Tie-dyed1 Encodes a Novel, Phloem-Expressed Transmembrane Protein that Functions in Carbohydrate Partitioning 1

Carbon is partitioned between export from the leaf and retention within the leaf, and this process is essential for all aspects of plant growth and development. In most plants, sucrose is loaded into the phloem of carbon-exporting leaves (sources), transported through the veins and unloaded into carbon-importing tissues (sinks). We have taken a genetic approach to identify genes regulating carbon partitioning in maize ( Zea mays ). We identified a collection of mutants, called the tie-dyed ( tdy ) loci, that hyperaccumulate carbohydrates in regions of their leaves. To understand the molecular function of Tdy1 , we cloned the gene. Tdy1 encodes a novel transmembrane protein present only in grasses, although two protein domains are conserved across angiosperms. We found that Tdy1 is expressed exclusively in phloem cells of both source and sink tissues, suggesting that Tdy1 may play a role in phloem loading and unloading processes. In addition, Tdy1 RNA accumulates in protophloem cells upon differentiation, suggesting Tdy1 may function as soon as phloem cells become competent to transport assimilates. Monitoring the movement of a fluorescent, soluble dye showed that tdy1 leaves have retarded phloem loading. However, once the dye entered into the phloem, solute transport appeared equal in wild-type and tdy1 mutant plants, suggesting that tdy1 plants are not defective in phloem unloading. Therefore, even though Tdy1 RNA accumulates in source and sink tissues, we propose that TDY1 functions in carbon partitioning by promoting phloem loading. Possible roles for TDY1 are discussed. EAFIXK where X is any residue. Using PSI-BLAST analysis, we found that this motif is present within a plant-specific domain of unknown function (DUF761; pfam05553). None of the proteins containing this domain have a characterized function. Two tdy1 mutant alleles encode proteins predicted to lack this domain due to a small deletion ( tdy1-shadow ) or an upstream frameshift mutation ( tdy1-PM ). These alleles confer a tdy1 mutant phenotype indistinguishable from the tdy1-D6 complete deletion allele, implying that this part of the protein is critical for its function. Carbohydrates were quantified according to Dinges from 100 mg of root tissue harvested from the middle of the root system of 10-d-old seedlings (n = 3). Tissue was collected every 4 hr over a diurnal cycle and the experiment replicated three times. Data presented is for all measurements. Root mass was determined by weighing entire root systems isolated at the seed attachment point of 10-d-old seedlings (n = 10). For both high- and low-light-grown plants, the experiment was replicated twice with similar results, and one repeat is shown.


INTRODUCTION
Assimilates, RNA and proteins are transported in the phloem tissue of veins (Lalonde et al., 2003a;van Bel, 2003;Lough and Lucas, 2006). Three distinct functional zones of the phloem, each with a different role in assimilate transport, have been described (van Bel, 2003).
The collection phloem is actively involved in sucrose (Suc) loading into veins in leaves. The transport phloem functions post-loading to distribute Suc through the vein system. The release phloem unloads Suc from veins into expanding cells in sink tissues. Suc movement within the phloem follows mass flow of solutes along a hydrostatic pressure gradient: high concentrations of nutrients and water enter the collection phloem and are conveyed through the transport phloem to the release phloem, where solutes and water are withdrawn, lowering hydrostatic pressure and maintaining the driving force (Lalonde et al., 2003a;van Bel, 2003).
The anatomy of a maize (Zea mays) leaf is intimately related to phloem loading and transport of assimilates (Evert et al., 1978). Leaf blades have three different orders of longitudinal veins with distinct functions in carbohydrate distribution. Phloem loading occurs in the small and intermediate veins (Fritz et al., 1983), which intergrade into large veins for transport of Suc out of the leaves and into the stem (Russell and Evert, 1985;Fritz et al., 1989).
The veins are surrounded by bundle sheath (BS) cells which are encircled by mesophyll (M) cells (Esau, 1977). The path of Suc movement from the photosynthetic cells into the vein is as follows. Suc moves symplastically from its site of synthesis in the M cells (Lunn and Furbank, 1999), diffuses through plasmodesmata into the BS cells and then into the vascular parenchyma (VP) cells (Russin et al., 1996). From the VP cells, Suc is exported to the apoplast by an unknown mechanism (Evert et al., 1978), prior to being imported into the phloem companion cells (CC) and/or sieve elements (SE). The pathway for carbon translocation in maize leaves was described 40 years ago (Hofstra and Nelson, 1969); however, we still know very little concerning the regulation of carbon partitioning.
Several classes of genes that contribute to phloem loading of Suc have been identified.
The genes directly responsible for Suc entry into the phloem are Suc transporters (SUTs) (Lalonde et al., 2004;Sauer, 2007). SUTs utilize the proton motive force generated by H + -ATPases to co-transport Suc and a H + across a membrane (Bush, 1993;Gaxiola et al., 2007).
Evidence demonstrating SUT function in phloem loading in vivo was provided by loss-of-6 function mutations which condition excess carbohydrate accumulation in leaves and reduced growth, delayed flowering and decreased yield due to the reduced delivery of photoassimilates to growing sink tissues (Riesmeier et al., 1994;Kühn et al., 1996;Bürkle et al., 1998;Gottwald et al., 2000;Srivastava et al., 2008). Mutations in other phloem-expressed genes, such as a H + -ATPase (Zhao et al., 2000), K + channel (Deeken et al., 2002) and aquaporin (Ma et al., 2004), which indirectly function in phloem loading of Suc, also lead to carbon accumulation in leaves and reduced growth. In all of these cases, the mutant phenotype affects the entire leaf and does not result in a variegated phenotype. In contrast to phloem loading, phloem unloading in vegetative sink tissues often occurs symplastically and does not require Suc passage across a membrane (Giaquinta et al., 1983;Haupt et al., 2001;Lalonde et al., 2003a;Stadler et al., 2005).
We have taken a genetic approach to identify genes regulating carbon partitioning in maize. We isolated mutants that develop chlorotic, carbon-hyperaccumulating regions in their leaves, and designated this new class of mutants tie-dyed (tdy) due to their striking variegated appearance (Fig. 1A, B) (Braun et al., 2006;Baker and Braun, 2008). Our rationale was that by characterizing variegated mutants, where some leaf regions contain excess carbohydrates while other regions appear similar to normal, the genes identified may function as regulators of carbon distribution rather than as transporters controlling sugar movement. The tdy mutants appear to be unique as most other characterized variegated leaf mutants do not hyperaccumulate carbohydrates (Yu et al., 2007). The only mutant with a similar phenotype is sucrose export defective1 in maize, which we have shown functions independently of Tdy1 and Tdy2 (Russin et al., 1996;Provencher et al., 2001;Baker and Braun, 2008;Ma et al., 2008). From our previous analyses, the most salient features of the tdy phenotype related to carbon allocation are that (1) soluble sugars and starch hyperaccumulate in leaf regions (Fig. 1C, D), (2) carbon accumulation is the earliest known defect and proposed to be responsible for the chlorotic phenotype, (3) Tdy1 acts within the middle tissue layer of a leaf which contains the veins, BS and interveinal M cells (Baker and Braun, 2007), (4) no blockages or plasmodesmatal structural perturbations are evident along the Suc symplastic transport path in tdy1 leaves, suggesting carbon accumulation does not result from a physical impediment (Ma et al., 2008), and (5) the tdy1 phenotype is independent of starch metabolism, indicating Tdy1 does not function in the photosynthetic cells in starch catabolism (Slewinski et al., 2008). Based on these and other data, we previously hypothesized that Tdy1 may function to promote Suc export from leaves by regulating SUT 8

