The cytoplasmic-localized, cytoskeletal-associated RNA binding protein OsTudor-SN: evidence for an essential role in storage protein RNA transport and localization

Authors


*(fax +1 509 335 7643; e-mail okita@wsu.edu).

Summary

Previous studies have demonstrated that the major storage protein RNAs found in the rice endosperm are transported as particles via actomyosin to specific subdomains of the cortical endoplasmic reticulum. In this study, we examined the potential role of OsTudor-SN, a major cytoskeletal-associated RNA binding protein, in RNA transport and localization. OsTudor-SN molecules occur as high-molecular-weight forms, the integrity of which are sensitive to RNase. Immunoprecipitation followed by RT-PCR showed that OsTudor-SN binds prolamine and glutelin RNAs. Immunofluorescence studies using affinity-purified antibodies show that OsTudor-SNs exists as particles in the cytoplasm, and are distributed to both the protein body endoplasmic reticulum (ER) and cisternal ER. Examination of OsTudor-SN particles in transgenic rice plants expressing GFP-tagged prolamine RNA transport particles showed co-localization of OsTudor-SN and GFP, suggesting a role in RNA transport. Consistent with this view, GFP-tagged OsTudor-SN is observed in living endosperm sections as moving particles, a property inhibited by microfilament inhibitors. Downregulation of OsTudor-SN by antisense and RNAi resulted in a decrease in steady state prolamine RNA and protein levels, and a reduction in the number of prolamine protein bodies. Collectively, these results show that OsTudor-SN is a component of the RNA transport particle, and may control storage protein biosynthesis by regulating one or more processes leading to the transport, localization and anchoring of their RNAs to the cortical ER.

Introduction

Rice seeds synthesize two major classes of storage proteins: prolamines and glutelins (Muench and Okita, 1997). These proteins are synthesized on the endoplasmic reticulum (ER), where they are translocated to the lumenal compartment (Krishnan et al., 1986; Tanaka et al., 1980; Yamagata and Tanaka, 1986). Prolamines are retained within this intracellular site, where they assemble to form spherical intracisternal inclusion bodies; whereas glutelins are exported to the Golgi, where they are packaged into dense vesicles and are then transported to storage vacuoles. Previous studies from this laboratory have established that the storage protein RNAs are not randomly distributed on the ER: prolamine protein body RNAs are localized specifically to the ER membranes (PB-ER) that surround the prolamine intracisternal inclusion bodies, whereas glutelin RNAs are distributed to the adjacent connecting cisternal ER (Li et al., 1993).

Targeting of the prolamine RNAs to the PB-ER is dependent on RNA cis-elements within the mRNA sequence (Choi et al., 2000; Hamada et al., 2003b). Two partially redundant RNA cis-elements are essential for restricted targeting of RNA to PB-ER, as a single cis-element causes prolamine RNA to be localized to both the PB-ER and the cisternal ER, whereas the absence of both cis-elements directed RNA localization solely to the cisternal ER (Choi et al., 2000; Hamada et al., 2003b). These results indicate that the prolamine RNAs are targeted to the PB-ER by a regulated process that is dependent on one or more specific trans-factors. Interestingly, glutelin RNAs are also transported to the cisternal ER by a regulated process, as addition of the glutelin 3′ sequences redirects the localization of prolamine RNA from the PB-ER to the cisternal ER (Choi et al., 2000; Hamada et al., 2003b). Hence, RNAs are transported by at least three pathways to the cortical ER in developing rice endosperm: two signal-dependent pathways and a third signal-independent pathway.

A modified two-gene system using GFP was employed to monitor RNA transport in real time in developing endosperm sections (Hamada et al., 2003a). GFP-tagged prolamine RNAs were observed as particles that moved in a stop-and-go pattern. Although movement was generally directional, oscillations around a stationary position and changes in direction were commonly observed. These movement properties, typical of a cytoskeleton-mediated process, were inhibited by drugs that destabilized actin filaments or inhibited the ATPase activity of myosin, but not by those that disrupt microtubules (Hamada et al., 2003a).

Current efforts are directed at identifying the trans-factors that comprise the RNA transport particle, and are responsible for storage protein RNA transport and localization. Previous studies identified several RNA binding activities detected in the cytoskeleton-enriched fraction from developing seeds (Sami-Subbu et al., 2000, 2001). One prominent RNA binding protein, previously called Rp120 and now renamed as OsTudor-SN, showed partial binding specificity, as it recognized the prolamine 3′ untranslated region (UTR), but not the 5′-UTR, coding sequence or polyA region (Sami-Subbu et al., 2001). On sucrose density gradients, OsTudor-SN co-sedimented with high-molecular-weight actin and tubulin, suggesting that OsTudor-SN is associated with the cytoskeleton.

Orthologs of Tudor-SN have been extensively studied in animals. Tudor-SN was first discovered as a transcriptional co-activator, p100, which stimulates the activity of several transcriptional factors (Leverson et al., 1998; Tong et al., 1995; Yang et al., 2002). In addition to its location in the nucleus, p100 is found in the cytoplasm as a major constituent of the ER, and is associated with lipid droplets in milk secreting animal cells (Keenan et al., 2000). In more recent years, Tudor-SN was found to be a component of the RNA interference silencing complex, RISC, in Caenorhabditis elegans, Drosophila and mammals (Caudy et al., 2003), and as a component of a cytoplasmic-localized nucleolytic activity of hyperedited RNAs (Scadden, 2005). In view of the varied intracellular locations, and roles, in transcription and double-stranded RNA processing, Tudor-SN has multiple roles in governing gene expression during cell growth and development in animals. Although Tudor-SN orthologs exist in Arabidopsis, rice and pea (Caudy et al., 2003), virtually nothing is known on the role of this RNA binding protein in governing gene expression in plants.

