A WD40 repeat protein from Medicago truncatula is necessary for tissue-specific anthocyanin and proanthocyanidin biosynthesis, but not for trichome development

WD40 repeat proteins regulate biosynthesis of anthocyanins, proanthocyanidins (PAs) and mucilage in the seed, and the development of trichomes and root hairs. We have cloned and characterized a WD40 repeat protein gene from Medicago truncatula ( MtWD40-1 ) via a retrotransposon tagging approach. Deficiency of MtWD40-1 expression blocks accumulation of mucilage and a range of phenolic compounds, including PAs, epicatechin, other flavonoids and benzoic acids, in the seed, reduces epicatechin levels without corresponding effects on other flavonoids in flowers, reduces isoflavone levels in roots, but does not impair trichome or root hair development. MtWD40-1 is expressed constitutively, with highest expression in the seed coat where its transcript profile temporally parallels those of PA biosynthetic genes. Transcript profile analysis revealed that many genes of flavonoid biosynthesis were down-regulated in a tissue-specific manner in M. truncatula lines harboring retrotransposon insertions in the MtWD40-1 gene. MtWD40-1 complemented the anthocyanin, PA and trichome phenotypes of the Arabidopsis ttg1 mutant. We discuss the function of MtWD40-1 in natural product formation in Medicago , and the potential use of the gene for engineering PAs in the forage legume alfalfa ( Medicago sativa ).


Introduction
Anthocyanins and proanthocyanidins (PAs, also called condensed tannins) are flavonoids which benefit both plant and human health. Anthocyanins attract pollinators, protect plant tissues from UV damage, and defend plants against predators (Stapleton and Walbot, other obvious phenotypes, such as altered density of glandular or non-glandular trichomes ( Fig. 2D) or root hairs (Fig 2E), were observed in the NF0977 mutant. Root hair density appeared to be unaffected on both young (4 days after germination, Fig. 2E) and mature roots (Supplemental Fig. S1).
One of 12 plants from the NF0977 R2 generation exhibiting the lack of pigmentation phenotype was allowed to undergo self pollination. All 29 plants from the R3 generation were homozygous, as confirmed by PCR with gene-specific primers and a primer for the Tnt1 insert, and retained the visible mutant phenotypes as characterized in Fig truncatula genome database, this Tnt1 insertion was found to be between the first and second nucleotides of amino acid residue Ser31 of the WD40 protein in the NF0977 mutant (Fig. 2F). A further 20 insertion sites in different regions of the genome were also recovered from NF0977 (Supplemental Table S1), typical for Tnt1 insertional mutagenesis in Medicago (Tadege et al., 2008). None of these insertions was in a gene that would be expected to affect flavonoid biosynthesis, although this does not rule out the possibility that the lack of pigmentation phenotype could have been the result of an insertion in one or more of these genes. A reverse genetic approach was therefore employed to screen the Tnt1 insertion mutant population for additional lines with insertions in the WD40 gene, and another mutant line, NF2745, was obtained. The insertion site in line NF2745 was between amino acid residues S46 and I47 (Fig. 2F).
Homozygous NF2745 plants exhibited the same phenotype as NF0977 (Fig 2A-E), strongly suggesting that the loss of function of the WD40 gene is responsible for the pigmentation phenotypes in the two mutants.

Molecular Cloning and Characterization of MtWD40-1
BlastX analysis of the partial WD40 sequence against the GenBank database showed that this gene was located on the M. truncatula BAC clone CR940305. Its full-length sequence was predicted to be 1363 bp in length with a 49 bp 5'UTR and a 285 bp 3'UTR region (designated as MtWD40-1, GenBank Accession No. EU040206). MtWD40-1 is a single copy gene lacking introns, as confirmed by DNA gel blot analysis and amplification of the MtWD40-1 open reading frame with genomic DNA as template (data not shown). MtWD40-1 encodes a predicted protein open reading frame of 343 amino acids, with a calculated isoelectric point of 4.99 and a molecular weight of 38 kDa.
The deduced amino acid sequence of MtWD40-1 showed 77-79% identity to other known WD40 repeat proteins from different plant species, such as TTG1 from Arabidopsis and AN11 from petunia (Fig. 3). The four WD40 repeat domains are highly conserved among all the WD40 repeat proteins including MtWD40-1, and the last two amino acids in each WD40 repeat are identical. Phylogenetic analysis (Fig. 4) showed that MtWD40-1 is most closely related to TTG1 from Arabidopsis. Another Medicago WD40-like protein, MtWD40-2, is less than 60% identical to MtWD40-1 at the amino acid level, and somewhat closer to PAC1 from maize.