Cloning of Tdy1
To understand the molecular function of Tdy1, we cloned the gene using a Mutator (Mu) transposon-tagging approach (Chandler and Hardeman, 1992;Bennetzen et al., 1993). Mu-active lines were crossed by tdy1-Reference (tdy1-R) plants and four new alleles were recovered with phenotypes indistinguishable from the tdy1-R allele. The new alleles were outcrossed to segregate away unlinked transposons. Southern blot analysis was performed to detect cosegregation of a linked Mu transposon with a tdy1 mutant allele. A Mu1 element tightly linked to the tdy1-D190 allele was identified (no recombinants/110 chromosomes, ~0.9 cM; Fig. 1E). The DNA flanking the Mu1 element was cloned by constructing a subgenomic DNA lambda library.
Sequencing the DNA flanking the transposon identified an open reading frame with homology to expressed cDNAs in rice (Oryza sativa) and sorghum (Sorghum bicolor). We verified the sequence cloned was the correct gene by analyzing independently derived alleles of tdy1. Two other alleles recovered from the transposon mutagenesis contained a Mu insertion at a different site in the same gene (tdy1-D24, -D46, Fig. 1F, Supp. Table S1). The tdy1-D6 allele was caused by a Mu element inserting into a nearby gene and resulting in a ~180 kb deletion encompassing the entire Tdy1 gene as well as several adjacent loci (Supp . Table S2) (Golubovskaya et al., 2006). Sequencing the tdy1-R allele identified a C to G substitution in the 112 th amino acid that changed a Pro to Arg. Additionally, we found a one base pair deletion in the tdy1-PM allele that caused a frameshift and premature termination. In total, we identified lesions in eight different tdy1 alleles, demonstrating that the gene we cloned corresponds to Tdy1.
BLAST analyses of nucleotide and deduced amino acid sequences determined that Tdy1 encodes a novel protein of 272 amino acids. Orthologous proteins having the highest sequence similarity were identified in sorghum and rice ( Fig. 2A). These genes showed homology throughout the protein coding region and mapped to syntenic genomic positions (Gale and Devos, 1998). The N-and C-terminal regions of the protein were most highly conserved with the middle portion of the protein showing the least conservation. TDY1 is predicted to be a membrane-localized protein (Fig. 3A, see below). The N-terminus contains two stretches of hydrophobic amino acids, the first predicted to be a signal peptide and the second a 9 transmembrane domain. The middle and C-terminal regions of TDY1 are predicted to be localized in the cytoplasm. The middle portion of TDY1 contains 10 interspersed Pro residues that are conserved in the grasses and may be important for protein folding. Interestingly, the second conserved Pro is mutated to an Arg residue in the tdy1-R allele, suggesting the mutant protein may not fold correctly. BLAST analyses also detected two domains conserved in TDY1 and proteins of unknown function from other plants (Fig. 2B, C). Domain I is 10 amino acids long, located within the predicted transmembrane domain, and is highly conserved in both monocots and dicots. Domain II is 27 amino acids in length, located near the C-terminus of the protein, and is also well conserved across angiosperms. We did not identify TDY1-like sequences in animal or prokaryotic genomes.
Although proteins similar in structure and showing limited sequence homology to TDY1 are found in dicots, several differences suggest that they may not have a similar function. The most closely related protein in Arabidopsis (Arabidopsis thaliana) shows 23% amino acid identity (42% similarity) to maize TDY1 over the entire protein (Fig. 2D). The Arabidopsis genome contains several genes predicted to encode proteins related to the closest TDY1 homolog, and all contain domains I and II (Supp. Fig. S1). The functions of these genes are unknown. Although the N-terminus is not as conserved in sequence, the At5g66440 protein also contains two stretches of hydrophobic amino acids and the protein is predicted to show a similar topology to TDY1. However, the contrasting calculated pI of the two proteins suggest that their functions may be different. The pI of TDY1 is basic at 11.4 whereas the Arabidopsis protein pI is acidic at 4.8. In addition, only four of the 10 Pro residues conserved in the middle portion of the TDY1-orthologous proteins from grasses are conserved between maize and Arabidopsis. Further, the Pro residue mutationally defined to be important for function in the tdy1-R allele is not conserved in Arabidopsis. Lastly, we have not observed any visible phenotype in Arabidopsis plants homozygous for a T-DNA or transposon insertion in the protein coding region of At5g66440 (TLS and DMB, unpublished results). Hence, it is not clear if TDY1 function is conserved in dicots, although disrupting the functions of multiple related genes may be required to uncover a phenotype.