To unravel the possible role(s) of OsTudor-SN during rice endosperm development, we studied its biochemical and cellular properties. OsTudor-SN exists as RNase-sensitive species that are found as stable in vivo complexes containing prolamine and glutelin RNAs. Using indirect immunofluorescence techniques, OsTudor-SN was observed as two forms: one present as particles and patches of fluorescence in the cytoplasm, and a second form tightly associated with microtubules. Several OsTudor-SN particles were observed in close contact with prolamine protein bodies, and were shown to co-localize with GFP-tagged prolamine RNA particles. Consistent with a role in RNA transport, GFP-tagged OsTudor-SN was observed as moving particles, the dynamic properties of which were dependent on intact actin filaments. Data from studies of antisense and RNAi plants are also consistent with a role for OsTudor-SN in prolamine biogenesis, with suppression in prolamine gene expression resulting in a reduction in the number, but not in the size, of mature prolamine protein bodies. Overall, our results support a role of this multifunctional RNA binding protein in one or more steps (transport, targeting and anchoring) of storage protein RNA localization in developing rice endosperm.

Results

Os Tudor-SN is present as a ribonucleoprotein complex containing storage protein RNAs

We had previously shown that OsTudor-SN displayed a polydispersed sedimentation behavior on sucrose density gradients, indicating that this RNA binding protein exists as a population of species of varying molecular weights (Sami-Subbu et al., 2001). OsTudor-SN was shown to interact with several different RNAs, including prolamine and glutelin RNAs, although interaction with prolamine RNAs was specific, as it bound the 3′-UTR, but not the 5′-UTR, coding or polyA sequences. To further characterize these biochemical properties, we examined the basis for its heterogeneous molecular size distribution. An extract from a crude cytoskeletal-enriched fraction was resolved by gel filtration chromatography, and fractions were analyzed for OsTudor-SN by immunoblotting. OsTudor-SN eluted broadly from the void volume at 2000 kDa to fractions of 158 kDa (Figure 1a). Pre-treatment of the post-cytoskeleton supernatant fraction with RNase resulted in a dramatic change in the elution profile, where OsTudor-SN was observed to elute with a molecular size of about 240 kDa, a molecular weight consistent with an OsTudor-SN dimer. These results show that OsTudor-SN is a component of RNase-sensitive ribonucleoprotein complexes, which exhibit considerable variation in molecular size.

Figure 1.

OsTudor-SN occurs as higher molecular weight complexes, and co-sediments with polysomes.
(a) The distribution of Os Tudor-SN before and after treatment of the post-cytoskeleton-enriched fraction supernatant with RNase, in fractions resolved by Sephacryl S300 chromatography. In addition to disrupting the large molecular complexes of OsTudor-SN to about 230 kDa (a dimer), the RNA binding protein was also more prone to degradation, forming a major product at 80 kDa (major lower band) and 40 kDa (not shown).
(b) Sucrose density gradient centrifugation of polysomes from developing endosperm. Fractions from the gradient were analyzed for OsTudor-SN by immunoblotting.

Abe et al. (2003) suggested that a fraction of the pea ortholog Tudor-SNc was bound to ribosomes. We therefore examined the possible association of OsTudor-SN with polysomes, which would account for the elution of OsTudor-SN as very large molecular weight complexes eluting in the void fractions of the gel filtration column. Figure 1(b) shows the sedimentation profile of polysomes resolved by sucrose density gradient centrifugation. Although the bulk of the OsTudor-SN remains in the upper soluble gradient fraction, detectable quantities of this RNA binding protein co-sediments with polysomes, suggesting a possible role in protein synthesis.

To assess whether these OsTudor-SN associated ribonucleoprotein complexes contain storage protein RNAs, coupled immunoprecipitation (IP) RT-PCR studies were carried out. To ensure specificity of our IP assays, we employed the use of transgenic pTO242 plants, which express a GFP–OsTudor-SN fusion and, thereby, enabled us to employ affinity-purified anti-GFP, which recognizes antigenic sites devoid in wild-type plants. Figure 2a depicts the polypeptides specifically immunoprecipitated by anti-GFP, as revealed by silver staining of an SDS polyacrylamide gel. In pTO242 transgenic plants expressing GFP–OsTudor-SN, but not in wild-type plants, a prominent high molecular band of ∼150 kDa is readily evident. This is the predicted size of the GFP–OsTudor-SN fusion protein, which was verified by immunoblot analysis using anti-OsTudor-SN and anti-GFP (Figure 2a). Other prominent polypeptide bands at 55 and 26 kDa are caused by IgG heavy and light chains.

Figure 2.

OsTudor-SN binds to both prolamine and glutelin RNAs in vivo.
(a) SDS-PAGE (left panel) and immunoblots (middle and right panels) of GFP–OsTudor-SN complexes immunoprecipitated (IP) by anti-GFP. The left panel depicts a silver-stained polyacrylamide gel of IP complexes from wild-type (WT) and pTO242 (242) extracts, whereas the middle and right panels depict immunoblots of wild-type and pTO242 input, and IP samples using anti-OsTudor-SN and anti-GFP, respectively.
(b) RT-PCR of RNA isolated from immunoprecipitated complexes. After reverse transcription, the cDNA was subjected to PCR using primers specific for prolamine and glutelin RNAs.
(c) RT-PCR for prolamine, ubiquitin, actin and 18S RNA from input RNA and IP complexes from the pTO242 transgenic line.

