Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
To identify genes controlling secondary cell wall biosynthesis in the model legume Medicago truncatula, we screened a Tnt1 retrotransposon insertion mutant population for plants with altered patterns of lignin autofluorescence. From more than 9000 R1 plants screened, four independent lines were identified with a total lack of lignin in the interfascicular region. The mutants also showed loss of lignin in phloem fibers, reduced lignin in vascular elements, failure in anther dehiscence and absence of phenolic autofluorescence in stomatal guard cell walls. Microarray and PCR analyses confirmed that the mutations were caused by the insertion of Tnt1 in a gene annotated as encoding a NAM (no apical meristem)-like protein (here designated Medicago truncatula NAC SECONDARY WALL THICKENING PROMOTING FACTOR 1, MtNST1). MtNST1 is the only family member in Medicago, but has three homologs (AtNST1–AtNST3) in Arabidopsis thaliana, which function in different combinations to control cell wall composition in stems and anthers. Loss of MtNST1 function resulted in reduced lignin content, associated with reduced expression of most lignin biosynthetic genes, and a smaller reduction in cell wall polysaccharide content, associated with reduced expression of putative cellulose and hemicellulose biosynthetic genes. Acid pre-treatment and cellulase digestion released significantly more sugars from cell walls of nst1 mutants compared with the wild type. We discuss the implications of these findings for the development of alfalfa (Medicago sativa) as a dedicated bioenergy crop.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Plant secondary cell walls constitute the most abundant biomass on earth, and can potentially be utilized for biofuel production via sugar release and fermentation (Somerville, 2007; Carroll and Somerville, 2009). Secondary walls are deposited inside the primary walls after the cells cease expansion (Turner et al., 2001). The ability to produce secondary walls has enabled vascular plants to form strong, thick-walled cells for water transport and mechanical support. Secondary wall thickening plays important roles in various biological processes, such as the formation of tracheary elements and fibers, anther dehiscence, and the shattering of seed pods.
Secondary cell walls are composed mostly of cellulose, lignin and hemicellulose. In recent years, the application of genetic and genomic strategies has led to significant progress in identifying and characterizing the genes involved in the biosynthesis of these cell wall components (Boerjan et al., 2003; Tanaka et al., 2003; Liepman et al., 2005; Somerville, 2006). Although secondary cell wall components are synthesized at different subcellular locations, the expression of the genes involved is highly coordinated, both temporally and spatially.
Recently, NAC (NAM, ATAF and CUC2) and MYB family transcription factors have been shown to play key roles in regulating secondary wall biosynthesis in Arabidopsis thaliana (Mitsuda et al., 2007; Zhong et al., 2007, 2008; Mitsuda and Ohme-Takagi, 2008; Zhou et al., 2009). Two Arabidopsis NAC genes, NAC SECONDARY WALL THICKENING PROMOTING FACTOR 1 (NST1) and NST2, were shown to function together in the promotion of secondary wall thickening in the anther endothecium, a process that is necessary for anther dehiscence (Mitsuda et al., 2005). In addition, the simultaneous knock-out of NST1 and another NST gene, NST3, resulted in the loss of secondary wall deposition in Arabidopsis stems, whereas single knock-outs in NST1 or NST3 show no phenotype (Mitsuda et al., 2007; Mitsuda and Ohme-Takagi, 2008). It was proposed that NST genes are master switches, capable of turning on the entire biosynthetic pathways for cellulose, xylan and lignin by initiating a transcriptional signaling network to direct downstream targets, including MYB transcription factors (Zhong et al., 2008).
Barrel medic (Medicago truncatula) has been developed as a model legume with extensive genomic and genetic resources (Town, 2006; Benedito et al., 2008; Young and Udvardi, 2009). It is very closely related to alfalfa (Medicago sativa), the world’s major forage legume, and potentially a dedicated bioenergy crop (Yang et al., 2009). Previous studies on engineering cell walls to improve biomass processing in alfalfa have directly targeted the enzymes of the lignin biosynthetic pathway (Reddy et al., 2005; Chen et al., 2006; Shadle et al., 2007). Here, we describe a non-biased, forward genetic approach to discover genes determining levels and spatial patterns of lignification in M. truncatula. Retrotransposon insertions in the MtNST1 gene result not only in the loss of lignification in the interfascicular region of the stem, and in the prevention of anther dehiscence, but also in alterations in the cell walls of stomatal guard cells. We describe the metabolic and transcriptional changes associated with the loss of MtNST1 function, and the impacts of NST1 downregulation on cell wall recalcitrance to saccharification. We also discuss the implications of these studies for the development of alfalfa as a dual forage/bioenergy crop.
Isolation and characterization of M. truncatula nst1 mutants
To identify mutants with defects in secondary cell wall formation, an R1 population of tobacco Tnt1 retrotransposon insertion-mutagenized M. truncatula (Tadege et al., 2005, 2008) was screened by UV microscopy of cross sections of the sixth stem internodes, where lignification is well developed in wild-type plants. Approximately 3400 independent R1 lines (around 9000 plants) were analyzed, and several lines with significantly altered extents and/or patterns of lignin autofluorescence were observed (Figure S1). One of these lines showed a dramatic loss of lignin in interfascicular fibers (Figure S1b). To better characterize developmental patterns of lignification in this mutant (line NF3056), we compared autofluorescence in cross sections from the second to fifth internodes (from top to bottom) of NF3056 and wild-type R108 M. truncatula (Figure 1). In the wild-type plant, lignin starts to accumulate in vascular tissues at an early stage (Figure 1e), and is most intensely located in secondary walls of vascular tissues, pith rays and fibers at late stages (Figure 1h). Compared with the wild-type plant, NF3056 shows no difference in the location or intensity of lignin autofluorescence at the early stage (Figure 1a,e), but there was clearly less lignin in the third internode and below, and the lignin was only deposited in vascular cells, in contrast to the strong lignification in the interfascicular region and the phloem fibers in the wild-type (compare Figure 1b–d, f–h). The level of lignin autofluorescence in vascular elements also appeared somewhat lower in mutant compared with wild-type plants. Although lignin fluorescence was absent from phloem fibers in the mutant, the phloem tissues appeared normal under the light microscope (Figure S2).