10
The protein sequence of TDY1 is novel. To gain insight into TDY1 function, we investigated the subcellular localization of the TDY1 protein. Bioinformatic programs (SignalP, Phobius, TM-HMM) were in agreement that the N-terminus of TDY1 contained a predicted signal peptide and a membrane-spanning domain. However, additional analyses utilizing multiple programs (ChloroP, MitoP, TargetP, PSORT) to predict the subcellular localization of TDY1 yielded contradictory results. To determine the subcellular localization of TDY1, we fused a red fluorescent protein (RFP) to the C-terminus of TDY1, which is predicted to be on the cytoplasmic face of the membrane (Fig. 3A). We used particle bombardment to transiently express the TDY1 fusion protein in onion (Allium cepa) epidermal cells. We also heterologously coexpressed TDY1-RFP with various fluorescent protein fusions targeted to known subcellular locations (nucleus, ER, Golgi, vacuole, plasma membrane and cytoplasm). Coexpression of TDY1-RFP with an endomembrane-targeted green fluorescent protein (GFP)-HDEL fusion determined that TDY1 localizes to the ER in onion cells (Fig. 3C, D). Fluorescence was observed in a net-like pattern of strands in the cytoplasm and on the perinuclear envelope, which is contiguous with the ER. We did not observe colocalization of TDY1 with other subcellular markers (data not shown).
Proteins that are localized to the secretory system are synthesized on and trafficked through the ER to reach their cellular destinations. High levels of protein expression have been found to cause ectopic localization of membrane proteins in the ER due to saturation of the target membrane (Gobert et al., 2006). To investigate if this possibility might explain the observed localization of TDY1 to the ER, we examined the localization of a TDY1 fusion to the yellow fluorescent protein (YFP) over time. We found that TDY1 localized exclusively to the ER in onion cells from the earliest time point after transformation, when a low level of protein expression was first detectable, and throughout our observations (Supp. Fig. S2). TDY1 was never visualized in the plasma membrane or other membrane locations. These data suggest that when heterologously expressed in onion cells, TDY1 resides on the ER membrane, and that the protein localization does not result from the fusion protein saturating a different subcellular location and ectopically "backing up" in the endomembrane system. However, in vivo subcellular localization of the endogenous protein in maize is required to confirm these results.
Bioinformatic programs indicated that the subcellular targeting determinants in TDY1 reside in the N-terminus of the protein. To test this prediction, we created a protein fusion of the 11 first 71 amino acids of TDY1, which contain the predicted signal peptide and transmembrane domain, translationally fused to YFP. Expressing this construct in onion epidermal cells demonstrated that the protein localized to the ER (data not shown). In addition, this construct was coexpressed with the full length TDY1-RFP fusion protein. The two fusion proteins showed perfect overlap in their localization (Fig. 3E, F), indicating that the N-terminus of TDY1 was sufficient to confer targeting to the ER in onion cells. To test the C-terminal region of TDY1 for subcellular targeting information, we translationally fused the remaining 201 amino acids of TDY1 to YFP and coexpressed the fusion with soluble RFP. Both proteins were similarly distributed in the cytoplasm and able to diffuse into the nucleus (Fig. 3G, H), indicating they did not contain any targeting signals. From these data we infer that the N-terminus of TDY1 targets the protein to the ER membrane in onion cells and the C-terminal domain is present in the cytoplasm.

