Selective lignin downregulation leads to constitutive defense response expression in alfalfa (Medicago sativa L.)

Summary

• •
  Downregulation of hydroxycinnamoyl CoA: shikimate hydroxycinnamoyl transferase (HCT) in alfalfa (Medicago sativa) reduces lignin levels and improves forage quality and saccharification efficiency for bioethanol production. However, the plants have reduced stature. It was previously reported that HCT-down-regulated Arabidopsis have impaired auxin transport, but this has recently been disproved.
• •
  To address the basis for the phenotypes of lignin-modified alfalfa, we measured auxin transport, profiled a range of metabolites including flavonoids and hormones, and performed in depth transcriptome analyses.
• •
  Auxin transport is unaffected in HCT antisense alfalfa despite increased flavonoid biosynthesis. The plants show increased cytokinin and reduced auxin levels, and gibberellin levels and sensitivity are both reduced. Levels of salicylic, jasmonic and abscisic acids are elevated, associated with massive upregulation of pathogenesis and abiotic stress-related genes and enhanced tolerance to fungal infection and drought.
• •
  We suggest that HCT downregulated alfalfa plants exhibit constitutive activation of defense responses, triggered by release of bioactive cell wall fragments and production of hydrogen peroxide as a result of impaired secondary cell wall integrity.

Introduction

Lignin provides mechanical strength and hydrophobicity to plant secondary cell walls and contributes to defense against biotic stress (Lewis & Yamamoto, 1990). However, lignin makes cell walls recalcitrant to digestibility for ruminant animals (Jung & Deetz, 1993; Guo et al., 2001b), and inhibits saccharification in bioethanol processing (Dien et al., 2006; Chen & Dixon, 2007).

Lignin is synthesized from hydroxyguaiacyl (H), guaiacyl (G) and syringyl (S) monolignol units (Boerjan et al., 2003). Genetic manipulation of the monolignol pathway improves forage digestibility and saccharification efficiency for liquid biofuel production, as illustrated in alfalfa with antisense downregulated expression of the hydroxycinnamoyl CoA: shikimate hydroxycinnamoyl transferase (HCT) gene (Shadle et al., 2007; Chen & Dixon, 2007). However, HCT downregulation in alfalfa also leads to changes in stem vascular tissue anatomy (Nakashima et al., 2008) and significant reductions in biomass yield (Shadle et al., 2007). It appears, however, that lignin levels can be downregulated without serious growth effects depending on which step in the monolignol pathway is targeted. For example, downregulation of cinnamoyl CoA reductase (CCR) causes strong growth defects in tobacco, whereas downregulation of the next enzyme in the pathway, cinnamyl alcohol dehydrogenase (CAD), does not (Chabannes et al., 2001).

HCT is the most recently discovered enzyme with a clear involvement in monolignol biosynthesis (Hoffmann et al., 2003). It catalyses the conversion of coumaroyl CoA to coumaroyl shikimate, the substrate for the introduction of the aromatic 3-hydroxyl group by the action of the cytochrome P450 enzyme coumaroyl shikimate 3′-hydroxylase (C3′H) (see the Supporting Information, Fig. S1). HCT then acts again, this time in the reverse direction, to convert caffeoyl shikimate to caffeoyl CoA, the substrate for O-methylation by caffeoyl CoA 3-O-methyltransferase (Fig S1).

Silencing HCT in Arabidopsis leads to accumulation of flavonoids and inhibition of auxin transport (Besseau et al., 2007). Restoring auxin transport by reducing flavonoid content was reported to overcome the dwarf phenotype while maintaining the reduced lignin phenotype (Besseau et al., 2007). However, this has recently been disproved (Li et al., 2010), and the restoration of lignin biosynthesis in HCT-silenced Arabidopsis through expression of a Selaginella gene encoding an enzyme that effectively bypasses the HCT step shown to restore normal growth (Li et al., 2010). However, this result does not distinguish between overall loss of lignin with altered cell wall integrity and spill-over of biosynthetic intermediates as reasons for the altered growth phenotype in HCT-downregulated plants. The mechanisms underlying growth inhibition in lignin-modified plants are therefore still unclear, as is the reason why some modifications to lignin content and/or composition inhibit plant growth whereas others do not.

Lignin reduction can induce wide-ranging pleiotropic alterations in gene expression and metabolism (Dauwe et al., 2007; Shi et al., 2009). HCT-downregulated alfalfa plants have increased branching (Shadle et al., 2007), suggesting changes in endogenous growth regulators. We here report a molecular and phenotypic characterization of HCT-downregulated alfalfa. We show that the plants possess normal auxin transport, but are gibberellin insensitive with increased levels of cytokinins and defense signals such as salicylic acid, jasmonic acid, abscisic acid and hydrogen peroxide. These changes result in defense response gene expression in the absence of pathogen attack, associated with enhanced tolerance to subsequent fungal infection and drought. Pectic material with the ability to induce these defense response genes is more readily extractable from cell walls of HCT-downregulated alfalfa, suggesting a model that links the changes in gene expression and growth to alterations in cell wall structure and integrity.

Materials and Methods

Chemicals

N-1-naphthylphtalamic acid (NPA) was from Gold Biotechnology (St Louis, MO, USA). Gibberellin GA₃, paclobutrazol (PCB), 3,3′-diaminobenzidine tetrahydrochloride (DAB), polygalacturonic acid (PGA), polygalacturonase, sodium tetraborate, m-hydroxybiphenyl, 4-hydroxybenzoic acid, 4-coumaric acid and apigenin were from Sigma Aldrich.

Plant material and growth conditions

Alfalfa (Medicago sativa cv Regen SY) plants were grown at 22°C under long-day photoperiod (16 h light, 8 h dark) with an average day light intensity of 240 μmol m⁻² s⁻¹ in the glasshouse with nitrogen fertilizer. All experiments were performed with clonally propagated transgenic material previously described (Reddy et al., 2005; Chen et al., 2006; Shadle et al., 2007). The antisense lines analysed were HCT (30a, 3a and 29a), C3′H (4a, 9a and 5A), an empty control vector (CK48-Wt1) and two nontransformed wild-type alfalfa lines (Ctl1-Wt2 and Ctrl49-Wt3).