Interacting with GL3
Hairy roots of M. truncatula R108 exhibit red anthocyanin pigmentation (Pang et al., 2008), but this was lacking in the NF0977 line. Hairy root transformation was therefore used as a rapid method to confirm that MtWD40-1 could complement the lack-of pigment phenotype of the NF0977 Tnt1 insertion mutant. Red pigmentation was seen in all 101 phosphinothricin (ppt)-resistant hairy root lines transformed with MtWD40-1, but in none of the 30 ppt-resistant NF0977 lines transformed with the GUS gene (Fig. 5A). qRT-PCR confirmed that MtWD40-1, ANS and the anthocyanin-specific glucosyltransferase UGT78G1 (Modolo et al., 2007;Peel et al., 2008) were expressed at higher levels in hairy roots of the MtWD40-1 transformed lines than in the GUS transformants ( Fig. 5B-D), thus accounting for the high levels of extractable anthocyanins in the MtWD40-1 expressing lines (Fig. 5E). No significant differences were observed in the levels of insoluble PAs (PAs that bind to the cell wall and can not be extracted by organic solvents such as 70% acetone) between the MtWD40-1-expressing NF0977 lines compared with the GUS control lines (Fig. 5G), or in the levels of transcripts encoding the PA pathwayspecific genes ANR and UGT72L1 (data not shown). In contrast, soluble PA levels decreased slightly in the mutant line complemented with MtWD40-1 (Fig. 5F), possibly as a result of flux into soluble PAs being diverted back into the anthocyanin pathway.
To determine whether MtWD40-1 is a functional ortholog of TTG1, the MtWD40-1 open reading frame under the control of the 35S promoter was transformed into the Arabidopsis ttg1-9 mutant, and expression of the foreign MtWD40-1 gene was confirmed by qRT-PCR (Supplemental Fig. S2). 35S:MtWD40-1 fully complemented the anthocyanin pigmentation, trichome deficiency and seed coat PA phenotypes ( Fig. 6A-C). We also tested the ability of MtWD40-1 to complement the Arabidopsis ttg1-9 mutant when expressed under control of the Arabidopsis Glabrous 2 (GL2) promoter, which is active in the shoots of ttg1 mutants (Szymanski et al., 1998). Again, the phenotype was fully rescued (Fig. 6D-F).
To further determine how MtWD40-1 might function to restore the trichome phenotype in ttg1-9 Arabidopsis, the yeast two-hybrid system was used to test the interaction of MtWD40-1 with Glabrous 3 (GL3), a bHLH protein that regulates trichome development in Arabidopsis through interaction with GL1 and TTG1 (Payne et al, 2000). truncatula.

Tissue-and Developmental-Specific Expression of MtWD40-1
To determine the developmental expression pattern of MtWD40-1, normalized data were retrieved from the M. truncatula gene expression atlas (Benedito et al., 2008) together with seed coat microarray data (Pang et al., 2008). The expression pattern of two probe sets for MtWD40-1 (TC105711 and AL372205, probe set locations shown in Fig. 2F) were essentially the same, confirming that, as is also the case for TTG1 in Arabidopsis (Walker et al., 1999), MtWD40-1 is expressed in all organs, with highest expression in the seed coat (Fig. 7A). During seed development, MtWD40-1 showed its highest expression level at or before 10 days after pollination (dap, Fig. 7B), with a subsequent decline towards seed maturity. This expression pattern parallels the expression of MtANR and UGT72L1 during seed development (Pang et al., 2008).
We also analyzed the expression pattern of MtWD40-2 in the M. truncatula gene expression atlas (Benedito et al., 2008), where it is represented by probe set Mtr.
22605.1.S1_at (http://bioinfo.noble.org/gene-atlas/v2/). The highest expression level is in roots 24 h after salt stress and in developing root nodules, but the expression level in these tissues is nearly two orders of magnitude lower than the maximum expression level of MtWD40-1 (in developing seeds). MtWD40-2 is expressed around 15-fold lower than