To identify the gene responsible for the tissue-specific loss of lignification, microarray analysis was performed with RNA isolated from lignifying stem internodes of NF3056. As Medicago Tnt1 retrotransposon lines generally contain 20–50 insertions (Tadege et al., 2008), progeny from the same parent plant segregating with the wild-type lignification phenotype were used as controls. Total RNA samples from triplicate biological replicates of the mutants and controls were subjected to Affymetrix microarray analysis. In total, 152 probe sets were downregulated in the mutant line by at least twofold (Table S1). To identify the gene with the Tnt1 retrotransposon insertion, PCR was performed with a Tnt1 primer and primers designed from the probe set sequences: this was done sequentially in order of the extent of downregulation of the 152 probe sets. PCR amplification using primers complementary to the three most downregulated genes in Table S1, each of which is annotated as encoding a gene involved in cell wall biosynthesis, did not lead to amplification of a PCR product, whereas the fourth candidate (Mtr.33913.1.S1_at, expression level decreased by 22-fold) was confirmed to contain a Tnt1 retrotransposon insert. The probe set Mtr.33913.1.S1_at is annotated as representing a NAM (no apical meristem)-like protein.
Using the Mtr.33913.1.S1_at probe sequence, a cDNA BLAST search was performed against the M. truncatula genome from the Dana-Faber Cancer Institute (DFCI) bioinformatics web site (http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=medicago). The first hit with the lowest e-value was TC141793, which contains a 1113-bp cDNA sequence, including the entire Mtr.33913.1.S1_at probe sequence. Using the TC141793 sequence, a nucleotide BLAST search was then performed against the NCBI database, and the first hit was Arabidopsis NST1 (At2g46770), a recently discovered NAC transcription factor responsible for secondary wall thickening (Mitsuda et al., 2005; Zhong et al., 2007). As TC141793 does not contain the full-length cDNA, 3′-RACE was performed to complete the cDNA sequence. Because NF3056 phenocopies the Arabidopsis nst1:nst3 double mutant, and TC141793 shares sequence similarity with Arabidopsis NST1, we named the full-length sequence MtNST1 (GenBank accession number GU144511), and the NF3056 mutant nst1-1.
Protein sequence alignment was performed using the deduced amino acid sequence of MtNST1, and Arabidopsis NST1, NST2 and NST3. Because Arabidopsis NST1 and NST2 are more similar to each other than either is to NST3, we first aligned MtNST1 with AtNST1 and NST2 (Figure 2a), and then aligned the Medicago sequence separately with AtNST3 (Figure S3). MtNST1 shares high sequence similarity with all three AtNST proteins at the conserved N-terminal NAC domain. The level of conservation is comparable with that between AtNST proteins. However, the C-terminal domain of MtNST1 is quite distinct, indicating that MtNST1 might have different functional characteristics. The overall sequence similarity between MtNST1 and AtNST proteins is only around 50%, even though the NAC domain is highly conserved.
Because of the problem of remobilization of Tnt1 elements during tissue culture (Tadege et al., 2005), we did not attempt to directly complement nst1–1 with the MtNST1 open reading frame. Instead, to confirm that the tissue-specific loss-of-lignification phenotype was indeed the result of insertional mutagenesis of MtNST1, we obtained three further independent alleles. The first (NF2911) was obtained by forward genetic screening as above, on the basis of the loss of interfascicular fluorescence, and microarray analysis confirmed the downregulation of many, but not all, of the genes shown to be downregulated in Mtnst1, including the NST1 gene itself (Table S1). PCR analysis then confirmed the insertion of a Tnt1 transposon in MtNST1. We then used MtNST1 gene-specific primers for the reverse genetic screening of DNA pools from the complete Tnt1 mutant population. Two additional independent lines (NF4041 and NF3687) were positive for the presence of Tnt1 insertions in MtNST1, and homozygous lines of both showed the altered lignification phenotype (Figure S4). Tnt1 flanking sequence analyses indicated that Tnt1 is inserted in the NST1 gene in the first exon, at 43 (nst1-1, NF3056) or 170 bp (nst1-2, NF2911), or in the second exon at 586 (nst1-3, NF4041) or 437 bp (nst1-4, NF3687) (Figure 2b). RT-PCR analysis confirmed that no MtNST1 transcript was present in any of the four mutant lines (Figure 2c and S4e).
As a final validation that the loss of lignin phenotype is caused by a loss of function of MtNST1, we obtained another line, by forward genetic screening, harboring a point mutation rather than a transposon insertion in MtNST1 (H. Wang, R.A. Dixon and F. Chen, unpublished data). Plants heterozygous for both the point mutation allele and the Tnt1 insertion allele at the MtNST1 locus showed loss of interfascicular lignification, further confirming NST1 as the gene controlling the lignification phenotype.
Cell wall composition of Mtnst1 mutant stems
The total lignin content of stem internodes 5–8, as determined by the acetyl bromide method (Fukushima and Hatfield, 2004), was reduced by around 50% in the Mtnst1 mutant compared with wild-type plants (Figure 3a). Thioacidolysis (Lapierre et al., 1985) revealed a reduction in G lignin levels in stems of Mtnst1 mutants, but a proportionately larger decrease in S lignin, such that the S/G ratio was dramatically reduced in the mutants (Figure S5).
The same cell wall samples used for lignin analysis were analyzed for total sugar content by phenol-sulfuric assay (Dubois et al., 1956). Total sugar levels were somewhat reduced in nst1 mutants compared with wild-type plants (Figure 3b). Analysis of monosaccharide composition indicated that glucose, xylose, galactose and arabinose were the major sugar components of the M. truncatula cell wall, with galactose levels being significantly higher than in Arabidopsis (Zhong et al., 2008) (Figure 3c). In comparison with the wild type, cell walls from nst1 mutants had slightly decreased levels of glucose and xylose, whereas there was no significant change in galactose or arabinose levels (Figure 3c). In contrast, Arabidopsis plants with a loss of NST function through expression of a chimeric NST repressor exhibit significantly reduced levels of xylose, and nearly double the levels of galactose and arabinose, compared with controls (Iwase et al., 2009). The change in overall monosaccharide composition in the Mtnst1 mutants is likely to stem from a reduced deposition of cellulose and hemicellulose in the defective secondary walls of the interfascicular fiber cells.
Levels of wall-bound phenolic compounds were determined by HPLC analysis of alkaline hydrolysates of cell wall material. The stem cell walls contained low levels of alkali-releasable vanillin, p-coumaric and ferulic acids. Apart from a small reduction in vanillin, the loss of function of Mtnst1 had little effect on wall-bound phenolic levels (Figure S6).
