The Folylpolyglutamate Synthetase Plastidial Isoform is Required for Postembryonic Root Development in Arabidopsis

A recessive Arabidopsis mutant with short primary roots and root hairs was identified from a forward genetic screen. The disrupted gene in the mutant encoded the plastidial isoform of folylpolyglutamate synthetase (FPGS) previously designated as AtDFB, an enzyme that catalyzes the addition of glutamate residues to the folate molecule to form folylpolyglutamates. The short primary root of atdfb was associated with a disorganized quiescent center (QC), dissipated auxin gradient in the root cap, bundled actin cytoskeleton, and reduced cell division and expansion. The accumulation of monoglutamylated forms of some folate classes in atdfb was consistent with impaired FPGS function. The observed cellular defects in roots of atdfb underscore the essential role of folylpolyglutamates in the highly compartmentalized one carbon transfer reactions (C1 metabolism) that lead to the biosynthesis of compounds required for metabolically active cells found in the growing root apex. Indeed, metabolic profiling uncovered a depletion of several amino acids and nucleotides in atdfb indicative of broad alterations in metabolism. Methionine and purines, which are synthesized de novo in plastids via C1 enzymatic reactions, were particularly depleted. The root growth and QC defects of atdfb were rescued by exogenous application of 5-formyl-tetrahydrofolate (5-CHO-THF), a stable folate that was readily converted to metabolically active folates. Collectively, our results indicate that AtDFB is the predominant FPGS isoform that generates polyglutamylated folate cofactors to support C1 metabolism required for meristem maintenance and cell expansion during postembryonic root development in Arabidopsis. no obvious primary root defects Fig. S3) our data indicate that AtDFB is the predominant FPGS isoform that generates bulk of the polyglutamylated folate cofactors for C1 metabolism required to sustain normal meristematic activity and cell expansion in actively growing root tips.

However, phenotypes were only reported in double FPGS mutants but not in single mutants leading to the conclusion of redundancy in compartmentalized FPGS activity (Mehrshahi et al., 2010). Also in Arabidopsis, double mutants to genes encoding two 10-formyl THF (10-CHO-THF) deformylases (PurU) that metabolize 10-CHO-THF to formate and THF, were smaller and paler compared to wild type plants. In addition, double PurU mutants exhibited delayed embryo development and shriveled non-viable seeds (Collakova et al., 2008).
In this study, a forward genetic screen led to the identification of a recessive Arabidopsis mutant that exhibited stunted primary root and root hair growth. The mutant contained a T-DNA insertion in the At5g05980 gene, which encodes the plastidial FPGS isoform designated as AtDFB for Arabidopsis thaliana DHS-FPGS isoform B (Ravanel et al., 2001) and recently renamed FPGS1 by Mehrshahi et al. (2010). The short primary root of atdfb was associated with a disorganized quiescent center (QC), reduced cell division and expansion, an extensively bundled actin cytoskeleton, and dissipation of the auxin gradient in the root cap. The AtDFB mutation led to changes in the glutamylation status of some folate classes that was coupled to a general depletion of amino acids and nucleotides, indicative of broad alterations in metabolism.
The disorganized QC in atdfb was consistent with previous transcript profiling studies that showed strong AtDFB expression in the QC (Nawy et al., 2005) and therefore point to a role for metabolically active folates in maintaining QC function. No obvious defects in root development were observed in single mutants to AtDFC and AtDFD. Our data indicate that AtDFB is the predominant FPGS isoform that generates physiologically active folate cofactors required to sustain postembryonic root growth in developing Arabidopsis seedlings.
www.plantphysiol.org on September 5, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved.