Phenotypic characterization

Measurements were performed starting on 10-wk-old plants. Stems with one visible node were labeled and their growth followed until flowering. Six stems were selected from each of three independent lines, and the numbers and lengths of internodes and total stem length were measured. Leaf area was measured using a LI-3000A portable area meter (LI-COR, http://www.licor.com) and stem diameter was measured using an automatic caliper (Fisher Scientific, http://www.fishersci.com/wps/portal/HOME).

Photosynthesis and transpiration rates were determined using an LI-6400 infrared CO₂ gas analyser (LI-COR) as described in Zhao et al. (2010).

Leaf epidermal cell numbers and areas were measured in folded mature leaves. Three plants from each wild-type and HCT-antisense line were taken for this analysis. Two pictures per leaf were taken from two leaves per plant under the light microscope at the same magnification. Each field measured contained at least 145 epidermal cells and 24 stomates.

Microarray and quantitative real-time polymerase chain reaction (qRT-PCR) analysis

The top six nodes of vegetative stems were harvested. Each sample consisted of pooled stems from six plants representing three independent lines. Three biological replicates were used for each line. Petioles and leaves were removed and stems frozen in liquid nitrogen and stored at −80°C. Total RNA was extracted and DNase-treated using an AMBION kit (Ambion, http://www.ambion.com/), then split into two aliquots for microarray and RT-PCR analysis. The Affymetrix GeneChip Medicago Genome Array (Affymetrix, http://www.affymetrix.com/) was used for expression analysis, with experimental and statistical procedures as described in Zhao et al. (2010). Differentially expressed genes between wild-type and HCT lines were selected using associative analysis (Dozmorov & Centola, 2003). Functional annotation of differentially expressed genes was performed manually to the most similar Arabidopsis protein by blast search (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Microarray data are available at ArrayExpresss (http://www.ebi.ac.uk/arrayexpress, ID = E-MEXP-2741). Three micrograms of total RNA for each sample was used for qRT-PCR as described in Zhao et al. (2010). Gene-specific primers are listed in Table S1.

Assessment of disease symptoms

Colletotrichum trifolii (ATCC#12126) was propagated in 20% V8 juice, 0.3% CaCO₃, 0.15% agar, 80% water, pH 7.2. Spores were scraped from agar plates flooded with sterile water, concentrated by centrifugation to 2 × 10⁶ spores ml⁻¹, and supplemented with 0.1% Tween 20. Stems and leaves were sprayed with spore suspension, or stem nodes of four plants were inoculated with 20 μl drops of spore suspension on at least three different sites. Control treatment was 0.1% Tween-20. Symptoms were recorded 18 d after inoculation. Disease severity was scored based on symptom incidence (number of visible lesions) and severity (lesion size).

Analysis of drought responsiveness

Eight-wk-old plants (nine replicates of both wild-type and HCT-downregulated lines) were subjected to drought stress for 9 d. After this period plants were rehydrated for a further 5 d. Total leaf water potential was measured used a Wescor (Logan, UT, USA) thermocouple psychrometer every 2 d during the drought period and after plants were rewatered. One leaf from the fully developed trifoliate was selected for this measurement. Five measurements of each wild-type and transgenic line were made at each time point. Soil moisture was also measured using a moisture meter HH2 Delta-T-Devices (Cambridge, UK). Two measurements per pot were made for all plants. This experiment was repeated three times.

Application of GA and PCB

Application of 10⁻⁵ M GA₃ and/or PCB was performed by root irrigation, twice per week for 1 month. Experiments were repeated twice.

Measurement of phytohormone levels

Extraction and purification of ABA, GA₁, GA₄, IAA, trans-zeatin (tZ), isopentenyl adenine (iP), salicylic acid (SA), jasmonic acid (JA) and JA-Ile (JA conjugated with isoleucine) were performed by solid-phase extraction. Stable isotope-labeled compounds used as internal standards were: D₆-ABA (Icon Isotopes, Summit, NJ, USA); d₂-GA₁, d₂-GA₄, d₆-iP, d₅-tZ (Olchemim Ltd, Olomouc, Czech Republic); d₂-IAA, d₆-SA (Sigma-Aldrich); d₂-JA (Tokyo Kasei, Tokyo, Japan). ¹³C₆-JA-Ile was synthesized as described in Jikumaru et al. (2004). For simultaneous measurement of hormones, 1 g fresh weight of stems was frozen and extracted with c. 10 times (v : w) 80% (v : v) methanol–1% acetic acid. After drying the extracts, residues were resuspended in 3 ml of 80% (v : v) acetonitrile: 1% (v : v) acetic acid and stable isotope labeled internal standards. Extracts were evaporated to 1% acetic acid, and applied to a pre-equilibrated Oasis HLB column cartridge (30 mg, 1 ml, Waters, Tokyo, Japan). After washing with 1 ml of 1% (v : v) acetic acid, hormones were eluted with 2 ml of 80% (v : v) acetonitrile: 1% (v : v) acetic acid, evaporated to 1% acetic acid, and applied to a pre-equilibrated Oasis MCX column cartridge (30 mg, 1 ml, Waters). After washing the cartridges with 1 ml of 1% acetic acid, the acidic and neutral fraction containing ABA, GA₁, GA₄, IAA, JA, JA-Ile and SA was eluted with 2 ml of acetonitrile. Two hundred microliters of this fraction was transferred, evaporated, and reconstituted with 1% acetic acid for analysis of SA. The MCX cartridges were further washed with 1 ml of 5% (v : v) aqueous ammonia, and the basic fraction containing tZ and iP was eluted with 2 ml of 60% (v : v) acetonitrile: 5% (v : /v) aqueous ammonia. Acidic and neutral fractions were further applied to a pre-equilibrated Oasis WAX column cartridge (30 mg, 1 ml, Waters). After washing with 1 ml of 1% acetic acid and 2 ml of 80% acetonitrile, the acidic fraction containing ABA, GA₁, GA₄, IAA, JA and JA-Ile was eluted with 2 ml of acetonitrile: 1% (v : v) acetic acid. This fraction was evaporated and reconstituted with 1% acetic acid. Hormones were analysed by LC-electrospray ionization (ESI)-MS/MS (Agilent 6410) on a ZORBAX Eclipse XDB-C18 column (Agilent) as described in Yoshimoto et al. (2009), and quantified using MassHunter v. B. 01. 02 spectrometer software (Agilent, Santa Clara, CA, USA).