Soluble phenolic compounds in M. truncatula nst1 mutants
HPLC analysis was used to compare the profiles of soluble phenolic compounds in wild-type and mutant plants. The major soluble phenolics in Medicago stems are a series of glycosides of the flavones apigenin and luteolin (Kowalska et al., 2007). Pairwise comparisons of the nst1-1 and nst1-2 lines with corresponding controls indicated no clear quantitative or qualitative differences in soluble phenolic profiles (Figure S7).
Growth characteristics of M. truncatula nst1 mutants
At first sight, the overall growth phenotypes of the four independent nst1 mutant lines appeared similar to those of wild-type plants, although the branches were less able to support themselves against gravity, presumably because of the defect in secondary wall formation (Figure S8a). Vegetative growth was not strongly affected in the Mt-nst1 mutant lines, but stem length was reduced by around 25%, and the average leaf area increased by around 14% (Figure S8b; Table 1). The reduction in stem length was mainly the result of a reduction of internode length in the three most mature internodes at the 12-internode stage (Figure S9). Flowering time and stem diameter were not affected in the mutant lines (Table 1).
Table 1. Growth parameters of independent nst1 insertion mutants
Stem height (cm)
Mean internode length (cm)
Leaf area (mm2)
Stem diameter (mm)
Results are the means of six stems per line ± SE. For each mutant line analyzed, a wild-type (Wt) was grown in parallel.
aAverage of the combined data for Wt1, 2 and 3 and nst1-1, -2 and -3.
70.4 ± 0.5
11.5 ± 0.6
5.9 ± 0.5
5.0 ± 0.2
1.6 ± 0.2
60.8 ± 0.6
11.0 ± 0.3
5.1 ± 0.2
6.4 ± 0.3
1.6 ± 0.1
67.7 ± 0.7
10.8 ± 0.6
5.7 ± 0.5
9.1 ± 0.5
2.0 ± 0.2
65.0 ± 0.6
12.0 ± 0.8
5.4 ± 0.4
11.0 ± 0.8
1.9 ± 0.3
70.3 ± 0.4
11.5 ± 0.6
5.9 ± 0.3
6.2 ± 0.4
1.6 ± 0.1
61.5 ± 0.4
11.0 ± 0.7
5.1 ± 0.3
7.3 ± 0.8
1.8 ± 0.2
69.5 ± 0.3
11.2 ± 0.5
5.8 ± 0.2
6.9 ± 0.6
1.73 ± 0.1
62.4 ± 0.4
11.3 ± 0.6
5.2 ± 0.3
8.1 ± 0.7
1.76 ± 0.2
Stomatal phenotype of Mtnst1 mutants
Examination of leaf epidermal cells revealed a significant loss of fluorescence signal in stomatal guard cells of homozygous mutant leaves, compared with the wild type (Figures 4a,b and S10). This was observed in all four independent nst1 alleles. Furthermore, after a light induction period, the guard cells of the nst1 mutant exhibited a striking reduction in stomatal aperture, of around 80% compared with controls (Figure 4c,d; Table 2). This was associated with a reduction in transpiration rate in the nst1 mutants, accompanied by a small but significant increase in the photosynthetic rate (Table 2).
Table 2. Stomatal aperture, transpiration rate and photosynthetic rate of independent nst1 insertion mutants
Photosynthetic rate (μmol CO2 m−2 sec−1)
Transpiration rate (mmol H2O m−2 sec−1)
Photosynthetic and transpiration rates are the means of six values (three replicates taken on two different days) ± SE. The stomatal aperture data (ratio between lengths and widths) are the means of 75 stomatal guard cells distributed in three different areas of the film strips ± SE. For each mutant line analyzed, a wild-type (Wt) was grown in parallel.
aAverage of the combined data for Wt1, 2 and 3 and nst1-1, -2 and -3.
13.1 ± 0.2
4.5 ± 0.1
1.5 ± 0.3
14.2 ± 0.3
3.9 ± 0.2
3.6 ± 0.3
14.2 ± 0.1
4.1 ± 0.2
1.7 ± 0.2
15.8 ± 0.2
3.5 ± 0.2
4.6 ± 0.3
14.2 ± 0.3
4.3 ± 0.1
1.8 ± 0.2
15.3 ± 0.1
3.6 ± 0.1
5.2 ± 0.3
13.8 ± 0.2
4.3 ± 0.1
1.6 ± 0.2
15.1 ± 0.2
3.6 ± 0.2
4.5 ± 0.3
The autofluorescence of stomatal guard cells has been proposed to result from the presence of wall-associated pectin-linked ferulic acid (Jones et al., 2005). To determine whether NST1 knock-out lines exhibited reduced wall-bound ferulic acid levels in leaf tissues, we first measured the total wall-bound phenolic levels in leaf panels (minus veins). Levels of coumaric, caffeic and ferulic acids were similar in extracts from the wild type and mutant lines. However, although the data suggested small increases in wall-bound coumaric and caffeic acid levels, and a small (less than 10%) reduction in wall-bound ferulic acid levels, these apparent changes were not statistically significant (data not shown).
To better address changes in phenolic compounds in the guard cells of Mtnst1 mutants, we used Coherent anti-Stokes Raman spectroscopy (CARS) (Zeng et al., 2010) to directly visualize ferulate signals in the walls. Previous Raman spectroscopy studies have demonstrated the capability of detecting ferulate by using specific Raman bands (Ram et al., 2003; Hélène et al., 2008). The sharp Raman band at 1600 cm−1 corresponds to the stretch vibration of the aromatic ring structure in ferulate. Similar to ordinary spontaneous Raman scattering, CARS offers chemical contrast based on the intrinsic Raman vibrational frequencies of the molecules of interest in the sample. However, because of the coherent nature, CARS offers orders of magnitude higher sensitivity than spontaneous Raman. The stomatal guard cells, and the central cavity between the cells of the complex, showed a bright CARS signal caused by the presence of ferulate in the epidermal peels of leaves of wild-type plants (Figure 4e,f). In contrast, this signal was largely absent on analysis of mutant plants (Figure 4g,h).