RNA from the IP material was subjected to RT-PCR using prolamine- and glutelin-specific primers. Figure 2(b) shows that prominent PCR products were observed for both prolamine and glutelin RNAs, following IP from extracts of pTO242-expressing plants, with very little signal observed for the wild-type controls. This result was reproducible and the yields were linear across multiple experiments, and we consistently observed a 20–50-fold enrichment for both prolamine and glutelin between the experimental and control PCR reactions. By contrast, RNAs for ubiquitin, actin and 18S RNA were not specifically immunoprecipitated under these conditions (Figure 2c). Collectively, these results indicate that OsTudor-SN exists as stable in vivo ribonucleoprotein complexes containing storage protein RNAs.

Os Tudor-SN is present as particles in the cytoplasm

To determine the intracellular localization of OsTudor-SN, indirect immunofluorescence studies using confocal microscopy were carried out (Figure 3). At low magnification, OsTudor-SN is observed as small fluorescent foci or particles, as well as irregular-shaped patches. As we had previously demonstrated that prolamine protein bodies are located predominantly in the cytoskeleton-enriched cortical region of the cell (Muench et al., 2000), we examined the spatial relationship between prolamine PBs and OsTudor-SN using affinity-purified anti-Tudor in endosperm sections from wild-type plants. Prolamine PBs were specifically labeled with rhodamine hexyl ester or antisera against BiP, a lumenal chaperone asymmetrically concentrated in the lumen of prolamine PBs (Muench et al., 1997).

Figure 3.

 The distribution of OsTudor-SN in 12-day-old rice developing endosperm.
OsTudor-SN was visualized by indirect immunofluorescence using anti-Tudor. Note the distribution of fluorescent particles and weaker patches of fluorescence.

Immunofluorescence analysis showed that OsTudor-SN particles were readily observed in the same focal plane as prolamine PBs. Closer examination showed that OsTudor-SN was associated with the cisternal ER and prolamine PBs (Figure 4a–d). Several of the prolamine PBs were nearly surrounded by small particles or speckles of Tudor-SN, whereas, in other instances, only a portion of the surface of the prolamine PB was covered (Figure 4a–d). This close spatial association was verified by three-dimensional reconstructions of a series of optical sections (see Video Clip S1). Likewise, the distribution of OsTudor-SN was not uniform on the cisternal ER, but was instead observed as patches (Figure 4a–d).

Figure 4.

 Spatial relationship between OsTudor-SN and prolamine protein bodies (PBs).
(a, c) The distribution of OsTudor-SN was visualized by indirect immunofluorescence using anti-Tudor.
(b, d) The same sections as in (a and c, respectively), but also depicting the distribution of prolamine PBs visualized with rhodamine hexyl ester (red).
(e) Immunofluorescence of the GFP–OsTudor-SN fusion protein using anti-GFP (mouse).
(f) Same section as in (e), but also depicting the distribution of prolamine PBs labeled with BiP antibodies.

To verify the close physical association between this RNA binding protein and prolamine PBs, we examined the distribution of GFP–OsTudor-SN in transgenic pTO242 plants using anti-GFP as a probe (Figure 4e,f). Nearly identical images were captured, as observed earlier for wild-type endosperm (Figure 4a,b), supporting a close spatial relationship to prolamine PBs and cisternal ER, and indicating that the addition of GFP to OsTudor-SN does not affect its function. Again, several of the prolamine PBs can be seen surrounded by small speckles of OsTudor-SN of varying fluorescence intensity. The binding of storage protein RNAs by OsTudor-SN particles, and its close physical association with prolamine PBs and the cisternal ER, supports a role for this RNA binding protein in RNA localization.

In addition to a high-molecular-weight form, biochemical evidence supported the existence of a second population of cytoskeletal-associated OsTudor-SN. Despite using conditions that were optimized to preserve these cytoskeletal structures in our tissue sections, the presence of intact actin filaments and microtubules was rarely detected. In the few instances where these cytoskeletal elements were preserved, OsTudor-SN was found to be dispersed among actin filaments, and appeared to essentially coat microtubules (Figure S1). This tight association was also evident when the microtubules were broken into small bundles by the depolymerizing agent oryzalin. Because of our inability to consistently preserve microtubules in developing endosperm sections, the relationship of this RNA-binding protein with microtubules was not further studied. Overall, these observations support our earlier conclusions that OsTudor-SN exists in two forms: one tightly associated with the cytoskeleton (microtubules), and a second smaller form dispersed as particles or small patches to the cytoskeletal-enriched cortical ER region of the cell.

Cytological evidence that Os Tudor-SN is a possible component of the RNA transport particle

We had previously demonstrated that prolamine RNAs are transported as particles to the prolamine PBs when viewed with the dual gene-hybrid GFP system (Hamada et al., 2003a). Because the OsTudor-SN particles observed in Figures 3 and 4 were similar in size to GFP-tagged prolamine RNA transport particles (∼0.5 μm in diameter), and were present as RNase-sensitive ribonucleoprotein SN complexes containing storage protein RNAs (Figure 2), we set out to determine whether OsTudor-SN was a component of the RNA transport particle. Double indirect immunofluorescence studies were carried out with seed sections prepared from a transgenic plant expressing a transcriptional fusion of 6XMS2-prolamine RNA sequences and a translational fusion of MS2 coat protein-GFP. Under these conditions, the MS2 coat protein–GFP fusion would bind to the MS2-prolamine RNA, enabling one to assess the location of fluorescent RNA-containing particles. Figure 5 depicts the immunofluorescence detection of GFP (panel A) and OsTudor-SN (panel B) of a developing endosperm cell from a transgenic plant expressing the GFP-tagged RNA transport system. GFP labels are present as small particles that are distributed mainly in the cortical (peripheral) region of the cell (panel A). OsTudor-SN show a similar distribution pattern (panel B), and merging the two images (panel C) readily confirms that many OsTudor-SN particles are coincident with GFP labeling, although the labeling distribution pattern for OsTudor-SN is much more extensive than that seen for GFP. This partial overlap of GFP and OsTudor-SN is also evident in images at higher magnification (Figure 5d–f).