Tdy1 RT-PCR and RNA in situ Hybridization Analyses
The primary phenotype of tdy1 mutants is observed in leaf blade tissue (Braun et al., 2006), and thus we predicted that Tdy1 would be expressed only in leaves. To determine which tissues accumulate Tdy1 RNA, we used semi-quantitative reverse transcription-PCR (RT-PCR) to measure relative transcript abundance. We isolated RNA from mature and immature leaves, developing ears and tassels, and from seedling roots. Tdy1 RNA accumulated in all tissues, although at much higher levels in young roots and at the lowest level in mature and immature leaves (Fig. 3B). This was a surprising result, and it suggests that Tdy1 may play a larger role than that defined by mutational analysis. and prior to protoxylem differentiation (Fig. 4N). The great majority of protophloem cells in maize are SE that lack associated CC (Sharman, 1942;Esau, 1943;Evert et al., 1996). To determine the identity of the Tdy1-expressing cells, we stained comparable sections with 4', 6diamidino-2-phenylindole (DAPI) which binds to DNA and therefore stains CC (Lalonde et al.,  We previously proposed that one potential function of TDY1 may be in a Suc-sensing signal transduction pathway (Ma et al., 2008). To test whether the Tdy1 gene was transcriptionally induced in response to excess carbohydrates, we utilized the tdy1-R allele, which has a single base pair change in the coding region (Fig. 1F, Supp. Table S1). This allele would be expected to normally produce RNA but not functional protein. As previously mentioned, tdy1 mutants produce variegated leaves with normal appearing green and chlorotic regions (Fig. 1B). The tdy1-R chlorotic leaf regions hyperaccumulate Suc, glucose, fructose and starch and therefore experience excess sugar stress, while the tdy1-R green regions contain carbohydrate levels similar to wild type (Braun et al., 2006). RNA was isolated from wild-type, tdy1-R green and tdy1-R chlorotic leaf regions. Quantitative RT-PCR (qRT-PCR) experiments showed that Tdy1 RNA was expressed at similar levels in all three tissues relative to Ubiquitin (Ubq), indicating that Tdy1 was not transcriptionally induced by sugar stress (Fig. 5A). As a 13 positive control, a gene which shows highly induced expression under excess sugar conditions (glucose 6-phosphate/phosphate translocator (G6PT); Lloyd and Zakhleniuk, 2004;Gonzali et al., 2006), was strongly up-regulated in tdy1 yellow tissue, but not in tdy1 green or wild type (Supp. Fig. S4). To examine if Tdy1 RNA might instead be induced by carbohydrate depletion, we transferred greenhouse-grown wild-type plants into a dark room for three days. This amount of time in darkness is sufficient to eliminate carbon reserves in leaves (Slewinski et al., 2008).
For this experiment, Glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used as the reference gene since we determined that Ubq was not a suitable control under these conditions, potentially due to increased proteolysis upon carbon starvation (see Methods). Comparing Tdy1 expression in wild-type leaves collected at the end of the day versus plants maintained in the dark revealed no change in Tdy1 RNA (Fig. 5B), suggesting that Tdy1 RNA is not induced by carbon starvation. Collectively, these data suggest that Tdy1 expression is not regulated by the carbohydrate status of the tissue. Therefore, if Tdy1 functions in a sugar-sensing pathway, it likely does so post-transcriptionally.

tdy1 Mutants Have Reduced Root Mass and Sugar Content When Grown in High Light
Tdy1 RNA accumulated to the highest level in roots, suggesting that Tdy1 may play an important role in this organ. To determine the consequences of the loss of Tdy1 function in roots, we compared wild-type and tdy1 root morphology, mass and starch accumulation. We grew plants for 10 days in a high-light growth chamber in which the tdy1 chlorotic leaf phenotype was strongly expressed. We found that the root system of tdy1 plants was significantly smaller than that of wild-type siblings (Fig. 6A, H). All classes of roots were present in tdy1 mutants, indicating that the reduced root mass was not caused by a failure to elaborate a particular type of root. To determine if the reduced root mass was due to altered carbohydrate partitioning, we stained root cross-sections and whole roots from wild-type and tdy1 mutant plants with iodine.
No differences in tissue histology or starch deposition were observed (Fig. 6B-G). To examine if there might be differences in the levels of sugars or starch in the roots of wild-type and tdy1 mutant plants, we quantified their abundance over a diurnal cycle. tdy1 mutant roots contained significantly less Suc at all time points except midnight compared to wild type (Fig. 6J). Glucose accumulation was also lower in tdy1 roots relative to wild type, except at 8 pm and 12 am (Fig.  14   6K). Fructose and starch levels were low in both tdy1 and wild-type roots and not significantly different at any time point (Supp. Fig. S5). To determine if tdy1 plants might have any root phenotype when carbohydrates did not hyperaccumulate in leaves, we grew wild-type and mutant plants under low-light conditions that did not induce any visible tdy1 phenotype.
Comparison of root biomass between wild-type and tdy1 plants showed no difference (Fig. 6I), suggesting that carbohydrates were normally translocated to the roots in mutant plants, and that there was no apparent phenotype when grown under low light. The reduced tdy1 root mass observed during growth in high light is consistent with the previously reported reductions in plant height and inflorescence size and the delayed time to flowering seen in tdy1 mutants in comparison to wild-type siblings (Braun et al., 2006). We propose that these phenotypes are due to retention of excess carbohydrates in tdy1 leaf tissues.