Phenylpropanoid/Flavonoid Pathway Gene Transcripts and Metabolites
To determine the impacts of the loss of WD40-1 function on gene expression in seed, we dissected seeds at 16 dap from both the NF0977 mutant line and the corresponding wildtype control (ecotype R108) for microarray analysis using the Affymetrix Medicago Genechip. We have previously shown that phenylpropanoid/flavonoid biosynthetic pathway genes are highly expressed at 16 dap (Pang et al., 2007). The microarray data showed that 152 probe sets were down-regulated more than 2-fold in the MtWD40-1 mutant line; among these, 3 probe sets were down-regulated by more than 100-fold, 25 by more than 5-fold, with the remainder between 2-5 fold (Supplemental Table 2E).
This latter class consisted primarily of phenylpropanoid/flavonoid pathway genes.
Among the 28 probe sets that exhibited a more than 5-fold reduction in expression level in the MtWD40-1 mutant (Table 1) Table   S2). CHS is encoded by a large gene family in Medicago, and nine different CHS probe sets were down-regulated more than 5-fold (Supplemental Table S2). The two later anthocyanin pathway genes, DFR and ANS, were down-regulated by 2.6-fold and 9.2/10.0-fold respectively ( Another 271 probe sets were up-regulated in seeds of the mutant, most of them associated with primary metabolism or stress responses, but no phenylpropanoid/flavonoid pathway genes were up-regulated (data not shown).
The large number of changes observed in non-phenylpropanoid/flavonoid pathway genes in the above experiment could potentially occur as a result of the additional retrotransposon insertions in line NF0977. We therefore re-examined changes in key flavonoid pathway gene transcripts in seeds and other organs, in both NF0977 and the independent retrotransposon insertion line NF2745, using quantitative RT-PCR (Table 2).
MtWD40-1 transcript levels were more strongly down-regulated in tissues of NF2745 than in NF0977 (Table 2, Supplemental Tables S3, S4). Compared to wild-type R108, PAL and CHI transcript levels were least affected in the two MtWD401 retrotransposon insertion mutants. The most consistent changes observed as a result of loss of MtWD40-1 function were strong reductions of CHS expression in flower (but only determined for one probe set corresponding to TC138581) and seed, DFR1 expression in leaf and flower, ANS expression in stem, leaf and seed, LAR and ANR expression in flower and seed, and UGT72L1 expression in seed (Table 2, Supplemental Tables S3, S4). Thus, although MtWD40-1 is most strongly expressed in the seed (coat), its loss of function can affect flavonoid pathway gene expression in multiple tissues.
To further investigate the impacts of loss of WD40-1 expression on flavonoid biosynthesis, levels of phenylpropanoid-derived secondary metabolites were measured by UPLC-QTOFMS in various tissues of wild-type R108 and the two independent retrotransposon insertion lines (Table 3). The greatest effects were seen in developing seed, where levels of epicatechin and its glucoside (Fig. 8), as well as cyanidin 3-Oglucoside, kaempferol 3-O-rutinoside and two benzoic acid derivatives were reduced to undetectable levels in the insertion lines. In contrast, although epicatechin and its conjugate were likewise undetectable in flowers of the two mutant lines, levels of cyanidin 3-O-glucoside and other flavonoids were increased (Table 3), in spite of the apparently strong reduction in CHS expression in these lines. Loss of function of WD40-1 had little effect on the levels of three flavonoids in leaves, but resulted in reduced isoflavone (biochanin A) and aurone levels in roots (Table 3). Flavonol (kaempferol 3-Orutinoside) levels were reduced in developing seed of the mutant lines, consistent with the reduction in flavonol synthase expression ( Table 1). The less consistent results of MtWD40-1 down-regulation in non-seed tissue could either be because natural product levels are more variable as a result of environmental factors in non-seed tissues, or because of effects of different additional retrotransposon inserts in the two mutant lines.

