Plant Physiol. (1997) 115: 41-50

Transcription

1 Plant Physiol. (1997) 115: Reduced Lignin Content and Altered Lignin Composition in Transgenic Tobacco Down-Regulated in Expression of L-Phenylalanine Ammonia-Lyase or Cinnamate 4-Hydroxylase Vincent J.H. Sewalt, Weiting Ni, Jack W. Blount, Hans C. Jung, Sameer A. Masoud2, Paul A. Howle~,~ Chris Lamb, and Richard A. Dixon* Plant Biology Division, Samuel Roberts Noble Foundation, P.. Box 21 80, Ardmore, klahoma (V.J.H.S., J.W.B., P.A.H., S.A.M., R.A.D.); Plant Biology Laboratory, Salk lnstitute for Biological Studies, 1001 North Torrey Pines Road, La Jolla, California (C.L.); and United States Department of Agriculture-Agricultura1 Research Station, Department of Agronomy and Genetics, University of Minnesota, St. Paul, Minnesota (W.N., H.G.J.) We analyzed lignin content and composition in transgenic tobacco (Nicotiana tabacum) lines altered in the expression of the early phenylpropanoid biosynthetic enzymes L-phenylalanine ammonia-lyase and cinnamate 4-hydroxylase (C4H). The reduction of C4H activity by antisense expression or sense suppression resulted in reduced levels of Klason lignin, accompanied by a decreased syringyl/guaiacyl monomer ratio as determined by pyrolysis gas chromatography/mass spectrometry. Similar reduction of lignin levels by down-regulation of i-phenylalanine ammonia-lyase, the enzyme preceding C4H in the central phenylpropanoid pathway, did not result in a decreased syringyl/guaiacyl ratio. Rather, analysis of lignin methoxyl content and pyrolysis suggested an increased syringyl/guaiacyl ratio. ne possible explanation of these results is that monolignol biosynthesis from L-phenylalanine might occur by more than one route, even at the early stages of the core phenylpropanoid pathway, prior to the formation of specific monolignol precursors. There is currently intense interest in modifying the content and / or composition of the cell wall structural polymer lignin as a means of improving the efficiency of the paper pulping process for forest trees or of increasing digestibility of forages for ruminant animals (Whetten and Sederoff, 1991; Boudet and Grima-Pettenati, 1996; Campbell and Sederoff, 1996). Recent studies have concentrated on attempts to downregulate the levels of enzymes involved in the reactions specific for lignin monomer synthesis by expression of This work was supported by the Samuel Roberts Noble Foundation. P.A.H. is a Noble Foundation/ Salk Institute plant biology postdoctoral fellow. Present address: Faculty of Agriculture, Jerash University, P.. Box 311, Jerash, Jordan. Present address: Department of Genetics and Developmental Biology, Monash University, Wellington Road, Clayton, Victoria, Australia * Corresponding author; radixon@noble.org; fax homologous or heterologous antisense genes in transgenic plants (Dwivedi et al., 1994; Halpin et al., 1994; Ni et al., 1994; Atanassova et al., 1995; Van Doorsselaere et al., 1995; Sewalt et al., 1997). Although the biosynthetic pathway to lignin monomers is relatively well understood, involving consecutive hydroxylation and -methylation reactions leading from p-coumaric acid via ferulic acid (the monomethoxylated precursor of the G residues of lignin) to sinapic acid (the dimethoxylated precursor of the S residues of lignin), it has recently been suggested that parallel pathways of monomer hydroxylation and methylation could occur at the level of the CA thioesters (Ye et al., 1994) or even at the level of the aldehydes formed after the first reduction of the CA thioesters (Matsui et al., 1994; Fig. 1). The existence of a metabolic grid for the -methylation of monolignols would complicate the interpretation of experiments in which a single enzyme of the pathway was down-regulated. Indeed, severa1 reports of the effects of antisense inhibition of enzymes involved in the late reactions of monolignol biosynthesis have presented unpredicted and sometimes contradictory results. Ni et al. (1994) reported that modest down-regulation of CMT activity in transgenic tobacco (Nicofiana fabacum) leads to a small reduction in lignin content with no significant change in lignin composition. However, other groups have shown that strong down-regulation of CMT in tobacco or poplar (Populus tremula X Populus alba) leads to a drastic reduction in S units, with corresponding incorporation of 5-hydroxy G units into lignin, the overall level of which is not reduced (Atanassova et al., 1995; Van Doorsselaere et al., 1995). The latter phenotype is similar to that reported for certain brown-midrib mutants with increased forage digestibility, and the Bm3 mutation in maize (Zea mays) and sorghum Abbreviations: CAD, cinnamyl alcohol dehydrogenase; C4H, cinnamate 4-hydroxylase; CMT, caffeic acid 3--methyltransferase; G, guaiacyl; NDF, neutra1 detergent fiber; PAL, L-Phe ammonia-lyase; S, syringyl Downloaded from on November 12, Published by