Figure 5.

 Spatial relationship between GFP-prolamine RNA transport particles and OsTudor-SN.
Double indirect immunofluorescence analysis was conducted on endosperm sections from a transgenic plant expressing GFP-prolamine RNA transport particles using anti-GFP (a, d) and anti-Tudor (b and e).
Panel c is a merged image of Panels a and b.
Panels a–c depict the labeling patterns of the midsection of a single cell from 12-day-old developing endosperm.
Panels d–f are a close-up view of the cortical region. Note that GFP fluorescence largely overlaps with the labeling of OsTudor-SN, although the latter labeling pattern is much more extensive.

The co-localization of the small GFP-labeled RNA particles with OsTudor-SN suggests that this RNA binding protein participates in RNA transport. If true, then the OsTudor-SN particles seen in Figures 3–5 would be expected to move with properties similar to those observed for RNA transport particles. Indeed, analysis of live developing endosperm sections from a transgenic plant expressing GFP–OsTudor-SN showed the existence of particles with movement behavior similar to that seen earlier for prolamine RNA transport particles. Figure 6 captures the movement of two GFP–OsTudor-SN particles that travel about 20 μm over a 3-min interval, while other particles remain stationary during this time frame. In the 60-sec frame, one of the particles moves out of the focal plane, but then returns to the same xy planar view when captured at 120 sec. Overall, the particles were observed to move in a stop-and-go fashion, reminiscent of the GFP-tagged RNA transport particles observed in our earlier study (Hamada et al., 2003a).

Figure 6.

 Movement of GFP–OsTudor-SN in live sections of developing rice endosperm.
Snapshots taken at 1-min intervals depict the movement of a pair of GFP–OsTudor-SN particles (arrows). Note that at 60 and 180 sec, only one of the particles is observed, as the other moved out of the focal plane. Scale bar = 5 μm.

Movement of GFP-tagged prolamine RNA transport particles was dependent on intact actin filaments (Hamada et al., 2003a). To determine the role of the cytoskeleton in the movement of GFP–OsTudor-SN particles, freehand sections of developing endosperm were incubated with inhibitors of microtubules and actin filaments for 15 min, and were then assessed for GFP–OsTudor-SN particle movement. In most instances, movement of GFP fluorescence was observed for up to 30 min under control conditions, and in the presence of the microtubule inhibitors oryzalin and nocodazole (Figure S2). However, pre-incubation of the endosperm slices with latrunculin B prevented GFP movement within 15 min or less. Cytochalasin D also inhibited movement, although it was not as effective as latrunculin B. Overall, these results indicate that OsTudor-SN forms particles that move only when microfilaments remain intact.

The relationship between Os Tudor-SN and prolamine gene expression

The preceding results indicate that OsTudor-SN is located in the cortical region of the cell and binds storage protein RNAs, and may serve as a component of the RNA transport particle. To determine whether this RNA binding protein is essential for prolamine gene expression, sense (overexpression) and antisense OsTudor-SN plants were generated. Figure 7 shows the results of an immunoblot of protein extracts prepared from mature seeds from these transgenic plant lines, as well as wild-type plants. The two antisense lines showed significant decreases in OsTudor-SN and prolamine polypeptides compared with wild type (Figure 7a). By comparison, ADPglucose pyrophosphorylase small subunit antigen levels were identical among these plant lines, ensuring that equal quantities of protein were used in the analysis. Consistent with the decline in prolamine polypeptide levels, northern analysis showed a significant decline in their steady state RNA levels (Figure 7b). A significant increase in prolamine RNA is also evident in the sense OsTudor-SN plants, although increases in corresponding polypeptide levels were only incremental. The close correlation between OsTudor-SN and prolamine levels suggests that OsTudor-SN is required for optimal prolamine gene expression.

Figure 7.

 Coordinate gene expression of OsTudor-SN and prolamine in sense (pMKM26) and antisense (Anti-, pMKM25) OsTudor-SN transgenic rice plants.
(a) Immunoblot analysis of OsTudor-SN, ADPglucose pyrophosphorylase (AGP) small subunit and prolamine in wild-type (WT), sense (2601) and antisense (2701, 2709 and 2711) OsTudor-SN plants. The arrow and arrowhead in the ‘sense’ lane denotes the intact GFP–OsTudor-SN and truncated fusion proteins, respectively, present in the transgenic pMKM26 plant. OsTudor-SN (120 kDa) is prone to proteolysis, which produces polypeptides of 80 and 40 kDa, the latter is not detected in this immunoblot using anti-Tudor.
(b) Northern analysis of prolamine RNAs in WT, sense (2601 and 2605) and antisense (2711 and 2701) OsTudor-SN transgenic rice plants. The upper panel depicts the levels of prolamine RNAs as assessed by hybridization of the 32P-labeled prolamine cDNA probe. The lower panel depicts the level of ribosomal RNA in each lane of the gel, as visualized by ethidium bromide staining. The data depicted from panel B was taken from a single agarose gel that was subjected to northern blotting.

