Characterization of RNA binding protein RBP-P reveals a possible role in rice glutelin gene expression and RNA localization
- First Online:
- Cite this article as:
- Doroshenk, K.A., Tian, L., Crofts, A.J. et al. Plant Mol Biol (2014) 85: 381. doi:10.1007/s11103-014-0191-z
- 533 Views
RNA binding proteins (RBPs) play an important role in mRNA metabolism including synthesis, maturation, transport, localization, and stability. In developing rice seeds, RNAs that code for the major storage proteins are transported to specific domains of the cortical endoplasmic reticulum (ER) by a regulated mechanism requiring RNA cis-localization elements, or zipcodes. Putative trans-acting RBPs that recognize prolamine RNA zipcodes required for restricted localization to protein body-ER have previously been identified. Here, we describe the identification of RBP-P using a Northwestern blot approach as an RBP that recognizes and binds to glutelin zipcode RNA, which is required for proper RNA localization to cisternal-ER. RBP-P protein expression coincides with that of glutelin during seed maturation and is localized to both the nucleus and cytosol. RNA-immunoprecipitation and subsequent RT-PCR analysis further demonstrated that RBP-P interacts with glutelin RNAs. In vitro RNA–protein UV-crosslinking assays showed that recombinant RBP-P binds strongly to glutelin mRNA, and in particular, 3′ UTR and zipcode RNA. RBP-P also exhibited strong binding activity to a glutelin intron sequence, suggesting that RBP-P might participate in mRNA splicing. Overall, these results support a multifunctional role for RBP-P in glutelin mRNA metabolism, perhaps in nuclear pre-mRNA splicing and cytosolic localization to the cisternal-ER.
KeywordsRiceRNA binding proteinRNA localizationZipcodeStorage proteinGlutelin
The fate of cytosolic mRNAs is first established in the nucleus, where events regulated by cis- and trans-acting factors determine stability, localization, and translation (Giorgi and Moore 2007). Trans-acting factors include RNA binding proteins (RBPs) that, together with their associated mRNAs and other interacting factors, form a ribonucleoprotein complex that is eventually exported to the cytosol (Glisovic et al. 2008). Although this complex is dynamic, some RBP components recruited in the nucleus remain with the mRNA well after export and can be found at the site of translation (Dreyfuss et al. 2002; Glisovic et al. 2008). This is true of some heterogeneous nuclear ribonucleoproteins (hnRNPs), RBPs that are involved in one or more aspects of nuclear and cytosolic RNA processing including RNA splicing, turnover, localization, and translation (Krecic and Swanson 1999; Dreyfuss et al. 2002). In fact, many RBPs have been reported to perform multiple duties within the cell and this functional diversity is due in part to the presence of one or more RNA binding domains, such as the RNA recognition motif (RRM), that ultimately influence protein activity (Lunde et al. 2007).
RNA localization is one process in which RBPs play an integral role. The transport of mRNAs to particular regions within the cell serves to control gene expression both spatially and temporally and is common to both eukaryotes and bacteria (Martin and Ephrussi 2009; Pratt and Mowry 2013). Results from extensive work in yeast and animals have provided a general model of active transport (St. Johnston 2005; Martin and Ephrussi 2009). In brief, cis-acting localization elements, or zipcodes, within the RNA recruit RBPs and associated factors, forming a ribonucleoprotein complex within the nucleus. After export, this complex undergoes remodeling where one or more components are lost while others are gained to form a transport particle that moves via the cytoskeleton in a translationally arrested state until it arrives at its final destination, where it is anchored and the RNA is subsequently translated.
Although not as well studied as in yeast and metazoans, RNA localization has been reported in a number of photosynthetic organisms (Doroshenk et al. 2012, and references within). Perhaps the best example of active RNA transport comes from work done in rice. In rice endosperm cells, storage protein mRNAs are asymmetrically targeted to the cortical endoplasmic reticulum (ER) (reviewed in Crofts et al. 2004; Doroshenk et al. 2012). Prolamine RNAs are transported to ER that delimits prolamine protein bodies (PB-ER), where after translation and translocation, the protein forms intracisternal inclusion granules in the ER lumen and is retained. Glutelin RNAs are instead localized to the cisternal ER (cis-ER) and after translation, the protein is sorted via the Golgi to protein storage vacuoles in the cytosol. Subsequent studies have shown that correct RNA localization is critical for proper storage protein deposition and disruptions may lead to altered cell architecture (Washida et al. 2009b; Doroshenk et al. 2010; Fukuda et al. 2011, 2013).