Over-Expression of MtWD40-1 in Medicago Hairy Roots
Ectopic expression of the Arabidopsis MYB transcription factor TT2 in M. truncatula hairy roots results in a massive induction of PAs accompanied by the up-regulation of several hundred genes, especially those of the anthocyanin/PA biosynthetic pathway (Pang et al., 2008), and TT2, at least in Arabidopsis, functions in a complex with TTG1 and TT8. We therefore introduced MtWD40-1 into hairy roots of wild-type M. truncatula to determine whether over-expression of this gene could modulate PA biosynthesis in the absence of TT2 over-expression. The MtWD40-1 over-expressing root lines did not exhibit obvious phenotypical differences compared with GUS control lines; both exhibited purple pigmentation but neither stained blue with DMACA reagent (data not shown).
Three independent MtWD40-1 over-expressing hairy root lines were selected for high

Expression of MtWD40-1 in alfalfa
The MtWD40-1 gene driven by the 35S promoter was introduced into alfalfa by Additional anthocyanin accumulation would be predicted in flowers in which ANR is strongly down regulated but ANS remains unaffected, since cyanidin is the immediate precursor of epicatechin (Xie et al, 2003).
Although WD40 proteins are known to regulate anthocyanin and PA biosynthesis, their potential involvement in other areas of phenylpropanoid biosynthesis is less clear.
Our data indicate that loss of function of MtWD40-1 also results in reduction in the levels of an aurone and an isoflavone glycoside in roots, and complete loss of benzoic acids in seeds. Levels of the latter compounds are likely directly regulated through the action of MtWD40-1, whereas the smaller change in isoflavone levels in roots might be an indirect effect of altered metabolic flux. We did, however, record a 2-fold decrease in isoflavone synthase transcripts in two independent mtwd40-1 alleles by qRT-PCR (data not shown).
Together, these data suggest a critical role for MtWD40-1 in the control of seed PA biosynthesis, with additional but less precise (and possibly indirect) effects on the formation of other flavonoid compounds in other tissues.

The Role of MtWD40-1 in Trichome Formation
WD40 repeat proteins are critical for trichome formation in Arabidopsis, but not in all plant species (Serna and Martin, 2006). In Arabidopsis, a regulatory complex consisting of GL1-GL3/EGL3 (Enhance Glabrous 3)-WD40 triggers expression of the downstream GL2 gene by binding to its promoter region, to regulate trichome formation in the epidermal cell layer (Oppenheimer et al., 1991;Payne et al., 2000;Zhang et al., 2003).
Lack of TTG1 expression in Arabidopsis leads to the loss of trichomes on aerial tissues (Walker et al., 1999 (Xie et al., 2006). However, none of the components of the TT2-TT8-WD40 transcription complex has been tested for engineering PAs in foliage of forage legumes.
In a previous study, we introduced the TT2 gene from Arabidopsis into M. truncatula hairy roots, and this alone led to massive accumulation of PAs (Pang et al., 2008). ETC2 function as suppressors of the GL1-GL3/EGL3-WD40 complex to repress trichome formation (Schnittger et al., 1999;Schellmann et al., 2002;Kirik et al., 2004aKirik et al., , 2004b) and possibly the anthocyanin/PA-promoting function of the complex. It is clear that the successful bioengineering of PAs in forage crops will depend largely on our gaining a better understanding of the endogenous regulatory controls for PA biosynthesis.

Insertion Mutant Screening and Molecular Confirmation by TAIL-PCR
Generation of the M. truncatula Tnt1 insertional mutant population and growth of R1 seed were as described previously (Tadege et al., 2005). The mutant line NF0977 was selected due to its lack of anthocyanins in the aerial tissues. Genomic DNA from the mutant was isolated using the Dellaporta method (Dellaporta et al., 1983). Tnt1 flanking sequences were recovered using Thermal Asymmetric Interlaced Polymerase Chain Reaction (TAIL-PCR) (Liu et al., 1995(Liu et al., , 2005. PCR fragments were purified using a PCR Purification Kit (Qiagen, Valencia, CA) and then cloned into pGEM-T easy vector (Promega, Madison, WI), followed by sequencing with the Tnt1-specific primer Tnt1-F2 (Supplemental Table S7). The sequenced fragments were then analyzed by Blastn against the M. truncatula genome at NCBI.
Seeds from the identified Tnt1 insertion lines were scarified with concentrated sulfuric acid, cold-treated for 3 days at 4°C on filter paper, and grown in Metro-mix 350 (Scott Company, OH), with an 18-h light/25°C and 6-h dark/22°C photoperiod in the greenhouse. Genomic DNA from the R2 and R3 progeny was extracted and analyzed as above, using the Tnt1-R1 and MtWD40-1F1 primers (Supplemental Table S7 (Tadege et al., 2005(Tadege et al., , 2008. A PCR approach was taken for reverse genetic screening to uncover MtWD40-1 mutants.
Briefly, two rounds of PCR were used to screen the superpools; the primers used for the primary PCR were Tnt1 reverse primer Tnt1-R and gene specific primer MtWD40-1F.
For nested PCR, Tnt1-R1 and MtWD40-1F1 were used (Supplemental Table S7). The PCR products from the final target plants were then purified with a QIAquick PCR Purification Kit (Qiagen) and sequenced with the primer Tnt1-R2.