Because the antisense plants did not exhibit reduced OsTudor-SN expression in subsequent generations, RNAi plants were created and evaluated. The RNAi plants showed a pronounced reduction in OsTudor-SN transcripts in 12–14-day-old developing seeds when measured by RT-PCR (Figure 8a). Likewise, there was a marked reduction in prolamine RNA levels, with the extent of gene suppression varying depending on the RNAi line analyzed. By contrast, RNAs for protein disulfide isomerase (PDI), hygromycin phosphotransferase and ubiquitin were unaffected. This reduction in prolamine RNA levels was also evident at the protein level, where the relative levels of prolamine polypeptide levels correlated with their RNA levels (Figure 8b). In addition, glutelin polypeptides were also significantly reduced in the RNAi plants, whereas PDI levels were unaffected (Figure 8b).

Figure 8.

 The effect of OsTudor-SN RNAi suppression (a and b) and overexpression (c) on prolamine gene expression.
(a) RNA from ∼12-day-old developing seeds from wild-type and transgenic rice plants expressing OsTudor-SN RNAi construct driven by the glutelin Gt1 promoter (Gt1-7 and Gt1-15) or Cauliflower mosaic virus (CaMV) 35S promoter (35S-6) were isolated and subjected to RT-PCR for the estimation of RNA transcripts for OsTudor-SN, PDI, hygromycin phosphotransferase, prolamine and ubiquitin. The lower two panels in (a) depict the results of a northern blot of prolamine RNA, and corresponding levels of ribosomal RNA in samples analyzed by Northern blot.
(b) Immunoblot analysis of wild-type and OsTudor-SN RNAi plant lines. Twenty micrograms of protein from ∼12-day-old developing seeds from wild-type and transgenic rice plants were subjected to immunoblot analysis using antibodies specific for PDI, prolamine and glutelin precursor. In addition to prolamine, glutelin protein levels were also lower in the RNAi plant lines. Lanes: a, wildtype; b, Gt1-7; c, Gt1-15; d, 35S-6; e, 35S-11.
(c) Enhancement of prolamine gene expression in pTO242 transgenic plants overexpressing GFP–OsTudor-SN. Note the increase in both prolamine RNA and protein in the pTO242 transgenic plant compared with the wild-type plant. By contrast, PDI protein levels are nearly identical in the two plant lines.

Microscopic analysis of 12-day-old developing seeds from one RNAi plant, Gt1-7, which showed the strongest reduction in prolamine polypeptide levels, revealed that this plant line contained a significant reduction in the number of prolamine-containing protein bodies (Figure S3). The RNAi rice line contained fewer protein bodies (up to 70%), with no significant difference in their size compared with those seen in the wild-type endosperm.

An apparent anomaly is that although prolamine and glutelin expression were lower in the RNAi plants, there remained significant expression of these storage proteins even when OsTudor-SNs were absent or were at very low levels. A closer examination of OsTudor-SN expression during seed development showed that significant quantities of this RNA binding protein are present in 8-day-old developing seeds, with subsequent suppression and significantly lower levels following as the seed matures (Figure S4), thereby accounting for its apparent absence in older seeds (Figure 8b). Hence, the significant expression of OsTudor-SN in very young developing seeds, a period where storage protein synthesis accelerates, may be sufficient to support storage protein biosynthesis during this period.

To further explore the relationship between OsTudor-SN and prolamine gene expression, we studied developing seeds from pTO242 plants expressing GFP–OsTudor-SN protein fusions, which was expressed at comparable levels to endogenous OsTudor-SN (Figure 8c). These plants contain significantly higher levels of prolamine RNAs and increased levels of prolamine polypeptides. As a control, PDI levels remain unaltered. Overall, these results show that the expression of OsTudor-SN and prolamine is coordinately regulated. The increase in prolamine gene expression in the overexpressed pTO242 line indicates that OsTudor-SN is a limiting factor in prolamine gene expression.

Discussion

Tudor-SN orthologs play diverse roles in gene expression in animal cells. In addition to its initial discovery as a transcriptional co-activator (p100) in mammals, it is a major component of RISC, and of the nucleolytic activity that degrades hyperedited RNAs. Although Tudor-SN is readily evident in higher plants (Abe et al., 2003; Chuong et al., 2004; Sami-Subbu et al., 2001), nothing is known about its role during plant growth and development.

OsTudor-SN contains a putative nuclear localization peptide signal, which suggests a possible nuclear function. Indeed, its initial discovery as a transcriptional co-activator (Tong et al., 1995) validates this intracellular location for this protein. However, several other studies support a dominant or solely cytoplasmic localization of this protein. In lactating mammalian cells, it is present on the ER and in lipid droplets (Keenan et al., 2000), whereas it is localized exclusively in the cytoplasm in Xenopus oocytes (Scadden, 2005). The available microscopic evidence in this study indicates that this RNA binding protein is non-nuclear in origin, and is readily conspicuous in the cytoplasm, and specifically in the cortical region of the developing endosperm cell. This dominant cytoplasmic location was seen for the cytoskeleton-associated pea homolog, Tudor-SNc (Abe et al., 2003). If present in rice endosperm nuclei, it is present at levels below the detection limits of immunofluorescence microscopy. The distribution of OsTudor-SN in the cortical region of the cell, and the apparent absence from the nucleus, indicates that its dominant role in developing rice endosperm entails cytoplasmic processes, and not transcription or nuclear export.

