Engineering traditional monolignols out of lignin by concomitant up-regulation of F5H1 and down-regulation of COMT in Arabidopsis

Ruben Vanholme1,2, John Ralph3

, Takuya Akiyama3, Fachuang lu3

, Jorge Rencoret Paz03, Hoon Kim3

, Jorgen Holst Christensen 1,2, Brecht Van ReuseI1,2, Véronique Storme 1,2, Riet De Rycke 1,2, Antje Rohde 1,2, Kris Morreel1,2 and Wout Boerjan 1,2,'

10epartment of Plant Systems Bíology, VIB, 9052 Gent, Belgíum,

20epartment of Plant Bíotechnology and Genetícs, Ghent Uníversíty, 9052 Gent, Belgíum, and

30epartment of Bíochemístry and Great Lakes Bíoenergy Research Genter, Uníversíty of Wísconsín, Madíson, W153706, USA

lignin engineering is a promising strategy to optimize lignocellulosic plant biomass for use as a renewable feedstock for agro-industrial applications, Current efforts focus on engineering lignin with monomers that are not normally incorporated into wild-type lignins, Here we describe an Arabidopsis line in which the lignin is derived to a major extent from a non-traditional monomer, The combination of mutation in the gene encoding caffeic acid O-methyltransferase (comt) with over-expression of ferulate 5-hydroxylase under the control of the cinnamate 4-hydroxylase promoter (C4H:F5H1) resulted in plants with a unique lignin comprising almost 92% benzodioxane units, In addition to biosynthesis of this particular lignin, the comt C4H:F5H1 plants revealed massive shifts in phenolic metabolism compared to the wild type, The structures of 38 metabolites that accumulated in comt C4H:F51 plants were resolved by mass spectral analyses, and were shown to derive from 5-hydroxy-substituted phenylpropanoids, These metabolites probably originate from passive metabolism via existing biochemical routes normally used for 5-methoxylated and 5-unsubstituted phenylpropanoids and from active detoxification by hexosylation, Transcripts of the phenylpropanoid biosynthesis pathway were highly up-regulated in comt C4H:F5H1 plants, indicating feedback regulation within the pathway, To investigate the role of flavonoids in the abnormal growth of comt C4H:F5H1 plants, a mutation in a gene encoding chalcone synthase (chs) was crossed in, The resulting comt C4H:F5H1 chs plants showed partial restoration of growth, However, a causal connection between flavonoid deficiency and this restoration of growth was not demonstrated; instead, genetic interactions between phenylpropanoid and flavonoid biosynthesis could explain the partial restoration, These genetic interactions must be taken into account in future cell-wall engineering strategies,

Keywords: cell wall, benzodioxane, NMR, phenolic profiling, monolignol, irx.

INTRODUCTION

Lignins are aromatic heteropolymers that are predominantly deposited in secondarily thickened plant cell walls, where they provide strength and imperviousness to the wall, and protect the cellulose microfibrils from microbial attack. At the same time, lignins limit the efficiency of several indus­trial processes that use plant cell walls as a feedstock. Engineering of lignin amount and structure through bio­technology is a promising strategy to improve the industrial processing of lignocellulosic plant biomass. During the last two decades, many details of the biochemical pathway

producing coniferyl and sinapyl alcohol, the primary monomers from which polymeric lignins derive, have been established. The genes for each of the steps have been identified, and the activities of the derived enzymes have been reasonably well characterized (Boerjan et al., 2003;

Ralph et al., 2004; Vanholme et al., 2010).

Many ofthe recent advances in lignin research have come from examining the consequences on overall lignin struc­ture of up- and down-regulating the various genes of the pathway. Such studies have also revealed the rather striking

metabolic malleability of the lignification process. Massive

compositional changes (i.e. changes in the ratio of guaiacyl

(G) and syringyl (S) moieties resulting from incorporation of

coniferyl and sinapyl alcohol monomers, respectively) in

angiosperm lignins can result from mis-regulation of single

genes in the pathway. For example, lignins consist largely

(or almost entirely) of G units when F5H, the gene encoding

ferulate 5-hydroxylase, is down-regulated (Figure 1) (Meyer

et al., 1998; Marita et al., 1999). Conversely, up-regulation of

F5H using a xylem-specific promoter in Arabidopsis, poplar

(Populus tremula · alba) or tobacco (Nicotiana tabacum)

leads to lignins with extraordinarily high S:G ratios (Meyer

et al., 1998; Marita et al., 1999; Franke et al., 2000; Stewart

et al., 2009). Despite containing essentially linear and appar-

ently low-molecular-weight lignins, poplar plants over-

expressing F5H displayed no visible growth phenotypes

(Stewart et al., 2009). However, the wood was markedly

easier to pulp than that of the wild type (Huntley et al., 2003),

demonstrating that altering lignin composition can be a

viable approach towards improved processing.

In addition to the compositional shifts in S:G ratios, it has

become apparent that lignification may tolerate the (partial)

substitution of monolignols by various other phenolic

precursors. Such substitution is compatible with the current

theory of lignification, in which the polymer derives via

chemical radical coupling reactions, primarily coupling of a

monomer with the growing polymer; any phenolic precursor

within the lignification zone may co-polymerize with other

available phenolic precursors to the extent allowed by

chemical compatibility (Ralph et al., 2004). Less obvious,

but apparently possible, is that phenolic precursors, which

are often formed in cells with perturbed phenylpropanoid

biosynthesis, can be transported to the wall for lignification.

For example, angiosperms deficient in caffeic acid O-meth-

yltransferase (COMT) incorporate 5-hydroxyconiferyl alco-

hol into their lignins, forming benzodioxane units that are

readily apparent by NMR, thioacidolysis and phenolic profil-

ing of lignin oligomers (Atanassova et al., 1995; Van Doorss-

elaere et al., 1995; Jouanin et al., 2000; Guo et al., 2001;

Ralph et al., 2001a,b; Marita et al., 2003; Morreel et al.,

2004b; Do et al., 2007; Lu et al., 2010).

The goal of the current study was to develop plants in

which the lignin composition approaches the state whereby

none of its constituents are derived from traditional mono-

lignols. Despite the already overwhelming evidence that

various non-traditional monolignols are readily incorpo-

rated into the lignin polymer in a variety of natural and

transgenic plants, a plant with a minor complement of

traditional monolignols would provide insight into how far

and in which directions lignin might be redesigned. One

simple strategy to produce non-traditional lignins consists

of up-regulating F5H while blocking COMT expression

(Figure 1). If there is no feedback inhibition on F5H specif-

ically, or within the phenylpropanoid pathway in general, it

is anticipated that the cells will primarily produce 5-hy-

droxyconiferyl alcohol for lignification. Substantial levels of

5-hydroxyconiferyl alcohol can apparently be tolerated in

lignification without obvious adverse effects on plant

growth and development (Figure S1). Here we describe

comt C4H:F5H1 plants in which the lignin comprises approx-

imately 92% benzodioxane units, far more than ever

reported previously and how plants respond to such a

massive replacement of traditional monolignols with

5-hydroxyconiferyl alcohol.

RESULTS

comt C4H:F5H1 plants have an altered development

In order to design lignins that are derived predominantly

from non-traditional 5-hydroxyconiferyl alcohol monomers,

Arabidopsis plants over-expressing F5H1 under the control

of the cinnamate 4-hydroxylase (C4H) promoter (C4H:F5H1

plants) were crossed with Arabidopsis comt mutants. Plants

homozygous for the comt mutation and the C4H:F5H1 con-

struct (comt C4H:F5H1 plants) were identified in the F2

progeny. Although the comt mutant grew normally and the

C4H:F5H1 over-expressing line showed only moderate

growth effects, plant development in comt C4H:F5H1 plants

was severely affected. When grown on agar medium, leaves

of the comt C4H:F5H1 seedlings were smaller compared to

those of the wild type (Figure 2a). The inflorescence stems of

agar-grown comt C4H:F5H1 plants remained small, with a

maximal observed final length of 8 cm, whereas wild-type

stems were over 30 cm at maturity. Extensive phenotypic

variation was observed for the length, number and color of

the inflorescence stems (Figure 2b–d), with the color rang-

ing from green to purple red. In many cases, additional

inflorescences developed, resulting in a bushy phenotype

(Figure 2d). The total stem biomass of senesced soil-grown

comt C4H:F5H1 plants was only a fraction of that of the wild

type (Table 1). Although comt C4H:F5H1 plants developed

flowers, they did not set seeds.

comt C4H:F5H1 plants have an irregular xylem phenotype

To reveal the morphological consequences of F5H1 over-

expression in the comt mutant background at the tissue

level, sections of the inflorescence stem were analyzed by

light microscopy. Toluidine blue-stained sections of comt

C4H:F5H1 plants showed collapsed xylem vessels, a phe-

notype typically known as the irregular xylem (irx) pheno-

type and observed in a range of mutants with perturbed

secondary cell-wall development (Figure 2f,g) (Turner and

Somerville, 1997; Jones et al., 2001; Franke et al., 2002;

Goujon et al., 2003; Hoffmann et al., 2004; Brown et al.,

2005; Besseau et al., 2007; Mir Derikvand et al., 2008;

Schilmiller et al., 2009; Li et al., 2010). No obvious differ-

ences were noted in the other stem tissues. In contrast to the

wild type, Maule staining, which is indicative of S units in

886 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

Figure 1. Biosynthetic pathway for coniferyl and sinapyl alcohol (green box) and the metabolites that accumulate during concomitant up-regulation of F5H1 and

down-regulation of COMT (red boxes).