Carboxyfluorescein Transport Studies
tdy1 mutants accumulate excess carbohydrates in their leaves (Braun et al., 2006). To determine whether any physical blockage of plasmodesmata could explain the carbon hyperaccumulation, we previously examined the symplastic transport route from the photosynthetic cells to the vein for callose deposition and plasmodesmatal ultrastructural changes that would prevent Suc movement. No alterations or structural perturbations were found, suggesting that the blockage to Suc transport occurs at the apoplastic step (Ma et al., 2008). If the apoplastic step is impeded, reduced solute mobility into the vein would be predicted. To test whether mobility into the phloem is reduced, we applied the fluorescent dye 5,6carboxyfluorescein (CF) directly to abraded leaves to monitor dye transport from the photosynthetic cells into the vein. In wild-type leaves, the dye was taken up by M cells, transported along the symplastic pathway into BS cells and preferentially accumulated in the VP cells ( Fig. 7A, C, E). CF staining in green regions of tdy1 mutant leaves appeared identical to wild-type leaves (data not shown). However, we found that the CF mobility into the vein was reduced in tdy1 chlorotic regions. CF was taken up by the M cells and moved into the BS cells where it accumulated, but less dye entered the VP cells (Fig. 7B, D, F). To examine phloem transport of the dye, we measured the rate of CF movement in wild-type and tdy1 leaves.
Although the value for tdy1 mutants was slightly slower than in wild type (tdy1 = 107.3 ± 9.7 15 mm hr -1 , wild type = 119.0 ± 7.6 mm hr -1 ), no significant difference was detected. These data suggest that the principal defect in CF movement in tdy1 chlorotic leaf regions is impeded solute flow from the photosynthetic cells toward the vein.
Tdy1 RNA accumulates in the phloem in all tissues. To investigate whether Tdy1 may function in the transport and release phloem, we fed wild-type and mutant leaves the phloemmobile tracer CF diacetate (CFDA). CFDA is a non-fluorescent form of the dye that can cross cell membranes. Upon entering a cell, non-specific esterases remove the diacetate groups, releasing CF which is fluorescent and charged and therefore remains in the cell (Wright and Oparka, 1996). We used CFDA to specifically image CF after entering into phloem cells. CF was transported through the veins to distal parts of the plants. In the transport phloem of the leaf sheath of the CF-fed leaf, the dye was similarly present in the phloem cells ( together, these data suggest that tdy1 mutants have phloem transport capabilities similar to wild type, and that reduced phloem transport capacity to distal sink tissues does not account for the hyperaccumulation of carbohydrates in leaves. These data also suggest that Tdy1 does not play a major role in the transport or release phloem.

DISCUSSION
The molecular mechanisms governing carbon partitioning in plants are largely unknown.
In this paper, we cloned and characterized the maize Tdy1 gene, which functions to promote carbohydrate export from leaves. We found that Tdy1 encodes a novel protein. Orthologous proteins were identified in grasses, but not in more distantly related plants. However, we identified two domains of TDY1 that are evolutionarily conserved across angiosperms, and therefore may be important for the function of these proteins. Additionally, using heterologous transient expression of a fluorescent reporter protein fusion, we determined that TDY1 localizes to the ER. Further, we showed that Tdy1 RNA is exclusively expressed in phloem cells.
Although Tdy1 RNA is highly abundant in sink tissues, dye labeling studies suggest that Tdy1 may function in phloem loading, but does not seem to play a substantial role in the transport or release phloem.

Subcellular Localization
The deduced amino acid sequence of TDY1 does not share similarity to any proteins of known function; however, two conserved domains were identified. Intriguingly, the first The middle portion of TDY1-like proteins is the most variable and does not contain a conserved domain. Within grasses, a series of conserved Pro residues in this region may be important for secondary structure. However, the majority of these Pro residues are not conserved in the closest Arabidopsis homolog. If this protein in Arabidopsis has a similar function to maize TDY1, the middle region of the protein might serve as a spacer or possibly a hinge region.
The C-terminus of TDY1 contains the second conserved domain. This domain is comprised of many charged residues (13/27 in maize TDY1) and is likely to be located on the surface of the protein. Within domain II is the conserved sequence EAFIXK where X is any residue. Using PSI-BLAST analysis, we found that this motif is present within a plant-specific domain of unknown function (DUF761; pfam05553). None of the proteins containing this domain have a characterized function. Two tdy1 mutant alleles encode proteins predicted to lack this domain due to a small deletion (tdy1-shadow) or an upstream frameshift mutation (tdy1-PM). These alleles confer a tdy1 mutant phenotype indistinguishable from the tdy1-D6 complete deletion allele, implying that this part of the protein is critical for its function.
18 Using a transient protein expression assay, we determined that TDY1 localized to the ER in onion cells. As the ER is the organelle responsible for the synthesis of proteins in the secretory system, it is possible that TDY1 functions in protein trafficking (Lee et al., 2003). The ER was found to be continuous between the CC and SE in tobacco (Nicotiana tabacum) (Martens et al., 2006), and protein movement between the CC and SE shown (Thompson and Wolniak, 2008).
These data suggest that the TDY1 protein could function in both of these cell types.
An interesting parallel to TDY1 localization and potential function may be found in comparison to the maize Floury1 (Fl1) gene. FL1 is a novel, plant-specific, ER-localized membrane protein that contains a domain of unknown function in its C-terminus that is different than the one present in TDY1 (Holding et al., 2007). fl1 mutants are so named because they have an opaque endosperm. In addition, fl1 mutants display uniform localization of the 22-kD α-zeins seed storage proteins throughout the protein bodies, in contrast to the specific localization seen in wild type. It was suggested that FL1 may act to facilitate the interaction of 22-kD α-zeins with chaperone proteins involved in their correct targeting within protein bodies. Though not related at the sequence level, it is tempting to speculate that since TDY1 has a similar protein architecture and putative localization, it might play a similar role as FL1. It is possible that TDY1 acts as a chaperone to assist in the correct protein targeting of phloem membrane/secretory proteins. Future work will characterize the functions of TDY1-interacting proteins and the localization of phloem membrane/secretory proteins in wild-type and tdy1 plants to test this hypothesis.

