Non-dispersive phloem-protein bodies (NPBs) of Populus trichocarpa consist of a SEOR protein and do not respond to cell wounding and Ca²⁺

Plant material

Ceiba pentandra (L.) Gaertn., Theobroma cacao L. (both Malvaceae, Malvales), Passiflora incarnata L. (Passifloraceae, Malpighiales), Pombalia (Hybanthus) communis [A.St.-Hil.] Paula-Souza and Viola tricolor L. (both Violaceae, Malpighiales) were permanently available through the instructional plant collection of the School of Biological Sciences, Washington State University. Populus trichocarpa Torr. & A. Gray ex Hook (Salicaceae, Malpighiales) was grown in the open on the WSU campus (46°44′34″N, 117°8′10″W; elevation 789 m). Vicia faba L. (Fabaceae) was grown in a greenhouse as described before (Knoblauch et al., 2003).

Light microscopy

Shallow periclinal cuts were made on the main vein of detached leaves to access active sieve tubes (compare Knoblauch & Van Bel, 1998). The leaves were fixed upside down on microscope slides with double-sided adhesive tape. Cut surfaces were covered with standard medium (10 mM KCl, 10 mM CaCl₂, 5 mM NaCl, 50 mM HEPES at pH 7.5) or Ca²⁺-free medium (10 mM Na₂-EDTA instead of CaCl₂). Sieve elements were observed with a Leica DM LFSA microscope equipped with water immersion lenses not corrected for coverslips (HCX Plan APO U-V-I series) and a Leica DFC 300 FX camera (Leica Microsystems, Wetzlar, Germany). Confocal laser-scanning microscopy was performed with a Leica TCS SP8 (Leica Microsystems, Wetzlar, Germany) on specimens stained with aniline blue and synapto red.

Electron microscopy

For scanning electron microscopy (SEM), pieces of P. trichocarpa mature petioles and leaf main veins 3 cm long were fixed for 24 h in 3% glutaraldehyde and 2% paraformaldehyde in either 50 mM cacodylate buffer, standard medium, or Ca²⁺-free medium (see above). The tissue was rinsed with cacodylate buffer and postfixed for 2 h in 2% OsO₄ in 50 mM cacodylate buffer. After rinsing two times in 50 mM cacodylate buffer and two times in water, the specimens were lyophilized for 12 h using a VirTis lyophilizer (VirTis, Gardiner, NY, USA) and mounted on aluminum specimen mounts with carbon Pelco tabs (TED Pella, Redding, CA, USA). Approximately 20 nm gold were applied with a Technics Hummer V sputter coater (Anatech, Hayward, CA, USA), and specimens were examined with a Quanta 200F SEM (FEI, Hillsboro, OR, USA).

For transmission electron microscopy (TEM), excisates of 3 × 3 × 3 mm from petioles and leaves were fixed as described above. Ultrathin sections (70–100 nm) were produced from tissue embedded in Spurr’s resin with an ultramicrotome (Reichert Ultracut R; Leica Microsystems, Wetzlar, Germany) and were transferred onto formvar-coated slot grids. After staining with 1% uranyl acetate and 0.01% KMnO₄ for 10 min and poststaining for 6 min in Reynolds’ lead citrate (Reynolds, 1963), the sections were imaged on a Philips CM200 and on an FEI Tecnai G2 20 Twin (FEI, Hillsboro, OR, USA).

For analysis by electron tomography, sections of about 150 nm were taken from the resin-embedded samples and placed onto formvar-coated grids. A single axis tilt series of ±55° was taken at 50,000 magnification on an FEI Tecnai G2 20 Twin (FEI, Hillsboro, OR, USA) using a high tilt holder. Images were taken every 2° to ±30°, and every 1° at higher angles. The pixel size of the original images (4,096 × 4,096 pixels) was 0.22 nm. The tomographic volume was acquired using FEI Automated Tomography software and reconstructed with Inspect3D after binning by four to produce a final tomography of 512 × 512 pixels with a pixel size of 0.88 nm. A voltex rendering of the reconstructed tomography was performed with Amira 2.0 (FEI, Hillsboro, OR, USA).