2 42 Sewalt et al. Plant Physiol. Vol. 11 5, 1997 Figure 1. Biosynthetic pathways for the formation of lignin monomers. C3H, 4-Coumarate hydroxylase; CMT, caffeic acid 3--methyltransferase; F5H, ferulate 5-hydroxylase; 4CL, coumarate: CA ligase; CCH, coumaroyl CA hydroxylase; CCMT, caffeoyl CA 3--methyltransferase; CCR, cinnamoyl CA reductase. Dotted arrows indicate reactions that are still uncertain in the context of lignin biosynthesis. 1 PAL C4H Cinnamic Acid 4 H H H H3 CY H I 4cL? I H H H?? Ccumaryl Aldehyde._ Sinapyl 1 CCR 1 CCR 1 CCR --.-,conlferyl Aldehyde CAD 1 CAD 1 --f- CAD Aldehyde Coumaryl Alcohol Coniferyl Alcohol Sinapyl Alcohol Peroxldase / Peroxldase / laccase I gluwsidase Peroxidase / laccase / gluwsidase H-lignin G-lignin gluwsidase S-lignin (Sorgkum bicolor) has recently been shown to be in the CMT structural gene (Vignols et al., 1995). To engineer plants with agronomically useful ligninrelated traits, it will be necessary to devise strategies that can flexibly and predictably yield reductions in lignin content and / or changes in lignin monomer composition. Because most reports suggest that reduced expression of the late enzymes of lignin monomer synthesis, CMT and CAD, affects lignin composition without affecting content, it may be necessary to reduce the flux into the lignin pathway at an earlier stage to reduce lignin content. It has recently been demonstrated that decreases in Phe pool size (Jones et al., 1995; Yao et al., 1995) or reduced activity of PAL, the entry enzyme into the phenylpropanoid pathway (Elkind et al., 1990; Bate et al., 1994), leads to decreased lignin content in transgenic plants. However, there are no reports to date of the effects of such manipulations on lignin composition. We describe the composition of lignin from transgenic tobacco plants with severely reduced lignin levels due to down-regulation of PAL or C4H activities. A reduction in PAL levels leads to an increase in the S/G ratio, whereas reduced C4H activity leads to a decrease in the S/G ratio. These observations support the existence of some sort of metabolic channeling between the enzymes of the central phenylpropanoid pathway and those of monolignol biosynthesis and provide a basis for the development of new strategies for lignin modification to improve digestibility of forage crops. MATERIALS AND METHDS Transgenic tobacco (Nicotiana tabacum L. cv Xanthi) plants originated from two independent sets of internally controlled experiments. For both sets of transgenic plants (PAL, C4H, and their respective controls), a11 plants were grown together under exactly the same environmental con- ditions and were harvested at the same time and physiological stage. PAL-modified and control transgenic plants were grown from seed and harvested after 5 weeks. C4H transgenic plants and corresponding controls were primary transformants cut back simultaneously and harvested after 4 weeks of regrowth. PAL lines evaluated were severely PAL sensesuppressed (160P3, second-generation progeny carrying a bean [Pkaseolus vulgaris] PAL transgene in the sense orientation), PAL-suppressed but recovering (274-T5 fifthgeneration selfed progeny), PAL-overexpressing (YEI- 6T1, first-generation selfed progeny), or operationally wild type (C17, first-generation progeny line that had lost the bean PAL transgene through segregation and therefore displayed a wild-type PAL phenotype), as described by Bate et al. (1994) and Howles et al. (1996). C4H lines were primary transformants carrying either an empty vector or an alfalfa (Medicago sativa) C4H transgene (Fahrendorf and Dixon, 1993) in the sense or antisense orientation, resulting in independent transformants with either normal, suppressed, or increased C4H activity. Designation of sense suppression was based on the presence of alfalfa C4H transcripts but reduced overall C4H enzymatic activity compared with the average and SD of values from a population of 15 control plants (Fig. 2). Enzyme Extraction and Assays Midstem sections (internodes 10 and 11 for PAL plants, internodes 8-11 for C4H plants, counting from the first fully opened leaf at the top) were collected and ground under liquid N,. Powdered tissue was divided into two tubes, one for assay of enzyme activities and the other for lignin analysis, and stored at -70 C. PAL (cytosolic) and C4H (microsomal) activities were assayed by the methods described by Edwards and Kessmann (1992).

3 Lignin Alterations in Transgenic Plants h E.r 100 E o.z 50 m 3 25 Histochemical Analysis The 10th internode from tobacco plants with selected PAL or C4H phenotypes was collected in a second sampling from regrown plants. Sections obtained by freehand sectioning were stained for lignin using phloroglucinol- HC1 or the Maule color reaction according to the method of Nakano and Meshitsuka (1992). Phloroglucinol-stained sections were photographed within 30 min. In addition, sections were stained according to the method of Srebotnik and Messner (1994) with 0.1% aqueous safranin- (color index no ; Sigma) and then with 1% aqueous astrablue (Sigma) for 3 min each. 150 t 125 h c E c o 75 2 LL 50 Z 25 r\ D l E E c 25 3 Individual plants Figure 2. Levels of C4H (A) and PAL (B) activities and NDF (C) and Klason lignin (D) levels in a control population of tobacco (cv Xanthi) plants. The mean (continuous line) and SD (dashed lines) are shown for each parameter. Lignin Analysis Powdered midstem samples were freeze-dried and ground in a cyclone mil1 to pass a 1-mm sieve. A portion of the sample was extracted with boiling neutra1 detergent solution (Van Soest et al., 1991). The resulting NDF was subjected to two-stage acid hydrolysis to determine Klason lignin (Sewalt et al., 1996). The residue after acid hydrolysis (Klason lignin) was used to determine lignin methoxyl groups (TAPPI, 1972; Zakis, 1994). Nonextracted dried stem material was subjected to pyrolysis GC-MS to determine the lignin S/G ratio (Ralph and Hatfield, 1991). Statistical Analysis Differences in enzymatic activity and lignin characteristics between groups of control and genetically modified plants were examined in internally controlled experiments by one-way analysis of variance (Snedecor and Cochran, 1989) using the data analysis tools in the program Excel (Microsoft, Redmond, WA). In