The black route is expected to be the favored route; the gray routes may also be followed depending on the genetic background and growth conditions. Dashed

arrow: route suggested by Mir Derikvand et al. (2008), Leple et al. (2007) and Nair et al. (2004); ?, conversion not demonstrated. Possible routes towards the four

classes of compounds are indicated by thin red arrows. CCR, cinnamoyl CoA reductase; COMT, caffeic acid O-methyltransferase; HCALDH, hydroxycinnamaldehyde

dehydrogenase; F5H, ferulate 5-hydroxylase. For convenience, metabolites are classified in four classes based on their chemical identity: class I, oligolignols; class II,

hexosylated oligolignols; class III, 5-hydroxyferulic acid-containing dimers; class IV, derivatives of 5-hydroxylated phenylpropanoid monomers (see Table 3).

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 887

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

lo

I~ pt20 ....:; OM.

OR' 2. 30 R' .. H; ~H: R:J..H

R'OJ:~ OR' 3.

4:R

'

I~ ~o ....:: M.

OR' 37 38 R'=Hex; ~=H or R'=H;~Hex

31 R'=H; ~H; R3:maas 155 D. 32 R'=Hex; ~H; ~H or

R'-H; ~Hex; ~H

.. R'",Hax; R"'H or R'=H: R2::Hex

R'cHax; R"=H; R'=H or R'",Hax; R"=Hax; R'zH or R'",H: ~=Hex: R'-=H or R'",Hex; ~=H: R'=Hex DI'

R'=H: RI",H; R'=Hex R'=H: ~=H&X: W=Hex

,

~o lol -- 10 F5H COMT

1: M ..... ~':H:O I : o... * ... 1: OM. OH OH OH Ferullc acld 5-Hydroxy- Sinapic acid

// ¡HCALDH ferul;I:~ALDH? ¡HCALDH

/ CCR ~O F5H lO -- COMT ZO Feruloyl=----- I "<:l ) I "<:! --.V.... I "<:!

CoA ~ ~M. ,. HO ~ M. ~M. ~ OM. OH OH ""\. OH

Conlferaldehyde 5-Hydroxy- I Slnapaldehyde coniferaldehyda

¡CAD ¡CA~I

I{H _F5.~~ I{' QlOM. ,. HO~M OH OH

Coniferyl alcohol 5-Hydroxy-conlferyl alcohol

¡CAD

lH

COMT * ... 1: OM. OH Sinapyl alcohol

Class I

, Class 111

19 R'=H; ~H: R'=H: ~H 20 R'=H" ~H· R'=H" R'-=Hu 21 R'=H; ~~; R'~H; R"=H 22 R'=OH; ~=H; R'=H; R'=Hex 23 R'=OHax; R2..H; R'=zH; R"=H or

R'=OH; ~=Hax; R'=H; ~H 24 R'=OMe; R2=H; R'=H; R .... Hex

25 R'=OMe; R2=Hex; R'=H; R'=H

OH O I~' M:.J:~~H M:.J:~:xt:t\ M'~O~ • OH OM. OH • OM. OH

Class 11

:Jt\ R' o~

12 R1"H; RJ,.Hex; R'=H or R1"'H; R2",H; R'=:Hex

13 R1=OHex; R2"H; R'''H or R1"OH; R2".Hex; R'-=H or R1"'OH; R2:H; R'=:Hex

15R"H 18R=OMa

Rl=Hax; ~=H; R'=H; R'=H or

R1"H; ~Hex; R'=H; R'=H or R1=H; ~H; R"=Hax; R'=H 01'

R1"'H, ~H, ¡¡J"'H, R'=Hex HO R ~ ~:~M. OH _ . 'Y:6 .. _ ¡. ~O~' xt:~. M'XJ('O-_ "'lv XY' I ~ I ..

~OH O~~. HO I 7 OH O 10M. M: I O OH: 10M. O ~ •. _ \::6.<>. _ ¡R'

... O I A OH O". OH 10 XJ('0-_~ ¿V HO R I 3, RR~H ~.x?°oi?~ O~__ x?oH O~~~· ~OH ~O.. J'H R'O 11 o"' OM. --u ...... R''''Hax; ~H: R'=H; R"=H or

5 R=OMe HO o M. O A O R'=H; ~Hex; R"=H; R"=H or 8 OH O... HO >:>.. I O I O I R'=H; ~H: R'=Hex; R'zH or

OH OM.11 OH OM. R'=H; Rz"H;W=H; R"=I-\ex

lignin, generally did not result in red coloration of the sec-

ondary cell wall, indicating a strong reduction in, or absence

of, S units in the comt C4H:F5H1 plants (Figure 2e). How-

ever, a slight red coloration was observed in a few of these

plants, suggesting the presence of residual COMT.

Transmission electron microscopy (TEM) of the cell walls

revealed multiple concentric cell-wall sub-layers in the comt

C4H:F5H1 plants (Figure 2i,k). This stratification appeared to

be restricted to the S2 region of the cell wall, and was more

obvious in interfascicular fibers than in xylem vessels.

However, a wide variability in the extent of cell-wall strat-

ification, ranging from absent to extensive, was observed

in the fibers. This remarkable phenotype, already described

for ccr1 Arabidopsis mutants and CCR down-regulated

poplars, has been suggested to be caused by an altered

assembly of cellulose microfibrils (Goujon et al., 2003; Leple

et al., 2007).

NMR analyses reveal massive incorporation of 5-hydroxy-

coniferyl alcohol into lignin

Although acetylbromide lignin levels in the senesced inflo-

rescence stems of either comt or C4H:F5H1 plants were not

significantly different from those of the wild-type, lignin

amounts in comt C4H:F5H1 plants were reduced to approx-

imately 56% of wild-type levels (Table 2). Cellulolytic

enzyme lignins (CEL) were isolated from senesced

inflorescence stems and analyzed by NMR. The aromatic

regions of the 2D 13C–1H correlation (heteronuclear single

quantum coherence, HSQC) spectra highlight differences in

the distribution of guaiacyl, 5-hydroxyguaiacyl and syringyl

(G, 5H and S) units in the lignins. The resulting spectra

(Figures 3 and S2) clearly show that G signals from comt

C4H:F5H1 plants are low, representing approximately 13.8%

of the total aromatic content (based on volume integration),

compared to 82.7% in the wild type. The dominant signals

are from 5H units. It is not immediately clear that S units are

depleted, because the S2/6 contours overlap with those from

(a)

(c) (d)

(e)

(b) (f) (h) (j)

(g) (i) (k)

(l)

Figure 2. Phenotype and inflorescence stem morphology of comt C4H:F5H1 and comt C4H:F5H1 chs plants.

(a) Fourteen-day-old seedlings, segregating from a comt/COMT C4H:F5H1/C4H:F5H1 parental line. The double homozygous comt C4H:F5H1 plant is indicated with

an arrow. (b–d) comt C4H:F5H1 plants with a wide phenotypic diversity: (b,c) 13-week-old and (d) 24-week-old plants.

(e) Maule staining of comt C4H:F5H1 plant grown for 28 weeks in vitro. A section of a greenhouse-grown wild-type plant is visible in the upper left corner.

(f, g) Light microscopy of toluidine blue-stained sections of wild-type and comt C4H:F5H1 tissue culture-grown plants, respectively. Arrows indicate xylem cells.

Note the collapsed xylem vessels in the comt C4H:F5H1 plants.

(h–k) TEMs of cell walls stained with uranyl acetate, showing interfascicular fibers of the wild type (h), interfascicular fibers of comt C4H:F5H1 (i), the xylem region of

the wild type (j), and the xylem region of comt C4H:F5H1 (k).

(l) Partially recovered phenotype of a comt C4H:F5H1 chs plant (see Table 1 for biomass data).

Table 1 Inflorescence height and the total inflorescence stem dryweight, as a measure for biomass, of the various lines

Line

Height (cm) Dry weight (mg)

Mean � SD n Mean � SD n

Wild type 47.7 � 3.5 29 104.9 � 32.0 10comt 44.2 � 4.9 30 125.4 � 27.1 10C4H:F5H1 29.4 � 4.7* 29 77.6 � 21.8 8chs 48.1 � 4.8 27 130.4 � 58.0 10comt C4H:F5H1 1.1 � 1.4* 14 1.4 � 3.4* 10comt C4H:F5H1 chs 11.0 � 2.1* 50 10.9 � 3.0* 12

n, number of plants analyzed.*Significantly different from the wild type (P < 0.001).

888 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

5H6. However, the level of the 5H2 contours suggests that

most of the 5H6/S2/6 contour region must belong to 5H6.