AtDFB Loss-of-Function Mutants are Impaired in Primary Root Growth
We isolated a mutant originally designated as drh2 from a T-DNA activation tagged population with root hairs that were wavy and short (Supplemental Fig. S1). It was also found that the mutant had significantly shorter primary roots than wild type ( Fig. 2A). The F2 generation segregated in a 3:1 (wild type to mutant) ratio after crossing the mutant to wild type consistent with a single recessive mutation. Closer examination of the primary roots of 7 day old seedlings showed a shorter growth zone in drh2 as evident from the emergence of root hairs very close to the root tip ( Fig. 2A). In 11 day old seedlings, roots of wild type were more than 8-fold longer than drh2 and formed about 2 adventitious roots ( Fig. 2B). At 3 to 8 days, primary root growth rate of drh2 was only 1.67±0.09 mm/day but increased to 4.40±0.31 mm/day by 15-17 days (Fig. 2C). In wild type seedlings, primary root growth rate was 8.35±0.46 mm/day in 3-8 day old seedlings and decreased to 7.02±0.37 mm/day in 15-17 day old seedlings (values are means of 20 roots ± S.E.).

TAIL-PCR of the drh2 mutant identified a T-DNA insert in the 5th intron of the
At5g05980 gene, which encodes AtDFB (Ravanel et al., 2001). SAIL_556_G08 and SALK_015472 were identified as additional lines with T-DNA insertions in the 6 th intron and 8 th exon of At5g05980 respectively and both lines exhibited root phenotypes similar to those of drh2 indicating that the defective primary root and root hair growth exhibited by the drh2 seedlings were due to mutations in the AtDFB gene. Drh2, SAIL_556_G08 and SALK_015472 were therefore designated as atdfb-1, atdfb-2 and atdfb-3 respectively (Fig. 3A, B). Reverse transcription (RT)-PCR analysis showed that atdfb-1, atdfb-2 and atdfb-3 had no detectable transcript using primers flanking the T-DNA insertion (Fig. 3C). Despite the short primary root of atdfb, no obvious defects in the development of the above ground organs were observed (Supplemental Fig. S2).
Real time quantitative (q) RT-PCR analysis revealed that AtDFB was expressed in both shoots and roots but expression in roots was higher compared to shoots (Fig. 4A). Whole mount in situ hybridization using gene specific probes confirmed stronger expression of AtDFB in the root apex compared to shoots ( Fig. 4B-D). Furthermore, previous transcript profiling of GFPmarked root cell types in Arabidopsis revealed that AtDFB was strongly expressed in the QC   Nawy et al., 2005). In agreement with these results, distinct AtDFB expression was observed in QC region of the root tip (Fig. 4C).
AtDFC and AtDFD encode two other FPGS isoforms that have been reported to localize to the mitochondria and cytoplasm respectively (Ravanel et al., 2001;Fig. 1). No obvious defects in root development were observed in single mutants to these other FPGS isoforms (Supplemental Fig. S3). The higher root expression of AtDFB compared to AtDFC and AtDFD as shown by publicly available Arabidopsis microarray expression data sets from Genevestigator (Supplemental Fig. S4; Zimmermann et al., 2004) supports our findings that AtDFB plays a more important role in root development than the other two FPGS isoforms.

Expansion
We next determined whether the short primary root of atdfb was the result of reduced cell division and/or expansion. Differential interference contrast (DIC) images of the epidermal cells from the root hair region were acquired and their lengths measured. From the DIC images, it was obvious that epidermal cells of wild type roots were two times longer than epidermal cells of atdfb-1 roots (Fig. 5A, B). The average epidermal cell length of wild type seedlings was 154.4±34.8 μm compared to 74.1±2.4 μm in atdfb-1 (Fig. 5C) (values are means ± SE of 40-50 cells from 10-12 seedlings).
Because the actin cytoskeleton is an important regulator of cell expansion (Blancaflor et al., 2006), we asked if the short primary roots of atdfb-1 was impaired in filamentous actin (Factin) organization. In wild type seedlings, cells in the root elongation zone consisted of randomly organized fine F-actin arrays, a typical feature of actin organization in wild type roots ( Fig. 5D; Wang et al., 2008). On the other hand, F-actin in the shorter epidermal cells of atdfb-1 roots was extensively bundled (Fig. 5E). Bundling and fluorescent aggregates were also observed in the interphase cells within the root meristem of atdfb-1 but not in wild type (Fig. 5F, G). In addition, cells that were undergoing cytokinesis in wild type roots could be easily identified by the presence of GFP-labeled F-actin in the phragmoplast (Fig. 5F). F-actin labeled phragmoplasts were rarely observed in the meristem of atdfb-1 roots (Fig. 5G).