the PAL experiment seedlings of independent transformants and control lines were used as replicates (Table 111). In the C4H experiment independent primary transformants were grouped into low (antisense/ sense-suppressed) and high (control/ overexpressor) C4H activity classes. RESULTS Transgenic Tobacco Plants with Altered Expression of Core Phenylpropanoid Pathway Enzymes The generation of transgenic tobacco plants with altered levels of C4H expression, using binary vector constructs containing the complete alfalfa C4H cdna sequence (Fahrendorf and Dixon, 1993) in both sense and antisense orientations, will be described in more detail elsewhere (J.W. Blount, S.A. Masoud, K.L. Korth, V.J.H. Sewalt, T. Fahrendorf, and R.A. Dixon, unpublished data). Plants transformed with the antisense construct expressed low levels of C4H transcripts and had significantly reduced C4H activity as initially determined in leaf tissue (approximately 35% of wild type on average). Plants transformed with the sense construct fel1 into two classes: overexpressors (470% of wild-type activity on average) and plants exhibiting reduced (45% of wild type on average) levels of C4H activity. The latter are designated as operationally sense-suppressed plants, although it should be cautioned that the exact mechanism for the reduced C4H activities has not been determined. The molecular and biochemical phenotypes of a subset of the C4H transformants that were selected for analysis of lignin content and composition are summarized in Table I. Further details of the phenotypes of the C4H transgenic plants will be described elsewhere (J.W. Blount, S.A. Masoud, K.L. Korth, V.J.H. Sewalt, T. Fahrendorf, and R.A. Dixon, unpublished data). We previously generated transgenic tobacco plants in which the leve1 of PAL activity was severely reduced as a result of sense suppression arising from the genomic inte-

4 NDF 44 Sewalt et al. Plant Physiol. Vol. 11 5, 1997 Table 1. Characteristics of plants used for analysis of effects of modified C4H expression on lignin content and composition Transgene Copy C4H Transcript C4H Activity in Plant Line Construct Status No." Levelb Stems 11 A Untransformed - N D~ 8A Untransformed - ND 2c Sense 1 +e 32C Sense ND + +' 25C Sense +>l +++g 726 Antisense Antisense +> C Sense h % of control a Transgene copy number was determined by Southern analysis using the alfalfa C4H-coding sequence as a probe and segregation of kanamycin resistance in the T, generation. bthe C4H Control Control Sense-suppressed Sense-suppressed Sense-suppressed Antisense Antisense verexpressor transcript leve1 was determined by northern analysis using the alfalfa C4H-coding sequence as a probe. -, No transgene. ND, Not detected. e +, Barely detectable. ++, Intermediate levei , Strong expression , Very strong expression. gration of a bean PAL transgene (Elkind et al., 1990) and showed that in one particular line (YE6-16) this sensesuppression phenotype was gradually lost in successive selfed generations (Bate et al., 1994). Plants with very low PAL activity have reduced levels of thioglycollic-acidextractable lignin (Bate et al., 1994). Another sensesuppressed transgenic tobacco line harboring the bean PAL transgene (YE10-6) changes to a PAL-overexpressing line in a single selfed generation (Howles et al., 1996). Thus, near-isogenic tobacco lines have been generated with a wide range of PAL activities, and representative independent plants from these various classes were chosen for analysis of lignin content and composition. To assess the extent of modification of C4H and PAL activity and lignin content in the above transgenic lines, we first analyzed a population of 15 independent control plants for C4H activity, PAL activity, NDF (pectin-free cell wall material), and Klason lignin levels in stem tissue. The results shown in Figure 2 demonstrate that the natural variation in C4H and PAL activity in stems was characterized by SDS of 2-22 and?26% of the mean, respectively. Means for NDF and Klason lignin had an SD of 26% each. Lignin Content and Composition in Transgenic Tobacco with Reduced or lncreased Expression of C4H The effects of alterations in C4H activity caused by expression of the alfalfa C4H transgene in the sense or antisense orientations on NDF, Klason lignin, and lignin methoxyl content are shown in Table 11. In the set of eight independent plants analyzed, overexpression resulted in approximately twice the wild-type C4H activity in stem tissues (similar to the increase in PAL in PALoverexpressing lines [Howles et al., 19961; see below), whereas the strongest antisense effect reduced C4H activity to approximately 20% of wild-type levels. verexpression of C4H had no effect on NDF, Klason lignin, or lignin methoxyl content measured by wet chemistry, whereas down-regulation to less than 50% of wild-type levels significantly reduced NDF and lignin levels in most lines (Table 11), the latter to as much as below 20% of wild type (expressed as a percentage of dry matter) in the most severely down-regulated line (Fig. 3). The only anomalous result from the plants analyzed in Table I1 was from line Table II. Relationship among C4H activity, NDF, Klason lignin, and lignin methoxyl group content in transgenic tobacco with modified C4H expression Plant No. 201 c 11A 8A 2c 32C C 13B P value Status' E C C ss ss AS ss AS nkat/g C4H NDF Klason Lignin Lignin Methoxyl Group Mean? SD for Mean 2 SD for ~~ Mean? SD for Mean? SD for % Dry matter % Klason lignin set set set set ? t ? a E, verexpressor; C, empty vector control; SS, sense-suppressed; AS, antisense. i i.oa ? ? ?