The S2/6 volume integral can be estimated by subtracting

the volume integral of the 5H2 region from that of the

5H6+S2/6 region; the volume integrals for 5H2 and 5H6

regions should be equal. This assumption results in a G:5H:S

ratio of approximately 13.8:71.0:15.2; i.e. we estimate

that the lignin is derived from approximately 71% of

5-hydroxylated monomers.

The inter-unit linkage distribution in the lignin fraction can

be determined from the aliphatic side-chain region of the

HSQC spectra (Table 2 and Figure S3). The correlation

contours in the spectra (Figure S3), such as those for b-aryl

ether (b–O–4) units (A), phenylcoumaran (b–5) units (B),

resinol (b–b) units (C), dibenzodioxocin (5–5/b–O–4) units (D)

and benzodioxane units (J), can be readily assigned by

comparison with model data and from lignin and cell-wall

spectra obtained previously (Ralph et al., 1999; Ralph and

Landucci, 2010). As for the aromatic regions in Figures 3 and

S2, the side-chain region of the comt C4H:F5H1 plants

(Figure S3e) is strikingly different from that of the wild type

(Figure S3a). Even though they are plotted at a lower

contour level (closer to noise level, to show the weak

residual signals from the traditional lignin structures), it

is evident that proportions of the various ‘normal’ lignin

structures (A–C) are minor. The obvious major difference is

the overwhelming presence of benzodioxane units (J) in the

comt C4H:F5H1 plants, for which the a- and b-correlations

are well resolved. No such correlations exist in the wild-type

spectrum. The volume integrals of the correlation contours

provide a measure of the relative contribution of each unit

type in the lignin polymer (Table 2). In the comt C4H:F5H1

plants, benzodioxanes account for approximately 92% of the

inter-unit linkage integrals measured. As noted, the 5H level

Table 2 Lignin composition as measured by NMR and total acetyl bromide lignin content

Line

Monomeric composition (%) Unit types (%) Acetyl bromide lignin(%)

G 5H S A B C D J Mean � SD n

Wild type 82.48 0.00 17.52 75.88 17.21 6.18 0.73 0.00 12.0 � 1.7 7comt 94.79 1.28 3.92 63.59 17.67 3.94 2.92 11.87 10.5 � 1.0 4C4H:F5H1 6.55 1.91 91.55 83.84 0.00 3.50 0.00 12.66 9.4 � 1.7 4chs 84.20 0.00 15.80 73.70 18.40 6.70 1.30 0.00 11.7 � 0.9 5comt C4H:F5H1 13.78 70.99 15.23 6.44 0.85 0.79 0.00 91.93 6.8 � 0.8*** 5comt C4H:F5H1 chs 33.54 20.02 46.44 59.07 1.42 1.50 0.00 38.02 8.8 � 1.1** 8

A, b-aryl ether; B, phenylcoumaran; C, resinol; D, dibenzodioxocin; J, benzodioxane (see Figure S3). n, number of plants analyzed.Statistical tests relative to the wild type were performed for total lignin only. Asterisks indicate values that are significantly different from the wildtype at **0.001 < P < 0.01 and ***P < 0.001, respectively.

(a) (b) (c) (d) (e)

Figure 3. Partial short-range 13C–1H (HSQC) spectra (aromatic regions) of acetylated enzyme lignins.

See Figure S2 for spectra of the complete set of transgenic lines and controls.

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 889

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

is approximately 71%; benzodioxane levels are higher

because they may result from addition of coniferyl or

sinapyl alcohol, as well as 5-hydroxyconiferyl alcohol, to

the 5-hydroxyguaiacyl end of the growing polymer, i.e. they

can be S, 5H or G benzodioxanes (see also Figure S1).

Phenolic profiling reveals benzodioxane-rich soluble

phenolics

To investigate whether perturbation of the monolignol bio-

synthetic pathway resulted in metabolites derived from

5-hydroxyconiferyl alcohol, a semi-quantitative comparison

of the profiles of methanol-soluble phenolics from comt

C4H:F5H1 and wild-type inflorescence stems was per-

formed. The profiles of wild-type stems were dominated by

sinapate esters and three flavonol glycosides (Figure S4). In

contrast, the comt C4H:F5H1 plants had lower levels of

sinapate esters and higher levels of flavonol glycosides, and

accumulated a large number of phenolic molecules that

were below the detection limit in wild-type plants (Fig-

ure S4). Elucidating their identities by MS2 spectral analyses

(Figure S5) (Morreel et al., 2004a,b; 2010a,b) revealed that

these metabolites were 5-hydroxylated phenylpropanoids.

Based on their chemical identity, they were grouped into

four main classes (Figure 1 and Table 3). Class I consists of

oligolignols with at least one and up to four 5H units (1–11).

Class II compounds (12–18) consist of hexosylated di- and

trilignols with at least one 5H unit. Class III contains

heterodimers of 5-hydroxyferulic acid with coniferyl,

5-hydroxyconiferyl or sinapyl alcohol, and their hexose

and malate derivatives (19–28), and class IV compounds

(29–38) are a series of derivatized 5-hydroxylated phenyl-

propanoids.

Chalcone synthase deficiency reduces C4H-driven gene

expression in comt C4H:F5H1 plants

Although the parental comt and C4H:F5H1 lines had a heal-

thy appearance, growth of the comt C4H:F5H1 plants was

severely affected. This phenotype is reminiscent of that of

hydroxycinnamoyl CoA:shikimate/quinate hydroxycinna-

moyl transferase (HCT) down-regulated Arabidopsis plants,

whose growth retardation was thought to result from inhi-

bition of auxin transport by the high levels of flavonoids in

these mutants (Besseau et al., 2007). Recent findings, how-

ever, disproved the existence of this link between flavonoids

and growth retardation in HCT down-regulated Arabidopsis

plants (Li et al., 2010). To test whether the increased levels of

flavonol glycosides (Figure S4) were responsible for growth

retardation in the comt C4H:F5H1 plants, a chalcone syn-

thase (chs) mutation was crossed into the comt C4H:F5H1

background. Soil-grown comt C4H:F5H1 chs mutants grew

significantly taller than comt C4H:F5H1 plants (Figure 2l and

Table 1), and produced viable seeds. However, such resto-

ration of the phenotype could also be caused by negative

feedback of the chs mutation on the C4H promoter and thus

on C4H:F5H1 expression. To investigate this, quantitative

RT-PCR was performed on inflorescence stems. Expression

of C4H and F5H1 was significantly reduced in comt C4H:F5H1

chs plants compared to C4H:F5H1 and comt C4H:F5H1 plants

(Figure 4a). The benzodioxane level in the lignin of comt

C4H:F5H1 chs plants was reduced to 38%, much less than the

92% found in comt C4H:F5H1 plants (Figure S3j and

Table 2). Furthermore, comt C4H:F5H1 chs plants had higher

lignin amounts compared to comt C4H:F5H1 plants, but less

than the wild-type (Table 2). Thus, the partial restoration in

growth can be explained by inhibition of C4H:F5H1 expres-

sion.

To further explore possible feedback regulation within the

phenylpropanoid pathway in comt C4H:F5H1 and comt

C4H:F5H1 chs plants, we measured the expression of

phenylpropanoid biosynthetic genes in 14-day-old seedling

leaves. At this stage, the visual phenotype is much less

pronounced (Figure 2a). Strikingly, the expression of many

general phenylpropanoid biosynthetic genes was higher in

C4H:F5H1, comt C4H:F5H1 and comt C4H:F5H1 chs plants

compared to the wild type, suggesting a higher flux through

the pathway (Figure 4b). In contrast to inflorescence stems,

however, C4H gene expression was similar between comt

C4H:F5H1 and comt C4H:F5H1 chs in the seedling leaves,

indicating developmental differences in feedback regula-

tion.

DISCUSSION

Lignins primarily consist of benzodioxane structures

Plants deficient in COMT and plants over-expressing F5H

were previously shown to incorporate small amounts of

5-hydroxyconiferyl alcohol into their lignins, demonstrating

that 5-hydroxyconiferyl alcohol can be transported, single-

electron-oxidized (‘radicalized’), and co-polymerized into

lignins (Atanassova et al., 1995; Van Doorsselaere et al.,

1995; Jouanin et al., 2000; Guo et al., 2001; Ralph et al.,

2001a,b; Sibout et al., 2002; Marita et al., 2003; Morreel

et al., 2004a,b; Do et al., 2007; Lu et al., 2010). Its incorpo-

ration gives rise to specific units, i.e. benzodioxanes. A

benzodioxane results from coupling of a monolignol at its

b-position with a 5-hydroxyguaiacyl unit at its 4-O position,

followed by internal trapping of the resulting quinone met-

hide by the 5-OH (Ralph et al., 2001a,b). Although these

benzodioxane units are the result of typical b–O–4 coupling

reactions that also take place in natural G/S-type lignins, the

less flexible benzodioxane ring structures alter the macro-

molecular configuration of the lignin polymer. Strikingly,

when Arabidopsis F5H1 is over-expressed in a comt mutant

background, the lignin almost exclusively comprises ben-

zodioxane structures, accounting for approximately 92% of

the units (Figure S3 and Table 2). This study shows that

lignin can be composed almost entirely from units that are

undetectable in wild-type plants.