Isolation of individual NPBs

Individual NPBs were isolated from sieve elements using fine-tipped (1–2 μm tip diameter) borosilicate glass capillaries (Clark Electromedical Instruments, Reading, UK) produced with a microelectrode puller (Model P-200; Sutter Istruments, Novato, CA, USA). The capillaries were mounted on a four-axis micromanipulator (MMW-204; Narishige, Tokyo, Japan) and the protein bodies were extracted from sieve elements as detailed by Knoblauch et al. (2003).

Mass isolation of NPBs and molecular characterization

Segments of 30 cm length from de-leafed P. trichocarpa stems of about 2 cm diameter were scored through the bark into the secondary xylem with a single edged razor blade. Bark strips were peeled from the wood; the phloem was scratched from the periderm using surgical scalpels and placed in extraction buffer (10 mM Tris–HCl at pH 8.5, 200 μM EDTA, 8 mM K₂SO₃) in a mortar. The isolated phloem was squeezed with a pestle for 3 min, and the tissue was strained using 20 μm CellMicroSieves (BioDesign, New York, NY, USA). The remaining tissue in the sieve was ground three more times in fresh extraction buffer, progressively applying more pressure with the pestle in each round. The crude NPB extraction was centrifuged at 100 × g for 10 min. The supernatant was discarded and the pellet re-suspended in the extraction buffer and 3.75 ml 80% Nycodenz (Axis-Shield PoC, Oslo, Norway) solution. Using another 15 ml Nycodenz solution, a density gradient (20–80% Nycodenz) was produced with a custom-built gradient mixer. The NPB solution was applied to the gradient and centrifuged at 1,650 × g and 20 °C for 3 h (5810R centrifuge; Eppendorf, Hamburg, Germany). One milliliter aliquots were transferred consecutively from the top of the gradient into 2 ml microcentrifuge tubes. Small sub-samples were checked microscopically. The tubes containing the samples with the highest NPB densities were centrifuged (22,000 × g at 20 °C) for 10 min to obtain purified NPBs.

The purified proteins were denatured at 98 °C for 10 min in 300 μl loading buffer (62.5 mM Tris–HCl at pH 6.8, 10% glycerol, 2% sodium dodecyl sulfate, 5% β-mercapto-ethanol, 0.05% bromophenol blue) and separated by SDS–PAGE. The gel consisted of a 4% stacking gel (pH 6.8) and a 12% separating gel (pH 8.8). The proteins were separated for 45 min at 15 mA and variable voltages, and then fixed in the gel and stained by incubation in five parts methanol/five parts distilled water/one part glacial acetic acid and 0.05% coomassie blue overnight at room temperature. The gel was destained with the same solution lacking coomassie blue, and rehydrated in distilled water. Protein bands were excised and a trypsin digest tandem mass spectrometry (MS–MS) analysis was performed at the Molecular Biology Core of the Center for Reproductive Biology (Washington State University, Pullman, WA, USA). The peptides obtained were blasted against the National Center for Biotechnology Information (NCBI; https://www.ncbi.nlm.nih.gov) database to determine possible protein matches. Two peptides matched sequences in the hypothetical protein referred to as Potri017G071000.1 in the current version (November 2017) of the P. trichocarpa genome database.

Based on the genomic sequence of the corresponding gene, primers (forward with a 5′ XbaI restriction site: 5′-TTCTCTAGAAGCCAATCTTCTTACTCAGC-3′; reverse with a 3′ HindIII restriction site: 5′-TTCAAGCTTCGCACTCAACAACATTATTTAG-3′) were designed to obtain the full-length DNA from our plants. Genomic DNA was isolated from 50 mg fresh tissue homogenized with liquid nitrogen. Genomic DNA was isolated with a commercial kit according to the manufacturer’s instructions (Promega #A1120, Madison, WI, USA). The gene was amplified from a 1:100 genomic DNA dilution (1 μM final primer concentration) for cloning into the AKK 1408 shuttle vector (Pélissier et al., 2008). The amplified band was gel purified (#20021, QIAGEN Sciences, Germantown, MD, USA) and digested with 1 U each of XbaI and HindIII fast digest enzymes (Fermentas, Baltimore, MD, USA) for 3 h at 37 °C. The AKK 1408 shuttle vector was also digested with the restriction enzymes and the gene was inserted; ligation was performed overnight at 4 °C with T4 DNA ligase (Invitrogen, Waltham, MA, USA) and the vector was transformed into DH5α Escherichia coli. Successful transformants were grown overnight at 37 °C on LB medium with 50 μg/ml ampicillin. When colony PCR confirmed insertion, plasmids were isolated from cultures incubated overnight with the Promega SV Wizard Miniprep kit according to manufacturer’s instructions (#A1330 Promega, Madison, WI, USA), and the gene was sequenced using the flanking M13F: 5′-GTAAAACGACGGCCAGT-3′ and M13R: 5′-GGAAACAGCTATGACCATG-3′ primer sites by Eurofins Genomics (https://www.eurofinsgenomics.com).