5 Lignin Alterations in Transgenic Plants 45 ZC, in which slightly reduced C4H activity was accompanied by an increase in NDF and Klason lignin levels. Increasing C4H activity above wild-type levels had no effect on the S/G ratio as determined by pyrolysis GC-MS analysis, whereas, surprisingly, reduction to approximately 40% of wild-type levels caused a large reduction to approximately 0.05 in antisense line 72B (Figs. 4 and 5). This resulted from a drastic decrease in S residues accompanied in the samples with the lowest lignin levels by a smaller decrease in G residues. In contrast, methoxyl analysis by wet chemistry suggested more modest reductions in methoxyl content as a percentage of Klason lignin leve1 in the down-regulated lines. Because of low levels of free iodine in the hydriodic acid used in the iodometric determination of methoxyl content, measurement by this method becomes less reliable at very low lignin levels. C4H-down-regulated plants displayed reduced phloroglucinol staining compared with the respective controls (Fig. 6, a and b), consistent with the reduction in Klason lignin. Maule staining of C4H-reduced plants showed a change in color from wine-red to dark-brown (Fig. 6, e and f), which is indicative of a reduction in S content. Safranin-, a basic dye, stains lignin red. Astra-blue, a phthalocyanin dye, is incorporated into cellulose fibers, B1 1 I S/G = S/G = , C4H activity (% of control) :::i 200 SIG = i21. o PAL activity (% of control) Figure 3. Relation between enzyme activities and Klason lignin levels in midstems of transgenic tobacco plants with altered expression of C4H (A) or PAL (6). A, Plants harboring an alfalfa C4H antisense or sense construct leading to reduced C4H activity (W), overexpressing C4H (), or untransformed control plants of the same physiological stage (). B, PAL sense-suppressed plants (M), PAL sense-suppressed but recovering plants (), control plants that had lost the bean PAL transgene through segregation (), or PALoverexpressing plants (). DM, Dry matter Time (min) Figure 4. Pyrolysis CC of stem material from tobacco plants expressing various alfalfa C4H constructs. A, Control line 8A harboring an empty vector construct; B, line 201 C overexpressing C4H activity; C, line 726 expressing a C4H antisense construct; and D, line 13B expressing a C4H antisense construct. Pyrolysis products derived from lignin monomers are indicated. Products from G lignin were guaiacol (GI), 4-methylguaiacol (G2), 4-ethylguaiacol (C3), vanillin (G4), and trans-isoeugenol (G5). Products from S lignin were 2,6- dimethoxyphenol (SI ), 2,6-dimethoxy-4-methylphenol (S2), 4-ethyl- 2,6-dimethoxyphenol (S3), 2,6-dimethoxy-4-vinylphenol (S4), and trans-2,6-dimethoxy-4-propenylphenol (S5). All assignations were based on MS analysis. See Tables I and II for further details of the phenotype of the individual lines.

6 46 Sewalt et al. Plant Physiol. Vol. 115, 1997 CD corresponding to G units decreased, whereas the level of S units remained relatively unaffected, resulting in an increase in the S/G ratio to approximately 1.9. In contrast, overexpression of PAL had little effect on the S/G ratio, consistent with the lack of any significant change in lignin levels. Analysis of additional PAL-suppressed lines by pyrolysis GC-MS confirmed the above findings, with S/G C4H activity (% of control) 250 Figure 5. Relation between C4H activity and S/G ratio of Klason lignin from stems of a range of transgenic tobacco plants harboring alfalfa C4H constructs or untransformed control plants of the same physiological stage., Control plants;, antisense and sensesuppressed plants; and, overexpressor plants. staining them blue only in the absence of lignin. Dual staining with safranin- and astra-blue confirmed the above results; C4H-suppressed plants showed distinct rows of blue cell walls protruding from the pith into the vascular region (ray cells) or patches of blue within the vascular region, which may be an indication that the antisense or sense-suppression effects are not evenly distributed over different cell types (Fig. 6, g and h). Lignin Content and Composition in Transgenic Tobacco with Reduced or Increased Expression of PAL In Table III the PAL activities, NDF, Klason lignin, and lignin methoxyl content of stem tissue from 11 independent transformants representative of four classes of PAL expression are shown: wild type (C17, a line that lost the bean PAL transgene, and therefore the sense suppression phenotype, through segregation), PAL-suppressed (160P3, T2 generation), PAL-suppressed but recovering (274-T5, fifth-generation selfed progeny of a strongly sensesuppressed primary transformant), and overexpressors (first-generation selfed progeny of line YE10-6). Although there was significant variation in extractable PAL activity within each class, the ranges are distinct and represent clearly defined PAL phenotypes. Reduction in PAL activity in the most severely sense-suppressed lines caused a large decrease in Klason lignin content and a corresponding effect on NDF value in two of the three plants analyzed. At the same time, however, there was a significant increase in methoxyl group content in the two most severely PALsuppressed plants. The increased methoxyl content relative to total lignin amount in the plants with reduced lignin levels suggests that inhibition of the flux into the phenylpropanoid pathway has qualitative as well as quantitative effects on lignin synthesis. The effect of altered PAL activity on lignin composition was confirmed by pyrolysis GC-MS (Fig. 7). In the control C17 line the S/G ratio was 1.102, which is comparable to that in untransformed control lines (data not shown). In the strongly PAL-suppressed line, pyrolysis product peaks Figure 6. Histochemical analysis of lignin in transgenic tobacco lines. Cross-sections of stems of transgenic tobacco lines (1 th internode from top) stained with phloroglucinol-hci (A-D), Ma'ule reagent (E and F), and safranin- and astra-blue (G-)). A, Untransformed control for C4H transgenics (11 A); B, C4H sense-suppressed transformant (32C); C, PAL overexpressor (10-6); D, PAL sensesuppressed (160-P3); E, untransformed control for C4H transgenics (11 A); F, C4H sense-suppressed (32C); G, untransformed control for C4H transgenics (11 A); H, C4H sense-suppressed (32C); I, PAL control line (105C); J, PAL sense-suppressed (160-P3). Magnification, X25 (G-)) and X50 (A-F).