When viewed by immunofluorescence microscopy, OsTudor-SN exists as particles and patches that are dispersed throughout the cortical region of the cell. This subcellular distribution was evident when OsTudor-SN antibodies were employed in the analysis of wild-type developing endosperm, or when using antibodies to GFP in the study of transgenic developing endosperm expressing a GFP–OsTudor-SN protein fusion. The OsTudor-SN particles are of similar size to GFP-tagged prolamine RNA transport particles, which move along actin filaments via myosin-based motor proteins (Hamada et al., 2003a). The overall evidence described in this study clearly supports the view that OsTudor-SN is a component of RNA transport particles (Figure 4), and is involved in transport, localization and/or anchoring of storage protein RNAs. First, soluble OsTudor-SN is present as RNase-sensitive ribonucleoprotein complexes. Second, OsTudor-SN can be isolated as stable ribonucleoprotein complexes with bound storage protein RNAs. Third, double indirect immunofluorescence studies of endosperm sections from transgenic plants expressing the GFP-RNA transport system, where the MS2 coat protein–GFP protein fusion is complexed to MS2-prolamine RNAs, showed that GFP-labeled prolamine RNA particles extensively co-localize with OsTudor-SN particles. Lastly, OsTudor-SN particles visualized as GFP fusions move with a path projection, stop-and-go movement and variable speed properties that are reminiscent of those observed for GFP-labeled prolamine RNA particles (Hamada et al., 2003a). In addition, similar to that observed for the GFP-labeled prolamine RNA transport particles, the movement of GFP–OsTudor-SN particles was dependent on intact actin filaments.

The role of this RNA binding protein in prolamine RNA transport and localization is also supported by its spatial relationship to prolamine protein bodies. OsTudor-SN particles can be observed in close proximity to the prolamine protein bodies, and can be seen in physical contact with these structures (Figure 4). In many of these instances, this RNA binding protein is observed as small speckles on the protein body surface, when three-dimensional images are reconstructed (Video Clip S1). Although less frequently observed, OsTudor-SN particles appear to completely encircle the prolamine PB (Figure 4). In live endosperm sections, GFP–OsTudor-SN particles can be seen to move to or close to prolamine PBs (results not shown). Overall, these cytological observations (Figures 2–5), together with the existence of storage protein RNA OsTudor-SN complexes (Figure 1), support a role for this RNA binding protein in binding prolamine RNAs and transporting them to the PB-ER.

In addition to the PB-ER, OsTudor-SN is also distributed on the cisternal ER, with some areas containing higher concentrations of this RNA binding protein, as indicated by the varying intensity of the fluorescence signals over this membrane complex. The association of this RNA binding protein with both PB-ER and cisternal ER, and its presence as stable in vivo ribonucleoprotein complexes with prolamine and glutelin RNAs, indicates that OsTudor-SN is not a trans-element for specific RNAs, but is more likely to be a general factor required for the transport and localization of storage protein RNAs to the cortical ER.

Antisense and RNAi transgenic plants expressing low levels of OsTudor-SN were generated, and showed a significant decrease in prolamine gene expression at both the RNA and protein levels, whereas the expression of other genes (e.g. PDI, ADPglucose pyrophosphorylase and ubiquitin) was unaffected (Figures 7 and 8). Hence, OsTudor-SN influences the expression of only a subset of genes in the rice endosperm. Glutelin polypeptide levels were also lower in the RNAi plants (Figure 8b), indicating that this RNA binding protein also functions in controlling the expression of this storage protein. The in vivo binding of OsTudor-SN to glutelin RNAs supports this view.

Immunoblot analysis shows that OsTudor-SN attains maximum steady state levels in 7-day-old developing seeds, and that these levels are sustained in later stages of seed development (Sami-Subbu et al., 2001). This early temporal gene expression pattern accounts for the significant levels of OsTudor-SN in 6- and 8-day-old seeds of the RNAi plants, which then show a significant drop-off in expression as the seed matures. The delay in RNAi suppression of OsTudor-SN expression presumably results from the use of the glutelin Gt1 promoter, which was used to drive expression of the antisense and RNAi gene constructs of OsTudor-SN. Efforts to identify an active promoter of the 5′ flanking region of the OsTudor-SN gene have been unsuccessful (data not shown).

Overexpression of OsTudor-SN as a GFP fusion elevated the steady state levels of prolamine RNAs and polypeptides, indicating that this RNA binding protein is a limiting factor in prolamine gene expression. The dependence on the relative levels of OsTudor-SN for optimal storage protein gene expression supports a non-catalytic (non-enzymatic) structural role for OsTudor-SN in this process. Overall, the direct relationship between OsTudor-SN and storage protein gene expression indicates that OsTudor-SN is an important factor for storage protein gene expression, possibly as an essential component of RNA transport and localization.

Microscopic studies of seeds from one RNAi plant, Gt1-7, which showed the strongest reduction in prolamine polypeptide levels, revealed that this plant line contained significantly less (∼70%) prolamine-containing protein bodies (Figure S3). This reduction in protein bodies is an adverse consequence mediated by the decrease in OsTudor-SN-dependent RNA transport. In developing seeds of 8-days old or younger, the available OsTudor-SN would mediate the transport and localization of prolamine RNAs to the protein body ER, whereupon active translation enables the formation and maturation of prolamine protein bodies. Between 8 and 10 days, when OsTudor-SN levels dramatically fall, the transport, localization and anchoring of new prolamine RNA transcripts to the PB-ER and their subsequent translation would not occur, resulting in the absence of new protein body formation, and thereby accounting for the reduction in prolamine protein bodies. Prolamine RNAs that are not transported to the cortical region of the cell are likely to be degraded, resulting in a decrease in the steady state levels, as is readily observed in the OsTudor-SN antisense and RNAi plant lines. If OsTudor-SN was not involved in prolamine RNA transport, but was instead required for its transcription or transcript stability, one would expect a decrease in the size of the prolamine protein bodies, but not in their number.

