transgenic stem cross-sections. Although no differences were seen between the overall Mäule staining intensity of wild-type and 35S–F5H stem cross-sections, C4H–F5H sections exhibited a more intense staining, consistent with the increased syringyl monomer content in these tissues as determined by NBO and DFRC analysis.
Effect of F5H over-expression on lignin quantity
To determine whether the alteration of lignin monomer composition in the C4H–F5H tobacco was accompanied by changes in lignin quantity, lignin content was measured in cell wall residues obtained by solvent extraction of ground freeze-dried stem tissue. The solvent extraction removes soluble compounds that might interfere with spectrophotometric lignin quantification. The proportion of ethanol-soluble compounds in stems was found to be similar in all plants (30–40% of the freeze-dried material), and the remaining cell wall residue was found to account for 5–6% of the fresh weight.
As no single technique is ideal for the determination of lignin content, two independent methods were used: the spectroscopic thioglycolic acid (TGA) derivatization technique ( Campbell & Ellis 1992) and the gravimetric Klason procedure ( Kaar et al. 1991 ). TGA lignin determination ( Fig. 5a) indicated the presence of comparable amounts of lignin within the wild-type samples assayed. Similar lignin contents were also measured in the cell walls of 35S–F5H transgenic plants (lines C and D) and the C4H–F5H transgenic line with undetectable F5H expression (line E). By contrast, lignin content as measured by the TGA method was significantly reduced in C4H–F5H transgenic lines G, J and K. These data indicate that F5H over-expression led to either a reduction in total lignin content or an altered extractability of the lignin polymer in these transgenic lines.
Similar results were obtained using the Klason method ( Fig. 5b). Lignin content was determined to be approximately 16–17% of the dry cell wall mass in wild-type plants, the 35S–F5H transgenic lines (C and D) and the C4H–F5H line with no detectable F5H transgene expression (E). The Klason lignin content of the other C4H–F5H transgenic lines (G, J and K) is only approximately 12–13% of the cell wall dry mass, consistent with the results derived from the TGA method.
Tissue-specific expression of a C4H–GUS fusion in transgenic tobacco
In agreement with previous work in Arabidopsis ( Meyer et al. 1998 ), the data described above indicate that the C4H–F5H construct is very effective in modifying lignin monomer composition. These findings suggest that the C4H promoter must drive high levels of F5H expression in cells that are lignifying or in cells that provide monolignols to adjacent lignifying cells. To evaluate the tissue specificity of C4H expression, and thus identify those cells likely to be generating monomers for lignin synthesis, we transformed tobacco plants with a C4H promoter–β-glucuronidase (GUS) fusion ( Bell-Lelong et al. 1997 ).
Histochemical staining for GUS activity in cross-sections of C4H–GUS transgenic tobacco stems ( Fig. 6) showed that GUS activity is relatively high in the xylem. By contrast, only modest expression of GUS could be detected in other cells, indicating that the C4H promoter is most active in lignifying tissues. A more detailed inspection revealed that within the xylem, the highest levels of C4H promoter-driven GUS expression occurred in cells of the ray parenchyma, adjacent to tracheary elements that were already at least partially lignified ( Fig. 6). GUS activity was not detected in older mature tracheary elements since these cells are dead and have lost their cytoplasm. GUS expression could also be detected in differentiating vessel elements ( Fig. 6, arrow); however, these were surprisingly few in number. Similar results were obtained when petioles of C4H–GUS transformants were stained and sectioned ( Fig. 6).
Modification of lignin in transgenic poplar
To determine whether high-syringyl lignins could be produced in a tree species, we generated over 40 transgenic C4H–F5H poplar lines by Agrobacterium-mediated transformation. The effect of the F5H transgene was assayed by NBO, and the results of a representative sample of nine lines are shown in Table 3. Transformants were recovered that had lignin syringyl monomer contents ranging from 55 mol% in the control lines to as high as 85 mol% in line I. Stem tissue samples from this subset of lines was also subjected to DFRC analysis to ensure that the products of the NBO analysis accurately represented the proportion of monomers found in the lignin polymer. The DFRC data were completely consistent with the NBO data, indicating that the C4H–F5H transgene leads to the deposition of high-syringyl lignins in tree species ( Fig. 7).