890 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

Table 3 Metabolites in comt C4H:F5H1 plants, arbitrarily classified for convenience

Peak number Retention time (min) m/z Molecule number Name

Class I: oligolignols1 18.5 373 1 G(b-O-4)5H2 18.4 403 2 S(b-O-4)5H3 21.6 371 3 G(b-O-4)5H’4 19.0 387 4 5H(b-O-4)5H’5 21.2 401 5 S(b-O-4)5H’6 21.1 567 6 G(b-O-4)5H(b-O-4)5H7 21.4 567 6 G(b-O-4)5H(b-O-4)5H8 24.0 565 7 G(b-O-4)5H(b-O-4)5H’9 24.2 565 7 G(b-O-4)5H(b-O-4)5H’10 21.4 569 8 G(b-O-4)5H(b-O-4)dihydro5H11 21.6 569 8 G(b-O-4)5H(b-O-4)dihydro5H12 22.8 761 9 G(b-O-4)5H(b-O-4)5H(b-O-4)5H13 23.0 761 9 G(b-O-4)5H(b-O-4)5H(b-O-4)5H14 25.4 759 10 G(b-O-4)5H(b-O-4)5H(b-O-4)5H’15 25.8 759 10 G(b-O-4)5H(b-O-4)5H(b-O-4)5H’16 26.4 953 11 G(b-O-4)5H(b-O-4)5H(b-O-4)5H(b-O-4)5H’

Class II: hexosylated oligolignols17 13.6 535 12 G(b-O-4)5H hexosidea

18 14.0 535 12 G(b-O-4)5H hexosidea

19 15.4 535 12 G(b-O-4)5H hexosidea

20 12.7 551 13 5H(b-O-4)5H hexosidea

21 14.0 551 13 5H(b-O-4)5H hexosidea

22 14.4 551 13 5H(b-O-4)5H hexosidea

23 16.6 549 14 5H(b-O-4)5H’ hexoside24 15.4 537 15 G(b-O-4)c-O-hexosyl dihydro5H25 16.3 567 16 S(b-O-4)c-O-hexosyl dihydro5H26 17.2 789 17 G(b-O-4)5H(b-O-4)5H hexoside27 17.5 789 17 G(b-O-4)5H(b-O-4)5H hexoside28 17.5 791 18 G(b-O-4)5H(b-O-4)dihydro5H hexoside29 17.6 791 18 G(b-O-4)5H(b-O-4)dihydro5H hexoside

Class III: 5-hydroxyferulic acid-containing dimers30 19.0 387 19 G(b-O-4)5-hydroxyferulic acid31 15.2 549 20 G(b-O-4)5-hydroxyferuloyl hexose32 15.9 549 20 G(b-O-4)5-hydroxyferuloyl hexose33 9.7 549 21 G(b-O-4)5-hydroxyferulic acid hexosidea

34 16.1 565 22 5H(b-O-4)5-hydroxyferuloyl hexose35 9.9 565 23 5H(b-O-4)5-hydroxyferulic acid hexosidea

36 16.0 579 24 S(b-O-4)5-hydroxyferuloyl hexose37 16.6 579 24 S(b-O-4)5-hydroxyferuloyl hexose38 10.0 579 25 S(b-O-4)5-hydroxyferulic acid hexosidea

39 10.1 503 26 G(b-O-4)5-hydroxyferuloyl malate40 10.7 503 26 G(b-O-4)5-hydroxyferuloyl malate41 8.0 519 27 5H(b-O-4)5-hydroxyferuloyl malate42 8.9 519 27 5H(b-O-4)5-hydroxyferuloyl malate43 9.6 519 27 5H(b-O-4)5-hydroxyferuloyl malate44 10.8 533 28 S(b-O-4)5-hydroxyferuloyl malate

Class IV: derivatives of 5-hydroxylated phenylpropanoid monomers45 2.7 371 29 5-hydroxyferulic acid hexoside46 2.8 371 29 5-hydroxyferulic acid hexoside47 8.3 371 30 5-hydroxyferuloyl hexose48 9.2 371 30 5-hydroxyferuloyl hexose49 10.5 525 31 5-hydroxyferuloyl hexose+155b

50 12.9 525 31 5-hydroxyferuloyl hexose+155b

51 7.7 533 32 5-hydroxyferuloyl hexose hexoside52 18.8 563 33 di-5-hydroxyferuloyl hexose53 3.4 373 34 dihydro 5-hydroxyferuloyl hexose54 7.4 369 35 5-hydroxyscopoletin hexoside55 10.3 355 36 5-hydroxyconiferaldehyde hexoside

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 891

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

The comt C4H:F5H1 plants accumulate large amounts

of 5-hydroxylated phenylpropanoids and their derivatives

Phenolic profiling of the methanol-soluble fraction revealed

many 5-hydroxylated phenylpropanoids (including alco-

hols, aldehydes, acids, and their derivatives) that were

undetectable in the extracts of wild-type plants (Figure 1

and Table 3). Class I oligolignols (1–11) are small lignin

polymers with chains of benzodioxane structures, reflect-

ing the high level of 5-hydroxyconiferyl alcohol monomers

incorporated into the lignin polymer. Although the phe-

nolic end unit can be a 5H unit, as seen in compound 4,

such units are inherently prone to oxidation to quinones

and beyond. Capping such end units by coupling with a

traditional monolignol stabilizes them. Indeed, most of the

oligomers identified were found to be capped in this way.

Interestingly, except for the cap, all other units in the

identified oligolignols were 5H units, i.e. no mixed oligo-

lignols were found with normal G or S units within the

chain. End capping was also observed in class II and III

compounds. Thus, although not all differential peaks were

structurally resolved, the structures of the most abundant

ones suggest that 5-hydroxylated monomers do not readily

couple b–O–4 to neither G or S units. Furthermore, it is

important to note that these small lignin oligomers involve

only b–O–4 binding, because b–5 coupling cannot occur

due to the hydroxylation at C5 in the 5H units. Because of

the instability of 5H end groups, the only b–b-coupled

compounds likely to be found would be tetramers at least,

in which both phenolic ends of a b–b-coupled dimer are

capped.

Classes II, III and IV comprise hexose and malate

derivatives of 5-hydroxyconiferyl alcohol, 5-hydroxyconif-

eraldehyde or 5-hydroxyferulate, with the latter compound

potentially derived from 5-hydroxyconiferaldehyde via

hydroxycinnamaldehyde dehydrogenase/reduced epidermal

fluorescence 1 (HCALDH/REF1) as described for the con-

version from coniferaldehyde and sinapaldehyde to ferulic

acid and sinapic acid, respectively (Nair et al., 2004). The

class III molecules (19–28) are dimers of 5-hydroxyferulate

with coniferyl or sinapyl alcohol and their hexose and

malate derivatives. These molecules are 5-hydroxy analogs

of the G(8-O-4)feruloyl malate that is found in wild-type

Arabidopsis (Rohde et al., 2004). Although the biological

role of this compound is unknown, G(8-O-4)feruloyl malate

has been suggested to be involved in the coupling of

hemicellulose to lignin (Rohde et al., 2004). The high

abundance of 5-hydroxyferulate esters (30–33) in class IV

is also noteworthy. The 5-hydroxyferulate esters might

functionally replace the biological role of sinapate esters

in vivo, because epidermal fluorescence upon illumination

with UV (366 nm) is observed in leaves of comt C4H:F5H1

plants, indicative of their UV protective role (Ruegger and

Chapple, 2001). This fluorescence is absent from f5h1

(fah1) mutants that produce neither 5-hydroxyferulate nor

sinapate esters (Ruegger and Chapple, 2001). We deduce

that 5-hydroxylated phenylpropanoids and their derivatives

originate partly from passive metabolism via existing

biochemical routes, and partly from active detoxification.

In passive metabolism, 5-hydroxylated phenylpropanoids

are surrogates for their 5-methoxylated and 5-unsubstitut-

ed analogs. For example, the presence of di-5-hydroxyfer-

uloyl hexose (33) suggests that the enzymes involved in

disinapoyl hexose biosynthesis also recognize and metab-

olize 5-hydroxylated analogs (Fraser et al., 2007). On the

other hand, UDP-glucose-dependent glycosyltransferases

(UGTs) have been shown to be involved in the detoxifica-

tion of xenobiotics (Welinder et al., 2002; Lim and Bowles,

2004; Brazier-Hicks et al., 2007). It is thus anticipated that

they will also detoxify the accumulating 5-hydroxylated

phenolics, thereby giving rise to the hexosylated com-

pounds of classes II, III and IV.

The subcellular location of synthesis of these compounds

remains obscure; oxidative coupling occurs in the cell wall

and possibly also in the vacuole, given the high number of

peroxidases in this compartment (Welinder et al., 2002).

However, glycosylation occurs in the cytoplasm but not in

the vacuole, and malate transferase activity is present in the

vacuole. Unraveling the subcellular routes for biosynthesis

of the class II–IV compounds requires a new research

approach that combines subcellular fractionation and phe-

nolic profiling.