Anther dehiscence in Mtnst1 mutants
Arabidopsis NST1 and NST2 together regulate secondary wall thickening in the anther endothecium, and nst1:nst2 double mutants show a defect in anther dehiscence (Mitsuda et al., 2005). Flowers of all four independent Mtnst1 mutants had indehiscent anthers, whereas the pollen grains are released normally from wild-type anthers (Figures 5a,b and S11). To examine pollen viability in the nst1 flowers, Alexander’s staining method was used (Alexander, 1969). Light microscopy showed that the pollen grains from Mtnst1-1 take up the stain, and are therefore viable; they also exhibit normal shape and size (Figure 5c,d). M. truncatula is a self-pollinating species: as a result of the impairment of anther dehiscence, mature plants of the nst1 mutants generally failed to develop seed pods, although a few very small pods were occasionally observed on old plants.
Potential downstream target genes for MtNST1 action
Because of the existence of multiple Tnt1 insertions in the Medicago mutant lines, the initial microarray analysis by which the NST1 gene was identified does not provide a definitive indication of co-regulated or potential downstream target genes for NST1 action. We therefore performed a parallel gene expression analysis using RNA from stems of an independent allele. Of the 152 probe sets that were downregulated by more than twofold in the nst1-1 mutant compared with its null segregant control, 94 were also downregulated by more than twofold in the nst1-2 mutant, compared with its null segregant control. Eighteen probe sets were also upregulated by more than twofold in both mutant lines (Table S1). In view of the lack of overlap in additional Tnt1 flanking sequences recovered from the two lines (data not shown), these genes represent probable targets for control via NST1. This is further supported by the observation that the same set of genes with increased or reduced expression in both the nst1-1 and nst1-2 lines was also regulated similarly in the independent nst1 point-mutation line.
Of the common downregulated genes in stems, 12 were involved in lignin biosynthesis, 17 were putatively involved in cellulose or hemicellulose biosynthesis, and 14 were tentatively involved in other aspects of cell wall formation, or with the cytoskeleton (Table 3). Quantitative real-time PCR was used to validate microarray results. This indicated that the pattern of MtNST1 expression paralleled that of lignin biosynthetic genes during stem development (Table 4), and that the expression of multiple genes involved in the biosynthesis of lignin, namely FERULATE 5 HYDROXYLASE (F5H), 4-COUMARATE CoA LIGASE (4CL), COUMAROYL SHIKIMATE 3-HYDROXYLASE (C3H), HYDROXYCINNAMOYL CoA: SHIKIMATE HYDROXYCINNAMOYL TRANSFERASE (HCT), CINNAMOYL CoA REDUCTASE (CCR), CAFFEIC ACID 3-O-METHYLTRANSFERASE (COMT) and CAFFEOYL CoA 3-O-METHYLTRANSFERASE (CCoAOMT) was indeed reduced in stems of the nst1-1 mutant (Figure 6a). Genes with high sequence similarity to IRREGUALR XYLEM 1 (IRX1) and FRAGILE FIBER 8 (FRA8) were selected to represent genes potentially involved in the cellulose and hemicellulose synthesis pathways, respectively. As the identities of these two genes in M. truncatula have not yet been experimentally confirmed, three different primer pairs were chosen from each gene for quantitative real-time PCR (qRT-PCR) to ensure the accurate determination of gene expression: this confirmed that the expression levels of the IRX1 and FRA8 homologs were decreased in the nst1 mutants (Figure 6b,c). The decreased expression of genes involved in secondary cell wall formation can explain the reduction of lignin and cell wall sugar in the nst1 mutant, similar to the situation in the Arabidopsis nst1:nst3 double mutant (Mitsuda et al., 2007).
Table 3. Cell wall-related genes downregulated by more than twofold in stems and leaves of two independent Medicago truncatula nst1 mutants
IRX14 | IRX14 (IRREGULAR XYLEM 14); glycosyl transferase that contributes to xylan biosynthesis
Glycosyl hydrolase family 3 protein 1e−68 same family as bxl2 and bxl3
Other cell wall related
Arabinogalactan protein like
ATMAP70-2, F5A9.19, F5A9_19, MICROTUBULE-ASSOCIATED PROTEIN 70-2
ARPC2B | ARPC2B (actin-related protein C2B), structural molecule; similar to ARPC2A/DIS2 (DISTORTED TRICHOMES 2)
Kinesin light chain-related
Kinesin light chain-related
KATC, ATK3 | ATK3 (ARABIDOPSIS THALIANA KINESIN 3); microtubule motor
Table 4. Quantitative real-time PCR analysis of NST1 and lignin pathway gene transcripts in Medicago truncatula stem internodes at different developmental stages (from 50-day-old plants)
Relative transcript level
Transcript level is expressed relative to actin.
41 ± 0.9
131 ± 28
477 ± 68
3315 ± 187
5108 ± 971
15 782 ± 1155
19 ± 2
33 ± 7
580 ± 101
1437 ± 317
2601 ± 498
8704 ± 302
3603 ± 355
5274 ± 579
8792 ± 285
In addition to the inactivated NST1 gene, probe sets annotated as encoding other transcription factors were also downregulated by more than twofold in Mtnst1-1 and Mtnst1-2 mutant lines, compared with controls: these included two basic helix loop helix (BHLH) genes, two NAC genes and two zinc finger protein genes (Table S1). The two NAC genes are homologs of Arabidopsis SND2, and are in a different phylogenetic subgroup from MtNST1 (Shen et al., 2009). Neither of these genes contained a transposon insertion in any of the four independent nst1 mutant alleles.
Because of the stomatal phenotype in leaf cells of the nst1 mutants, we also performed microarray analysis with RNA samples from leaf panels (major veins removed) of wild-type and mutant plants. The loss of MtNST1 function also has a major impact on polysaccharide biosynthesis gene expression in leaves (Table 3). The most downregulated genes were represented by eight cellulose synthase probe sets, in addition to arabinogalactan protein, peroxidase, glycoside hydrolase, chitinase-like and proline-rich protein. Five of the cellulose synthase probe sets were also downregulated in stems of the nst1 mutant, three were only downregulated in leaves and two were only downregulated in stems. The arabinogalactan protein probe sets were downregulated in leaves but not stems of the nst1 mutant. The other cell wall biosynthesis genes that were downregulated in stems of the nst1 mutants were not downregulated in leaves (Table 3).