RNA Localization and CF Transport Studies Suggest TDY1 Functions in Phloem Loading
Several inferences can be drawn from the Tdy1 RNA localization pattern: (1) Tdy1 is expressed in protophloem cells upon differentiation, suggesting that Tdy1 function may be required as soon as these cells become competent for phloem transport. To our knowledge, Tdy1 is one of the earliest known maize genes specifically expressed in protophloem. Tdy1 expression may be useful to mark this tissue, and the Tdy1 promoter could have applications in driving gene expression from the onset of phloem function. (2) We found that Tdy1 is expressed strongly in the phloem cells of sink tissues and is also expressed in mature source leaves. This suggests that Tdy1 could function in sink and source tissues, and possibly play a role in both phloem loading 19 and unloading. However, the tdy1 carbohydrate-hyperaccumulation phenotype is only evident in leaves, and no defect in CF unloading in tdy1 mutants was observed. One explanation for these results could be that genetic redundancy masks any phenotype in sink tissues (see below).
Another intriguing possibility is that the Tdy1 RNA is mobile in the phloem, which could account for the high level of expression detected in roots, ears and other sink tissues. Analysis of where the Tdy1 promoter is transcriptionally active will be required to explore this speculation.
(3) Tdy1 expression in phloem cells indicates that the decrease in chlorophyll observed in tdy1 chlorotic leaf regions is an indirect effect likely caused by excess carbohydrates repressing photosynthetic gene expression (Sheen, 1990;Goldschmidt and Huber, 1992;Krapp and Stitt, 1995;Koch, 1996). This is in agreement with an earlier study that showed starch accumulation preceded leaf chlorosis in tdy1 emerging leaves (Braun et al., 2006).
Dye transport assays suggest that Tdy1 functions in the collection phloem, but does not have an essential role in the transport or release phloem. In wild-type leaves, the high level of CF detected in the VP cells reflects solute flow from the M cells toward the vein following symplastic Suc movement. As no transporter to export the xenobiotic dye to the apoplast is likely to exist, CF accumulates in the VP cells. In contrast, in tdy1 mutant leaves, we observed that CF movement from tdy1 chlorotic tissue into the veins was retarded. CF accumulated to higher levels in both the tdy1 BS and M cells, but had diminished levels in the VP cells compared to wild type. These data suggest that tdy1 chlorotic leaf regions have reduced symplastic solute flow leading to decreased phloem loading, which would explain the reduced amount of Suc, glucose and biomass observed in tdy1 mutant roots relative to wild type. One possibility to account for these data is that some type of plasmodesmatal restriction at the BS-VP cell interface constrains symplastic flow. However, from former investigations, no alterations or blockages of the plasmodesmata were detected in tdy1 leaves (Ma et al., 2008). Thus, these data suggest that any limitation in symplastic solute movement may be caused by a partial closure of the plasmodesmata rather than by a physical obstruction. In addition, Tdy1 RNA is expressed in the CC and SE, suggesting that a BS-VP plasmodesmatal constriction would likely be an indirect and downstream effect. Further studies are necessary to establish the molecular function of TDY1 and its role in promoting Suc loading into the collection phloem.
Based on the carbohydrate hyperaccumulation and other previous phenotypic characterizations, we hypothesized that Tdy1 may function to promote carbon export from 20 leaves, possibly by inducing SUT activity (Braun et al., 2006). However, a surprising finding was the high level of Tdy1 RNA accumulation detected in roots. Phloem unloading in maize roots occurs symplastically (Giaquinta et al., 1983), and therefore does not require SUT function.
These data raise the possibility that, at least in roots, TDY1 function is independent of SUTs, or that SUT regulation may be but one component of Tdy1 function. The expression data also suggest the alternative possibility that the tdy1 mutation may have uncovered only a limited aspect of Tdy1 function, i.e., no phenotype was observed in sink tissues with high levels of Tdy1 RNA due to genetic redundancy. We previously characterized an independent locus, Tdy2, with a nearly identical mutant phenotype, and proposed that it may have a related function to Tdy1 (Baker and Braun, 2008). In addition, genetic modifiers of tdy1, distinct from tdy2, are currently being isolated that enhance the severity of the tdy1 phenotype and may represent genes with partially overlapping functions. Finally, sequences related to Tdy1 are present in the maize genome. Future research to characterize the function of these genes and their overlap with Tdy1 may uncover additional roles for Tdy1 in sink tissues.
In summary, Tdy1 encodes a novel protein expressed in the phloem. Orthologous proteins are present only in grasses, but two domains conserved in angiosperms were identified. Based on the wild-type Tdy1 RNA localization pattern, the putative TDY1 protein subcellular localization, the defects in CF movement into the veins, and the carbohydrate hyperaccumulation in tdy1 chlorotic leaves, we hypothesize that Tdy1 acts in an ER-localized process that promotes phloem loading of Suc. If so, it is somewhat surprising that TDY1 sequence would not be more conserved in dicots and perhaps more distantly related vascular plants. In this respect, it is interesting that monocots and dicots contain evolutionarily distinct clades of SUTs (Sauer, 2007) (Braun and Slewinski, submitted). TDY1 may function in a grass-specific aspect of phloem biology, possibly by interacting with and having co-evolved with, grass-specific SUTs or other phloem proteins.
22 for all primers). Alleles not containing a Mu insertion were PCR-amplified and sequenced using gene-specific primers. Tdy1 PCR reactions typically contained 10% DMSO and 3% glycerol due to the high GC%, and all sequencing was performed by the Penn State Nucleic Acids Facility.