Determination of auxin transport

Polar auxin transport was measured using a modification of published protocols (Brown et al., 2001; Besseau et al., 2007). Apical shoots were removed and 9 cm stem segments placed in 15 ml Falcon tubes with the apical ends submerged in 250 μl of 5 mM 2-(N-morpholino) ethanesulfonic acid (MES), 1% sucrose, pH 5.5 containing 234 nM 5-³H-IAA (20 Ci mmol⁻¹; American Radiolabeled Chemicals, St Louis, MO, USA) in the presence or absence of 10 mM NPA (an auxin transport inhibitor) as a negative control to monitor external radioactivity contamination. After 2, 6, 10 and 18 h of incubation, stem segments were removed and the nonsubmerged portions excised and placed in 3 ml of 80% ethanol. After overnight extraction at 4°C, 2.5 ml of liquid scintillation cocktail (MP solution, Beckman, Brea, CA, USA) was added and radioactivity measured by scintillation counting (Beckman LS6500).

HPLC analysis of phenolic compounds

Freeze-dried stems (70 mg) were extracted in 3 ml of chloroform/methanol (2 : 1 v : v) containing 22 μg ml⁻¹ of docosanol overnight at 20°C. Ribitol (100 μg ml⁻¹, internal standard, 1.1 ml in H₂O) was added and the mixtures incubated for a further 4 h. After centrifuging at 3000 g for 10 min at 4°C, the chloroform and aqueous phases were separated with a microsyringe. The aqueous phase (1 ml) was split into two: 25 μl of the first 500 μl were used directly for high-pressure liquid chromatography (HPLC) profiling and the other 500 μl was subjected to enzymatic hydrolysis. The aqueous fraction to be hydrolysed was resuspended in 0.6 volumes citrate-phosphate buffer, pH 5.2, 25 units of a crude almond enzyme extract containing β-glucosidase and β-glucuronidase (Sigma) was added per reaction, and the mixtures were incubated overnight at 37°C and extracted twice with one volume of ethyl acetate. The ethyl acetate extracts were dried under nitrogen, and residues resuspended in 500 μl of methanol for HPLC profiling (25 μl of sample injected). The HPLC was carried out on a Beckman System Gold HPLC system with a Waters Spherisorb ODS-2 5 μ reverse phase column (5-μm particle, 250 × 4.6 mm). Solvent A was 0.1% phosphoric acid in water, solvent B acetonitrile, gradient 8–30% B for 44 min and 30–60% B for a further 5 min. Compounds were identified by comparing UV spectra and retention times with authentic standards, and the apigenin peak was confirmed by LC-MS.

Determination of lignin content and composition

The lignin content/composition was determined by thioacidolysis as described previously (Zhao et al., 2010).

Determination of anthocyanin content

Anthocyanins were extracted and quantified used cyanidin as a reference by the colorimetric method described previously (Peel et al., 2009)

Hydrogen peroxide staining

Ten leaves from 8-wk-old glasshouse grown plants were detached, vacuum-infiltrated with DAB solution (1 mg ml⁻¹), and incubated in the dark for 10 h. The leaves were bleached with absolute ethanol and observed by light microscopy.

Isolation of cell wall polysaccharides

Five grams fresh weight of plant material (aerial portions) were ground in liquid nitrogen_, homogenized with two volumes of 80% ethanol, and incubated overnight at 4°C. The homogenate was centrifuged at 1500 g for 5 min and the alcohol insoluble cell wall residue (AIR) washed twice with 20 ml of absolute ethanol and dried under N₂. A 1 g aliquot of AIR was homogenized in 20 ml of water and shaken overnight at 20°C. The homogenate was centrifuged at 1500 g for 15 min and washed twice with 15 ml water. The supernatants (crude cold water-soluble pectin elicitor extracts (WS)) were pooled and heat-treated (100°C, 5 min) before assay of elicitor activity in cell cultures. The WS residue was subjected to two consecutive extractions (in 20 ml with two successive 15 ml washes) to generate the EDTA-soluble and HCl-soluble pectin fraction as described in Rosli et al. (2004). Uronic acid concentrations of all fractions were estimated by the m-hydroxydiphenyl method (Blumenkrantz & Asboe-Hansen, 1973), using PGA as standard. Total pectin was determined by addition of the uronic acid equivalent units of the WS, EDTA- and HCl-soluble fractions.

Elicitation of cell cultures

Eight days after subculture, 35- ml batches of dark-grown cell suspension cultures derived from M. truncatula (‘Jemalong A17’) roots (Suzuki et al., 2005) were treated with 10 ml of the WS fraction from wild-type or HCT-downregulated plants in 50-ml Erlenmeyer flasks. The negative control was distilled water, and the positive control PGA (final concentration 0.5 mg ml⁻¹). The WS fractions and PGA were also treated with polygalacturonase from Aspergillus niger (0.5 mU 10 ml⁻¹ WS) for 8 h at 30°C. Reactions were stopped by heating at 100°C for 5 min. Triplicate culture batches were harvested for each group of elicited samples at 10 h post-elicitation, and frozen at −80°C for further analysis. This experiment was performed twice with similar results.