0
The scarcity of F-actin labeled phragmoplasts (Fig. 5G) in atdfb-1 roots strongly suggested that cell division was impaired. Consistent with the F-actin labeling result, the average number of mitotic cells in 5 day old DAPI-stained roots of atdfb-1 seedlings was significantly less than in wild type (Fig. 5H). Taken together, these cellular assays indicated that the reduced primary root growth of atdfb-1 can be attributed to defects in both cell division and cell expansion.

Postembryonic Quiescent Center Organization and Auxin Gradients are Altered in Roots of atdfb
Because the QC plays a crucial role in root meristem maintenance by serving as a source for replenishing expiring cell initials (van den Berg et al., 1997), it is possible that altered QC function is one cause of the impaired cell division in atdfb-1 given the strong expression of However, no gross differences in the organization of cells in the QC region of mature embryos were observed between wild type and atdfb (Fig. 6C, D), which appeared to persist for up to 3-4 days post-germination (Fig. 6E, F). Clear differences in QC organization between wild type and atdfb-1 only became obvious in seedlings older than 4-5 days.
Another important regulator of root development is the plant hormone auxin. In Arabidopsis roots, local auxin concentration gradients have been shown to direct patterns of cell division, expansion and differentiation in the root apex (Overvoorde et al., 2010). We therefore investigated whether the root defects of atdfb were associated with altered local auxin gradients.
For these studies we used the synthetic auxin responsive reporter, DR5:GFPm, which indirectly reports patterns of local auxin accumulation in Arabidopsis roots (Ottenschläger et al., 2003).
Roots from 8 day old wild type seedlings showed DR5:GFPm expression in both the QC and central cells of the columella (Fig. 7A). In contrast, the expression of DR5:GFPm in roots of atdfb-1 was confined to only a few cells in the columella (Fig. 7B). The number of cells 1 1 expressing DR5:GFPm in median confocal sections of the root cap of 8 day old atdfb-1 seedlings was about 3-fold less than that of wild type (Fig. 7C ).
Another function of the QC is the generation of cell autonomous signals that suppress differentiation and therefore maintain the adjacent stem cells (van den Berg et al., 1997). In Arabidopsis roots where the QC was removed by laser ablation or mutants that were unable to specify a QC, the ability to maintain the surrounding columella initials was compromised (van den Berg et al., 1997;Sabatini et al., 2003). In the root cap, differentiation into columella cells is marked by the formation of starch filled amyloplasts that can be readily imaged by Lugol staining. In this and other studies, it was shown that columella initials of wild type roots proximal to the QC typically lacked amyloplasts ( Fig. 7D; Sabatini et al., 2003). However, similar to roots with a defective QC (Sabatini et al., 2003), cells immediately below the disorganized QC region of atdfb acquired the amyloplast differentiation markers that are typically only expressed in mature columella cells (Fig. 7E). Furthermore, real time qRT-PCR of the AGL42 gene, which is a marker for the QC (Nawy et al., 2005) was significantly reduced in root tips of atdfb-1 (Fig. 7F).