Saccharification potential of Nst1 mutants
To evaluate the impact of cell wall changes on recalcitrance of biomass to saccharification, mature stems of Mtnst1-1 and Mtnst1–2 plants were subjected to dilute sulfuric acid pre-treatment and enzymatic hydrolysis with cellulases. Compared with the wild type, cell walls from the nst1 mutants released more glucose and less xylose during sulfuric acid pre-treatment (Figures 7a and S12). Without pre-treatment, the cellulase mixture could release 64% of the structural sugars from the nst1-1 mutant, significantly higher than in the wild type (46%). Following acid pre-treatment, enzymatic hydrolysis could release 90% of the total sugars (723 mg sugar per gram of cell wall residue) from nst1-1 compared with 67% (572 mg sugar per gram cell wall residue) from the wild type (Figure 7b). Similar results were obtained with a second allele (nst1-2) (Figure S12).
Loss of function of a single NAC gene impacts lignification in Medicago
We have identified an NAC family transcription factor, MtNST1, from UV microscopy-based screening for secondary cell wall formation defects in an M. truncatula mutant collection generated by tobacco Tnt1 retrotransposon insertion mutagenesis. Four independent nst1 mutants all show secondary cell wall formation defects in stems, anther dehiscence defects, and reduced ferulic acid fluorescence in stomatal guard cell walls. The nst1 mutants phenocopy the recently reported Arabidopsis nst1:nst3 double mutant and nst1:nst2 double mutant, as regards the stem and anther phenotypes, respectively (Mitsuda et al., 2007; Zhong et al., 2007), but the guard cell phenotype has not been reported previously.
Unlike Arabidopsis, which contains three partially redundant NST genes, M. truncatula only appears to possess one NST gene. M. truncatula NST1 shares high sequence similarity with its Arabidopsis counterparts, but only in the conserved N-terminal NAC domain. A recently published phylogenetic analysis of 1232 NAC proteins from 11 different plants places the proteins into eight subfamilies, which are then divided into different clades based on motif patterns in the highly diverse C-terminal regions (Shen et al., 2009). MtNST1 and the three Arabidopsis NST proteins all fall into the NAC-c subfamily, subgroup c-3. Interestingly, Medicago is the only organism with a single representative NAC protein in the C-3 subfamily.
The M. truncatula nst1 mutants have severe secondary wall biosynthesis defects, whereas none of the single Arabidopsis NST gene knock-outs shows such phenotypes. Redundancy appears to be a common theme for other genes involved in cell wall component biosynthesis. For example, 10 CesA (cellulose synthase) genes are present in Arabidopsis, and poplar has 18 (Djerbi et al., 2005; Keegstra and Walton, 2006). At least nine cinnamyl alcohol dehydrogenase (CAD) enzymes catalyzing reduction of various phenylpropenyl aldehyde derivatives in lignin biosynthesis have been functionally characterized in Arabidopsis (Kim et al., 2004). Arabidopsis CAD4 and CAD5 are functionally redundant, and only a double knock-out shows striking lignin reduction (Sibout et al., 2005). A cad1 mutant from M. truncatula shows severe lignin reduction when the single gene is disrupted, indicating that, like MtNST1, CAD1 also does not have a second functionally redundant gene family member in Medicago (Q. Zhao, F. Chen and R.A. Dixon, unpublished data). Overall, however, phenylpropanoid pathway gene families are as complex, or more complex, in Medicago than in Arabidopsis: for example, Arabidopsis possesses a single functional chalcone synthase (CHS) gene, whereas the Medicago CHS gene family contains at least 23 members.
MtNST1 orchestrates cell wall synthesis in multiple cell types
Based on the microarray analysis of gene expression patterns at different developmental stages of stems, MtNST1 is expressed at a very low level in early stages, when most cells are still dividing and expanding. The expression level increases as cells mature, in parallel with the expression of lignin biosynthetic genes.
Knock-out of MtNST1 appears to block the development of secondary cell walls in the interfasicular region of the stem, and in the phloem fibers, and may also lead to a reduction of total lignin in xylem vessel cell walls. However, the loss of fluorescence and ferulate-related CARS signal from the guard cell walls in all of the nst1 mutants suggests that NST1 may also control modification of primary cell walls in guard cells. Substitution of the arabinan portion of pectin with ferulic acid is essential for providing the necessary physical properties to guard cell walls, to allow for the expansion and contraction necessary for stomatal opening and closing (Jones et al., 2003), and the removal of ferulates from guard cell walls by incubation of epidermal strips with feruloyl esterase results in an inability of guard cells to open properly (Jones et al., 2005). This is exactly mirrored by the phenotype of the nst1 mutants, in which guard cell aperture is dramatically reduced.
The CARS analysis provides spatial and semi-quantitative information on ferualte levels in the stomatal complex. Direct measurement of pectin-linked ferulate levels in guard cells of nst1 mutants, compared with wild-type M. truncatula will probably require laser capture microdissection, a technique that we have recently used for analysis of cell type-specific deposition of lignin in transgenic alfalfa (Nakashima et al., 2008).
The lack of significant reduction in the total leaf wall-bound ferulate pool in the nst1 mutants, despite the total loss of guard cell wall fluorescence, can be readily explained by the low frequency of guard cells in the samples. This would also explain our inability to detect changes in pectin-related cell wall biosynthetic gene transcripts in microarray analysis of transcripts from whole leaf panels. It is possible that the loss of function of MtNST1 also leads to a loss of wall-bound ferulate in the interfascicular region of the stem: this would not be observed because of the massive autofluorescence from lignin in wild-type plants.
We cannot rule out the possibility that the reduction in the stomatal aperture response in nst1 mutants might also result from disruptions to other cell wall components, such as cellulose, as suggested by the downregulation of a specific class of putative cellulose synthase genes in nst1 mutant leaves.
Potential downstream targets for NST1 action
In addition to the downregulation of over fifty genes involved in the biosynthesis of the major components of the secondary cell wall, genes encoding six putative transcription factors were either down- or upregulated in stems of two independent Mtnst-1 mutant lines. Previous work from Arabidopsis has suggested that NST genes function as master switches of a transcriptional network involved in the regulation of secondary wall biosynthesis, and several transcription factors have been shown to be regulated by NST genes (Zhong et al., 2008; Zhou et al., 2009). Among the six transcription factors with modified expression in the nst1 mutants, one is the homolog of AtMYB103, which was previously reported to be regulated by AtNST1 and NST3 (Zhou et al., 2009). The other transcription factors include two NAC genes, two Dof-type zinc finger domain-containing genes and one homeobox gene. These could potentially control downstream branches of the secondary wall biosynthesis pathway in Medicago. For example, the two NAC genes downregulated as a result of loss of MtNST1 function are homologs of AtSND2, which is a downstream target of NST1 in Arabidopsis (Zhong et al., 2008). The zinc finger and homeodomain genes represent novel candidates for cell wall regulation.