Examination of RNA localization mutants suggests the two storage protein mRNA transport pathways to the cortical ER are interrelated (Crofts et al. 2004; Doroshenk et al. 2012). Considering this, one possible scenario for transport is that prolamine and glutelin RNAs are assembled together with interacting RBPs within the nucleus as part of a single ribonucleoprotein transport particle. This particle is then exported and moves via the cytoskeleton to the cortical ER. At some point, the complex disassembles and prolamine and glutelin RNAs are targeted to either the PB- or cis-ER, respectively. Alternatively, each RNA may bind a core of common RBPs as well as unique RBPs that specify separate transport pathways to the cortical ER. Both hypotheses suggest some components of the transport particles are shared, which is supported by the finding that certain RBPs interact with distinct, yet functionally related RNAs that are ultimately targeted to the same region within the cell (Lécuyer et al. 2009). Other factors such as cis-localization elements and RBPs are specific to each pathway and necessary to direct the RNAs to different subdomains of the ER. In an attempt to characterize these trans-acting components, our lab previously utilized prolamine zipcode RNA as bait to isolate and identify 15 interacting RBPs, many of which contain RNA binding domains and include putative hnRNPs (Crofts et al. 2010). These proteins are currently being investigated to determine what, if any, role they play in prolamine RNA localization. Here, we describe our efforts to identify glutelin zipcode RBPs as a way of better understanding the mechanism of RNA localization in rice.
Materials and methods
Extraction and fractionation of rice seed proteins
All steps were performed at 4 °C or on ice unless indicated. To obtain protein fractions for Northwestern blotting analysis, 20 g of mid-developing wild type rice seed (Oryza sativa cultivar Kitaake) harvested 12–14 days after flowering and immediately frozen in liquid nitrogen were ground in buffer (100 mM Tris–Cl pH 7.5, 100 mM KCl, 50 mM Mg acetate, 1 mM EDTA, 0.2 M sucrose, 1 mM PMSF, 10 µM leupeptin, 1 µM pepstatin, 1 mM DTT) and the resulting extract centrifuged to pellet starch (3,000g, 15 min). The supernatant was clarified by centrifugation at 100,000g for 45 min and protein concentration determined by 5× Protein Assay Reagent (Cytoskeleton, Inc, www.cytoskeleton.com). The extract was diluted with column buffer (50 mM Tris pH 7.5, 1 mM EDTA) to a final protein concentration of 1 mg/ml and 20 ml were passed through a Bio-Rad UNO Q1 Column (www.bio-rad.com) previously equilibrated with column buffer. The column was washed with 10 ml column buffer and proteins eluted in 2 ml fractions by a 30 ml linear NaCl gradient from 0 to 500 mM.
Nuclear and cytosolic proteins from mid-developing seeds were obtained using the CelLytic PN Isolation/Extraction Kit (Sigma Aldrich, www.sigmaaldrich.com). Two g of dehulled seed were extracted in 3 vol Nuclear Isolation Buffer (NIB) containing 1 mM DTT. The extract was passed through a mesh filter and centrifuged twice to pellet starch (25g, 10 min). The supernatant was then centrifuged at 1,260g for 10 min and the resulting cell pellet homogenized in 400 µl NIB with 0.3 % NP-40 and 1× protease inhibitor cocktail (Sigma-Aldrich). The lysate was overlaid on a 2.3 M sucrose cushion and centrifuged at 12,000g for 10 min. The upper aqueous cytosolic fraction was collected and the nuclear pellet washed with NIB. Nuclear proteins were extracted in 60 µl SDS sample buffer with vortexing for 60 min and the soluble fraction collected after centrifugation. For Western blot analysis, the percent of total volume loaded for the nuclear protein fraction was approximately 10× more than that of the cytosolic protein fraction to verify there was no cytosolic cross-contamination.
Crude protein extracts were prepared from seeds harvested during various stages of development using 6 M urea SDS sample buffer and analyzed by SDS-PAGE as in Crofts et al. (2010).