Table 3 (Continued)

Peak number Retention time (min) m/z Molecule number Name

56 8.6 357 37 5-hydroxyconiferyl alcohol hexoside57 9.2 357 37 5-hydroxyconiferyl alcohol hexoside58 5.9 579 38 hexosyl 5-hydroxyconiferyl alcohol hexoside59 7.6 579 38 hexosyl 5-hydroxyconiferyl alcohol hexoside

The shorthand naming convention for oligolignols is described by Morreel et al. (2004a,b). In short, G, 5H, S and 5H’ are used for units derived fromconiferyl alcohol, 5-hydroxyconiferyl alcohol, sinapyl alcohol and 5-hydroxyconiferaldehyde, respectively. Here, b-O-4 (normally used for lignins)is used instead of 8-O-4 (normally used for metabolites) to ensure consistent nomenclature throughout the text.aThe hexose is not on the c-O position of the b-coupled unit.bThe structure of the 155 Da subunit has not yet been resolved.

892 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

Genetic interactions within phenylpropanoid biosynthesis

Although the comt and C4H:F5H1 parental lines have nor-

mal lignin levels, comt C4H:F5H1 plants deposit signifi-

cantly less lignin. Moreover, comt C4H:F5H1 plants remain

small and their xylem vessels have an irx phenotype. One

hypothesis to explain the lower lignin levels is that

5-hydroxyconiferyl alcohol is not efficiently translocated

towards the cell wall and/or not efficiently incorporated into

lignin. The reduction in the lignin levels might be sufficient

to explain the irx phenotype (Jones et al., 2001; Franke

et al., 2002; Goujon et al., 2003; Hoffmann et al., 2004;

Besseau et al., 2007; Mir Derikvand et al., 2008; Schilmiller

et al., 2009; Li et al., 2010), but the altered structural and

physico-chemical properties of the 5H lignin may also

contribute to the irx phenotype. In addition to the collapsed

xylem, accumulating phenolic precursors might cause

aberrant growth, because they are toxic to cellular

processes. Indeed, the phenolic profile of comt C4H:F5H1

plants revealed numerous hexosylated (i.e. detoxified)

compounds. Accumulation of these compounds is logical

given that the sink, lignin, is reduced. However, we also

observed a massive up-regulation of multiple phenylprop-

anoid pathway transcripts in the comt C4H:F5H1 plants

(Figure 4). It is tempting to believe that the reduction in

cell-wall integrity, possibly due to the shortage of lignin in

the cell wall, results in a greater demand for biosynthesis of

coniferyl and sinapyl alcohols, and thus a higher flux

through the phenylpropanoid pathway (Figure 4). Up-

regulation of the phenylpropanoid pathway might thus be

part of the plant’s response to an aberrant cell-wall structure

(Cano-Delgado et al., 2003; Rogers and Campbell, 2004).

We also tested whether the growth abnormalities in the

comt C4H:F5H1 plants could have been caused by the higher

flavonoid levels, as suggested by Besseau et al. (2007).

However, in contrast with their results, and in support of the

recent data by Li et al. (2010), our data indicate that

flavonoids are certainly not the sole cause for the growth

defect in comt C4H:F5H1 plants, as crossing in a chs

mutation was not able to fully restore the phenotype.

Instead, the main reason for the partial recovery in growth

observed in comt C4H:F5H1 chs plants was reduction of

C4H-driven gene expression caused by the chs mutation,

resulting in lower over-expression of the F5H1 gene and a

consequent a drop in the frequency of benzodioxane units

(resulting from a reduction in the proportion of 5-hydroxy-

coniferyl alcohol monomers) and an increase in total lignin

content (Table 2). The reason for reduction of this C4H-

driven gene expression may either be negative feedback on

the C4H promoter, or positive feedback on the C4H promoter

that subsequently triggered gene silencing. In either case,

the data clearly demonstrate a genetic interaction between

the flavonoid and phenylpropanoid pathways. Note that this

interaction in inflorescence stems was dependent on the

genetic background: the chs mutation reduced C4H-driven

gene expression only in comt C4H:F5H1 chs plants, but not

in the single chs mutant. In addition, the reduction of C4H-

driven gene expression appeared to be dependent on the

developmental stage, as it was observed in stems but not in

seedling leaves (Figure 4). It is furthermore important to

note that the C4H promoter is one of the promoters of choice

to drive transgene expression in lignifying cells in order to

modify lignin biosynthesis in crop species (Huntley et al.,

2003; Stewart et al., 2009). The observation that the expres-

sion driven by this promoter is regulated, possibly in

response to altered concentrations of pathway intermedi-

ates, has repercussions on its use in gene engineering

strategies. Such poorly studied genetic interactions within

the phenylpropanoid pathway, for which full elucidation

requires a systems approach, need to be taken into account

during rational design of plant cell walls with improved

properties for agro-industrial applications.

(a)

(b)

Figure 4. Expression of phenylpropanoid biosynthetic genes in the various

mutants as determined by quantitative RT-PCR.

(a) Expression of C4H and F5H1 in inflorescence stems. Error bars are

standard deviations. Significant differences in expression between the wild

type and the mutant (P < 0.05) are indicated by an asterisk.

(b) Expression in 14-day-old seedling leaves. Values are the ratio of the

expression in the mutant line versus that in the wild type: a value >1 indicates

up regulation. Standard deviations are given in parentheses. Red and blue

backgrounds indicate higher and lower gene expression relative to the wild

type, respectively. The color intensity reflects the strength of induction and

reduction. Values that are significantly different from those of the wild type

(P < 0.05) are shown in bold.

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 893

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

EXPERIMENTAL PROCEDURES

Plant material

All experiments were performed with Arabidopsis ecotype Colum-bia (Col-0). The comt mutant (At5g54160, SALK_002373) and chsmutant (At5g13930, SALK_020583) were obtained from the SALKcollection (Alonso et al., 2003). Zygosity was analyzed using prim-ers 1 (5¢-ATGCCAAAGAGTCATGTTTTA-3¢) and 2 (5¢-TGGTG-TCTCTGGAAGTATAC-3¢) to amplify the endogenous COMT gene,and primers 1 and 3 (5¢-GCCCTTTGACGTTGGAGTC-3¢) to amplifythe comt T-DNA left-border insertion region. For CHS, primers 4(5¢-CTCGTCGGTCAGGCTCTTT-3¢) and 5 (5¢-TTCCATACTCGCTCAA-CACG-3¢) were used for the endogenous gene, and primers 3 and 4for the chs T-DNA left-border insertion region. The C4H:F5H1transgenic line (C4H-F5H) in the fah1-2 mutant background (Meyeret al., 1998) was kindly provided by Clint Chapple (Department ofBiochemistry, Purdue University, IN). This line over-expresses F5H1(At4g36220) under the control of the C4H (At2g30490) promoter.The presence of the C4H:F5H1 construct was analyzed usingprimers 5¢-CGTCGTGGGATCTGATAGTT-3¢ and 5¢-CTCCCTCTTA-TAAAATCCTCTC-3¢. To confirm the presence of the fah1-2 pointmutation, primers 5¢-ATGGAGAACTTTGCCTCCTG-3¢ and 5¢-GGA-TAAACAAGCGGCTCGT-3¢ were used. The C4H:F5H1 transgene isnot amplified by these primers. The 907 bp PCR product was furtherdigested using MseI (New England Biolabs Inc., http://www.neb.com), resulting in DNA fragments of 398, 225, 221, 56 and 7 bp forthe wild-type F5H1 and 267, 225, 221, 131, 56 and 7 bp for themutant fah1-2 allele. The comt C4H:F5H1 fah1-2, comt C4H:F5H1and C4H:F5H1 lines were isolated from the F2 progeny of a comt xC4H-F5H1 fah1-2 cross. Because comt C4H:F5H1 fah1-2 andcomt C4H:F5H1 plants are sterile, these genotypes were generatedfrom the parental plants heterozygous for comt and homozygousfor C4H:F5H1 and fah1-2. All experiments described were performedon comt C4H:F5H1 plants. However, comt C4H:F5H1 fah1-2 plantshad similar phenotypes.

Plant growth conditions

Agar-grown plants were grown at 23�C, with 16 h of light per day,on the following medium: 0.43% Murashige and Skoog vitaminmedium/basal salt mixture (M0221.0050, Duchefa Biochemie, http://www.duchefa.com), 1% sucrose (177140010, Acros, http://www.acros.com), 0.05% MES (M1503.0100, Duchefa Biochemie), pH 5.7,0.725% agar (A1296, Sigma, http://www.sigmaaldrich.com/) and0.1% 1000· Murashige and Skoog vitamin solution (M3900, Sigma).Soil-grown plants were germinated directly in Saniflor soil (VanIsrael; http://vanisrael.testvds.com) supplemented with 10% v/vvermiculite (16 h light, 22�C, 55% humidity).

Biomass

For biomass measurements, plants were grown in soil until senes-cence. Cauline leaves and siliques were removed from the inflo-rescence before weighing. Statistical analysis was performed usingANOVA followed by the Scheffe post hoc test.