5-Formyl Tetrahydrofolate Rescues the Root Growth Defects of atdfb
We next tested whether the exogenous application of a stable form of THF such as 5-CHO-THF could rescue the atdfb root defects. 5-CHO-THF has been used to rescue the inability of cultured gla1 embryos, which are defective in the gene encoding AtDFA, to form calli (Ishikawa et al., 2003) and partially rescue the developmental defects of Arabidopsis FPGS double mutants (Mehrshahi et al., 2010). When all three mutant alleles of AtDFB were planted on media supplemented with 500 μM 5-CHO-THF, their primary roots were restored to wild type lengths (Fig. 8A, B). The restoration of root growth of atdfb-1 to wild type upon exposure to 5-CHO-THF was also accompanied by the reformation of wild type QC organization (Fig. 8C) and a corresponding increase in the number of mitotic cells (Supplemental Fig. S5A). In addition, the average root hair length of atdfb increased significantly compared to untreated seedlings and was almost restored to wild type lengths (Supplemental Fig. S5B, C).
To better understand why 5-CHO-THF, which is not a C1 donor, was able to chemically complement the root phenotype of atdfb, we analyzed total folates in roots of the rescued 1 2 seedlings. It was found that a massive (>2300 fold) increase in total folate occurred in 5-CHO-THF supplemented roots compared to non-supplemented roots in both wild type and atdfb-1 seedlings. Individual folate classes accumulated in 5-CHO-THF-treated seedling roots showed similar contributions to the folate pool than those observed in solvent controls with 5-CH 3 -THF as the most predominant folate that accumulated (Supplemental Fig. S5D).

The Folate Glutamylation Profile is Altered in atdfb Seedlings
To determine the impact of the AtDFB mutation on folate metabolism, folate analyses were performed on shoots and roots of 15 day old seedlings. Total folate content did not significantly change between wild type and atdfb-1 in both shoots and roots. However, we did find differences in the accumulation patterns of some folate classes and general changes in the contribution of each folate class to the total folate pool (Supplemental Table S1).
Because AtDFB catalyzes the extension of the glutamate tail of the folate molecule (Ravanel et al., 2001; Fig. 1), differences in the glutamylation state of folates between wild type and atdfb-1 seedlings might be predicted rather than differences in total folate content. Indeed, a considerable increase in total monoglutamylated folates in atdfb roots when compared to wild type roots was found (Fig. 9B). This difference was not observed in shoots ( Fig. 9A) and polyglutamylated total folate content was not altered in either tissue (Fig. 9C, D). However, when we looked at the individual folate classes, the glutamylation state of some folate classes was significantly different between wild type and atdfb-1 seedlings ( Fig. 9A-D). The most significant difference was an increase in monoglutamylated 5-CH 3 -THF from 9% of the total 5-CH 3 -THF pool in wild type roots to 53% in atdfb-1 roots (Student's t-test, α =0.05, p≤0.0006; Fig. 9B). The root glutamylation profile of 5,10-methenyl-THF (5,10-CH=THF), which is the sum of the conversion of 10-CHO-THF at acidic pH and the existing 5,10-CH=THF pool (Quinlivan et al., 2006), was also impacted by the AtDFB mutation. Roots of atdfb-1 had 77% of the 5,10-CH=THF pool in the monoglutamylated form as opposed to wild type, which had only 44% of the 5,10-CH=THF pool in the monoglutamylated form (Fig. 9B).
Although the level of monoglutamylated 5-CH 3 -THF in both shoots and roots of atdfb-1 was generally higher than wild type (Fig. 9A, B), the polyglutamylation level of this folate class was not significantly different between wild type and atdfb-1 (Fig. 9C, D). However, 1 3 polyglutamylation levels of 5,10-CH=THF were significantly lower in both shoots and roots of atdfb-1 compared to wild-type ( Fig. 9C, D). A 7-fold increase in levels of 5-CHO-THF was also detected in atdfb-1 roots compared to wild type with more than 90% of this increase in the polyglutamylated form (Fig. 9D).