TDY1 Sequence Alignments
Tdy1 DNA and protein sequences were used to query public EST, genomic DNA and protein databases by BLAST analyses (Altschul et al., 1990). Protein alignment was performed using Align X (Vector NTI, Invitrogen).

Protein Subcellular Localization
Tdy1 contains no introns. The protein coding region was PCR-amplified from B73 genomic DNA using a proofreading DNA polymerase (Pfu, Stratagene), cloned into the pENTR/SD/D-TOPO vector (Invitrogen) and sequenced. To generate YFP translational fusions to TDY1, the entry clone was recombined with the pEarleyGate 101 destination vector (Earley et al., 2006).
For RFP fusions, a modified pEarleyGate vector containing mCherry was used (Skirpan et al., 2008). Construction of the N-terminal 71 amino acids or C-terminal 201 amino acids fusions was similarly performed. Constructs were transformed into onion bulb (Allium cepa) epidermal cells by biolistic particle bombardment using a Bio-Rad PDS-1000/He system according to the manufacturer's protocol. Fluorescent proteins were observed using a Nikon Eclipse 80i fluorescent microscope with a 100-W mercury lamp, and images were captured using a DXM1200F digital camera. YFP/GFP were imaged using a 465-495 nm excitation filter and a 515-555 nm bandpass emission filter. RFP was imaged using a 530-560 nm excitation filter and a 590-650 nm bandpass emission filter. Control experiments determined that no YFP/GFP fluorescence was observed using the RFP filter set, nor was RFP detected using the YFP/GFP filter set (data not shown).

RT-PCR and RNA in situ Hybridization Analyses
RNA was isolated using the RNeasy Plant Mini Kit (Qiagen) from fully mature adult leaves, young, etiolated, immature leaves not emerged from the whorl, ~2.5 cm long developing ears and tassels, and one-week-old seedling roots from seeds germinated in the dark on moist paper towels. RNA was treated with RQ1 DNAse (Promega), the integrity and purity assessed by 23 agarose gel electrophoresis and the concentration determined with a NanoDrop spectrophotometer (Thermo Scientific). cDNA was synthesized following the manufacturer's instructions from 1 μg of total RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). For semi-quantitative RT-PCR assays, 1 μl of cDNA was used as template in a PCR reaction with 20 cycles of 95 o C 15 sec, 60 o C 15 sec, 72 o C 15 sec. The products were separated on a 1.5% agarose gel, transferred to a nylon membrane, hybridized with a corresponding 32 P-labeled DNA probe, washed at high stringency and the blots exposed to X-ray film. For qRT-PCR assays, the cDNA was diluted five-fold and used as templates in an ABI 7500 Fast Real-Time PCR System with SYBR Green detection. In each qRT-PCR experiment, similar threshold cycle values of the reference genes were observed across all samples, indicating their suitability as appropriate standards. For testing sugar induction of Tdy1 expression, Ubq was used as the reference gene. However, Gapdh was used as the reference gene for testing Tdy1 regulation by carbohydrate depletion as we found that the threshold cycle values for Ubq varied between treatments. Melting curve and gel electrophoresis analyses determined that a single product of the expected size was amplified for all PCR reactions. Three independent RNA isolations (biological replicates) were analyzed with three technical replicates each for qRT-PCR experiments. As a negative control, separate reactions containing RNA isolated from The plasmid was linearized with NdeI and transcribed with T7 RNA polymerase to produce a DIG-labeled antisense probe. For producing the sense-strand probe, the plasmid was linearized with NcoI and transcribed with SP6 RNA polymerase. RNA in situ hybridization was carried out as described (Jackson et al., 1994;Langdale, 1994). To identify phloem cell types, tissue sections were stained with DAPI (1 μg mL -1 in water) or 0.05% aniline blue ([w/v] in potassium 24 phosphate buffer, pH 9.0) and viewed under UV light. DAPI was visualized with a 340-380 nm excitation filter and a 435-485 nm bandpass emission filter. Aniline blue was imaged with a 360-370 nm excitation filter and a 420 nm longpass emission filter.