Sequence data analysis

Amino acid sequences of proteins and genomic nucleotide sequences were downloaded from the NCBI (https://www.ncbi.nlm.nih.gov) and phytozome databases (https://phytozome.jgi.doe.gov) for further analyses. Sequence alignments for conserved structure comparisons were conducted using CLC Sequence Viewer v.7.8.1, or earlier versions (QIAGEN, Aarhus, Denmark) with default settings. Phylogenetic analyses of the SEO/SEOR gene family for angiosperms used all putative gene family members from the above databases with gene copies aligned with MUSCLE (Edgar, 2004). Maximum likelihood (ML) analyses were performed using the JTT model in RAxML (Stamatakis, 2006) with 100 bootstraps. The tree was rooted on the branch placing the single SEOR gene copy from Amborella sister to the rest of the angiosperm gene copies, based on the strong independent evidence for that relationship (Angiosperm Phylogeny Group et al., 2016).

NPBs in sieve elements of P. trichocarpa

We applied conventional light microscopy to study the abundance and distribution of NPBs in the phloem of P. trichocarpa. A few cell layers were removed from the abaxial side of main veins of detached leaves to enable the observation of phloem tissue in situ. NPBs showed as spherical inclusions in live sieve elements (Fig. 2). Their diameters ranged from 1.8 to 5.5 μm and occupied between 25% and 50% of the diameter of their sieve elements (n = 28). Of over 120 sieve elements examined, about 90% contained one single NPB that was found close to a sieve plate (Fig. 2A). There was no consistency regarding the side of the sieve plate on which the NPB was located. Repeatedly we saw one NPB each on the upstream and downstream side of the same sieve plate (Fig. 2A), and NPBs at the upstream and downstream ends of sieve elements in parallel sieve tubes (Fig. 2B). About 3% of the sieve elements harbored two NPBs, and in roughly 10% an NPB was located at a distance from a sieve plate; an example of both features in the same cell is presented in Fig. 2B. No movements of NPBs ever were noticed.

Structure of P. trichocarpa NPBs

In sieve tubes prepared for scanning electron microscopy, NPBs often appeared embedded in an irregular, fibrillar network of unclear identity that varied from relatively dense (Fig. 3A) to almost absent (Fig. 3B). Corresponding features were evident in transmission electron-micrographs. The lumen of the sieve elements around the NPBs exhibited various degrees of occupation by fibrillar structures from densely filled (Fig. 3C) to almost empty (Fig. 3D). We were unable to correlate this variation unequivocally with any obvious physiological or preparation-related parameter. Regardless of the presence of the irregular network in the sieve element, NPBs in both the TEM and SEM showed a rugged spiny surface (Fig. 3) that had not been resolved by in situ light microscopy (Fig. 2).

The spines on the surface of the P. trichocarpa NPBs consisted of groups of chimney-like rods, as revealed in the SEM. These rods differed in length and diameter, and terminated in blunt rectangular tips (Fig. 4A). In the TEM, the rods showed longitudinal striations (Fig. 4B). The striations continued into the interior of the protein bodies where bundles of rods appeared to be condensed into a solid mass. Consequently, cross-sections showed different zones of striations that varied in orientation (Fig. 4C). In rods that were sectioned perpendicular to their long axes it became evident that the longitudinal striations corresponded to tubules, since an electron-dense tubule wall could be distinguished clearly from an apparent cavity in the tubule center (examples marked by arrows in Fig. 4C). These tubules had outer and inner diameters of 20.8 ± 2.3 and 10.2 ± 1.7 nm (n = 50), respectively, suggesting a wall thickness of 5.3 nm. Electron tomography corroborated these findings and showed that groups of tubules in their most highly ordered state were packed in hexagonal arrays (Figs. 4D and 4E).