Amino Acids and Nucleotides are Depleted in Whole Seedlings of atdfb
Non-targeted metabolic profiling of 7 day old whole Arabidopsis seedlings was performed using GC-MS to determine how the AtDFB mutation impacted overall metabolism. A total of 776 polar and 415 non-polar compounds were detected in atdfb-1 and wild type seedlings respectively, of which 193 polar compounds and 92 non-polar compounds could be assigned a chemical structure based on EI-MS (electron ionization mass spectrometry) spectral matching to authentic compounds (Supplemental Table S2 and S3).
Hierarchical clustering analysis (HCA) using a green-black-red diagram was used to illustrate the distribution of normalized signal intensities of cellular metabolites in wild type and atdfb-1. This analysis showed that nucleotides and amino acids were generally depleted in atdfb-1 (Fig. 10A). Amino acids that showed a statistically significant reduction in atdfb-1 compared to wild type included cysteine, arginine, glutamine, asparagine, methionine, phenylalanine, lysine and tyrosine (Fig. 10B). Folates serve as cofactors for methionine biosynthesis and serineglycine interconversion (Hanson and Roje, 2001; Fig. 1). However, we did not find any significant differences in serine and glycine levels between wild type and atdfb-1 seedlings ( Fig.   10C). For nucleotides, statistically significant reductions in the levels of guanosine, adenine, uridine, uridine monophosphate and 5-methyl pyrimidine were noted ( Fig. 10D, E).

Exogenous Methionine and Guanosine Partially Alleviate the Root Growth Defects of atdfb
Metabolic profiling showed that endogenous methionine levels in atdfb-1 seedlings were lower relative to wild type (Fig. 10B). In yeast, met7, a mutant disrupted in a gene encoding a cytosolic FPGS, requires a source of external methionine for growth (Cherest et al., 2000).
Expressing the AtDFB gene in met7 abolished its dependence on external methionine (Ravanel et al., 2001). We therefore tested whether the application of exogenous methionine could rescue the root defects of atdfb. It was found that methionine concentrations of 10-25 μM induced a more 1 4 than 2 fold increase in root growth of atdfb-1 compared to the non-treated atdfb-1 controls (Supplemental Fig. 6A). Although root growth of atdfb-1 seedlings was clearly promoted upon exposure to methionine, their roots remained significantly shorter than wild type roots grown with or without external methionine (Supplemental Fig. 6B).
We also tested whether supplying seedlings with external adenine or guanosine had any effect on roots of atdfb-1 given the reduction in the levels of these metabolites in the mutant compared to wild type (Fig. 10E). Exogenous guanosine at 10 nM induced a slight but statistically significant increase in root length of atdfb-1. However, 10 µM external guanosine did not impact root growth of atdfb-1 (Supplemental Fig. 6C). Unlike methionine and guanosine, adenine did not promote root growth of the mutant (Supplemental Fig. S6C). A combination of adenine, guanosine and methionine induced a statistically significant increase in atdfb-1 root length compared to controls but the induction of growth was similar to either methionine or guanosine alone (Supplemental Fig. S6C). was found that cells in the QC region of atdfb were disorganized and this was coupled to altered gradients of the auxin sensitive DR5:GFPm, reduced expression of the QC-expressed gene,