Sequence Alignment and Phylogeny Analysis
A multiple alignment of the deduced amino acid sequences of MtWD40-1 and other WD40 repeat domain proteins was constructed using Clustal X 1.81 (Thompson et al, 1997). For phylogeny analysis, the alignment was performed by using MAFFT (Katoh et al, 2005). The resulting alignment was further edited manually using Mesquite (Maddison and Maddison, 2009). The un-rooted consensus tree was constructed by using PAUP* 4.0b10 with 1000 bootstrap replicates (Swofford, 2003).

Sample Collection, RNA Extraction, qRT-PCR and Microarray Analysis
Root, stem, leaf, flower and seed samples from three independent homozygous NF0977 and NF2745 R3 generation and wild-type R108 plants were collected one month after planting in soil. Additional flowers were labeled individually according to pollination date, and seed pods harvested at 16 dap; the seeds were collected and stored at -80°C.
RNA was extracted from triplicate biological replicates of the above samples using the  name "Medicago truncatula TTG1 over-expressing hairy root"; E-MEXP-1759, experiment name "MtTTG1 over-expression transgenic alfalfa gene profiling".

Staining Seeds for PA and Mucilage
To determine the presence of PAs in the seed coat, seeds were soaked in DMACA reagent (0.1% w/v DMACA in methanol-3N HCl) for 1 h and then destained with ethanol/acetate acid (75:25). To stain for mucilage, seeds were imbibed in sterilized deionized water for 1 h, transferred to 0.01% ruthenium red solution for 10 min, and then washed twice with water.

Scanning Electron Microscopy
Young developing leaves with attached petioles were mounted on copper stubs, frozen in liquid nitrogen, sputter coated with gold using an Emitech K1150 cryo preparation system (Emitech, Houston, TX), and imaged with a Hitachi S3500N scanning electron microscope as described by Ahlstrand (1996).

Analysis of Anthocyanins, PAs and Total Flavonoids
For extraction of anthocyanins, 2-3 mL 0.1% HCl/methanol was added to 0.1 g ground fresh samples, followed by sonication for 30 min and standing overnight at 4°C.
Following centrifugation at 2,500g for 10 min, the extraction was repeated once and the supernatants pooled. An equal volume of water and chloroform was added to remove chlorophyll, and the absorption of the aqueous phase was recorded at 530 nm. Total anthocyanin content was calculated based on the molar absorbance of cyanidin-3-Oglucoside.
For PA analysis, 0.5-0.75 g ground samples were extracted with 5 mL of 70% acetone/0.5% acetic acid (extraction solution) by vortexing, and then sonicated at room temperature for 1 h. Following centrifugation at 2,500g for 10 min, the residues were reextracted twice as above. The pooled supernatants were then extracted three times with chloroform and once with hexane, and the supernatants (containing soluble PAs) and residues (containing insoluble PAs) from each sample were freeze dried separately. The dried soluble PAs were suspended in extraction solution to a concentration of 3 mg/mL.
Total soluble PA content was determined spectrophotometrically after reaction with DMACA reagent (0.2% w/v DMACA in methanol-3N HCl) at 640 nm, with (+)-catechin as standard. For quantification of insoluble PAs, 2 ml of butanol-HCl (95:5, v/v) was added to the dried residues and the mixtures were sonicated at room temperature for 1 h, followed by centrifugation at 2,500g for 10 min. The absorption of the supernatants was measured at 550 nm; the samples were then boiled for 1 h, cooled to room temperature, and the absorbance at 550 nm was measured again, with the first value being subtracted from the second. Absorbance values were converted into PA equivalents using a standard curve generated with procyanidin B1 (Indofine, NJ).
For determination of total flavonoids, 0.1 g batches of ground samples were extracted with 2 ml 80% methanol, sonicated for 1 h, and then kept at 4°C overnight. The extract was centrifuged to remove tissue debris and the supernatant was dried under nitrogen gas, followed by hydrolysis in 2 ml of 5 mg/mL β -glucosidase (34 units  Bridge was used to convert the raw data files to NetCDF. Relative abundances were calculated using MET-IDEA (Broeckling et al, 2006) and the peak areas were normalized by dividing each peak area by the value of the internal standard peak area.