Table 3. . Impact of C4H promoter-driven F5H expression on lignin monomer composition in poplar stems as determined by nitrobenzene oxidation
Line designations refer to those shown in Fig. 1. Data represent the mean of at least three replicates ± SD.
Sum of vanillin and vanillic acid;
b sum of syringaldehyde and syringic acid.
172 ± 54
248 ± 68
420 ± 120
59.3 ± 3.58
97 ± 27
128 ± 52
225 ± 79
55.5 ± 5.50
71 ± 16
141 ± 41
212 ± 57
66.2 ± 2.18
76 ± 5.2
160 ± 15
236 ± 20
67.8 ± 1.15
88 ± 36
215 ± 118
303 ± 154
69.5 ± 3.27
50 ± 11
125 ± 31
175 ± 41
71.0 ± 1.51
32 ± 4.2
114 ± 16
146 ± 18
77.7 ± 2.09
35 ± 14
148 ± 93
183 ± 108
80.1 ± 2.24
38 ± 18
167 ± 83
205 ± 100
81.1 ± 1.86
37 ± 25
206 ± 96
243 ± 118
85.2 ± 3.28
Generation of a high-syringyl lignin in transgenic tobacco and poplar plants
The biochemical and histochemical results presented here demonstrate that lignin monomer composition in xylem of tobacco and poplar is regulated at the level of F5H expression. They also indicate that the Arabidopsis F5H can be functionally expressed in these plants and that this strategy can be used to alter the deposition of syringyl lignin in plants that undergo secondary growth. By comparing the effects of the 35S–F5H and C4H–F5H constructs, it is clear that the use of a lignification-associated promoter contributes substantially to the efficacy of this metabolic engineering strategy. At the same time, we cannot exclude the possibility that other constitutive promoters, such as the double 35S promoter which has been used with success in other lignin metabolic engineering strategies ( Hu et al. 1999 ; Piquemal et al. 1998 ), might also have utility in F5H over-expression.
Two independent methods for lignin monomer determination revealed that the C4H–F5H transgenic plants deposit lignin with a syringyl monomer content as high as 85 mol%. Despite this substantial redirection of carbon flow in the lignin biosynthetic pathway, the total amount of uncondensed lignin subunits in F5H over-expressing plants as measured by NBO ( Table 1) or DFRC analysis (data not shown) appears to be unaffected. This suggests that total carbon flux through the phenylpropanoid pathway is not affected by F5H over-expression, and that F5H is not rate-limiting for overall lignin biosynthesis. Since collapsed vessel elements such as those seen in CCR and CCoAOMT antisense plants ( Piquemal et al. 1998 ; Zhong et al. 1998 ) were not observed in stem cross-sections of F5H over-expressing plants, it appears that the deposition of syringyl-rich lignin does not compromise tracheary element function. Consequently, F5H over-expression appears to represent an ideal metabolic engineering strategy for the modification of lignin in forest species used in the pulp and paper industry.
These data also emphasize that F5H is the key branch-point enzyme determining the fate of guaiacyl-substituted lignin precursors in the pathway. Until recently, the efficacy of F5H over-expression was difficult to reconcile with the observation that down-regulation of CCoAOMT expression leads to a decrease in lignin deposition in tobacco. These data suggest that the alternative pathway provides a quantitatively important input of guaiacyl units that should escape hydroxylation by F5H. The finding that F5H is a multi-functional P450 that can also hydroxylate coniferaldehyde and coniferyl alcohol now places F5H downstream of the alternative pathway ( Humphreys et al. 1999 ; Osakabe et al. 1999 ). Thus, there are as many as three possible routes by which F5H could redirect guaiacyl precursors into syringyl lignin. Considering that F5H exhibits a much lower Km towards coniferaldehyde and coniferyl alcohol than it does towards ferulate, and that sinapoyl CoA ligase activity is not found in most plants, it is very likely that F5H over-expression results in the redirection of carbon flux through the two later routes to syringyl lignin.