7 Lignin Alterations in Transgenic Plants 47 Table 111. Relationship among PAL activity, NDF, Klason lignin, and lignin methoxyl group content in transgenic tobacco with modified PAL expression PA L NDF Klason Lignin Lignin Methoxyl Croup Plant No. Statusa Mean t SD for Mean t SD for ~/~ NDF Mean t SD for nkat/g % Dry matter set set set % Klason lignin Mean t SD for set C17a WT C17c WT C17b WT i 67'3d 34.0? 2.5b ' 12.5 i 1.5b 160P3b SS P3a SS P3c SS i 12b 29.6 i 7.1b 5.68 i C 5.8' 274T5b SS/R T5a SSIR T5c SS/R i 10' 32.9 i 5.5b bsC 15.4 i 3.bSC 10-6a E ~ E i lld 36.4 i 2.5b 9.92 i 3.31' 12.1 i 2.6' P value a WT, Wild-type plants having lost the bean PAL2 transgene through segregation; SS, sense-suppressed; SS/R, recovering sense-suppressed; E, overexpressor. b,c,d Significant (P < 0.05) differences between sets (WT, SS, SS/R, or E) are indicated by dissimilar superscripts. ratios of 1.09? 0.08 for control plants (n = 3), 1.13? 0.21 for overexpressors (n = 3), and 1.59? 0.21 for sensesuppressed plants (n = 3). The S/G ratio in the control set of transformants for the PAL lines was higher than in the control lines for the C4H experiment (see above). This is either because the samples analyzed from the PAL transgenics consisted of slightly older stem material harvested at a slightly later stage or, less likely, because the PAL transgenics were derived from a different original source of cv Xanthi seed. The very poor staining of PAL-suppressed tobacco stem sections with phloroglucinol confirmed the reduction in lignin content (Fig. 6d), whereas PAL-overexpressing plants stained strongly (Fig. 6c). PAL suppression resulted in less intense Maule staining than in plants with wild-type PAL levels and a slight color shift from plain brown (indicative of predominance of G lignin in the vascular tissue of wild-type tobacco) to a patterned wine-red / brown staining (indicative of a shift to S lignin in xylem ray cells and / or sclerenchyma fibers in PAL-suppressed plants; data not shown). Double staining with safranin- and astra-blue confirmed the results obtained with phloroglucinol; shifts from red to purple or blue in the vascular ring in plants with reduced PAL activity were indicative of reduced lignin leve1 and increased accessibility of cellulose to the stain (Fig. 6, i and i). DISCUSSIN Cenetic Manipulation of Lignin Content in Transgenic Tobacco We are attempting to develop genetic engineering strategies for reducing lignin content to improve digestibility of forage species. However, severa1 reports of targeted downregulation of late enzymes of lignin biosynthesis have failed to demonstrate reductions in lignin content (Dwivedi et al., 1994; Halpin et al., 1994; Atanassova et al., 1995; Van Doorsselaere et al., 1995), highlighting our lack of understanding of the regulatory control points for monolignol biosynthesis (Lewis and Yamamoto, 1990). In the case of down-regulation of CAD, the last enzyme in the classical monolignol pathway, reducing the formation of hydroxycinnamyl alcohols results in the incorporation of the corresponding aldehydes into lignin, leading to a wine-red lignin with increased extractability (Halpin et al., 1994; Higuchi et al., 1994; Hibino et al., 1995; Bernard-Vailhé et al., 1996). Reduction in the activity of the bispecific caffeic acid/ 5-hydroxyferulic acid -methyltransferase in transgenic plants to less than 10% of wild-type levels resulted in qualitative changes in lignin composition (reduced S/G ratio), with no apparent effect on overall lignin content (Atanassova et al., 1995; Van Doorsselaere et al., 1995). The potential complications of a metabolic grid associated with the ring substitution reactions in the later stages of monolignol synthesis would at first sight make the approach of reducing lignin by limiting flux at an earlier stage in the pathway more attractive, in spite of the ultimate need to engineer such modifications under tight temporal and spatial control to avoid pleiotropic effects such as the increased disease susceptibility observed in plants with reduced PAL activity (Maher et al., 1994; Pallas et al., 1996). ur preliminary results with PAL-suppressed plants indicated a reduction in lignin based on toluidine blue staining of stem cross-sections (Elkind et al., 1990), and this was later confirmed by analysis of thioglycollic acid lignin levels in a series of progeny lines representative of the wide variation in PAL activity levels (Bate et al., 1994). We have now shown that reduction of PAL activity to approximately 15% of wild-type levels gives an approximately 2-fold decrease in Klason lignin as a percentage of

8 48 Sewalt et al. Plant Physiol. Vol. 11 5, s1 I I S/G=1.102 composition. Furthermore, the PAL activity (measured in the supernatant of the same extract used for determination of C4H activity) in stem tissues of various C4H-suppressed lines (e.g nkat/g protein for line 13B) combined with wild-type C4H activity would not, on the basis of the data shown in Table 111, be predicted to give a lignin reduction of the magnitude observed G1 I 63, G'. s' I / F4 S/G = S/G= Time (min) Figure 7. Pyrolysis CC of stem material from tobacco plants with modified PAL expression. A, Control line C17a that had lost the PAL-suppressed phenotype by segregation of the bean PALZ transgene; B, line 160P3a exhibiting strong sense-suppression of PAL; C, line 10-6a, a PAL-overexpressing line. The S/G ratio is indicated. See Table 111 for further details of the phenotype of the individual