Microscopy

Electron microscopy and lignin staining were performed essentiallyas described previously (Rohde et al., 2004). Stem segments werecut into small pieces and fixed with 4% paraformaldehyde and 2.5%glutaraldehyde, and post-fixed in 1% OsO4 with 1.5% K3Fe(CN)6 in0.1 M sodium cacodylate buffer, pH 7.2. Samples were dehydratedthrough a graded ethanol series, including bulk staining with 2%uranyl acetate at the 50% ethanol step, followed by embedding inSpurr’s resin. For tissue morphology, semi-thin sections (0.5 lm)

were cut using an Ultracut E microtome (Reichert-Jung, http://www.leica.com), stained in a 1% toluidine blue and 2% boratesolution in distilled water, and viewed using a Leica DM LB micro-scope (http://www.leica.com/).

For electron microscopy, ultra-thin sections of a gold interferencecolor were produced on an Ultracut E microtome (Reichert-Jung),and then post-stained in an ultrastainer (Leica) using uranyl acetateand lead citrate. They were viewed using a JEOL 1010 transmissionelectron microscope (JEOL, http://www.jeol.com).

Lignin histochemistry was performed by Maule staining aspreviously described (Rohde et al., 2004). Briefly, agar-imbeddedstem segments were sectioned with a vibroslicer, incubated in 4%KMnO4 for 5 min, rinsed with water, incubated in 37% HCl/H2O (1:1)for 2 min, and observed after addition of a drop of aqueous NH3.Sections were viewed under a binocular microscope (Leica MZ16).

Phenolic profiling and MS analysis

Single inflorescences, lacking cauline leaves, flowers and siliques,were sampled and immediately frozen in liquid nitrogen. Ten wholetissue-culture grown stems between 2.2 and 6.0 cm long wereobtained from comt C4H:F5H1 plants and compared with five wild-type whole stems of approximately 8.0 cm, as well as five apical andfive basal stem segments of 5.0 cm from approximately 20 cm tallwild-type plants. Samples were transferred to 1.5 ml Eppendorftubes. After grinding, the samples were extracted in 600 ll THF/isopropanol/methanol (15:15:70). The extract (550 ll) was freeze-dried in a SAVANT Speedvac (SC21A SpeedVac Concentrator,Thermo, http://www.thermo.com), and subjected to extractionusing water/cyclohexane (100 ll of each). Each individual sample(60 ll) was injected by means of a SpectraSystem AS1000 auto-sampler (Thermo Separation Products) onto a reverse-phase LunaC18(2) column (150 · 2.1 mm, 3 lm; Phenomenex, http://www.phenomenex.com).

Gradient separation was performed on a SpectraSystemP1000XR HPLC pump (Thermo Separation Products), using agradient of two buffers: buffer A (water/acetonitrile (ACN)/aceticacid, 100:1:0.1, pH 2) and buffer B (ACN/water/acetic acid, 100:1:0.1,pH 2). A flow rate of 0.25 ml min)1 was used, with a buffercomposition starting at 5% B, linearly increasing to 17% at 1 min,77% at 33 min, and finally 100% at 36 min. A SpectraSystemUV6000LP detector (Thermo Separation Products) was used tomeasure UV/vis absorption between 200 and 450 nm with a scanrate of 2 Hz. Atmospheric pressure chemical ionization (APCI)operated in the negative-ion mode, was used as an ion source tocouple HPLC to LCQ Classic MS instrument (ThermoQuest) (vapor-izer temperature 350�C, capillary temperature 140�C, source current5 mA, sheath gas flow set at 27, auxiliary gas flow set at 3, m/z range140–1000, normalized collision energy for MS2 35).

NMR sample preparation

NMR samples were prepared essentially as described previously(Weng et al., 2010). Senesced inflorescence stems of Arabidopsis(approximately 200 mg) from tissue culture- and soil-grown plants(Figures S1 and S2) were pre-ground in a Retsch MM400 mixer mill(http://retsch.com), equipped with a 10 ml ZrO2 vessel and two ZrO2

ball-bearings (12 mm diameter), for 1 min at 30 Hz. The groundmaterial was extracted three times with water and subsequentlythree times with 80% ethanol by sonicating in an ultrasonic bath for30 min each time. The freeze-dried extractive-free Arabidopsissamples (approximately 100 mg) were ball-milled using a RetschPM 100 planetary ball mill in a 50 ml ZrO2 vessel with 10 ZrO2 ball-bearings (10 mm diameter) at 600 rpm, with 5 min breaks afterevery 5 min of milling. The total ball-milling time for the sampleswas 16.5 min. Cellulolytic enzyme lignins (CEL) were then isolated

894 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

by enzymatically saccharifying (most of the) polysaccharides asdescribed by Chang et al. (1975). Cellulysin cellulase (EC 3.2.1.4,Calbiochem, http://www.calbiochem.com), a crude cellulase prep-aration containing hemicellulase activities, from Trichoderma viridewas used. Its activity was 10 000 units g)1 dry weight. The ball-milled material was suspended in 20 mM NaOAc buffer (30 ml, pH5.0) in a 50 ml centrifuge tube, 8 mg of Cellulysin was added, andthe reaction slurry was incubated at 35�C for 48 h. The solids werepelleted by centrifugation (1500 g, 21�C, 40 min), and the processwas repeated with fresh buffer and enzyme. CEL was recovered byfiltration through a nylon membrane (0.2 lm). The product waswashed with water and dried under vacuum at room temperature,yielding 16–28 mg CEL, depending on the sample. CEL (12 mg) wasacetylated using 2 ml DMSO, 1 ml N-methylimidazole and 0.6 mlacetic anhydride, stirring for 1.5 h at room temperature, beforequenching the reaction by pouring the solution into 1 L of distilledwater. The precipitated material was recovered by filtration througha nylon membrane (0.2 mm). Acetylated CEL, typically approxi-mately 15–18 mg from 12 mg CEL, was dried under vacuum anddissolved in 0.5 ml CDCl3 for NMR analysis. For the comt C4H:F5H1line, only approximately 3–4 mg of acetylated CEL was obtainedafter filtering out unidentified gelatinous CDCl3-insoluble material.Nevertheless, the 2D HSQC spectrum was of excellent quality, and itwas even possible to obtain a 3D TOCSY-HSQC spectrum from thissample.

NMR experiments

NMR spectra were acquired on a Bruker-Biospin AVANCE 500 MHzspectrometer (Bruker-Biospin, http://www.bruker-biospin.com)fitted with a cryogenically cooled 5 mm Bruker TCI (triple resonancecryoprobe optimized for 1H and 13C observation) gradient probewith inverse geometry (proton coils closest to the sample). The 2D13C–1H correlation spectra were acquired using an adiabatic HSQCpulse program (Bruker standard pulse sequence ‘hsqcetgpsisp2.2¢)(Kupce and Freeman, 2007) and the following parameters: spectrawere acquired from 10 to 0 ppm (5000 Hz) in F2 (1H) using 1000 datapoints for an acquisition time (AQ) of 200 msec, an interscan delay(D1) of 500 msec, and from 196 to )23 ppm (25 154 Hz) in F1 (13C)using 400 increments (F1 acquisition time 8 msec) of 90 scans, for atotal acquisition time of 17 h 17 min. The number of scans can beadjusted depending on the signal-to-noise required from a sample.The 1JCH used was 145 Hz. Processing used typical matchedGaussian apodization in the 1H dimension and squared cosine-bellapodization in the 13C dimension. Prior to Fourier transformation,the data matrices were zero-filled to 1024 points in the 13C dimen-sion. The central chloroform solvent peak at a dC/dH of 77.0/7.26 ppm was used as an internal reference. The 3D TOCSY-HSQCspectrum was acquired and processed exactly as described previ-ously, providing similar evidence for the benzodioxane units(Marita et al., 2001).

G:5H:S integral ratios were obtained by integrating the contoursfrom the 2 or 2,6 positions of each type of aromatic unit (with Sintegrals being logically halved as the 2 and 6 positions are identicalin the symmetrical S units). In samples with 5H correlations, wherethe 5H6 correlations overlap with the S2/6 correlations, the S integralwas calculated by subtracting the 5H2 integral (which should be thesame as the 5H6 integral) from the integral of the overlapping S2/6 +5H6 contours (Figure 3b–e).

Lignin quantification

Cell-wall material was prepared from 5 mg of senescent inflores-cence stems of soil-grown plants, washed sequentially with 0.5 mlwater, ethanol, chloroform and acetone for 30 min each at 98, 76, 59

and 45�C, respectively. Acetyl bromide lignin extraction was per-formed according a downscaled method similar to that describedpreviously (Dence, 1992). Briefly, 0.1 ml acetyl bromide (25% inacetic acid) and 4 ll 60% perchloric acid were added to the dry cellwall and the mixture was incubated at 70�C for 30 min. NaOH (2 M,0.2 ml) and 0.5 ml acetic acid were then added. After centrifugation(15 000 g, room temperature, 15 min), the supernatant was sepa-rated from the pellet. The pellet was further washed by adding0.5 ml acetic acid (vortex, 5 s), followed by centrifugation (15 000 g,room temperature, 15 min). The second supernatant was combinedwith the first. Acetic acid was added to the combined supernatantsto a volume of 2 ml, and absorption was measured at 280 nm usinga NanoDrop ND-1000 spectrophotometer (Nanodrop, http://www.nanodrop.com). An extinction coefficient for lignin at 280 nm of23.35 L g)1 cm)1 was used (Chang et al., 2008). As for every ligninquantification method, the response varies with the composition ofthe lignin; the acetyl bromide protocol gives only an estimate of thetotal lignin amount (Hatfield and Fukushima, 2005) and we used thesame lignin UV280 extinction coefficient for all samples here.Statistical tests were performed as indicated for biomassmeasurements.