Statistical methods

Statistical treatment of data was performed by analysis of variance using Fisher’s LSD procedure for multiple comparison tests (Statgraphics Plus program, version 5.1 for windows, Rockville, MD, USA)

Results

Phenotypic and physiological characterization of HCT antisense alfalfa

As described (Chen et al., 2006; Shadle et al., 2007), levels of HCT transcripts were strongly reduced in the T1 clonally propagated HCT antisense alfalfa lines used in the present study, and levels of both G and S lignin monomer units were reduced, but H monomer units increased (Fig. S2a,b). Analysis of all nine monolignol pathway gene transcripts by qRT-PCR indicated that HCT downregulation also resulted in downregulation (by about twofold) of transcripts encoding C3′H, the enzyme that co-acts with HCT in the 3-substitution of the monolignol skeleton (Fig. S3), but that the expression of the other monolignol pathway genes was essentially unaffected.

Stem height, internode length and leaf area were significantly lower in HCT antisense lines compared with controls (Table S2). Reduction in height was caused by reduced internode elongation rather than reduced node number, and reduction in internode length was observed in internode three (from the top of the stem) and below (Fig. 1). Continued production of vegetative nodes in HCT antisense lines reflected delayed flowering (Table S2). Increased branching of HCT antisense lines was observed as documented previously (Shadle et al., 2007); on average, the number of vegetative stems produced was 8 ± 2 for wild type plants and 28 ± 3 for HCT antisense lines.

Figure 1. Internode lengths of control (black rhombs) and HCT-downregulated (open squares) alfalfa. Nodes are numbered from the top to the bottom of the stem.

Photosynthetic and transpiration rates were reduced in HCT antisense lines compared with controls, as was epidermal cell area (Table S2). By contrast, stomatal and epidermal cell numbers were increased (Table S2), suggesting that the overall effect of HCT downregulation on growth was a result of reduced cell growth rather than reduced cell divisions.

Phenolic compounds in HCT antisense plants

The major flavonoid compounds in alfalfa stems and leaves are flavone conjugates (Stochmal et al., 2001). Comparison between HPLC traces of stem extracts from wild-type and HCT antisense lines revealed changes in the levels of a number of phenolic compounds, including several peaks that represented differently substituted glycosides of the flavone apigenin (Fig. S4a). To simplify profiling and facilitate compound identification, stem extracts were hydrolysed with a crude β-glucosidase/β-glucuronidase preparation before analysis. The HPLC profiles then revealed increases in 4-hydroxybenzoic acid, 4-coumaric acid and apigenin aglycone in HCT antisense lines (Table 1 and Fig. S4b). Increased coumaric acid likely results from build-up and subsequent hydrolysis of coumaroyl CoA as a direct result of reduced HCT expression, and coumaroyl CoA could also spillover into the synthesis of flavones. In addition, we found an increase in total anthocyanin level in HCT lines compared with control (Table 1), supporting the spillover from lignin to flavonoid pathway in those plants.

Table 1.   Soluble phenolic aglycones and total anthocyanin content of stems of wild-type and HCT-downregulated alfalfa (μg g⁻¹ FW)

Compound	WT	HCT
Total anthocyanin levels are expressed as cyanidin equivalents. The results are the mean ± SD of three biological replicates.
4-Hydroxybenzoic acid	0.43 ± 0.03	1.27 ± 0.02
4-Coumaric-acid	0.37 ± 0.05	6.06 ± 0.40
Apigenin	2.17 ± 0.40	12.5 ± 1.20
Anthocyanin	3.21 ± 0.30	6.15 ± 0.50

HCT downregulation does not affect auxin transport

To test whether flavonoid accumulation affects auxin transport in HCT antisense alfalfa lines, we applied radiolabeled IAA, with and without the auxin transport inhibitor NPA, to the apical ends of cut stem segments, and determined accumulation of label within the stem at different times post-application. There was no significant difference in the rate of auxin transport between control and HCT antisense lines (Fig. 2).

Figure 2. Auxin transport in stems of control and HCT-downregulated alfalfa. Each value represents the average of the radioactivity from [H³]-IAA determined in four independent stem segments. Means are not significantly different (P < 0.05). This experiment was performed twice in the dark and twice in the light with similar results.

Lignin modification affects phytohormone levels

To determine whether the stunted/branched phenotypes of HCT antisense lines result from alterations in levels of plant growth regulators, we profiled a range of hormones. Levels of three cytokinins, tZ, iP and dihydrozeatin (DHZ) were higher in stems of HCT lines compared with controls (Table 2), and levels of IAA were lower. Increased cytokinin to auxin ratio is consistent with the branched phenotype of the HCT lines (Perilli et al., 2010).

Table 2.   Phytohormone levels in stems of control and HCT-downregulated alfalfa

Hormone (ng g^−1 FW)	WT	HCT	Change compared with Wt
ABA, abscisic acid; JA, jasmonic acid; JA-Ile, jasmonic acid conjugated to isoleucine; SA, salicilic acid; GA, gibberellins; IAA, indole acetic acid; tZ, trans-zeatin; iP, isopentenyl adenine; DHZ, dihydrozeatin. Results are the means of three biological replicates ± SD. All values comparing wild-type and HCT-downregulated lines were significantly different (P < 0.05).
ABA	41.6 ± 3.0	57.5 ± 12.0	↑
JA	49.0 ± 6.0	72.5 ± 11.0	↑
JA-lle	1.9 ± 0.1	2.4 ± 0.3	↑
SA	134 ± 30	352 ± 48	↑
GA1	0.8 ± 0.1	0.5 ± 0.1	↓
GA4	0.20 ± 0.01	0.10 ± 0.01	↓
IAA	59.2 ± 8.0	37.7 ± 5.0	↓
tZ	0.06 ± 0.01	0.24 ± 0.03	↑
iP	0.21 ± 0.03	0.32 ± 0.04	↑
DHZ	0.14 ± 0.03	0.43 ± 0.05	↑

The concentrations of gibberellins GA₁ and GA₄ were reduced in HCT lines (Table 2). We applied GA (10⁻⁵ M) by root irrigation to determine whether this could restore growth, and PCB (10⁻⁵ M), an inhibitor of GA biosynthesis, independently and along with GA, to confirm that inhibition of GA biosynthesis reduces stem elongation in alfalfa (Table 3). Application of PCB to control lines reduced stem length, leaf area and petiole length whereas only a small reduction in stem length was recorded in HCT lines (Table 3). The HCT-downregulated lines showed no response to 4 wk of GA treatment, whereas the stems of control lines increased in length by 42% (Fig. S5, Table 3). The GA treatment also increased leaf area and petiole length in control lines. Thus, HCT antisense lines have strongly reduced responsiveness to GA.