NPBs in Malpighiales and Malvales do not respond to sieve tube wounding

To establish whether P. trichocarpa NPBs were reactive to sieve element wounding like the forisomes of the Fabaceae (Fabales), we studied their responsiveness in situ by CLSM. NPBs appeared as spherical bodies of a diameter less than half that of their sieve element; the spiny surface structure was resolved (Fig. 5). We injured the sieve elements by puncturing individual sieve tubes with a dissection laser or with micro-pipettes. NPBs in the affected sieve elements never exhibited any response (Fig. 5 shows a representative example). To see whether this finding could be generalized for the order Malpighiales, we tested three additional members (Viola tricolor and Pombalia communis, Violaceae; Passiflora incarnata, Passifloraceae) with identical results. Moreover, NPBs in two members of the Malvales (Ceiba pentandra and Theobroma cacao, Malvaceae), more remotely related to Fabales but still in the rosid clade (Fig. 1), also failed to respond to any of the stimuli (see Fig. S1 for examples).

Isolated P. trichocarpa NPBs are insensitive to Ca²⁺

Forisomes, the P-protein bodies of leguminous plants, respond to injuries of the sieve tube by a Ca²⁺-mediated swelling reaction. The fact that P. trichocarpa NPBs had not responded to sieve tube wounding (Fig. 5) did not necessarily rule out the possibility that they were reactive to Ca²⁺. We isolated individual NPBs mechanically and exposed them to Ca²⁺ concentrations of up to 10 mM. The NPBs remained unresponsive under conditions that triggered the full response in forisomes from Vicia faba (Fig. 6).

P. trichocarpa NPBs consist of a SEOR protein

When isolated and purified P. trichocarpa NPBs were subjected to SDS–PAGE, they gave rise to two bands of just below and above 80 kDa, respectively. These bands were excised and analyzed by ESI MS–MS after trypsin digestion, leading to the identification of four peptides (1: LVSDSDIRK; 2: IPLEMVYVGK; 3: NKPLVVLDPQGK; 4: LVSDSDIYKLPYLHAIVK). BLAST searches failed to detect peptides 1 and 4 in the P. trichocarpa genome or any other genome database, and the origin and nature of these peptides remains obscure at this time. On the other hand, peptides 2 and 3 were identified in the amino acid sequence of the hypothetical protein, POPTR_0017s10720g, encoded by the gene called Potri.017G071000.1 in the current version 3.0 of the P. trichocarpa genome (https://phytozome.jgi.doe.gov). We cloned the gene from our plant material and sequenced it. The protein derived from the resulting DNA sequence had 716 amino acids and a molecular mass of 81.9 kDa. This result agreed well with the molecular masses of the protein bands derived from isolated NPBs, which supported the conclusion that the protein was a major component of P. trichocarpa NPBs.

Our NPB protein was almost identical to the hypothetical product of Potri.017G071000.1 (Fig. S2), which in the genome database (https://phytozome.jgi.doe.gov) is annotated as an ortholog of At3g01680, the gene encoding the SEOR1 protein in Arabidopsis. Our rooted phylogenetic hypothesis of relationships among almost 200 hypothetical SEOR and SEO proteins from 29 species (Document S1) detected two clusters consisting of six and seven gene copies, respectively, of P. trichocarpa, while another gene sequence of this species assumed a position remote from the clusters (Fig. 7). Potri.017G071000.1 was localized in the six-protein cluster that was most closely related to AtSEOR1. We concluded that P. trichocarpa NPBs consist mainly or entirely of a P-protein whose primary structure closely resembles that of previously characterized SEOR proteins, in particular AtSEOR1. Consequently, we designated the new protein PtSEOR1.