DISCUSSION
AGL42 (Nawy et al., 2005), and premature formation of amyloplast differentiation markers. In this regard, it is noteworthy that genetic disruption of DR5 expression and mutations that resulted in loss of QC identity led to short primary roots and cell patterning abnormalities reminiscent of atdfb defects (Sabatini et al., 1999;2003). Furthermore, the adventitious root phenotype of atdfb mirrored that of short root (shr), which is consistent with the loss of root apical meristem activity (Lucas et al., 2010).
The lack of any obvious embryonic and shoot defects in atdfb support recent reverse genetic studies suggesting that AtDFC or AtDFD are able to compensate for the absence of However, a novel finding from our studies is that these other FPGS isoforms do not always function redundantly at least within a specific window of postembryonic root development.
During this developmental window, the QC-expressed AtDFB isoform plays a major role in maintaining QC identity to support normal meristematic function during postembryonic root growth. We cannot rule out the possibility that QC identity in atdfb is already impacted before defects in cell organization within the QC region become apparent. Future studies will examine the spatial and temporal expression of several QC identity markers (Sabatini et al., 2003;Nawy et al., 2005) in atdfb mutants at different stages of embryonic and postembryonic root development to better understand the dynamics of QC function in relation to folate metabolism.
It also is not clear whether AtDFB acts exclusively within the QC since expression data suggest that it could be required for normal function of other cell types within the root meristem ( Fig. 4;   Supplemental Fig. S4).
Because folate monoglutamates that accumulated in atdfb cannot be efficiently utilized by folate requiring enzymes (Appling, 1991;Sahr et al., 2005), biochemical processes in the root apex that require C1 metabolites may not function optimally leading to the root defects of atdfb.
On the other hand, differences in polyglutamylated folates between wild type and atdfb, which 1 6 were absent in whole root folate profiling experiments, could be more pronounced in plastids where AtDFB was shown to localize (Ravanel et al., 2001). Indeed, Mehrshahi et al. (2010) found that chloroplasts in mature leaves of atdfb had lower polyglutamylated folates than wild type but even within this compartment, polyglutamylated folate content was not totally abolished in the mutants. This led to the proposal that different FPGS isoforms might be targeted to more than one compartment or that plants have a mechanism for transporting folate polyglutamates (Mehrshahi et al., 2010). Given that the root developmental defects of atdfb can be traced in part to a defined group of cells that comprise the QC, whole root folate quantification could dilute differences in the folate polyglutamylation profiles between wild type and atdfb. Thus, it will be necessary to develop sensitive methods to quantify folates from plastids of specific root cell types to better understand how metabolic changes within a defined group of cells translate into whole organ development.
Closer examination of individual folate classes revealed higher monoglutamylated and lower polyglutamylated 5,10-CH=THF in atdfb-1 roots compared to wild type (Fig. 9B, D). In this regard it is worth noting that the main form of folate found in chloroplasts is  Fig. 1). Indeed, it was found that endogenous levels of adenine and guanosine were reduced in atdfb seedlings (Fig. 10).
We did not find significant changes in THF + 5,10-CH 2 -THF polyglutamylation levels in atdfb roots compared to wild type (Fig. 9D). However, mutant shoots showed an increase in polyglutamylated forms of these folate classes (Fig. 9C)  in mutant whole seedlings (Fig. 10C), which could be due to other SHMT isoforms acting outside plastids.
We also found a 7-fold increase of 5-CHO-THF in atdfb roots with more than 90% of this increase in the polyglutamylated form (Fig. 9D). Polyglutamylated 5-CHO-THF was found to be a major folate in leaf mitochondria (Orsomando et al., 2005;Goyer et al., 2005). Thus, it is possible that reduction of polyglutamylation in plastids induced a redirection of folate flux, which then caused an accumulation of this folate class in roots. In fact, the contribution of each folate class to the total folate pool changed in the mutant in both shoots and roots (Supplemental Table S1), suggesting that the absence of AtDFB generates changes in the distribution and interconversion of folate pools within the cell. In this respect, previous work in Chinese hamster ovary cells showed that changes in mitochondrial and cytosolic FPGS activities caused clear folate redistribution between cell compartments (Lin et al., 1993;Lowe et al., 1993;Qi et al., 1999). 5-CHO-THF does not participate in C1 reactions and its role in planta is not very well understood. In fact this folate species has been shown to inhibit the activity of many folate utilizing enzymes (Stover and Schirch, 1993;Roje et al., 2002). Therefore, it was interesting to find that exogenous 5-CHO-THF rescued the root defects of atdfb (Fig. 8). When we analyzed these chemically complemented mutant seedlings, we found that they accumulated more than 1,900 nmol/g of folates, which is by far above all natural folate levels reported (Supplemental represented less than 18% of the accumulated folates with 5-CH 3 -THF being the most predominant folate class in the total folate pool (Supplemental Fig. S5D). Thus, the recovery of root growth in atdfb treated with 5-CHO-THF was most likely due to the excess of folates in the monoglutamylated form, which might be able to complement the mutation by accomplishing the same functions as a small amount of polyglutamlylated folates. Additionally, these results demonstrate the high capacity of the seedling to metabolize and accumulate large amounts of folates without apparent negative effects.
The depleted levels of amino acids and nucleotides are likely contributors to the defective root growth of atdfb. However, because metabolic profiling was done on whole seedlings, only  Table S2 and S3), and the cytoskeletal, cell division, and cell expansion defects in roots, as these processes are highly dependent on products of cellular methylation reactions.
This notion is supported by the observation that exogenous application of methionine caused a partial restoration of root growth in atdfb (Supplemental Fig. S6).
The reduced levels of adenine and guanosine that may result from inadequate 10-CHO-THF polyglutamates, could also contribute to defective atdfb root growth given their important role as building blocks of DNA and as major energy donors in numerous metabolic reactions of the cell (Zrenner et al., 2006). However, only guanosine caused a promotion in root length of atdfb. It is possible that certain metabolites that are applied exogenously including methionine and guanosine, which were only able to partially restore root growth in the mutant, cannot reach a population of cells such as the QC where they are needed most. Alternatively, the applied compounds could be rapidly metabolized into forms that the cell cannot utilize for restoring complete biochemical function. Moreover, a combination of depleted levels of other metabolites combined with the toxic accumulation of others (e.g. adenosine) likely defines the overall root defects of atdfb. A more targeted analysis of metabolites within a defined population of cells 1 9 particularly those that participate in C1 transfer reactions such as AdoMet and homocysteine will be needed to fully understand how changes in folate glutamylation, impact root development.
In conclusion, we provide genetic evidence that AtDFB plays a pivotal with T-DNA insertions at the AtDFB locus (SALK_015472, and SAIL_556_G08) were obtained from the ABRC (Alonso et al., 2003). These lines were renamed atdfb-2 and atdfb-3 respectively after confirming the T-DNA insertion site and observing that the root phenotypes of these additional T-DNA lines were similar to atdfb-1 (Fig. 3). Mutants to the AtDFC (SALK_008883) and AtDFD (SAIL_151_E09) gene were also obtained from ABRC and genotyping of these lines was essentially as described above (Supplemental Fig. S3).