RNA extraction, cDNA preparation and quantitative RT-PCR

Total RNA was extracted from frozen soil-grown inflorescencestems and tissue-cultured seedling leaves using a plant RNeasyextraction kit (Qiagen, http://www.qiagen.com/), and quantifiedusing a NanoDrop ND-1000 spectrophotometer. Total RNA (4.5 lg)was reverse-transcribed in a final volume of 190 ll using a Super-script II reverse transcription kit (Invitrogen, http://www.invitro-gen.com/). Quantitative RT-PCR was performed using a SYBR GreenI master kit (Roche, http://www.roche.com) on a Lightcycler 480(Roche). The primer sequences used for quantitative RT-PCR aregiven in Table S1. Gene expression was normalized on the basis ofACTIN expression. Statistical tests were performed using a procglmnested model with unequal variances (SAS 9.2; SAS Institute Inc.,http://www.sas.com).

ACKNOWLEDGEMENTS

The authors thank Clint Chapple (Department of Biochemistry,Purdue University, West Lafayette, IN) for kindly providing theC4H:F5H1 fah1-2 line, Bart Ivens and David Casini for practicalassistance, and Martine De Cock for help in preparing the manu-script. We gratefully acknowledge partial funding through the Uni-ted States Department of Energy (DOE) Energy Biosciencesprogram (grant number DE-AI02-00ER15067) and the DOE GreatLakes Bioenergy Research Center (DOE Office of Science, grantnumber BER DE-FC02-07ER64494) to J.R., the Research Foundation-Flanders (grant number G.0352.05N), the European Community’s7th Framework Programme (FP7/2007) under grant agreement no.211982 (RENEWALL), the Global Climate and Energy Project (GCEP)(grants to W.B. for ‘Towards New Degradable Lignin Types’ and toJ.R. for ‘Efficient Biomass Conversion: Delineating the Best LigninMonomer Substitutes’), and the Multidisciplinary Research Part-nership ‘Biotechnology for a Sustainable Economy’ (01MRB510W)of Ghent University. Some of the NMR experiments on the BrukerDMX-500 cryoprobe system made use of the National MagneticResonance Facility at Madison (http://www.nmrfam.wisc.edu). R.V.is indebted to the Agency for Innovation by Science and Technologyfor a pre-doctoral fellowship.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the onlineversion of this article:

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 895

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

Figure S1. Percentages of 5H and benzodioxane units in lignins ofvarious mutant and transgenic plant lines.Figure S2. Partial short-range 13C–1H (HSQC) spectra of acetylatedenzyme lignins in aromatic regions.Figure S3. Partial short-range 13C–1H (HSQC) spectra of acetylatedenzyme lignins in aliphatic regions.Figure S4. UV/vis absorption chromatogram of inflorescence stemextracts of comt C4H:F5H1 and wild-type plants.Figure S5. MS spectra of 38 5-hydroxylated phenylpropanoids thatwere identified in comt C4H:F5H1 plants.Table S1. Primers used in quantitative RT-PCR experiments.Please note: As a service to our authors and readers, this journalprovides supporting Information supplied by the authors. Suchmaterials are peer-reviewed and may be re-organized for onlinedelivery, but are not copy-edited or typeset. Technical supportissues arising from supporting Information (other than missingfiles) should be addressed to the authors.

REFERENCES

Alonso, J.M., Stepanova, A.N., Leisse, T.J. et al. (2003) Genome-wide inser-

tional mutagenesis of Arabidopsis thaliana. Science, 301, 653–657.

Atanassova, R., Favet, N., Martz, F., Chabbert, B., Tollier, M.T., Monties, B.,

Fritig, B. and Legrand, M. (1995) Altered lignin composition in transgenic

tobacco expressing O-methyltransferase sequences in sense and antisense

orientation. Plant J. 8, 465–477.

Besseau, S., Hoffmann, L., Geoffroy, P., Lapierre, C., Pollet, B. and Legrand, M.

(2007) Flavonoid accumulation in Arabidopsis repressed in lignin synthesis

affects auxin transport and plant growth. Plant Cell, 19, 148–162.

Boerjan, W., Ralph, J. and Baucher, M. (2003) Lignin biosynthesis. Annu. Rev.

Plant Biol. 54, 519–546.

Brazier-Hicks, M., Offen, W.A., Gershater, M.C., Revett, T.J., Kim, E.-K.,

Bowles, D.J., Davies, G.J. and Edwards, R. (2007) Characterization and

engineering of the bifunctional N- and O-glucosyltransferase involved in

xenobiotic metabolism in plants. Proc. Natl Acad. Sci. USA, 104, 20238–

20243.

Brown, D.M., Zeef, L.A.H., Ellis, J., Goodacre, R. and Turner, S.R. (2005)

Identification of novel genes in Arabidopsis involved in secondary cell wall

formation using expression profiling and reverse genetics. Plant Cell, 17,

2281–2295.

Cano-Delgado, A., Penfield, S., Smith, C., Catley, M. and Bevan, M. (2003)

Reduced cellulose synthesis invokes lignification and defense responses in

Arabidopsis thaliana. Plant J. 34, 351–362.

Chang, H.-M., Cowling, E.B., Brown, W., Adler, E. and Miksche, G. (1975)

Comparative studies on cellulolytic enzyme lignin and milled wood lignin

of sweetgum and spruce. Holzforschung, 29, 153–159.

Chang, X.F., Chandra, R., Berleth, T. and Beatson, R.P. (2008) Rapid, micro-

scale, acetyl bromide-based method for high-throughput determination of

lignin content in Arabidopsis thaliana. J. Agric. Food Chem. 56, 6825–6834.

Dence, C.W. (1992) Lignin determination. In Methods in Lignin Chemistry:

Springer Series in Wood Science (Lin, S.Y. and Dence, C.W., eds). Berlin:

Springer-Verlag, pp. 33–61.

Do, C.-T., Pollet, B., Thevenin, J., Sibout, R., Denoue, D., Barriere, Y., Lapierre,

C. and Jouanin, L. (2007) Both caffeoyl coenzyme A 3-O-methyltransferase

1 and caffeic acid O-methyltransferase 1 are involved in redundant func-

tions for lignin, flavonoids and sinapoyl malate biosynthesis in Arabidop-

sis. Planta, 226, 1117–1129.

Franke, R., McMichael, C.M., Meyer, K., Shirley, A.M., Cusumano, J.C. and

Chapple, C. (2000) Modified lignin in tobacco and poplar plants over-

expressing the Arabidopsis gene encoding ferulate 5-hydroxylase. Plant J.

22, 223–234.

Franke, R., Hemm, M.R., Denault, J.W., Ruegger, M.O., Humphreys, J.M. and

Chapple, C. (2002) Changes in secondary metabolism and deposition of an

unusual lignin in the ref8 mutant of Arabidopsis. Plant J. 30, 47–59.

Fraser, C.M., Thompson, M.G., Shirley, A.M., Ralph, J., Schoenherr, J.A.,

Sinlapadech, T., Hall, M.C. and Chapple, C. (2007) Related Arabidopsis

serine carboxypeptidase-like sinapoylglucose acyltransferases display

distinct but overlapping substrate specificities. Plant Physiol. 144, 1986–

1999.

Goujon, T., Ferret, V., Mila, I., Pollet, B., Ruel, K., Burlat, V., Joseleau, J.-P.,

Barriere, Y., Lapierre, C. and Jouanin, L. (2003) Down-regulation of the

AtCCR1 gene in Arabidopsis thaliana: effects on phenotype, lignins and cell

wall degradability. Planta, 217, 218–228.

Guo, D., Chen, F., Inoue, K., Blount, J.W. and Dixon, R.A. (2001) Downregu-

lation of caffeic acid 3-O-methyltransferase and caffeoyl CoA 3-O-methyl-

transferase in transgenic alfalfa: impacts on lignin structure and

implications for the biosynthesis of G and S lignin. Plant Cell, 13, 73–88.

Hatfield, R. and Fukushima, R.S. (2005) Can lignin be accurately measured?

Crop Sci. 45, 832–839.

Hoffmann, L., Besseau, S., Geoffroy, P., Ritzenthaler, C., Meyer, D., Lapierre,

C., Pollet, B. and Legrand, M. (2004) Silencing of hydroxycinnamoyl-

coenzyme A shikimate/quinate hydroxycinnamoyltransferase affects phe-

nylpropanoid biosynthesis. Plant Cell, 16, 1446–1465.