Modification of lignin extractability in F5H over-expressing plants
The analysis of total lignin content by TGA derivatization or the Klason method appears to indicate that the lignin content in C4H–F5H over-expressing lines is reduced compared to wild-type and the 35S–F5H transgenic lines ( Fig. 4). The interpretation of these data is complicated by the observation that solubilized TGA–lignin complexes from C4H–F5H lines with high lignin syringyl monomer content exhibit different absorbance spectra compared to those of the wild-type and 35S–F5H transgenic lines (data not shown). The altered UV spectrum in the region of the 280 nm wavelength used for TGA lignin quantification suggests that lignin with a high syringyl monomer content may lead to artefactual under-estimation of lignin content by the TGA method. These differences in TGA results could be caused by a difference in the UV spectra of the TGA–syringyl lignin complexes, or by a failure to precipitate these complexes during sample preparation. The lower apparent Klason lignin content could also be explained by the differences in lignin monomer composition among the various lines. Similar results were obtained in Arabidopsis where TGA lignin analysis revealed a strongly reduced TGA lignin content in C4H–F5H transgenic lines compared to wild-type ( Meyer et al. 1998 ; R. Franke and C. Chapple, unpublished results). Because syringyl residues cannot be cross-linked at the 5-position, these aromatic subunits are less condensed compared to more cross-linked guaiacyl monomers. Consequently, these uncondensed residues may be more readily released during the Klason lignin procedure, leading to a smaller acid-resistant residue and a corresponding under-estimation of lignin content. Differences in lignin yield may be a common property for all syringyl monomer-rich lignins and should be considered when interpreting TGA and Klason data from experiments that have an impact on lignin monomer composition.
Lignin in C4H–F5H tobacco contains reduced levels of hydroxycinnamaldehydes
The staining of lignin with phloroglucinol, also known as the Wiesner reaction, has long been used as a non-specific stain to detect the presence of lignin. This reagent reacts with hydroxycinnamaldehyde groups incorporated into the polymer. The use of commercially available standards for coniferaldehyde, sinapaldehyde, coniferyl alcohol, sinapyl alcohol, vanillin and syringaldehyde in the Wiesner reaction verified that the reagent displays a similar sensitivity for coniferaldehyde and sinapaldehyde, and gives no reaction with the other phenolics (data not shown). In this context, the relatively weak phloroglucinol staining in the C4H–F5H transgenic lines was unexpected, and suggested that their syringyl monomer-rich lignin contains fewer aldehyde residues than wild-type lignin. There are a number of possible explanations for this observation. The step catalysed by CAD may be less of a metabolic bottleneck for the synthesis of syringyl- substituted moieties than for their guaiacyl counterparts. A lower Km or higher Vmax of CAD toward sinapaldehyde might lead to this difference in flux. This hypothesis would be at odds with the kinetic constants determined for CAD in vitro ( Goffner et al. 1992 ; Sarni et al. 1984 ), and with the findings of Halpin et al. (1994) who reported that antisense inhibition of CAD activity had a greater impact on syringyl than guaiacyl monomers. Alternatively, 5-hydroxyguaiacyl and/or syringyl-substituted aldehydes may be more slowly transported out of cells, providing more opportunity for their reduction by CAD. Although it is difficult to determine which, if either, of these explanations is correct, it is clear that phloroglucinol, which is traditionally used for semi-quantitative estimations of lignin content, may not be a useful test for the presence of lignin in samples that show extreme differences in lignin monomer composition.