Phenotypic Characterization of atdfb
For phenotypic analyses, seeds of wild type (Col-0) and mutants were surface-sterilized with 95% ethanol (v/v), 20% bleach (v/v) and washed extensively with sterile water before  and methionine, a stock solution of 10 mM was made using deionized water. Seeds were either planted directly or seedlings transplanted on the growth media and root growth assays were performed as described earlier.

Transcript Analysis and Whole Mount in situ Hybridization
Total RNA was isolated using an RNeasy mini kit (Qiagen, Hilden, Germany) and reverse transcribed using a first strand cDNA kit with Oligodt 20 and SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA). For semiquantitative PCR, first strand cDNA was amplified with Platinum Taq High Fidelity (Invitrogen). Cycling parameters were 94 o C for 10 s, 60 o C for 15 s, and 72 o C for 50 s.
For quantitative two-step RT-PCR of AtDFB and AGAMOUS-LIKE 42 (AGL42), 1 μ g of total RNA from wild type and atdfb-1 was reverse-transcribed to first-strand cDNA with the Qiagen cDNA synthesis kit (Qiagen). First-strand cDNA was used as a template for quantitative PCR using gene-specific primers. Arabidopsis eukaryotic protein synthesis initiation factor 4A2 (EIF4A2) served as a control for constitutive gene expression in plants. Primers used are shown in Supplemental Table S4. Relative expression levels (2 −ΔCt ) were calculated according to Ramakers et al., (2003). Expression levels of AtDFB were calculated based on the relative level 1 of EIF4A2 expression in each sample. Values are the means of three biological with three technical replicates for each.
Four day old seedlings were processed for in situ hybridization essentially as described by Hejátko et al., (2006) using AtDFB specific probes (Supplemental Table S4) Roots were examined with a Nikon Microphot FX compound microscope and images were acquired using a Nikon DXM 1200 camera running on ACT-1 software (Nikon Instruments, Melville, NY).