NPB localization in sieve tubes

Free-floating large objects exposed to the sieve tube flow will move to the downstream ends of sieve elements and accumulate there, at the upstream sides of sieve plates. In organs with a clearly defined expected flow direction, for instance mature photosynthesizing leaves, we would expect the objects to accumulate at the proximal ends of the sieve elements (i.e., the ends directed toward the petiole). This effect should be reinforced when a petiole is cut to obtain a leaf for microscopic examination (as in Fig. 2), because the opening of the sieve tube system will induce a turgor-driven surge of the sieve tube contents toward the cut. The NPBs in detached leaves of P. trichocarpa showed no such consistent localization. While they mostly were found close to sieve plates (Figs. 2 and 5), they were located randomly on the expected upstream or downstream sides of the plates. Consequently, sieve plates with one NPB on each side were observed regularly in P. trichocarpa (Fig. 2A). Similar observations have been reported from the related species Passiflora coerulea (Oberhäuser & Kollmann, 1977), Salix nigra (Deshpande & Evert, 1970) and S. caprea (Mishra & Spanner, 1970). The latter authors argued that the presence of NPBs (referred to as extruded nucleoli at the time) on both sides of sieve plates showed that no turgor-driven surge had been triggered by the preparation procedure. This seemed to imply further that no bulk flow had occurred in the sieve tubes prior to preparation, an important argument in favor of Spanner’s electro-osmotic theory of phloem transport (Spanner, 1979; for a review of the historical context, see Knoblauch & Peters, 2017). However, the unexpectedly inconsistent localization of NPBs under an assumed rapid bulk flow could be explained by structural anchors, as shown for various sieve element organelles (Ehlers, Knoblauch & Van Bel, 2000). It remains to be established whether an anchoring function can be ascribed to the irregular fibrillar networks in which NPBs seemed embedded to various degrees in preparations for electron microscopy (Fig. 3). The nature of these fibrils is obscure. One may suspect that they consist of structural cytoskeletal or SEOR proteins. However, the only such proteins detected in phloem exudates from hybrid poplar (P. trichocarpa × P. deltoides) were α- and β-tubulin (Dafoe et al., 2009). While hybrid poplar is not necessarily a valid model for its parent taxa, we note that the fibrils do not resemble microtubules (Fig. 3).

NPB structure

Non-dispersive P-protein bodies of P. trichocarpa consist of numerous bundles of tubular protein fibrils (Fig. 4), resulting in a spiny surface of the NPB (Figs. 3A, 3B and 4A) and conspicuous striations visible in sections of NPBs in the TEM (Figs. 4B and 4C). Within each bundle, the fibrils form arrays with a hexagonal symmetry (Figs. 4D and 4E). All these structural characters have been documented previously in other species of the Malpighiales, including Passiflora coerulea (Passifloraceae, Kollmann, 1960; Oberhäuser & Kollmann, 1977) and the members of the Salicaceae, Idesia polycarpa (Behnke, 1991), Salix caprea (Mishra & Spanner, 1970), S. sachalinensis (Nehls, Schaffner & Kollmann, 1978), S. nigra and Populus deltoides (Deshpande & Evert, 1970). For the tubules forming NPBs in S. sachalinensis, Nehls, Schaffner & Kollmann (1978) reported an outer diameter of 17–18 nm and a wall thickness of 5–6 nm, similar to the values of 20.8 and 5.3 nm, respectively, we found in P. trichocarpa. These authors applied a negative staining protocol that allowed them to detect oblique striations on the walls of the tubules, prompting the hypothesis of a helical arrangement of substructures (Nehls, Schaffner & Kollmann, 1978). We were unable to reproduce this result in P. trichocarpa, despite significant efforts. Nonetheless, the pronounced similarities between NPBs of all Malpighiales examined suggest that these protein bodies are essentially the same structures in the species of this order.

The maximum packing density of cylinders is achieved by co-alignment in hexagonal arrays (Bezdek & Kuperberg, 1990). This conclusion holds when the cylinders are curved or flexible as is the case with cytoskeletal filaments and other proteinaceous fibrils (Starostin, 2006). Therefore the occurrence of hexagonal arrays in NPBs (Fig. 4) as such is not surprising. Somewhat irregular hexagonal patterns can be found even in the P-protein masses in Arabidopsis sieve tubes (Fig. S3; compare Froelich et al., 2011), which are capable of flowing along sieve tubes (Knoblauch & Peters, 2017). What calls for an explanation, though, is the difference in the stability of the filament arrays in P. trichocarpa NPBs as compared to the viscous sieve tube “slime” in Arabidopsis.