Tissue specificity and efficacy of the C4H promoter
RNA gel blot hybridization demonstrated that F5H transgene mRNA levels are much higher in leaves of 35S–F5H plants than in the C4H–F5H lines. In contrast, although syringyl monomer content was increased in petioles, the 35S–F5H transgene had no effect on stem lignin monomer composition, whereas the C4H–F5H construct was highly efficacious in increasing lignin syringyl monomer content. Similar results were previously reported in Arabidopsis, where the 35S–F5H transgene expression increased syringyl monomer content to a maximum of 30 mol%. These results may be explained by a previous study in which 35S–GUS fusions in transgenic aspen were shown to lead to strong GUS activity in all tissues except for the cambium and the xylem ( Nilsson et al. 1996 ). Thus, while 35S promoter-driven F5H expression led to high levels of mRNA in leaves, and presumably in non-lignified cells of stem tissue, this promoter may not efficiently target cells synthesizing monolignols. If so, only a small proportion of the mRNA detected in gel blot hybridization analyses is capable of having a phenotypic impact on lignin biosynthesis.
The GUS staining pattern observed in C4H–GUS transformants is consistent with the role of C4H in lignification. In petioles, GUS activity was detected mainly in xylem parenchyma, although presumably early in their development tracheary elements also express C4H. In stems, GUS activity was observed only rarely in developing tracheary elements, and was most consistently seen in xylem ray parenchyma. Similar results have been reported for phenylalanine ammonia lyase, 4CL and CAD promoter–GUS fusions ( Bevan et al. 1989 ; Feuillet et al. 1995 ; Hauffe et al. 1991 ) in transgenic tobacco and poplar, and are consistent with a conserved mechanism of phenylpropanoid gene regulation ( Tamagnone et al. 1998 ). These data strongly suggest that C4H expression plays a role during the early stages of tracheary element wall synthesis, and that the continued expression of phenylpropanoid genes in adjacent cells may permit the modification of the lignin polymer after tracheary element autolysis has occurred.
With regard to the C4H–F5H transgenics, these data are consistent with a model in which xylem parenchyma cells are competent to express the F5H gene under the control of the C4H promoter. As a result, they synthesize and secrete primarily sinapyl alcohol for polymerization into adjacent lignified cell walls. This model would explain the efficacy of the C4H–F5H construct since it would impact upon lignification throughout the entire period of cell wall synthesis. In contrast, the 35S–F5H construct increases petiole lignin syringyl monomer content, but does not significantly influence lignin monomer composition in stems. These data suggest that 35S promoter-driven F5H expression influences lignification only during primary xylem development, perhaps only until autolysis occurs. These findings also further substantiate the utility of tissue-specific promoters, such as the C4H promoter, in the metabolic engineering of lignin biosynthesis.
The data presented here demonstrate that F5H expression determines the developmental- and tissue-specific regulation of lignin monomer composition in plants that undergo secondary growth, including tree species. F5H over-expressing plants deposit a lignin highly enriched in syringyl units. Whereas some other attempts to genetically manipulate lignin profiles were accompanied by the deposition of unusual cell wall phenolics, abnormal xylem pigmentation, and/or deformation of vessel elements ( Atanassova et al. 1995 ; Baucher et al. 1996 ; Halpin et al. 1994 ; Piquemal et al. 1998 ; Van Doorsselaere et al. 1995 ; Zhong et al. 1998 ), F5H over-expressing plants appear to tolerate the chemical and structural modifications in their lignin with no obvious deleterious effects. The lack of phenotypic impact of the lignin with a high syringyl monomer content suggests that water conduction and mechanical strength are not significantly disrupted in C4H–F5H transgenics. Thus, it seems possible to modify syringyl lignin content in herbaceous and woody species of economic value for the benefit of agriculture and industry without consequences on plant viability.
Plant growth conditions
Wild-type and transgenic tobacco plants (Nicotiana tabacum cv. Wisconsin #38) were cultivated in ProMix potting mixture (Premier Horticulture, Red Hill, Pennsylvania, USA) in a greenhouse with supplemental fluorescent lighting to provide a 20 h photoperiod. For selection of transgenic individuals, seeds were surface-sterilized and distributed on plates containing MS medium ( Murashige & Skoog 1962) supplemented with 200 mg l−1 kanamycin and 200 mg l−1 Timentin (SmithKline Beecham).