Cell biological Studies of Roots
The atdfb-1 mutants were transformed with a green fluorescent protein ( μM propidium iodide. In a parallel set of experiments, mature embryos from wild type and atdfb-1 were dissected from seeds imbibed for 24 hours in water and the QC was imaged in aniline blue-stained embryos as described in Bougourd et al. (2000).

2
number of mitotic cells from the apical 100 μm of the root apex was counted from projected images of 26 confocal optical sections taken at 1 μm intervals.
To visualize amyloplasts, 7 day old seedlings were fixed in 4% formaldehyde and stained with potassium iodide (Lugol's solution) as described by Willemsen et al. (1998).

Folate Analysis
Shoots and roots from 15 day old wild type and mutant seedlings were analyzed for folates from one gram total tissue by high performance liquid chromatography (HPLC) as described previously (Goyer et al., 2005;Orsomando et al., 2005)

Metabolite Analyses by Gas Chromatography-Mass Spectrometry
Metabolite analyses were carried out using a gas chromatography-mass spectrometry The dried polar extracts were methoximated in pyridine with methoxyamine-HCl, briefly sonicated, and incubated for 1h at 50 °C. Metabolites were then derivatized with MSTFA+1% TMCS for 1 h at 50 °C and analyzed by GC-MS. The GC system used was an Agilent 6890 coupled to a 5973 MSD quadrupole mass spectrometer scanning from m/z 50-650. Acquired mass spectra were deconvoluted using AMDIS software, and metabolite identifications were achieved by mass spectral matching to the Noble Foundation's in-house EIMS spectral library of authentic compounds, the publicly available GOLM library (http://csbdb.mpimpgolm.mpg.de/csbdb/dbma/msri.html), and the NIST08 library. Peak selection and alignment were performed using MET-IDEA software (Broeckling et al., 2006). The area of each peak was normalized against the area of the internal standard, and absolute quantification for selected metabolites was achieved using authentic standard calibration curves.

Accession Numbers
Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: AtDFB (At) NP_196217; At5g05980.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. AtDFB knockouts have no obvious defects in shoot development when compared to wild type.
Supplemental Figure S3. Identification of mutants to AtDFC and AtDFD.   shorter primary root of drh2 was due to a reduced growth zone as indicated by the emergence of root hairs closer to the root tip compared to wild type (white vertical bars; inset in A). B, In 11 day old seedlings, differences in root architecture between wild type and mutant become more apparent. In addition to the short primary root, drh2 seedlings had about 2 adventitious roots (arrowheads, inset in B) while wild type seedlings had none. C, At 16 days, drh2 primary roots started to elongate faster compared to the earlier stages of seedling development but were still significantly shorter than wild type. Bars = 200 μm (for insets in A and B).      Data are means from 18 to 25 roots ± SE. Means with different letters are statistically significant (Tukey's test, p < 0.05). C, QC organization in atdfb is also restored to wild type patterns when grown on 500 µM 5-CHO-THF. 3 4 means±SE of three independent preparations. Asterisks indicate statistically significant differences (Student's t-test, p<0.05*, p<0.001**). 5-CH3-THF (5-methyl-THF);10-CHO-THF(10-formyl-THF); 5,10-CH 2 -THF (5,10-methylene THF) ; 5,10-CH=THF (5,10-methenyl-THF); 5-CHO-THF (5-formyl-THF).