Table 3.   Effects of gibberellin (GA₃) and Paclobutrazol (PCB) on growth of HCT antisense and control alfalfa

Line	Stem height (mm)	Average internode length (mm)	Leaf area (mm^2)	Petiole length (mm)
Data were collected after 1 month of hormone treatment by irrigation. Results are the means of nine replicates ± SD from each treatment.
WT untreated	310 ± 45	44 ± 4	49 ± 1	20 ± 1
WT + PCB	230 ± 45	37 ± 4	34 ± 1	11 ± 3
WT + GA	440 ± 60	58 ± 1	50 ± 1	25 ± 1
WT + PCB + GA	300 ± 30	40 ± 3	45 ± 1	12 ± 1
HCT untreated	195 ± 45	33 ± 4	35 ± 2	13 ± 2
HCT + PCB	190 ± 30	31 ± 4	30 ± 1	12 ± 1
HCT + GA	212 ± 36	34 ± 5	33 ± 1	13 ± 1
HCT + PCB + GA	205 ± 45	34 ± 4	36 ± 1	14 ± 1

Abscisic acid, salicylic acid and jasmonic acid are stress response signals in plants (Bari & Jones, 2009). The concentrations of ABA, SA, JA and JA-Ile (believed to be the bioactive form of JA perceived by plants; Koo et al., 2009) were significantly increased in HCT antisense lines (Table 2).

Transcriptome profiling

The high concentrations of SA and JA suggest that HCT antisense plants are responding to perceived biotic stress. To determine whether these signals are transduced into downstream responses, we performed genome-wide transcript profiling of stem nodes using the M. truncatula Affymetrix gene chip, which has been validated for use with alfalfa (Tesfaye et al., 2006). Sixty nine genes were downregulated and 269 upregulated in HCT antisense lines (Tables S3, S4). The gene ontology groups with most members upregulated in HCT antisense plants were pathogenesis-related (PR; 55 genes, 21% of differentially expressed genes), transport (24 genes, 9% of differentially expressed genes), secondary metabolism (20 genes, 8% of differentially expressed genes), protein kinase receptor-like proteins (21 genes) and abiotic stress-related (19 genes, 7% of differentially expressed genes) (Table S5). The categories most downregulated were related to cell wall biosynthesis (16 genes, 24%) and hormone biosynthesis and signaling (11 genes, 16% of differentially expressed genes) (Table S5).

Most over-expressed genes in the PR category were annotated as encoding PR1, PR10, chitinases (PR4), β-1,3-glucanases and thaumatin (PR5) (Tables 4, S4, S5). These were generally highly induced, with six members of the PR5/thaumatin family, five chitinases, two glucanases, three PR1 genes, two PR4 genes and two PR10 genes being expressed more than fivefold higher than in controls (Table 4). The most highly differentially expressed gene in HCT lines (nearly 56-fold higher than controls) was a PR1 gene (Table 4). Two WRKY family transcription factors were induced between 4- and 4.5-fold (Table S4); WRKY genes have been implicated in the activation of PR gene expression (Eulgem et al., 1999).

Table 4.   Genes with functional annotations that are upregulated by at least fivefold in HCT-downregulated lines compared with controls

Probesets	Target description	Ratio (HCT/WT)	P	Q
Note that 87% (28/32) are genes related to plant defense responses.
Mtr-U57609-2_s_at	REPORTER neomycin phosphotransferase	490.4098	0	0.051677
Mtr.8977.1.S1_at	Pathogenesis-related protein PR1	55.9847	0	0.119736
Mtr.10332.1.S1_at	Chitinase class II	26.9985	0	0.246081
Mtr.7638.1.S1_at	Endo-1,3-beta-glucanase	26.5578	0	0.044739
Mtr.42775.1.S1_at	Thaumatin-like protein PR-5a precursor	26.2216	0	0.239276
Mtr.7638.1.S1_s_at	Endo-1,3-beta-glucanase	25.0000	0	0.191748
Msa.3085.1.S1_at	Thaumatin-like protein PR-5a	24.2623	0	0.035761
Mtr.40147.1.S1_s_at	Pathogenesis-related protein PR 10	22.9521	0	0.139670
Mtr.49448.1.S1_at	Glycoside hydrolase, family 18; chitinase II	20.0130	0	0.057616
Msa.2848.1.S1_at	Chitinase III	19.5222	0	0.085511
Msa.3171.1.S1_at	Pathogenesis-related protein PR1a	15.7973	0	0.054690
Msa.3006.1.S1_at	Pathogenesis-related protein PR4A	15.1646	0	0.072343
Msa.1009.1.S1_at	Chitinase class III	14.2190	0	0.165848
Msa.2942.1.S1_s_at	Pathogenesis-related protein PR 10	13.3518	0	0.127170
Mtr.10968.1.S1_at	Pathogenesis-related protein PR5-1	11.5648	0	0.045259
Msa.1598.1.S1_at	Thaumatin-like protein	11.2369	0	0.068088
Mtr.43565.1.S1_at	Pathogenesis-related protein PR-1b	8.36289	0	0.166606
Mtr.27695.1.S1_at	Glycine-rich protein	7.38461	0	0.163871
Mtr.6738.1.S1_at	Transmembrane receptor protein kinase	7.32103	0	0.050456
Msa.1621.1.S1_at	Pathogenesis-related protein PR4A	7.04091	0	0.028609
Mtr.12525.1.S1_at	Chitinase	6.94950	0	0.061873
Msa.2952.1.S1_at	DNAJ heat-shock protein	6.85863	0	0.061822
Msa.1259.1.S1_at	Glutathione S-transferase GST; plant defense related	5.95885	2.5E-174	0.073503
Mtr.13988.1.S1_at	Secretory protein-like	5.56025	0	0.028819
Msa.2700.1.S1_at	Class IV chitinase	5.52389	0	0.085179
Mtr.39358.1.S1_at	Sinapate 1-glucosyltransferase activity	5.39876	2.5E-224	0.140221
Mtr.42902.1.S1_s_at	ProB glutamate phosphate reductase; salt resistance related	5.39394	0	0.062806
Mtr.244.1.S1_at	DnaJ-like protein	5.30398	0	0.010218
Msa.686.1.S1_at	DNAJ heat shock protein	5.19531	0	0.110513
Mtr.246.1.S1_at	DnaJ-like protein	5.16725	4.5E-58	0.098516
Msa.1765.1.S1_at	Thaumatin-like protein	5.15802	0	0.133232
Mtr.6757.1.S1_at	Thaumatin-like protein	5.13318	0	0.040183