lines. dry matter. A similar relationship was observed following the reduction in C4H activity. Both types of modification resulted in a numerical decrease in the level of NDF and in lignin staining patterns consistent with the reduced levels of the polymer. Reduced NDF values in plants with reduced lignin levels may reflect both the reduction in lignin content and alterations in the cell wall polysaccharide content or extractability as a result of reduced lignincarbohydrate interactions. Reducing the level of C4H activity in tobacco stems results in a corresponding decrease in PAL activity (J.W. Blount, S. Masoud, and R.A. Dixon, unpublished results). However, the reduced lignin levels in C4H transgenic plants are unlikely to be simply the result of the reduced PAL activity, because of the difference in effect on lignin Genetic Manipulation of Lignin Composition in Transgenic Tobacco Severely PAL-suppressed plants produced low levels of lignin with increased methoxyl content and S/G ratio indicative of a reduction in G lignin. Contrary to the situation with severe PAL suppression, the S/G ratio was reduced by C4H suppression, indicating that the altered S/G ratio in C4H-down-regulated plants was the result of the change in C4H activity rather than of the associated decrease in PAL activity observed in these plants (J.W. Blount, S. Masoud, and R.A. Dixon, unpublished results). verexpression of either enzyme resulted in no change in lignin composition, which is indicative of downstream control points in the lignin biosynthetic pathway. Critical to the interpretation of the data concerning lignin composition is the nature of the method used to determine the S/G ratio. Because of the complexity and heterogeneity of the lignin polymer, most methods for the determination of lignin composition have some limitations. Pyrolysis MS and pyrolysis GC-MS combine direct depolymerization of organic material by rapid heating in vamo and visualization of dissociation products (Ralph and Hatfield, 1991). In the case of lignin, dissociation products include monomeric, dimeric, and trimeric structural elements. Recently, the use of pyrolysis GC-MS has become routine in agricultura1 chemistry and plant biology research (Mulder and Emons, 1993; Niemann et al., 1993; Boudet et al., 1995) to elucidate differences between lignin assays (Reeves and Galletti, 1993; Hatfield et al., 1994) and for characterization of tobacco lignin in wild-type (Faix et al., 1992) and transgenic plants (Halpin et al., 1994; Sewalt et al., 1997). Pyrolysis is a unique and rapid ligninfingerprinting tool capable of determining lignin monomer composition, but it lacks the capability of thioacidolysis, a commonly used method of lignin analysis, to provide detailed information about lignin-bonding patterns and functionality (Boudet et al., 1995). However, thioacidolysis specifically targets p-0-4 linkages in uncondensed lignin moieties (Lapierre et al., 1985). A direct comparison of the two methods can be drawn from the results of Halpin et al. (1994), who reported a sharp decrease in the S/G ratio in CAD-antisense tobacco from 0.83 (control) to 0.46 as determined by thioacidolysis and a similar decrease in S/G as determined by pyrolysis MS. We tested the reproducibility of the pyrolysis GC-MS method for tobacco lignin by determining the levels of the G- and S-derived residues in five independent control lines (of the cv Xanthi type used in the PAL-suppression experiments). The values (in arbitrary units relative to the interna1 standard) were for G units and for S units, giving an S/G ratio for the group of

9 Lignin Alterations in Transgenic Plants We therefore conclude that the pyrolysis procedure is a sensitive and reproducible method for determining the S / G ratio in control and transgenic tobacco lines. Although pyrolysis is a relatively efficient depolymerization method that targets more than just the uncondensed lignin portion, it still does not characterize the entire lignin polymer. There may be highly condensed lignin moieties that are not depolymerized by pyrolysis but that are quantified by the Klason lignin method and are also represented in the quantification of methoxyl groups. Such unavoidable difficulties in lignin analysis may explain why the highly down-regulated C4H line 138 shows virtually no G or S residues in the pyrogram (Fig. 4D) but nevertheless still contains approximately 30% of the Klason lignin of wild-type plants. The methoxyl determination, which involves exhaustive demethylation with hydriodic acid (TAPPI, 1972; Zakis, 1994), is most inclusive with regard to different lignin portions. Therefore, the methoxyl value includes the entire residual lignin, which, in the case of C4H suppression, seems to be of a more condensed nature (as judged by the less efficient pyrolysis and only slightly reduced total methoxyl content). PAL suppression results in reduced lignin content, with the remaining lignin being relatively uncondensed (as judged by efficient pyrolysis and increased S/G ratio and methoxyl content). ur data provide new information about the effects of reduction of flux into phenylpropanoid synthesis on lignin composition. The observation that the reduction of PAL or C4H activities leads to altered lignin composition in addition to reduced lignin levels contrasts with previous observations of altered composition but no change in lignin level following down-regulation of later enzymes in the monolignol pathway (Atanassova et al., 1995; Van