In addition to its suggested role in storage protein RNA transport and localization, OsTudor-SN may also be involved in other cellular processes. OsTudor-SN co-sediments with polysomes (Figure 1b), which may infer an additional role in protein synthesis, possibly by activating translation, as suggested by recent studies on the fragile X mental retardation protein (FMRP). OsTudor-SN and FMRP share several structural and functional properties. Like OsTudor-SN, FMRP has a tudor domain (Maurer-Stroh et al., 2003), binds polysomes, and has been suggested to play diverse roles in the transport and localization of a subset of mRNAs and their subsequent translation in dendrites (Miyashiro et al., 2003; Zalfa et al., 2003). This latter role in regulating translation has been recently demonstrated for the mammalian FMR-like protein. This RNA binding protein, together with Argonaute 2, upregulates the translation of tumor necrosis factor α RNA by interacting with the AU-rich elements located in the 3′-UTR (Vasudevan et al., 2007). Interestingly, OsTudor-SN binds specifically to the prolamine 3′-UTR, but not to 5′-UTR, coding or polyA sequences, suggesting that it binds to a specific cis-element in the prolamine 3′-UTR. As OsTudor-SN also binds glutelin RNAs, the cis-element is likely to be highly conserved and common in 3′-UTRs. The ubiquitous 3′-UTR AU-rich elements, which stimulate translation directly or indirectly via RNA stability, satisfy these requirements.

In addition to its presence in individual particles, OsTudor-SN is tightly associated with microtubules (Figure S1). This co-localization of OsTudor-SN with microtubules supports our earlier conclusion that this RNA binding protein is cytoskeleton-associated, based on its sedimentation properties under different ionic strength conditions (Sami-Subbu et al., 2001). The Arabidopsis Tudor-SN was found to be a tubulin-binding protein (Chuong et al., 2004). Microtubules appear to be a site rich in RNA binding proteins, including polyadenylation binding protein, RNA helicases, translational factors and ribosomal proteins (Chuong et al., 2004). The microtubule association of this latter group of proteins is consistent with the suggested role of the cytoskeleton in regulating translational efficiency (Davies et al., 1996; Jansen, 1999).

Collectively, our biochemical and cytological data demonstrate that OsTudor-SN plays a fundamental role in the transport of storage protein RNAs, and in regulating the expression of the proteins they encode. The modular domain structure of OsTudor-SN not only supports its role in binding storage protein RNAs, but also supports its role as a scaffolding protein in interactions with other RNA binding proteins or accessory proteins for the formation, transport, localization and/or anchoring of the RNA transport particle. Studies to elucidate these protein interactions, which will identify other components of the RNA transport and localization machinery, are underway.

Experimental procedures

Plant material and immunofluorescence labeling

Free-hand sections of endosperm tissue (300-μm thick) from developing rice (Oryza sativa cv. Kitaake) seeds, 6–15-days old, were subjected to immunolabeling and protein body labeling with Rhodamine B hexyl ester, as described previously (Muench et al., 2000). Antibodies to α-tubulin (N356; Amersham, http://www.amersham.com), rice OsTudor-SN (rabbit), Tudor domain of OsTudor-SN (rabbit) and GFP (mouse and rabbit) were used at a dilution of 1/200. The secondary antibodies, Alexa 488- or 595-conjugated anti-mouse antibodies and Alexa 594- or fluorescein-conjugated anti-rabbit secondary antibodies (Molecular Probes, http://probes.invitrogen.com), were used at 1/300 dilutions.

For imaging GFP native fluorescence in live endosperm tissue, the sections (< 500 μm) were mounted on microscope slides and bathed in N6 medium (Caisson Laboratories, Inc., http://www.caissonlabs.com) containing 100 mg l−1 myo-inositol, 300 mg l−1 casamino acid, 2.8 g l−1 l-proline and 2% sucrose.

Confocal microscopy was carried out on a Zeiss 410 series laser scanning confocal microscope (http://www.zeiss.com) or a Bio-Rad View Scan DVC-250 laser scanning confocal microscope (http://www.bio-rad.com) using the fluorescein and rhodamine filter sets. Image processing was performed using Adobe Photoshop software (http://www.adobe.com).

Analysis of OsTudor-SN by gel filtration

A cytoskeletal-enriched protein fraction was obtained by extracting a crude cytoskeletal pellet (Sami-Subbu et al., 2001), prepared from 13.5 g of developing seeds, in 3 ml of buffer containing 25 mm Tris–HCl, pH 7.5, and 200 mm NaCl. After incubation on ice overnight, the solution was centrifuged at 21 000 g for 10 min and was then fractionated onto a Sephacryl S-300 column (50 × 1.5 cm) at 4°C. A nuclease-treated sample was prepared by incubating the sample with 100 μg of RNase A at room temperature (21–23°C) for 30 min before fractionation on the gel filtration column. Fractions were collected and analyzed by immunoblot analysis using anti-OsTudor-SN sera.

Sucrose density gradient centrifugation of polysomes

Polysomes were extracted from a cytoskeletal-enriched protein body fraction, and were resolved on a 10–40% sucrose density gradient, as described previously (Muench et al., 1998).

Construction and expression of antisense, sense and RNAi Os Tudor-SN and GFP– Os Tudor-SN plants

The translational stop of mGFP from pBIN mGFP-ER (Haseloff, 1999) was removed by amplifying the coding sequence with 5′ sense CGGGATCCAAGGAGATATAACAATG and 3′ antisense CTTACTCCCGGGTTTGTATAGTTCATCCATGCC primers. The DNA was then digested with BamHI and SmaI (present at the 5′ and 3′ primer sequences), and was then inserted into pYW502 (Hamada et al., 2003b) to give pMKM25. The open reading frame sequences of OsTudor-SN cloned into the EcoRV site of pSL1180 were removed by EcoRV digestion, and were inserted into SmaI-digested pMKM25 to give pMKM26 (sense orientation) and pMKM27 (antisense orientation). A second DNA sense construction lacking the chitinase signal peptide was obtained by amplification of the mature coding sequence for GFP–OsTudor-SN containing BamHI termini. The resulting DNA was cloned into the BamHI sites of pBluescript II, and then later was cloned into pYW502 between the Gt1 promoter and 3′nos terminator sequences (Hamada et al., 2003b).