Flavonoid biosynthetic pathway genes were induced in HCT antisense lines (Table S4, S5). This was subsequently validated by quantitative RT-PCR (Fig. S6), and suggests that increased flavonoid accumulation is not simply a result of metabolic spillover, but also has an underlying gene activation component leading to increased flavonoid biosynthesis.

In the abiotic stress subclass, the gene categories most represented were drought-related, proline rich proteins and DNAj heat shock proteins (Table S5). Phosphate, sugar and amino-acid transporters were most over-represented in the transport subclass (Table S5).

Downregulated cell wall subclass genes in HCT antisense plants were most represented by xyloglucan endotransglycosylase/hydrolase (XTH), and in the hormone-related subclass by auxin response (SAUR) and genes associated with GA metabolism (GA20 oxidase) and signaling (GAST) (Table S3, S5).

Coumaroyl shikimate 3′-hydroxylase co-acts with HCT in the generation of 3-hydroxy-substituted monolignol precursors, and C3′H downregulation also leads to strongly reduced lignin levels with a strikingly increased level of H monomers in alfalfa (Reddy et al., 2005). The qRT-PCR analysis of C3′H antisense alfalfa lines confirmed that downregulation of C3′H also induced expression of selected PR proteins (Fig. S7).

HCT antisense plants have improved stress tolerance

Upregulation of biotic and abiotic stress response genes suggests that HCT antisense plants might have increased tolerance to infection or drought. We therefore evaluated the response of the plants to the fungus Colletotrichum trifolii, the causal agent of alfalfa anthracnose (Yang et al., 2007). Eighteen days post-inoculation, HCT antisense lines exhibited less disease incidence and severity than corresponding control lines, and the size of necrotic lesions at inoculated sites was clearly reduced (Fig. 3a,b, Table 5).

Figure 3. Biotic and abiotic stress responses of HCT-downregulated alfalfa lines compared with controls. Experiments were repeated twice with different vegetatively propagated plant groups of the same age grown in parallel. (a) Ten-wk-old plants inoculated with Colletotrichum trifolli, 18 d post infection (dpi). Arrows indicate anthracnose symptoms. (b) Details of symptoms at inoculated sites. (c) Eight-wk-old plants subjected to drought stress for 9 d. Soil moisture was monitored throughout to confirm a similar soil drying rate for all plants (see the Supporting Information, Fig. S8). (d) Recovery from drought stress after 5 d rehydration.

Table 5.   Evaluation of disease tolerance to Colletotrichum trifolli in HCT-downregulated and wild-type control alfalfa lines

Line	Symptoms			Disease severity
	(Number of stems affected/number of stems inoculated)	(Number of visible lesions per plant)	Disease incidence (%)	(% of stem affected)
Disease evaluation was recorded at 18 days post infection (dpi).
Wt1	9/12	10 ± 1.5	83	50
Wt2	8/12	9 ± 1.2	75	40
Wt3	9/12	7 ± 0.9	59	35
HCT3a	3/12	2 ± 0.7	17	25
HCT30a	4/12	3 ± 0.9	25	20
HCT29a	3/12	3 ± 0.6	25	15

We next tested the drought response. After 9 d without irrigation, control stems exhibited wilting whereas HCT antisense lines showed few symptoms of drought stress (Fig. 3c). Moreover, these plants recovered completely after 5 d of rehydration, whereas wild-type plants exhibited extensive damage on stems and leaves, and did not survive to the end of the experiment (Fig. 3d). This experiment was repeated with measurement of leaf water potential throughout the drought and rehydration periods. The soil moisture content values for the wild-type and HCT antisense lines were not significantly different during the experiment (Fig. S8), whereas the leaf water potential of the HCT antisense lines decreased at a slower rate than that of the controls during the drought period, and recovered more rapidly following rewatering (Fig. 4).

Figure 4. Leaf water potential of control (solid line) and HCT- antisense (dashed line) plants after watering was suspended for 1, 3, 5, 7 and 9 d, and at 1, 3 and 5 d after rehydration (RW). Numbers in parenthesis indicate numbers of surviving plants from n = 9. All HCT plants survived.

Initial signals for defense gene activation?

Hydrogen peroxide is a signal for both local and systemic SA-mediated PR gene expression (Wu et al., 1995; Alvarez et al., 1998). Staining with DAB, a histochemical reagent for H₂O₂ detection (Thordal-Christensen et al., 1997), revealed that leaf panels from HCT antisense lines, but not wild-type plants, accumulated H₂O₂ under nonstress conditions, and that the H₂O₂ appeared to be associated with the vascular system (Fig. 5a,b).