Doorsselaere et al., 1995). Furthermore, the apparently opposite effects of PAL and C4H down-regulation on the S/G ratio were totally unexpected. This finding suggests two possibilities: (a) down-regulation of PAL or C4H could lead to differentia1 feed-forward effects on later downstream enzymes (this is not the case for CMT, which is not reduced in C4H-down-regulated plants [V.J.H. Sewalt and R.A. Dixon, unpublished results]) and (b) the route for monolignol formation at the level of the early, "core" reactions of the phenylpropanoid pathway may, like the later reactions of -methylation and hydroxylation, not be a single linear pathway, although it is unlikely that parallel pathways utilizing enzymes in addition to PAL and C4H exist. ne possible explanation for the different effects of PAL and C4H down-regulation could be that these enzymes are organized into more than one complex or metabolic channel, perhaps also involving later enzymes such as the CA ligase or -methyltransferases. Such putative channel complexes might be associated with specific isoforms of PAL, which is encoded by a multigene family in tobacco (Fukasawa-Akada et al., 1996) and most other species studied. Such complexes might direct Phe or cinnamate specifically into the production of G or S units, and partitioning of flux into these complexes might not be equally affected by down-regulation of PAL compared with C4H, particu- larly if different tobacco PAL forms are differentially sensitive to co-suppression by the bean PAL transgene. Such hypotheses remain to be tested. However, channeling of Phe through the PAL and C4H reactions has been previously demonstrated by metabolic labeling experiments in microsomes from potato tubers and cucumber and buckwheat hypocotyls (Czichi and Kindl, 1975, 1977; Hrazdina and Wagner, 1985), and we have recently confirmed this in tobacco (S. Rasmussen and R.A. Dixon, unpublished results). It will be interesting to study the effects of PAL down-regulation on lignin composition in monocots, in which the PAL enzyme also has Tyr ammonia-lyase activity (Rosler et al., 1997) and can therefore bypass the C4H reaction in the synthesis of 4-coumarate. In conclusion, the present results indicate the feasibility of reducing lignin content for forage improvement, with targeted changes in lignin monomer composition, by transgenic strategies. At the same time, they reveal the possibility of a hitherto unexpected complexity in the functioning of the early stages of the core phenylpropanoid pathway. ACKNWLEDGMENTS We thank Drs. Richard Nelson and Alenka Hlousek-Radojcic for critically reviewing the manuscript and Dr. Ponsamuel Jayakumar for assistance with preparing freehand sections. Received March 20, 1997; accepted June 6, Copyright Clearance Center: /97/ 115/0041/10. LITERATURE ClTED Atanassova R, Favet N, Martz F, Chabbert B, Tollier M-T, Monties B, Fritig B, Legrand M (1995) Altered lignin composition in transgenic tobacco expressing -methyltransferase sequences in sense and antisense orientation. Plant J 8: Bate NJ, rr J, Ni W, Meroni A, Nadler-Hassar T, Doerner PW, Dixon RA, Lamb CJ, Elkind Y (1994) Quantitative relationship between phenylalanine ammonia-lyase levels and phenylpropanoid accumulation in transgenic tobacco identifies a rate determining step in natural product synthesis. Proc Natl Acad sci USA 91: Bernard-Vailhé M-A, Cornu A, Robert D, Maillot MP, Besle J-M (1996) Cell wall degradability of transgenic tobacco stems in relation to their chemical extraction and lignin quality. J Agric Food Chem 44: Boudet AM, Grima-Pettenati J (1996) Lignin genetic engineering. Mo1 Breeding 2: Boudet AM, Lapierre C, Grima-Pettenati J (1995) Tansley review no. 80. Biochemistry and molecular biology of lignification. New Phytol 129: Campbell MM, Sederoff RR (1996) Variation in lignin content and composition-mechanisms of control and implications for the genetic improvement of plants. Plant Physiol 110: 3-13 Czichi U, Kindl H (1975) Formation of p-coumaric acid and o-coumaric acid from L-phenylalanine by microsomal membrane fractions from potato: evidence of membrane-bound enzyme complexes. Planta 125: Czichi U, Kindl H (1977) Phenylalanine ammonia-lyase and cinnamic acid hydroxylase as assembled consecutive enzymes on microsomal membranes of cucumber cotyledons: cooperation and subcellular distribution. Planta 134: Dwivedi UN, Campbell WH, Yu J, Datla RSS, Bugos RC, Chiang VL, Podila GK (1994) Modification of lignin biosynthesis in transgenic Nicotiana through expression of an antisense -methyltransferase gene from Populus. Plant Mo1 Biol26: 61-71

10 50 Sewalt et al. Plant Physiol. Vol. 11 5, 1997 Edwards R, Kessmann H (1992) Isoflavonoid phytoalexins and their biosynthetic enzymes. In SJ Gurr, MJ McPherson, DJ Bowles, eds, Molecular Plant Pathology-A Practical Approach. IRL Press, xford, UK, pp Elkind Y, Edwards R, Mavandad M, Hedrick SA, Ribak, Dixon RA, Lamb CJ (1990) Abnormal plant development and down regulation of phenylpropanoid biosynthesis in transgenic tobacco containing a heterologous phenylalanine ammonia-lyase gene. Proc Natl Acad Sci USA 87: Fahrendorf T, Dixon RA (1993) Stress responses in alfalfa. XVIII. Molecular cloning of the elicitor-inducible cinnamic acid 4-hydroxylase cytochrome P450 from alfalfa. Arch Biochem Biophys 305: Faix, Bremer J, Meier D, Fortmann I, Scheijen MA, Boon JJ (1992) Characterization of tobacco lignin by analytical pyrolysis and Fourier transform-infrared spectroscopy. J Anal Appl