CF Transport Assays
To monitor CF symplastic movement, the adaxial surface of mature leaves was gently abraded with sandpaper and 75 μL of 50 μg mL -1 CF prepared according to Grignon et al. (1989) was applied for 1 hr as described (Haupt et al., 2001;Botha, 2005). To measure CF transport rates in leaves, the fourth or fifth mature leaf from three to four-week-old plants was abraded 200 mm distal to the ligule, 50 μL CF applied, and the application site covered with parafilm. Preliminary experiments examined CF movement at multiple time points after application. One hr was selected as the optimal assay time as the dye had not exited the leaf and was predominantly present in phloem cells two to three cm proximal to the application site. After 1 hr, the leaf was dissected and analyzed by free-hand cross-sections to assess dye movement in the phloem. For both wild-type and tdy1 plants, three leaves were assayed and the experiment was repeated three times. Data for one replicate is presented. For studies of CF movement in the transport and release phloem, 50 μg mL -1 5,6-CFDA (Invitrogen) was fed to the cut end of mature leaf four as described (Aoki et al., 2004). Leaf sheaths and roots were examined 3 hr post-feeding, and immature sink leaves were inspected after 6 hr. To visualize CF, free-hand sections were examined using either a 450-490 nm excitation filter and a 515 nm longpass emission filter or a 465-495 nm excitation filter and a 515-555 nm bandpass emission filter to reduce tissue autofluorescence.

Starch Staining, Carbohydrate Quantifications and Root Mass Determinations
Starch was detected in cleared leaves and roots by staining with iodine potassium iodide (Braun et al., 2006). Carbohydrates were quantified according to Dinges et al. (2001) from 100 mg of root tissue harvested from the middle of the root system of 10-d-old seedlings (n = 3). Tissue was collected every 4 hr over a diurnal cycle and the experiment replicated three times. Data presented is for all measurements. Root mass was determined by weighing entire root systems isolated at the seed attachment point of 10-d-old seedlings (n = 10). For both high-and low-lightgrown plants, the experiment was replicated twice with similar results, and one repeat is shown.

25
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number xxxxxxx.

Supplemental Material
Supplemental Figure S1. Alignment of Arabidopsis TDY1-like proteins.

Supplemental
Supplemental Table S3. List of primers used.

32
shows region deleted in the shadow allele and dotted line indicates that the entire gene is deleted in the D6 allele.     shown in dark gray and tdy1 mutant in white. Bars represent the mean (n=10) and error bars the SE. An asterisk indicates that the value is significantly different than wild type at P < 0.05 using Student's t-test. J, K, sugar accumulation in roots over a diurnal cycle. Wild type is shown by black triangles and tdy1 by orange squares (n=9). J, Suc. K, Glucose. At all time points, tdy1 34 values were significantly different than wild type except at 12 am for both Suc and glucose and 8 pm for glucose as determined by Student's t-test at P < 0.05.    and Ubiquitin (Ubq) as a loading control in B73 wildtype tissues. m, mature leaves, im, immature leaves, e, developing ear, t, developing tassel, r, one-week-old roots. C-H, fluorescent proteins transiently expressed in onion epidermal cells. C, ER membrane-targeted GFP-HDEL (Goodin et al., 2007). D, same cell as in C coexpressing full-length TDY1-RFP fusion, showing colocalization to the ER. Arrow indicates perinuclear envelope which is contiguous with the ER. E, N-terminal amino acids 1-71 of TDY1 fused to YFP showing ER localization. F, same cell as in E coexpressing full-length TDY1-RFP fusion. G, C-terminal amino acids 72-272 of TDY1 fused to YFP localizes in the cytoplasm and nucleus. H, same cell as in G coexpressing soluble RFP in the cytoplasm and nucleus. . qRT-PCR analyses of Tdy1 RNA accumulation. A, Tdy1 RNA levels relative to Ubq. Wild type (WT) is shown in dark gray, tdy1-R green (G) regions in light gray, and tdy1-R chlorotic (Y), carbon-hyperaccumulating regions in white. B, Tdy1 RNA levels relative to Gapdh. RNA was isolated from leaves of wild-type plants grown in the greenhouse at the end of the day (dark gray) or from carbohydrate-depleted plants kept in a dark room for three days (white). For both A and B, bars represent the mean and error bars the SE. Samples were not statistically different at P < 0.05 using Student's t-test. Bars represent the mean (n=10) and error bars the SE. An asterisk indicates that the value is significantly different than wild type at P < 0.05 using Student's t-test. J, K, sugar accumulation in roots over a diurnal cycle. Wild type is shown by black triangles and tdy1 by orange squares (n=9). J, Suc. K, Glucose. At all time points, tdy1 values were significantly different than wild type except at 12 am for both Suc and glucose and 8 pm for glucose as determined by Student's t-test at P < 0.05.