Figure 5. 3,3′-Diaminobenzidine (DAB) staining of 8-wk-old leaves of uninfected, HCT-downregulated alfalfa lines compared with controls. The experiment was performed three times with similar results.

Cell wall pectic polysaccharides induce PR gene expression in plants via the intermediacy of H₂O₂ (Legendre et al., 1993). To determine whether cell walls of HCT antisense alfalfa plants contain easily-releasable elicitors of PR gene expression, we extracted cell walls isolated from combined aerial tissues with water at room temperature. This very mild extraction should only release a fraction of the soluble wall components such as pectin (Rosli et al., 2004), and walls from HCT antisense material released 3.7-fold more cold water-soluble (WS) pectic material (measured as uronic acid units) than from wild-type material (Table S6). Moreover, when the crude cell wall extracts were added to Medicago suspension cultured cells, PR5, PR1 and endoglucanase genes were more strongly induced by extracts from HCT antisense plants than from wild type (Fig. 6). Pre-treatment of cell wall preparations with polygalacturonase destroyed PR activation activity, confirming that this activity resulted from the facile release of pectic material from the walls. The remainder of the pectic material in the cell wall preparations was then extracted using harsher methods (EDTA and HCl); importantly, the total pectic material released from the cell walls of the wild-type and HCT antisense lines was essentially the same, indicating that the increase in readily extractable pectic material in the HCT antisense lines is a result of reduced cell wall integrity rather than an overall increase in the pectic fraction.

Figure 6. PR transcript levels in alfalfa suspension cultures treated with cold water extracts of cell walls from control and HCT-downregulated plants. PGA was positive control. Extracts were also digested with polygalacturonase (PGase), and all samples heat-treated before assay. Fold changes of PR transcript level normalized to Actin were determined by quantitative real-time polymerase chain reaction, and compared with values from water-treated controls. Results are mean ± SD of three biological and three analytical replicates.

Discussion

Hormonal changes and growth of HCT antisense alfalfa

The HCT-downregulated alfalfa plants accumulate higher levels of apigenin conjugates and anthocyanins than wild-type plants through a combination of metabolic spillover plus activation of flavonoid biosynthesis, but have normal auxin transport. The flavonoids ascribed a role in growth reduction of HCT down-regulated Arabidopsis through inhibition of auxin transport are flavonols and anthocyanins (Besseau et al., 2007). Auxin transport is known to be inhibited by flavonols (Jacobs & Rubery, 1988; Brown et al., 2001), but there is no evidence for inhibition by flavones.

The overall reduction of auxin levels in stems of HCT antisense plants correlated with the downregulation of several small auxin-up RNA (SAUR) family genes (Chapman & Estelle, 2009). Auxin is transported downwards in shoots and inhibits bud outgrowth, whereas cytokinins move upwards to activate bud outgrowth (Ruzicka et al., 2009). The increase in cytokinin to auxin ratio in the transgenic plants likely accounts for the observed branching phenotype.

Levels of GA₁ and GA₄ were reduced in stems of HCT antisense plants, consistent with the downregulation of GA response/biosynthesis genes including GAST, extensins, XTH and GA-20 oxidases. Independent suppression of these GA-regulated genes has been reported to result in reduced cell elongation (Lee & Kende, 2001). More striking was the near-complete loss of GA sensitivity in the HCT lines, which may also contribute to the reduction in cell elongation.

Constitutive defense gene activation in lignin-modified plants

Multiple PR proteins were expressed from 5- to 56-fold higher in HCT antisense lines than in controls, and genes associated with heat and drought responses were also highly induced. These responses likely result from the elevated levels of SA, JA, JA-Ileu, and ABA in the HCT transgenics.

Pathogenesis-related proteins play important roles in pathogen defense in alfalfa and other species (Alexander et al., 1993; Salles et al., 2002). Constitutive over-expression of PR genes enhances resistance to a broad range of pathogens (Alexander et al., 1993; Broglie et al., 1991), as shown here for C. trifolii, and PR gene expression during infection is largely coordinated through SA accumulation (Bari & Jones, 2009).

The drought tolerance of the HCT lines may be attributed to several factors in addition to the induction of abiotic stress response genes. These include increased ABA levels, reduction in leaf transpiration rate and increased stomatal density. The last two changes are associated with drought adaptation responses (Xu & Zhou, 2008).

Although the hormonal and gene expression profiles of the lignin downregulated lines explain both the growth patterns and stress response phenotypes of the plants, a major question remains as to how these changes are triggered by lignin reduction. Lignin reduction per se does not directly correlate with altered plant growth phenotypes. For example, downregulation of CCR causes strong growth defects in tobacco, whereas down-regulation of the next enzyme in the monolignol pathway, CAD does not, and neither does co-downregulation of CCR and CAD (Chabannes et al., 2001). There is no clear correlation between the growth phenotypes observed in the present work and the changes in lignin content and composition recorded here and elsewhere for the same lines (Chen et al., 2006; Shadle et al., 2007). Furthermore, the compounds that may be produced by spillover of metabolic flux will differ depending on where the monolignol pathway is downregulated; for example, downregulation of the pathway at the HCT or coumaroyl shikimate 3′-hydroxylase (C3′H) steps will lead, initially, to accumulation of coumarate esters, which can be channeled into flavonoid biosynthesis. Downregulation of caffeoyl CoA 3-O-methyltransferase (CCoAOMT) results in accumulation of caffeoyl glucoside (Chen et al., 2003; Meyermans et al., 2000), and CAD downregulated plants produce hydroxycinnamyl aldehydes, which become incorporated into lignin (Dauwe et al., 2007). These intermediates all have different metabolic fates when further progress into monolignol biosynthesis is blocked. A detailed transcriptomic and metabolomic analysis of CCR downregulated tobacco plants revealed extensive transcriptional reprogramming and evidence for oxidative stress (perhaps arising as a result of increased photorespiration), but the pathways involved were very different from those reported here, and constitutive PR protein expression was not recorded (Dauwe et al., 2007), confirming that there is probably no single mechanism to explain altered growth phenotypes in lignin-modified plants based on metabolic spillover.