The transcript levels of two putative laccases were decreased as a result of the NST1 downregulation in stems. An additional microarray analysis using RNA samples from different developmental stages indicates that the two laccases exhibit a similar expression pattern to that of lignin biosynthetic genes (H. Wang, Q. Zhao, F. Chen and R.A. Dixon, unpublished data). Despite many years of effort, no specific laccase genes have been unequivocally demonstrated to be involved in lignin polymerization. The finding that the two laccases are regulated by MtNST1, and that they have similar developmental expression patterns as lignin genes, suggests that both laccases are potentially involved in lignin polymerization.
MtNST1 as a target for development of alfalfa as a bioenergy crop
Mutation of MtNST1 in M. truncatula resulted in the abolition of lignification in interfascicular and phloem fibers, and is likely to affect other aspects of secondary cell wall synthesis in these fiber cells. Cell wall xylose levels were significantly reduced, which probably reflects changes in the hemicellulose component of the secondary cell wall. The overall consequence of these changes in cell wall composition is that the mutant is more amenable to chemical and enzymatic treatment for release of sugars for ethanol fermentation. Whereas the lower level of xylose in the pre-treatment hydrolysate is likely to result from impaired biosynthesis of hemicelluloses in nst1 mutants, the increased glucose content may be caused by the decreased crystalinity of cellulose.
Knock-out of NST1 in M. truncatula greatly improved the enzymatic hydrolysis efficiency of cell walls with or without acid pre-treatment. Similar to our previous report on transgenic plants modified in the monolignol biosynthetic pathway (Chen and Dixon, 2007), the sugar release efficiency of cell walls from mutants that were not pre-treated is close to that of the pre-treated wild type cell walls. Recent studies indicate that biomass from NST-downregulated Arabidopsis plants is approximately twice as efficiently processed by enzymatic and physicochemical treatments as is biomass from control plants (Iwase et al., 2009). The fact that the walls from the mutants are more responsive to chemical and enzymatic treatment suggests that milder pre-treatment or lower enzyme loading could be used to release sugar from these materials. Although there is a small overall reduction in cell wall polysaccharide in the Mtnst1 mutants, this is more than compensated for as regards saccharification yields by the reduction in lignin content. Furthermore, the growth reductions observed in lines harboring a complete knock-out of NST1 are significantly less than those observed in transgenic plants, in which saccharification efficiency is similarly improved through the knock-down of monolignol biosynthetic enzymes (Chen and Dixon, 2007).
Our results clearly demonstrate the potential of NST1 as a target for genetic engineering of plant cell walls for bioenergy feedstock improvement. Furthermore, NST1 downregulation has the added advantage (from the transgenic release perspective) of resulting in pollen containment, and the consequent lack of seed production in the self-pollinating M. truncatula resulting from the impairment of anther dehiscence; it is, however, easy to obtain pollen for crossing during breeding through manual dehiscence of the anthers. Full exploitation of nst1 mutants for bioenergy and/or forage crop development might, however, require crossing into lodging tolerant genetic backgrounds to compensate for the weakened stems as a result of the blocked development of the fiber cell walls.
Plant materials and growth conditions
A tobacco (Nicotiana tabacum) Tnt1 retrotransposon-tagged mutant collection of M. truncatula (Tadege et al., 2005, 2008) was screened for defects in secondary cell wall formation. Plants were grown in MetroMix 350 soil mix at 24°C during the day and 20°C during the night, with a 16-h day/8-h night photoperiod, 70–80% relative humidity, and 150 μmol m−2 sec−1 light intensity. Sixth internodes of each plant were harvested when the plants had reached around eight internodes, and were stored at −80°C.
Four nst1 alleles were identified from forward and reverse genetic screening of a segregating population of just over 9000 independent R1 lines.
Identification and molecular cloning of MtNST1
To identify the gene that is linked to the lack-of-lignification phenotype, total RNA samples from fifth to eighth internodes were subjected to Affymetrix microarray analysis, as described below. Segregating progeny without the loss-of-lignification phenotype from the same parent plant as the mutants were used as control. Downregulated probe sets were chosen as candidates for RT-PCR analysis, prioritized in order of the extent of downregulation, as compared with lines segregating with the wild-type phenotype. PCR was performed using Tnt1 (forward primer, 5′-TCCTTGTTGGATTGGTAGCCAACTTTGTTG-3′; reverse primer, 5′-AGTTGGCTACCAATCCAACAAGGA-3′) and gene specific primers to confirm that the insertion was linked to the phenotype.
To clone the full-length NST1 gene, BLAST analysis of the M. truncatula genome from DFCI (http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=medicago) was performed using Mtr.33913.1.S1_at as the query probe sequence: this led to a partial cDNA sequence. The sequence was completed by 3′ RACE using the SMART RACE cDNA Amplification Kit (Clontech, http://www.clontech.com) using the NST1 forward primer (5′-CCCTAGACAGCCCTTCTGGAGAGGGAAGAAGAAGCC-3′) to complete the 3′ end of the NST1 cDNA. The NST1 genomic sequence was also PCR amplified and sequenced from M. truncatula ecotype R108.
To confirm NST1 as the disrupted gene causing the lignification phenotype, reverse genetic screening was performed to uncover additional alleles. Genomic DNA was extracted from individual lines and pooled together with 500 independent lines, contributing to one super pool. PCR screening of superpools, and subsequently the screening of smaller pools to individual lines, was performed using the Tnt1 forward primer 5′-TCCTTGTTGGATTGGTAGCCAACTTTGTTG-3′, the Tnt1 reverse primer 5′-AGTTGGCTACCAATCCAACAAGGA-3′ and two pairs of NST1 gene-specific primers.
megalign protein alignment software (DNASTAR, http://www.dnastar.com) was used for multiple sequence alignments using the Clustal algorithm.
Identification of Tnt1 insertion sites in MtNST1
The Tnt1 flanking sequences from the nst1-1, nst1-2, nst1-3 and nst1-4 alleles were PCR amplified using a combination of the NST1 forward primer (5′-ATGCCTGATAACATGAGTATATCTGTTAATGGAC-3′) and the Tnt1 primer above. The PCR product sequences were aligned with the NST1 genomic DNA sequence to determine the retrotransposon insertion sites.