An RNAi plasmid to downregulate expression of OsTudor-SN was generated by cloning the Tudor domain (221-bp fragment starting from nucleotide 2329 of the coding sequence) using Gateway technology (Invitrogen, http://www.invitrogen.com). The specific primers used were GGGGACAAGTTTGTACAAAAAAG, GGGGAGGAGTTTGTACAAGAAAG, ACAAAAAAGCAGGCTaatcctgtgaaggggga and ACAAGAAAGCTGGGTgctgcacaactgagcaa, where the capital letters are homologous to Gateway attB1/attB2, whereas the lower case letters are complimentary to the Tudor sequence. The RNAi gene sequences separated by the Chloramphenicol resistance gene were then cloned into pYW502 containing either the constitutive Cauliflower mosaic virus (CaMV) 35S promoter or the endosperm-specific glutelin-1 (Gt1) promoter.

Analysis of RNAi transgenic plants by RT-PCR

RNA samples were prepared from ∼100 mg of 12-day-old developing seeds using a modified guanidine extraction buffer (Li & Trick 2005). The total RNA yield was about 200 μg per 100 mg of developing seed. The 12-day-old developing seeds from RNAi plants were smaller than wild type (∼25 mg per seed), in the range of 15, 20 and 17 mg for Gt1-7, Gt-15 and 35S-6, respectively, per seed. Total RNA (1 μg) was reversed transcribed using MLV Reverse Transcriptase (M1701; Promega, http://www.promega.com), and then 1/20th of the reaction was used for RT-PCR with the following gene-specific primers: ttagtgctgacaactcctgg and gatagactgctgcttcatgg for OsTudor-SN (pos. 2575–2823); gagctaagcaagcacgatcc and ttgcgtcttctggtgactt for PDI (pos. 301–538); tctccgacctgatgcagctc and cgcgaccggttgtagaacag for hygromycin phosphotransferase (pos. 65–348); aggcagcagtatggcatagc and attacaagacaccgccaagg for prolamine (pos. 358–655); and aaggctaagatccaggac and ggttcaacaacatccagg for ubiquitin (pos. 106–509). The PCR conditions were 95°C for 2 min followed by 30 cycles of 95°C for 15 sec, 60°C for 30 sec and 72°C for 30 sec.

Northern blot analysis

About 20 μg of total RNA was analyzed by northern blot analysis using HyBond N+ nylon membrane. The prolamine 7 specific DNA probe was labeled using DIG High Prime DNA labeling kit (Roche, http://www.roche.com). Detection of hybridized digoxygenin-labeled cDNA was accomplished using the PCR DIG Labeling kit (Roche Applied Science).

Protein extraction and immunoblot analysis

The protein yield from 100 mg of 12-day-old developing seeds varied, with ∼300 μg of protein obtained from wild-type endosperm, whereas Gt1-7, Gt1-15 and 35S-6 contained 180, 250, and 200 μg, respectively. Twenty micrograms of protein for each sample was subjected to SDS-PAGE and immunoblot analysis using specific antibodies against OsTudor-SN, Tudor, PDI, ADPglucose pyrophosphorylase small subunit, prolamine or glutelin. Immunoreactions were detected using the horseradish peroxidase conjugated secondary anti-rabbit and the SuperSignal West Pico reagent, (Pierce, http://www.piercenet.com).

RNA immunoprecipitation

Freshly harvested mid-stage developing rice seeds were sectioned and then fixed for 10 min in 10 ml 0.5% formaldehyde in mCSB (Abe et al., 2003) at room temperature. After washing, the sections were homogenized in 10 volumes of IP buffer (0.4 g in 4 ml) supplemented with 0.1 mm PMSF and 20 U μl−1 RNase inhibitor, and were then centrifuged at 15 000 g for 10 min. Typically, 1 ml of the resulting supernatant was incubated with 5 μl of affinity purified anti-GFP- protein A resin. The resin was then washed five times with 1 ml IP buffer, twice with 1 ml DNase buffer (10 mm Tris-HCl, pH 7.5, 2.5 mm MgCl2, 0.1 mm CaCl2) and was then incubated with 0.25 μl RNase inhibitor and 0.2 μl DNase (1 U μl−1) at 37°C for 30 min. The sample was then treated with 0.5% SDS and 20 μg proteinase K for 1 h at 60°C. Total RNA was obtained using five volumes of Trizol (Sigma-Aldrich, http://www.sigmaaldrich.com) according to the manufacturer’s protocol. Total RNA was resuspended in 5 μl of water, and was subjected to reverse transcription using oligo dT primers in a total volume of 10 μl for 1 h at 37°C. PCR was performed using either prolamine (AK240910-107f 5′-GTCAAAGTTATAGGCAATATCAGCTGC-3′ and AK240910-428r 5′-CCTAGGGTAGATACCATATCTAGATGGC-3′) or glutelin (AK242245-2080f 5′-GTACCGGAGTATCTGTTGTCCGTC-3′ and AK242245-2502r 5′-CTCACGCCTGTATGCTTGAGG-3′) primers. Prolamine cDNA was amplified with 20 cycles, and glutelin RNA was amplified with 30 cycles.

Acknowledgements

This project is supported by the National Research Initiative (NRI) Plant Biology: Gene Function and Regulation of the USDA Cooperative State Research, Education and Extension Service (CSREES) grant numbers 2003-35301-13270 and 2006-35301-17043.

Ancillary