Responsiveness of NPBs

Forisomes, the P-protein bodies of legumes, are SEO protein complexes that respond to injuries of the sieve tube by a Ca²⁺-mediated swelling (Knoblauch et al., 2001, 2003), which probably results in an occlusion of the tube (Knoblauch et al., 2012). Forisomes consist of fibrils arranged in pseudo-crystalline arrays (Laflèche, 1966; Lawton, 1978; Knoblauch et al., 2001). Poplar NPBs possess a similarly regular ultrastructure (Fig. 4), but seem to consist of a SEOR protein (Fig. 7). SEOR proteins have failed to produce any responses to Ca²⁺ or sieve element injury so far (Knoblauch et al., 2014). Populus trichocarpa NPBs showed no response to sieve element wounding in vivo (Fig. 5), and neither did NPBs in three other members of the Malpighiales. Together with Celastrales and Oxalidales the Malpighiales form the COM clade, which is part of either the fabid branch (as shown in Fig. 1) or the malvid branch of the rosids, depending on whether the analysis is based on plastid DNA or low-copy nuclear and mitochondrial genes (Angiosperm Phylogeny Group et al., 2016). Therefore it is of interest that NPBs in two species of the Malvales also failed to respond to sieve tube injury. Considering also that NPBs in members of the Rosales, part of the sister clade of the Fabales, exhibited the same lack of responsiveness (Knoblauch et al., 2001), we conclude that NPBs of P. trichocarpa are representative of P-protein bodies of the entire rosid clade, except for Fabaceae, with respect to wound responses in vivo. Furthermore, P. trichocarpa NPBs in vitro did not respond to Ca²⁺ concentrations of up to 10 mM (Fig. 6), over a 100-fold the level required for the maximum forisome reaction under similar conditions (Knoblauch et al., 2005). Taken together, our findings provide no support for the assumption that SEOR proteins are involved in active wound responses or in any kind of Ca²⁺-mediated activities in live plants. They rather are in line with the hypothesis that Ca²⁺-dependent contractility of P-proteins is a synapomorphic trait of the Fabaceae (Peters et al., 2010), implying that the resulting possibility of an involvement of P-proteins in active wound responses is also restricted to this family. Unfortunately, this implies that the biological function of NPBs in P. trichocarpa and other species remains in the dark at this time.

Relations between PtSEOR1 and other SEOR and SEO proteins

Two peptides derived from isolated P. trichocarpa NPBs also were found in the amino acid sequence of the hypothetical P. trichocarpa protein, Potri.017G071000.1. From the experimentally established sequence of the gene in our plants, we derived an amino acid sequence that was slightly shorter than that of Potri.017G071000.1 (716 compared to 752 residues), but exhibited identity with Potri.017G071000.1 in 712 of its 716 residues (99.4%; Fig. S2). Both sequences were closely related to that of AtSEOR1 (Fig. S2; Fig. 7). Taken together, these findings support the conclusion that an important or even the main component of NPBs in P. trichocarpa is a SEOR protein. This protein derived from live plants evidently represents the hypothetical Potri.017G071000.1, and we call it PtSEOR1.

The P. trichocarpa genome database contains several hypothetical proteins exhibiting sequence similarities with Arabidopsis SEORs, and thus with PtSEOR1. Phylogenetic analysis places most of these hypothetical proteins in two dense clusters of proteins with closely related primary structures (Fig. 7), which may reflect recent gene multiplication events. The promoter region of one of the corresponding genes, Potri.001G340200.1, recently has been shown to be active specifically in phloem tissue (Nguyen et al., 2017). Unfortunately, the resolution of the histological approach did not allow these authors to differentiate between the cell types of the phloem, and it remains unclear whether the promoter, denoted PtrDP3, is sieve element-specific like the promoters of VfSEO1 (Noll et al., 2007), AtSEOR1 (Froelich et al., 2011), AtSEOR2 (Rüping et al., 2010) and MtSEO1 (Bucsenez et al., 2012). Nguyen et al. (2017) determined that the sequence located within −200 to −300 bp from the translation initiation codon was sufficient to render the PtrDP3 promoter phloem-specific. An alignment of the putative promoter sequence of the PtSEOR1 gene Potri017G071000.1 and that of PtrDP3 revealed a high degree of homology between the 100-bp sequence conferring phloem-specificity to the latter and the corresponding sequence of PtSEOR1, including a conserved TATA-box motif (Fig. S4). Therefore it seems a plausible hypothesis that the PtSEOR1 promoter might be active specifically in the phloem as well. In the future, the comparative functional analysis of different SEOR promoters can be expected to help elucidating the mechanisms of phloem- and sieve element-specific gene expression.