Huntley, S.K., Ellis, D., Gilbert, M., Chapple, C. and Mansfield, S.D. (2003)

Significant increases in pulping efficiency in C4H–F5H-transformed pop-

lars: improved chemical savings and reduced environmental toxins.

J. Agric. Food Chem. 51, 6178–6183.

Jones, L., Ennos, A.R. and Turner, S.R. (2001) Cloning and characterization of

irregular xylem4 (irx4): a severely lignin-deficient mutant of Arabidopsis.

Plant J. 26, 205–216.

Jouanin, L., Goujon, T., de Nadaı, V. et al. (2000) Lignification in transgenic

poplars with extremely reduced caffeic acid O-methyltransferase activity.

Plant Physiol. 123, 1363–1373.

Kupce, E. and Freeman, R. (2007) Compensated adiabatic inversion pulses:

broadband INEPT and HSQC. J. Magn. Reson. 187, 258–265.

Leple, J.-C., Dauwe, R., Morreel, K. et al. (2007) Downregulation of cinnamoyl-

coenzyme A reductase in poplar: multiple-level phenotyping reveals effects

on cell wall polymer metabolism and structure. Plant Cell, 19, 3669–3691.

Li, X., Bonawitz, N.D., Weng, J.-K. and Chapple, C. (2010) The growth

reduction associated with repressed lignin biosynthesis in Arabidopsis

thaliana is independent of flavonoids. Plant Cell, 22, 1620–1632.

Lim, E.-K. and Bowles, D.J. (2004) A class of plant glycosyltransferases

involved in cellular homeostasis. EMBO J. 23, 2915–2922.

Lu, F., Marita, J.M., Lapierre, C., Jouanin, L., Morreel, K., Boerjan, W. and

Ralph, J. (2010) Sequencing around 5-hydroxyconiferyl alcohol-derived

units in COMT-deficient poplar lignins. Plant Physiol. 153, 569–579.

Marita, J.M., Ralph, J., Hatfield, R.D. and Chapple, C. (1999) NMR character-

ization of lignins in Arabidopsis altered in the activity of ferulate

5-hydroxylase. Proc. Natl Acad. Sci. USA, 96, 12328–12332.

Marita, J.M., Ralph, J., Lapierre, C., Jouanin, L. and Boerjan, W. (2001) NMR

characterization of lignins from transgenic poplars with suppressed caffeic

acid O-methyltransferase activity. J. Chem. Soc. Perkin Trans. 1, 2001,

2939–2945.

Marita, J.M., Ralph, J., Hatfield, R.D., Guo, D., Chen, F. and Dixon, R.A. (2003)

Structural and compositional modifications in lignin of transgenic alfalfa

down-regulated in caffeic acid 3-O-methyltransferase and caffeoyl coen-

zyme A 3-O-methyltransferase. Phytochemistry, 62, 53–65.

Meyer, K., Shirley, A.M., Cusumano, J.C., Bell-Lelong, D.A. and Chapple, C.

(1998) Lignin monomer composition is determined by the expression of a

cytochrome P450-dependent monooxygenase in Arabidopsis. Proc. Natl

Acad. Sci. USA, 95, 6619–6623.

Mir Derikvand, M., Berrio Sierra, J., Ruel, K., Pollet, B., Do, C.-T., Thevenin, J.,

Buffard, D., Jouanin, L. and Lapierre, C. (2008) Redirection of the phenyl-

propanoid pathway to feruloyl malate in Arabidopsis mutants deficient for

cinnamoyl-CoA reductase 1. Planta, 227, 943–956.

Morreel, K., Ralph, J., Kim, H., Lu, F., Goeminne, G., Ralph, S., Messens, E.

and Boerjan, W. (2004a) Profiling of oligolignols reveals monolignol

coupling conditions in lignifying poplar xylem. Plant Physiol. 136, 3537–

3549.

Morreel, K., Ralph, J., Lu, F., Goeminne, G., Busson, R., Herdewijn, P.,

Goeman, J.L., Van der Eycken, J., Boerjan, W. and Messens, E. (2004b)

Phenolic profiling of caffeic acid O-methyltransferase-deficient poplar

reveals novel benzodioxane oligolignols. Plant Physiol. 136, 4023–4036.

Morreel, K., Dima, O., Kim, H. et al. (2010a) Mass spectrometry-based

sequencing of lignin oligomers. Plant Physiol. 153, 1464–1478.

Morreel, K., Kim, H. and Lu, F. et al. (2010b) Mass spectrometry-based

fragmentation as an identification tool in lignomics. Anal. Chem. 82, 8095–

8105.

Nair, R.B., Bastress, K.L., Ruegger, M.O., Denault, J.W. and Chapple, C. (2004)

The Arabidopsis thaliana REDUCED EPIDERMAL FLUORESCENCE1 gene

896 Ruben Vanholme et al.

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897

encodes an aldehyde dehydrogenase involved in ferulic acid and sinapic

acid biosynthesis. Plant Cell, 16, 544–554.

Ralph, J. and Landucci, L.L. (2010) NMR of lignins. In Lignin and Lignans:

Advances in Chemistry (Heitner, C., Dimmel, D.R. and Schmidt, J.A., eds).

Boca Raton, FL: CRC Press, pp. 137–234.

Ralph, J., Marita, J.M., Ralph, S.A. et al. (1999) Solution-state NMR of lignins.

In Advances in Lignocellulosic Characterization (Argyropoulos, D.S., ed.).

Atlanta, GA: TAPPI Press, pp. 55–108.

Ralph, J., Lapierre, C., Lu, F., Marita, J.M., Pilate, G., Van Doorsselaere, J.,

Boerjan, W. and Jouanin, L. (2001a) NMR evidence for benzodioxane

structures resulting from incorporation of 5-hydroxyconiferyl alcohol into

lignins of O-methyltransferase-deficient poplars. J. Agric. Food Chem. 49,

86–91.

Ralph, J., Lapierre, C., Marita, J.M. et al. (2001b) Elucidation of new structures

in lignins of CAD- and COMT-deficient plants by NMR. Phytochemistry, 57,

993.

Ralph, J., Lundquist, K., Brunow, G. et al. (2004) Lignins: natural polymers

from oxidative coupling of 4-hydroxyphenylpropanoids. Phytochem. Rev.

3, 29–60.

Rogers, L.A. and Campbell, M.M. (2004) The genetic control of lignin depo-

sition during plant growth and development. New Phytol. 164, 17–30.

Rohde, A., Morreel, K., Ralph, J. et al. (2004) Molecular phenotyping of the

pal1 and pal2 mutants of Arabidopsis thaliana reveals far-reaching con-

sequences on phenylpropanoid, amino acid, and carbohydrate metabo-

lism. Plant Cell, 16, 2749–2771.

Ruegger, M. and Chapple, C. (2001) Mutations that reduce sinapoylmalate

accumulation in Arabidopsis thaliana define loci with diverse roles in

phenylpropanoid metabolism. Genetics, 159, 1741–1749.

Schilmiller, A.L., Stout, J., Weng, J.-K., Humphreys, J., Ruegger, M.O. and

Chapple, C. (2009) Mutations in the cinnamate 4-hydroxylase gene

impact metabolism, growth and development in Arabidopsis. Plant J.

60, 771–782.

Sibout, R., Baucher, M., Van Doorsselaere, J. et al. (2002) Expression of a

poplar ferulate-5-hydroxylase in Arabidopsis thaliana. Plant Physiol. Bio-

chem. 40, 1087–1096.

Stewart, J.J., Akiyama, T., Chapple, C., Ralph, J. and Mansfield, S.D. (2009)

The effects on lignin structure of overexpression of ferulate 5-hydroxylase

in hybrid poplar. Plant Physiol. 150, 621–635.

Turner, S.R. and Somerville, C.R. (1997) Collapsed xylem phenotype of

Arabidopsis identifies mutants deficient in cellulose deposition in the

secondary cell wall. Plant Cell, 9, 689–701.

Van Doorsselaere, J., Baucher, M., Chognot, E. et al. (1995) A novel lignin in

poplar trees with a reduced caffeic acid/5-hydroxyferulic acid O-methyl-

transferase activity. Plant J. 8, 855–864.

Vanholme, R., Demedts, B., Morreel, K., Ralph, J. and Boerjan, W. (2010)

Lignin biosynthesis and structure. Plant Physiol. 153, 895–905.

Welinder, K.G., Justesen, A.F., Kjærsgard, I.V.H., Jensen, R.B., Rasmussen,

S.K., Jespersen, H.M. and Duroux, L. (2002) Structural diversity and tran-

scription of class III peroxidases from Arabidopsis thaliana. Eur. J. Bio-

chem. 269, 6063–6081.

Weng, J.-K., Akiyama, T., Bonawitz, N.D., Li, X., Ralph, J. and Chapple, C.

(2010) Convergent evolution of syringyl lignin biosynthesis via distinct

pathways in lycophyte Selaginella and flowering plants. Plant Cell, 22,

1033–1045.

Concomitant F5H1 up-regulation and COMT down-regulation in Arabidopsis 897

ª 2010 The AuthorsThe Plant Journal ª 2010 Blackwell Publishing Ltd, The Plant Journal, (2010), 64, 885–897