Phenotypical analysis and physiological measurements
All measurements were performed on plants of M. truncatula at the time when the first flower appeared. Six stems were harvested for each line. The number and lengths of internodes and total stem length were recorded, and their means analyzed in Excel (Microsoft, http://www.microsoft.com). 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 calibrator (Fisher Scientific, http://www.fishersci.com).
Photosynthesis and transpiration rates were determined using an LI-6400 infra-red CO2 gas analyzer (LI-COR). Cylinders of pre-mixed gases were supplied by LI-COR. Measurements were made at a light intensity of 600 μmol m−2 sec−1, a leaf temperature of 22°C and a vapor pressure deficit of 0.7 kPa. A fluorescence chamber head (Li-6400-40; LI-COR) was integrated with the LI-6400 open gas exchange system to measure steady-state fluorescence (F′s) and maximum fluorescence during a light-saturating pulse (F′m). The photochemical quantum yield of electron transport through photosystem II (PSII) was calculated as ΦPSII = (F′m−F′s)/F′m (Genty et al., 1989). All measurements were performed on two different days from 11:00 until 13:00 h in the glasshouse chamber.
Scanning electron microscopy
Samples were vacuum infiltrated with 3% glutaraldehyde in 1xPBS overnight at 4°C, then washed four times with fresh PBS solution. Post-fixation was in 1% osmiun tetroxide in H2O for 2 h on ice, followed by washing in fresh PBS solution. Samples were dehydrated in an ethanol series from 50 to 100%, then placed in 100% tert-butanol and freeze dried. Dried samples were mounted on aluminum stubs, and petals of individual flowers were removed manually. The samples were then sputter-coated with gold palladium. Flower structures were examined under a Zeiss DSM-960A SEM at an accelerating voltage of 5 kV. Digital photographs were collected and assembled using Photoshop (Adobe, http://www.adobe.com).
Analysis of stomata
Epidermal cells strips were peeled with nail varnish from mature leaves of 8-week-old plants. The thin film was peeled off from the leaf surface, mounted on a glass slide and observed under the microscope. For measurement of stomatal opening, the plants were kept at 20°C in the dark for 1 h, and then transferred to light for 2 h. Epidermal cell film strips were taken under both conditions, and observed directly under the microscope. Strips harvested during the dark phase were kept in 10 mm KCl, and those harvested in the light were kept in 75 mm KCl. Stomatal aperture was defined as the ratio between the length and the width of guard cells (in μm). A ratio of 1 indicates the maximum opening potential of the stomatal pore. Length analysis was performed using ImageJ (http://rsbweb.nih.gov/ij).
For CARS microscopy, we used a mode-locked Nd:VAN laser (High Q Laser, Inc., http://www.highqlaser.at) to generate 7-ps, 76-MHz pulse trains of both 1064- and 532-nm laser beams. The 1064-nm beam was used directly as the Stokes beam. The 532-nm beam pumped an optical parametric oscillator (OPO) (Levante Emerald; APE-Berlin, http://www.ape-berlin.de) to generate the CARS pump beam. The pump beam was tuned to 910 nm to effectively detect the 1600-cm−1 Raman band. Collinear and temporally overlapped pump and Stokes beams were focused onto the sample by an Olympus UPlanSapo 60× water immersion objective (http://www.olympus-global.com). Epidermal layers of the leaves were carefully peeled off and spread out between two coverslips. The CARS images of ferulate in this epidermal layer were obtained at 1600 cm−1 on an Olympus IX 81 microscope. The focused beams were raster scanned over the sample. The anti-Stokes light was collected from the Epi-direction and filtered by a clean-up filter 800/40 (Thorlabs, http://www.thorlabs.de). Each time, an off-resonance image at 1550 cm−1 was also acquired and subtracted from the 1600 cm−1 image to remove the non-resonance background.
The second to fifth internodes, counting from the top of stems, of wild-type M. truncatula and nst1 mutants were cross-sectioned to 100 μm using a Leica RM 2255 Microtome (http://www.leica.com). For UV-fluorescence microscopy, the UV intensity was kept constant between samples. Sections were observed under the Nikon Microphot-FX system with a Nikon DXM 1200 color camera (http://www.nikon.com).
To determine pollen viability, flowers were picked at similar stages from wild-type and nst1 mutant M. truncatula. Entire flowers were incubated in Carnoy’s fixative for 1 h at room temperature, and then transferred to Alexander solution (Alexander, 1969) for 4 h at room temperature. The flowers were then de-stained in 10% glycerol for 30 min prior to observation.
Total RNA was isolated with Tri-reagent according to the manufacturer’s protocol (Invitrogen, http://www.invitrogen.com). RNA was prepared from stem or leaf tissue from 2-month-old plants. Veins were removed from leaf tissue prior to RNA isolation. RNA was cleaned and concentrated using the RNeasy MinElute Cleanup Kit (Qiagen, http://www.qiagen.com) and 10 μg of purified RNA was used for microarray analysis of three biological replicates. In both stem and leaf microarrays, RNA was isolated from two independent mutant alleles (nst1-1 and nst1-2). Probe labeling, hybridization and scanning for microarray analysis were conducted according to the manufacturer’s instructions (Affymetrix, http://www.affymetrix.com). Data normalization was conducted using robust multi-chip average (RMA) (Irizarry et al., 2003). The presence/absence call for each probe set was obtained from dCHIP (Li and Wong, 2001). Genes with significantly different expression levels between the wild-type control and mutants were selected using associative analysis, as described by Dozmorov and Centola (Dozmorov and Centola, 2003). The type-I family-wise error rate was reduced by using a Bonferroni corrected P-value threshold of 0.05/N, where N represents the number of genes present on the chip. The false discovery rate was monitored and controlled by Q value (false discovery rate), calculated using Extraction of Differential Gene Expression (edge, http://www.biostat.washington.edu/software/jstorey/edge) (Storey and Tibshirani, 2003; Leek et al., 2006).
cDNA samples were used for qRT-PCR, with technical duplicates. The 10-μl reaction included 2 μl of primers (0.5 μm of each primer), 5 μl Power Sybr (Applied Biosystems, http://www.appliedbiosystems.com), 2 μl of 1:20 diluted cDNA from the reverse transcription step and 1 μl of water. qRT-PCR data were analyzed using sds 2.2.1 (Applied Biosystems). The PCR efficiency was estimated using LinRegPCR (Ramakers et al., 2003), and the transcript levels were determined by relative quantification (Pfaffl, 2001) using the M. truncatula actin gene as a reference.