RNA extraction and reverse transcription (RT)-PCR
Total RNAs were extracted from rice immature seeds harvested at varying days after flowering (DAF) using Plant RNA Reagent (Invitrogen, www.invitrogen.com) and used to synthesize first strand cDNA using M-MLV reverse transcriptase (Invitrogen) according to the manufacturer’s protocol. cDNA transcribed from 1 or 100 ng total RNA was used as a template to PCR amplify glutelin or actin cDNA, respectively, using gene specific primers (Glutelin-f 5′-GTACCGGAGTATCTGTTGTCCGTC-3′ and Glutelin-r 5′-CTCACGCCTGTATGCTTGAGG-3′; Actin-f 5′-TCCATCTTGGCATCTCTCAG-3′ and Actin-r 5′-GTACCCGCATCAGGCATCT-3′) with 24 cycles. RT-PCR reactions were repeated in triplicate to verify results.
Plasmid construction and in vitro transcription
Glutelin RNAs were in vitro transcribed using the following constructs. For exon 1 and the combined zipcode regions in exon 4 and the 3′ UTR (regions 1 and 2, Fig. 1a), glutelin cDNA subcloned into pBluescript II KS(-) was PCR amplified using primers to introduce SacI sites (exon 1 primers: Gt2 ATGS and Gt2 CDSR, Washida et al. 2009b; exon 4/3′ UTR primers: gt2-1087f 5′-TTGAGCTCAGTGTTCAACGGCGAGCTTCG-3′ and gt2-1692r 5′-TCGAGCTCAAGGGGTGTTATATTTTTAT-3′) and the resulting fragments cloned into pBSII KS. Glutelin exons 2, 3, and the combination of exons 2 and 3 with their intervening intron were excised with EclI from constructs 21, 22, and 20 (Washida et al. 2009b), respectively, and subcloned into the SacI sites of pBSII KS. pG22 (Okita et al. 1989), pSamG1 and pSamG2 were used as templates to transcribe glutelin full length, 3′ UTR, 5′ UTR, and coding region RNA as described in Sami-Subbu et al. (2000).
For Northwestern blotting experiments, 32P-labeled RNA was synthesized from 1 µg Not1-digested plasmid DNA containing either glutelin Exon 3 or zipcode regions 1 and 2 in a 20 µl reaction using Ambion’s T7 MAXIscript Kit (www.invitrogen.com/ambion) according to the manufacturer’s instructions. RNA integrity was verified by 5 % acrylamide, 8 M urea denaturing PAGE and autoradiography and subsequently purified with ammonium acetate. To generate digoxigenin (DIG)-labeled RNA, the indicated glutelin cDNA plasmids and empty vector were linearized with appropriate restriction enzyme and in vitro transcribed using Ambion’s MAXIscript Kit according to the manufacturer’s instructions with slight modifications. One μg cDNA template was mixed with 1 mM each of ATP, CTP, and GTP, 0.65 mM UTP, 0.35 mM DIG-UTP (Roche Applied Sciences), 20 U SUPERase In (Ambion), 1× MAXIscript buffer and 2 μl T7 or T3 RNA polymerase per 20 μl reaction. The reaction was incubated at 37 °C for 2 h and then terminated by the addition of 2 U of RNase-free DNase. Two microliter of the reaction were resolved by formaldehyde-agarose gel electrophoresis to verify proper RNA synthesis and the remaining reaction was purified by ammonium acetate precipitation.
For competition assays, double-stranded DNA oligonucleotides with sense sequences of 20 bp poly(A) or poly(G) were cloned into XbaI and XhoI sites of pBSII KS and used for the production of unlabeled transcripts of poly(A), poly(U), poly(G) or poly(C) using Ambion’s MAXIscript Kit.
Northwestern analysis was based in part on Ham et al. (2009). Protein fractions obtained from anion exchange chromatography or immunoprecipitation experiments were resolved on 10 % acrylamide SDS-PAGE gels and transferred to nitrocellulose membranes. Membranes were washed briefly in NW buffer (10 mM Tris–Cl pH 7.5, 50 mM KCl, 1 mM EDTA, 1 mM DTT, 0.1 % Triton) and blocked in NW buffer plus 0.1 mg/ml tRNA, 0.5 % polyvinylpyrrolidone, and 0.05 % ultrapure BSA (Ambion) overnight at 4 °C. Following a brief wash in NW buffer, membranes were incubated with either 32P-labeled glutelin zipcode or control RNA (2 × 105 cpm/ml) in NW buffer containing 0.5 mg/ml heparin and 10 U/ml SUPERase In (Ambion) for 1 h at room temperature. Membranes were washed extensively in NW buffer with RNase inhibitor, air dried, and exposed to film.