RNA was extracted from leaf material ( Goldsbrough & Cullis 1981), electrophoretically separated, transferred to Hybond N+ membrane (Amersham), and probed with radiolabelled probes prepared from the F5H cDNA according to standard protocols.
Histochemistry and reporter gene analysis
Tobacco petioles and stems were hand-sectioned using a double-edged razor blade and stained as follows. For general anatomical observation, sections were briefly incubated in a 0.05% (w/v) aqueous solution of TBO in water, rinsed, and examined using bright-field microscopy. For histochemical analysis of lignification, sections were stained with 1% (w/v) phloroglucinol in 12% HCl ( Clifford 1974). Aldehyde groups in control samples were reduced by treating sections with 2% (w/v) sodium borohydride in DMSO for 10 min followed by washing with DMSO, 1 m NaOH, DMSO and water for 30 sec. Sections were then stained with phloroglucinol as described above. For the histochemical analysis of lignin monomer composition, Mäule staining was performed as described previously ( Chapple et al. 1992 ). Assays for β-glucuronidase activity in C4H–GUS transformants were conducted as described by Bell-Lelong et al. (1997) .
Cell wall analysis
For cell wall preparation, petiole and stem tissue from 3–6-month-old tobacco and poplar plants was frozen in liquid nitrogen immediately after harvesting and lyophilized. The dry tissue was ground to a fine powder in a Tekmar analytical mill and soluble components removed from the powdered tissue by successive extraction with neutral phosphate buffer, 80% ethanol and acetone ( Meyer et al. 1998 ). One set of samples from T0 plants was used for the analysis of lignin monomer composition by NBO. An independent set of samples harvested from younger T1 plants was used for the analysis of lignin monomer composition by the DFRC method, and for lignin quantification by the TGA and Klason methods.
To measure lignin content, cell wall samples were analysed using the TGA method ( Campbell & Ellis 1992), or the microscale Klason method ( Kaar et al. 1991 ). Lignin monomer composition was determined by nitrobenzene oxidation ( Meyer et al. 1998 ) and by the DFRC method ( Lu & Ralph 1997) modified as follows. Cell wall polysaccharides interfering in the GC analysis of DFRC products were reduced by digestion with Driselase (Sigma), a crude mixture of endo- and exo-glucanases. Solvent-extracted cell walls (100 mg) were suspended in 1% Driselase in 50 m m pyridinium acetate buffer, pH 4.7, and incubated for 3 days at 37°C. The undigested cell wall residue was precipitated by centrifugation (1000 g, 5 min) and the supernatant was discarded. The pellet was washed twice with 50 m m pyridinium acetate buffer, pH 4.7 and the Driselase digestion was repeated for an additional 2 days. The undigested residue was collected by centrifugation, washed in 50 m m pyridinium acetate buffer, water, 3% (w/v) SDS in 1% (w/v) ammonium bicarbonate, water and finally acetone, and was dried overnight at room temperature. The Driselase-digested cell wall residue (20 mg) was derivatized essentially as described by Lu & Ralph (1997) using 0.2 mg of 4,4-ethylidene-bisphenol (Aldrich) as an internal standard. The acetylated lignin monomers, coniferyl diacetate and sinapyl diacetate were quantified by comparison with authentic standards.
We thank John Ralph for providing helpful advice and standards for DFRC analysis. This work was supported by grants from the Frasch Foundation and the Division of Energy Biosciences, United States Department of Energy to C.C., a Post-doctoral Fellowship from the Alexander von Humboldt Foundation (Feodor Lynen Fellowship) to K.M., and a start-up grant from the state Rheinland-Pfalz to R.F. This is journal paper number 16252 of the Purdue University Agricultural Experiment Station.