Pyrolysis 22: Fukasawa-Akada T, Kung S, Watson JC (1996) Phenylalanine ammonia-lyase gene structure, expression, and evolution in Nicotiana. Plant Mo1 Biol 30: Halpin C, Knight ME, Foxon GA, Campbell MM, Boudet AM, Boon JA, Chabbert B, Tollier M-T, Schuch W (1994) Manipulation of lignin quality by down-regulation of cinnamyl alcohol dehydrogenase. Plant J 6: Hatfield RD, Jung HJG, Ralph J, Buxton DR, Weimer PJ (1994) A comparison of the insoluble residues produced by the Klason lignin and acid detergent lignin procedures. J Sci Food Agric 65: Hibino T, Yakabe K, Kawazu T, Shibata D, Higuchi T (1995) Increase of cinnamaldehyde groups in lignin of transgenic tobacco plants carrying an antisense gene for cinnamyl alcohol dehydrogenase. Biosci Biotech Biochem 59: Higuchi T, Ito T, Umezawa T, Hibino T, Shibata D (1994) Redbrown color of lignified tissues of transgenic plants with antisense CAD gene: wine-red lignin from coniferyl aldehyde. J Biotechnol37: Howles PA, Paiva NL, Sewalt VJH, Elkind NL, Bate Y, Lamb CJ, Dixon RA (1996) verexpression of L-phenylalanine ammonialyase in transgenic tobacco plants reveals control points for flux into phenylpropanoid biosynthesis. Plant Physiol Hrazdina G, Wagner GJ (1985) Metabolic pathways as enzyme complexes: evidence for the synthesis of phenylpropanoids and flavonoids on membrane associated enzyme complexes. Arch Biochem Biophys 237: Jones JD, Henstrand JM, Handa AK, Herrmann KM, Weller SC (1995) Impaired wound induction of 3-deoxy-~-arabinoheptulosonate-7-phosphate (DAHP) synthase and altered stem development in transgenic potato plants expressing a DAHP synthase antisense construct. Plant Physiol 108: Lapierre C, Monties B, Rolando C (1985) Thioacidolysis of lignin: comparison with acidolysis. J Wood Chem Technol5: Lewis NG, Yamamoto E (1990) Lignin: occurrence, biogenesis and biodegradation. Annu Rev Plant Physiol Plant Mo1 Biol 41: Maher EA, Bate NJ, Ni W, Elkind Y, Dixon RA, Lamb CJ (1994) Increased disease susceptibility of transgenic tobacco plants with suppressed levels of preformed phenylpropanoid products. Proc Natl Acad Sci USA 91: Matsui N, Fukushima K, Yasuda S, Terashima N (1994) n the behavior of monolignol glucosides in lignin biosynthesis. Holzforschung 48: Mulder MM, Emons AMC (1993) Cell-wall development in maize somatic embryos studied by pyrolysis mass-spectrometry. J Anal Appl Pyrol 25: Nakano J, Meshitsuka (1992) The detection of lignin. In SY Lin, CW Dence, eds, Methods in Lignin Chemistry. Berlin, Springer Verlag, pp Ni W, Paiva NL, Dixon RA (1994) Reduced lignin in transgenic plants containing an engineered caffeic acid -methyltransferase antisense gene. Transgenic Res 3: Niemann GJ, Eijkel GB, Konings H, Pureveen JBM, Boon JJ (1993) Chemical differences between wildtype and gibberellin mutants of tomato determined by pyrolysis-mass spectrometry. Plant Cell Environ 16: Pallas JA, Paiva NL, Lamb CJ, Dixon RA (1996) Tobacco plants epigenetically suppressed in phehylalanine ammonia-lyase expression do not develop systemic acquired resistance in response to infection by tobacco mosaic virus. Plant J 10: Ralph J, Hatfield RD (1991) Pyrolysis-GC-MS characterization of forage materials. J Food Agric Chem 39: Reeves JB, Galletti GC (1993) Use of pyrolysis-gas chromatography mass-spectrometry in the study of lignin assays. J Anal Appl Pyrolysis 24: Rosler J, Krekel F, Amrhein N, Schmid J (1997) Maize phenylalanine ammonia-lyase has tyrosine ammonia-lyase activity. Plant Physiol 113: Sewalt VJH, Glasser WG, Fontenot JP, Allen VG (1996) Lignin impact on fiber degradation. 1. Quinone methide intermediates formed from lignin during in vitro fermentation of corn stover. J Sci Food Agric 71: Sewalt VJH, Ni W, Jung H, Dixon RA (1997) Lignin impact on fiber degradation: increased enzymatic digestibility of genetically engineered tobacco (Nicotiana tubucum) stems reduced in lignin content. J Agric Food Chem 45: Snedecor GW, Cochran WG (1989) Statistical Methods, Ed 8. Iowa State University Press, Ames, IA Srebotnik E, Messner K (1994) A simple method that uses differential staining and light microscopy to assess the selectivity of wood delignification by white rot fungi. Appl Environ Microbiol 60: TAPPI (1972) Methoxyl content of pulp and wood; TAPPI standard method 209 su-72. Technical Association of the Pulp and Paper Industry, Atlanta, GA Van Doorsselaere J, Baucher M, Chognot E, Chabbert B, Tollier M-T, Petit-Conil M, Leplé J-C, Pilate G, Cornu D, Monties B, and others (1995) A nove1 lignin in poplar trees with a reduced caffeic acid / 5-hydroxyferulic acid -methyltransferase activity. Plant J 8: Van Soest PJ, Robertson JB, Lewis BA (1991) Methods for dietary fiber, neutra1 detergent fiber, and non-starch polysaccharides in relation to animal nutrition. J Dairy Sci Vignols F, Rigau J, Torres MA, Capellades M, Puigdoménech P (1995) The brown midrib3 (bm3) mutation in maize occurs in the gene encoding caffeic acid -methyltransferase. Plant Cell 7: Whetten R, Sederoff R (1991) Genetic engineering of wood. For Eco1 Manage 43: Yao KN, Deluca V, Brisson N (1995) Creation of a metabolic sink for tryptophan alters the phenylpropanoid pathway and the susceptibility of potato to Phytophthova infestam. Plant Cell 7: Ye ZH, Kneusel RE, Matern U, Varner JE (1994) An alternative methylation pathway in lignin biosynthesis in Zinnia. Plant Cell 6: Zakis GF (1994) Functional Analysis of Lignins and their Derivative. Technical Association of the Pulp and Paper Industry Press, Atlanta, GA, p 94