Determination of lignin content and composition
The lignin content of stem material (internodes 5–8) was determined by the acetyl bromide method using ∼15 mg of extractive-free material (Hatfield et al., 1999). The same molar extinction coefficient of 17.2 (as determined for lignin from wild-type alfalfa) was used for samples of all the transgenic lines.
Lignin composition was determined by thioacidolysis (Lapierre et al., 1985, 1995). Approximately 15–20 mg of extractive-free samples was reacted with 15 ml of 0.2 M BF3 etherate in an 8.75:1 dioxane/ethanethiol mixture. Lignin-derived monomers were identified by gas chromatography mass spectrometry (GC/MS), and quantified by GC as their trimethylsilyl derivatives. GC/MS was performed on a Hewlett-Packard 5890 series-II gas chromatograph with a 5971 series mass selective detector (column:HP-1; 60 m × 0.25 mm; 0.25-μm film thickness; http://www.hp.com), and mass spectra were recorded in electron impact mode (70 eV) with a 60–650 m/z scanning range.
Profiling of soluble and wall-bound phenolics
Glasshouse-grown stem samples were collected from two independent mutant alleles (nst1-1 and nst1-2). The top second to ninth stem internodes were harvested from three individual mutants and corresponding heterozygous control plants, and ground under liquid nitrogen. For soluble phenolics, freeze-dried samples (100 mg) were incubated with 2 ml of chloroform/methanol (2:1, v/v) at room temperature overnight with gentle shaking. The samples were incubated on the shaker for another 4 h at room temperature. After centrifugation at 12 000 g for 10 min, the supernatant (methanol/water fraction) was taken for HPLC profiling of soluble phenolics.
For analysis of wall-bound phenolics, the cell wall residues were further extracted with 100% methanol, 50% methanol and water (three times each), and then freeze dried. Freeze-dried cell wall residue (60 mg) was hydrolyzed with aqueous 2 N NaOH (1.8 ml) at 37°C for 5 h in the dark. The mixtures were acidified to pH 3 with 1:1 (v/v) HCL (600 μl). After extraction with ethyl acetate (1.6 ml × 3), the combined organic phases were dried under a stream of N2 and re-suspended in 300 μl 70% methanol for HPLC analysis.
Compounds were identified by comparing the UV spectra and retention times with those of authentic standards, and quantified by means of standard curves. HPLC was carried out on a Beckman System Gold HPLC system, consisting of a programmable solvent module 126, a System Gold 508 autosampler and a System Gold 168 diode array detector (http://beckman-coulter.com). A Phenomenex Luna 5μ C18 reverse phase column (5-μm particle, 250 × 4.6 mm) was used (http://www.phenomenex.com): solvent A, 0.1% phosphoric acid in water; solvent B, acetonitrile. The gradient for soluble phenolics was: 8% B for 1 min; 8% B to 35% B in 45 min; at 46 min, B from 35% to 100% in 1 min. The gradient for wall-bound phenolics was: 8% B for 1 min; 10% B to 27% B in 50 min; at 51 min, B from 27% to 100% in 1 min.
Analysis of cell wall sugar levels and composition
The determination of carbohydrates was conducted according to the Laboratory Analytical Procedure of the National Renewable Energy Laboratory (LAP-019). Biomass (∼300 mg) was first hydrolyzed in 72% sulfuric acid at 30°C, then in dilute acid (4%) at 130°C for 1 h. The solubilized sugars were analyzed spectrophotometrically using the phenol-sulfuric acid assay (Dubois et al., 1956). Monosaccharide compositions were determined by HPLC (Agilent 1200 Series LC System with 1200 Series Refractive Index Detector; Agilent, http://www.chem.agilent.com). An Aminex HPX-87P column was used at 70°C, and the sugars were eluted with Milli-Q filtered water at a flow rate of 0.6 ml min−1. The detector temperature was 50°C. Peaks were identified and quantified by comparison with authentic standards.
Chemical pre-treatment and enzymatic saccharification
Dried stem material at a solid loading of 2% (w/w) was mixed with dilute sulfuric acid (final concentration 1.5%, w/w) and pre-treated in an autoclave at 130°C for 2 h. After pre-treatment, the hydrolysates were separated and collected by filtration from residual biomass, and the biomass residues were washed with water. For enzymatic hydrolysis, Celluclast 1.5 L (cellulase from Trichoderma reesei) and Novozyme 188 (cellobiase from Aspergillus niger) (Sigma-Aldrich, http://www.sigmaaldrich.com) were mixed (equal volumes) and used at a loading of 31.5 filter paper units per g cell wall residue. Enzymatic saccharification of lignocellulosic material was performed according to the Laboratory Analytical Procedure of the National Renewable Energy Laboratory (LAP-009). About 0.125 mg of cell wall residues were hydrolyzed with a cellulase/cellobiase mix in a total volume of 10 ml by adding appropriate quantities of enzyme mixture and sodium citrate buffer (0.1 m, pH 4.8) for 72 h. Enzyme blanks and Whatman #1 filter paper were digested alongside the samples (http://www.whatman.com). Hydrolysis of filter paper was always more than 95%. The total sugar and sugar composition in hydrolysates from chemical pre-treatment and enzymatic hydrolysis were determined using the phenol-sufuric acid assay and HPLC, as described above.
We thank Lisa Jackson, Gail Shadle, and Liying Qi for assistance with mutant screening and lignin analyses, Drs Jiangqi Wen and Xiaofei Cheng for forward and reverse genetic screening, Dr Yuhong Tang for microarray analysis, and Drs Jin Nakashima, Elison Blancaflor and Li Quan for help with pollen staining and light microscopy, Dr Scott Russell (University of Oklahoma) for assistance with SEM analysis, and Drs Rujin Chen, Catalina Pislariu and Jianghua Chen for their critical reading of the manuscript. The M. truncatula plants utilized in this research project, which are jointly owned by the Centre National de la Recherche Scientifique, Gif-sur-Yvette, France, and the Samuel Roberts Noble Foundation, Ardmore, OK, USA, were created through research funded, in part, by grant 703285 from the National Science Foundation. This work was supported by grants to RAD from the USDA-DOE Feedstock Genomics program (award number DE-FG02-06ER64303) and the State Regents of Oklahoma (Oklahoma Bioenergy Center), the US Department of Energy Bioenergy Research Centers, through the Office of Biological and Environmental Research in the DOE Office of Science, and by the Samuel Roberts Noble Foundation.