An alternative hypothesis (Dauwe et al., 2007), places altered cell wall integrity as an important factor in determining growth phenotypes. In HCT-downregulated alfalfa lines, the reduced lignin levels confer increased extractability of pectic (oligogalacturonide) fractions from the cell walls. Galacturonides are well known as elicitors of plant defense responses, including PR protein gene expression (Roco et al., 1993), and this was confirmed in the present work by the susceptibility of the elicitor-active fractions from cell walls of HCT-downregulated plants to degradation by polygalacturonase. It is therefore possible that continuous leaching of bioactive oligosaccharides from improperly lignified secondary cell walls can, in some cases, act as the trigger for induced defense. This may be amplified by the release of H₂O₂ from the same cell types, perhaps as excess to requirement for polymerization of the reduced levels of monolignol units in the wall, or else in direct response to the released galacturonides (Legendre et al., 1993). Hydrogen peroxide preinduces PR gene expression and SA accumulation, both locally and systemically (Alvarez et al., 1998), and salicylate acts to enhance H₂O₂ accumulation and downstream defenses (Shirasu et al., 1997).

Release of galacturonides is more likely to be a result of altered cell wall integrity than increased production of pectic compounds in the transgenic plants; in fact, total pectin content was similar in HCT antisense and wild-type stems (Table S6). In addition, microarray analysis failed to reveal upregulation of pectic polysaccharide biosynthesis genes; the few pectin related genes that were upregulated (by about threefold) are all annotated as being involved in pectin degradation. These include two α-galacturonidases and a polygalacturonase inhibitor; it is currently not clear whether or how these changes in gene expression might contribute to the production and release of pectic oligosaccharides from the cell walls of HCT-downregulated plants. It has recently been shown that the H-rich lignin that appears in HCT- or C3′H-downregulated alfalfa is of a lower molecular weight and thus itself more extractable than the bulk lignin (Ziebell et al., 2010). This similarity in cell wall modification is reflected in the observation that PR gene expression is activated in C3′H- as well as HCT antisense lines. However, we did not observe PR gene activation in alfalfa lines downregulated in expression of caffeic acid 3-O-methyltransferase, in which lignin reduction is associated with a drastically decreased S : G ratio with no increase in H-lignin (Guo et al., 2001a; L. Gallego-Giraldo & R.A. Dixon, unpublished data). It is, however, unlikely that the extractable H oligomers are inducing the defense response, as the PR gene inducing activity of the cell wall extracts from HCT antisense lines was completely destroyed by pretreatment with polygalacturonase. The recent observation that the growth defects in HCT-downregulated Arabidopsis plants can be overcome by restoring lignin accumulation through expression of a novel Selaginella enzyme that bypasses the HCT and C3′H reactions (Li et al., 2010) is consistent with the lignified secondary wall itself being the origin of the growth defects, but does not formally rule out the possibility of the effect originating from metabolic spillover.

Plants exhibiting constitutive defense responses tend to exhibit reduced growth (Igari et al., 2008), and this may result in part from SA-mediated downregulation of XTH genes, as observed in the present work, with subsequent effects on cell expansion and elongation (Miura et al., 2010). Salicylic acid accumulation also blocks responsiveness to GA (Xie et al., 2007), as observed here. Reduced GA levels or responsiveness may themselves promote biotic and abiotic stress tolerance (Achard et al., 2006). Clearly, the means by which plants integrate growth and defense signal pathways are complex and still poorly understood (Bari & Jones, 2009). Fig. 7 summarizes possible links between signaling components and downstream responses in HCT antisense alfalfa plants. It accounts for most of the changes observed, but does not explain the increase in cytokinin levels. The model will ultimately require verification through genetic approaches.

Figure 7. Model integrating genetic and metabolic alterations and phenotypes of HCT antisense alfalfa. V, vascular element; IF, interfascicular fiber; P, phloem; PGA, polygalacturonide; SA, salicylic acid; SP/PP, shikimate/phenylpropanoid pathway; XTH, xyloglucan endotransglycosylase/hydrolase. Blue arrows, products of processes (biosynthesis or cell wall leaching); red arrows, induction; green line, repression; black arrows, downstream phenotypic responses. Processes, with literature citations, are: 1, diffusion of hydrogen peroxide from vascular tissues; 2, diffusion of polygalacturonide elicitors (PGAs) from secondary cell walls; 3, induction of hydrogen peroxide formation by PGAs (Legendre et al., 1993); 4, induction of pathogenesis-related (PR) proteins locally and systemically by hydrogen peroxide (Alvarez et al., 1998); 5, generation of SA (Wildermuth et al., 2001); 6, induction of PR proteins by SA (Alexander et al., 1993); 7, downregulation of XTH expression by SA (Miura et al., 2010); 8, potentiation of hydrogen peroxide production by SA (Shirasu et al., 1997); 9, induction of hydrogen peroxide via jasmonate signaling; 10, SA-induced loss of GA responsiveness (Traw & Bergelson, 2003; Xie et al., 2007).

Because of the role of induced lignification as a plant defense mechanism, downregulation of lignin biosynthesis might be expected to weaken plant disease resistance. However, the present data, and other recent reports (Funnell-Harris et al., 2010) indicate that lignin reduction can paradoxically have the opposite effect. Our results provide a model to direct future studies of enhanced resistance in lignin-modified plants.

Acknowledgements

We thank Jack Blount for assistance with cell cultures, David Human for mass spectrometry, Dr Fang Chen for lignin analysis, and Drs Elison Blancaflor and Rujin Chen for critical reading of the manuscript. This work was supported by the Oklahoma Department of Energy’s Bioenergy Center (OBC).