Protein fractions were analyzed by SDS-PAGE and Western blotting using indicated primary antibodies and goat anti-rabbit HRP-conjugated secondary antibody (Pierce Biotechnology, www.piercenet.com) except in the case of crosslinked RNA–protein samples, which were probed with anti-DIG antibody (HRP-conjugated, Roche Applied Science, www.roche-applied-science.com). Protein-antibody complexes were detected after incubation with SuperSignal West Pico Chemiluminescent Substrate (Pierce Biotechnology) with a FujiFilm LAS-3000 image analyzer (www.fujifilm.com).
Excised gel plugs from Coomassie stained SDS-PAGE gels were destained, in-gel trypsin digested, extracted, and analyzed by liquid chromatography tandem mass spectrometry as in Doroshenk et al. (2009). Proteins were identified by searching the Oryza sativa NCBI non-redundant database (139482 sequences) using Mascot (www.matrixscience.com) as previously published (Doroshenk et al. 2009).
Immature seeds were collected 10–12 days after flowering and fixed in 0.1 M PBS buffer (pH 7.2) containing 4 % (v/v) paraformaldehyde and 0.1 % (v/v) glutaraldehyde at 4 °C for 24 h. The fixed samples were then dehydrated and embedded in LR White resin. The embedded samples were sectioned (1 μm) on an ultratome and placed onto Teflon printed slides (Electron Microscopy Sciences, www.emsdiasum.com). Samples were blocked with 5 % BSA in PBS at room temperature for 2 h before labeling. Immunofluorescence labeling was performed using standard protocols with anti-RBP-P antibody (1:200 dilution) and rhodamine-conjugated anti-rabbit IgG (1:100 dilution; Molecular Probes, www.invitrogen.com). After labeling, the sections were stained by DAPI and fluorescence images were obtained by a Zeiss LSM 510 META confocal microscope.
Immunodepletion and RNA-immunoprecipitation (RNA-IP)
All experiments were performed at 4 °C under RNase-free condition unless indicated. 100 µl of 20 % Protein A resin (Invitrogen) were incubated overnight with 80 µl of PBS (no antibody control) or partially purified antibody to RBP-P or reticulon. The resin was washed 3 times with 1 ml of IP buffer (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1 % NP-40). Samples (750 µl) from an ion exchange chromatography fraction containing RBP-P (as assessed by Western blot) were diluted to 1 ml with IP buffer and 1× protease inhibitor cocktail (Sigma-Aldrich) and added to each of the antibody bound or no antibody control Protein A resins. After incubating for 2 h, unbound proteins were collected and the resin was washed three times with IP buffer and then resuspended in 40 µl 1× SDS sample buffer. Five microliter of bound and 25 µl of unbound proteins were analyzed by Northwestern and Western blotting as described earlier.
For RNA-IP, 200 µl of 20 % Protein A resin (Invitrogen) were incubated overnight with 50 µl PBS (no antibody control) or 50 µg of partially purified antibody to RBP-P or GFP in 1× coupling buffer (0.01 M sodium phosphate, 0.15 M NaCl). The resin was washed with 10 bed volumes of 1× coupling buffer and equilibrated with 1 ml of RIP buffer (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, 1 mM PMSF, 0.5 % NP-40, 50 U/ml Ambion SUPERase In). Five g of dehulled rice seed were sectioned into about 1 mm thick pieces using a razor blade, immersed immediately in 20 ml 1 % formaldehyde in 1× PBS buffer and fixed for 40 min with gentle rotation at room temperature. After washing three times with 1× PBS buffer and quenching with 10 ml RIP buffer, the crosslinked seed samples were extracted in 15 ml IP buffer and the resulting extract clarified by centrifuging at 1,000g for 5 min to remove starch and then twice at 12,000g for 10 min. The extract was incubated with naked Protein A resin for 30 min to eliminate non-specific binding and the unbound protein fraction was collected and diluted with 0.5 vol RIP buffer supplied with 0.02 mg/ml heparin. Five hundred microliter of the pre-cleared extract were reserved for total RNA extraction as described below. Three milliliter of the extract were loaded onto each previously prepared antibody column and incubated at 4 °C for 2 h with gentle rotation. After washing with 1 ml RIP-H (50 mM Tris–Cl, pH 7.5, 250 mM NaCl, 1 mM EDTA, 0.2 mM PMSF, 0.1 mM DTT, 1 % NP-40, 0.05 % sodium dodecylsulfate (SDS), 10 U/ml SUPERase In), the resins were incubated with 50 µl 1× DNase buffer supplied with 5 U DNase I (Ambion) at 37 °C for 15 min to remove DNA contamination. The captured protein-RNA complexes were finally eluted with 500 µl of elution buffer (50 mM Tris–Cl, pH 7.0, 5 mM EDTA, 1 % β-mecaptoethanol and 2 % SDS). The eluted samples and 500 µl of the reserved extract were incubated at 70 °C for 45 min to reverse crosslinking. RNA was extracted with 3 vol TRIzol (Invitrogen) according to the manufacturer’s instructions, purified with Qiagen’s RNeasy Cleanup Kit (www.qiagen.com) and dissolved in 20 µl DEPC H2O supplied with 1 µl SUPERase In. RNA sample from the reserved extract was further diluted 10 times for cDNA synthesis. First strand cDNA was synthesized from 4 µl of purified total RNA using M-MLV reverse transcriptase (Invitrogen) and oligo dT primers in a final volume of 20 µl according to the manufacturer’s protocol. PCR was performed as described in Wang et al. (2008) using the aforementioned glutelin and actin gene specific primers with 24 or 30 cycles of amplification.