Construction of Binary Vectors for MtWD40-1 Expression in Plants
The ORF of the MtWD40-1 gene was amplified from cDNA produced from total RNA isolated from M. truncatula seed coats, using the primers MtWD40-1CF and MtWD40-1R1 and DNA polymerase with proofreading activity. The PCR product was purified and cloned into the Gateway Entry vector pENTR/D-TOPO (Invitrogen, Carlsbad, CA), and the MtWD40-1 ORF in the resulting vector pENTR-MtWD40-1 was confirmed by sequencing.
The primers MtWD40-1NF (with an NcoI site) and MtWD40-1BR (with a BstE II site) (Supplemental Table S7) were used to amplify the ORF region (with added NcoI and BstEII restriction sites) from pENTR-MtWD40-1 template with proofreading DNA polymerase. The resulting fragment was digested, purified and ligated into plasmid pCAMBIA3301-HP (Xiao et al., 2005)

Rescue of the Arabidopsis ttg1-9 Mutant
The construct used to generate Arabidopsis expressing GL2::MtWD40-1 was derived from pGL2:: GUS (Szymanski et al, 1998). This plasmid was modified by removal of the GUS coding sequence by SmaI/SacI digestion followed by blunt ending with Klenow.
The RFA Gateway recombination fragment RFA from Invitrogen was inserted into this site. The coding region of MtWD40-1 was derived from cDNA using total RNA isolated from M. truncatula (Jemalong A17) shoots as a template. Primers flanking the MtWD40-1 coding region were used to generate a double stranded DNA product via PCR that was first subcloned into pCR8 (Invitrogen) before being moved into the Gateway GL2 promoter vector.
The Arabidopsis ttg1-9 mutant (Walker et al, 1999) was transformed by the floral dip infiltration method (Clough and Bent, 1998). Selection of transformants was conducted on 0.5 × Murashige and Skoog (MS) medium supplied with 7.5 mg/L ppt. The pptresistant seedlings were then transferred into soil to set seed. Progeny from self-fertilized primary transformants were grown in soil for observation of trichome phenotype.

Yeast Two-Hybrid Assay
For the yeast two-hybrid assays, PCR was used generate a copy of the MtWD40-1 coding region with leading and tailing EcoRI and BamHI restriction enzyme sites. The coding region was then moved into the corresponding sites of pBridge (Clontech) to create pMtWD40-1DB. The empty vector pGAD424 was from Clontech and pGL3-AD was as described previously (Esch et al, 2003). Beta-galactosidase activity was detected as adapted from Duttweiler (1996) and further described at (http://www.fccc.edu/research/labs/golemis/betagal/plates_vs_overlay.html).

Supplemental Data
The following materials are available in the online version of this article. Figure S1. MtWD40-1 transcript levels in the Arabidopsis ttg1-9 mutant and two lines complemented with MtWD40-1. Figure S2. Gene functional categories of probe sets that were down-regulated by more than 2-fold in the NF0977 mutant compared with wild type R108.     Table S1. BlastN analysis of all Tnt1 flanking sequences retrieved from the NF0977 mutant. Table S2. Probe sets that were down-regulated more than 2-fold in developing seed of the M. truncatula NF0977 mutant.

ACKNOWLEDGEMENTS
We thank Dr. Ji He for blast analysis, Ms Darla Boydston for assistance with artwork, and Dr Elison Blancaflor and Alan Sparks for help with root hair analysis..

39
The data represent the peak area corresponding to each compound divided by that of the internal standard and multiplied by 1000. Results are presented as mean and standard deviation from biological triplicates. a R108 columns to the right of the mutant lines represent independent sets of plants grown in parallel with the corresponding mutants. b ND: Not detected