Preparation of recombinant protein for UV-crosslinking assay
A cDNA clone of RBP-P (AK099896) was requested from the Rice Genome Resource Center (http://www.rgrc.dna.affrc.go.jp/). The plasmid was cut with NcoI and the resulting RBP-P fragment subcloned into pET30a for expression of His-tagged RBP-P, which was designated P-30a. To construct truncated P-NR-30a, P-30a was digested with SacI to remove a fragment containing 155 amino acids from the C-terminus. To construct P-RC-30a, P-30a was digested with EcoRI to remove the first 122 N-terminal amino acids. The digested fragment containing truncated RBP-P was then blunt cloned into the EcoRV site of pET30a. Recombinant His-tagged full length or truncated RBP-P protein was expressed in E. coli strain EA3457 (Hwang et al. 2008) and purified by Ni-chelated Sepharose column chromatography.
In vitro RNA–protein UV-crosslinking assay
RNA–protein UV-crosslinking reactions (15 µl) containing 1 μg of purified His-tagged protein, 1 μg of DIG-labeled RNA, 10 mM Tris–HCl, pH 8.0, 50 mM NaCl, 0.1 mM DTT, 10 % glycerol, 1 μg yeast tRNA, and 1 U SUPERase In (Ambion) were incubated at 4 °C for 30 min, transferred to a parafilm sheet on ice, and then exposed to UV light for 7 min using an Ultralum UV crosslinker (Stratagene, www.genomics.agilent.com). The crosslinked samples were digested with 100 U of RNase T1 or 2 μg RNase A at 37 °C for 30 min to remove unbound RNAs and subjected to Western blot as described above. For competition experiments, equal molar amounts (50 pmol) of labeled 3′UTR and unlabeled competitor RNAs were used in crosslinking reactions. Unlabeled competitor RNAs were synthesized without DIG-UTP using Ambion’s MAXIscript Kit as described above. Competition with different homoribopolymers was carried out by addition of unlabeled transcripts of poly(A), poly(U), poly(G) or poly(C) in 5, 10 and 100-fold molar excess to DIG-labeled glutelin RNA prior to UV-crosslinking.
Identification of glutelin RNA binding proteins
In an effort to identify proteins that specifically interact with glutelin cis-localization elements and therefore may be involved in RNA localization, we employed a Northwestern blot approach. A crude rice seed protein extract was fractionated by anion exchange chromatography and eluted protein fractions were subjected to SDS-PAGE and transferred to nitrocellulose membrane. 32P-labeled RNA containing both zipcode regions in exon 4 and the 3′ UTR (Fig. 1a) was in vitro transcribed and used to probe possible RNA binding targets on the membrane. Incubation with 32P-labeled RNA of a similar length from glutelin exon 3 that contained no zipcode sequence served as a control. Sequence and predicted secondary structure of both RNA probes used can be found in Supplementary Figs. 1 and 2. Results from Fig. 1b show numerous proteins of various sizes that seem to have an increased binding affinity for the 32P-labeled zipcode RNA (middle panel) compared to the control (lower panel). Because background binding seemed to be slightly higher on the zipcode RNA Northwestern blot, we chose to only focus on those proteins exhibiting an obvious preference for the zipcode RNA to avoid false positives. We targeted an approximately 55 kDa protein in eluted fraction #12 that appeared to have greater binding affinity to the glutelin zipcode RNA probe versus the control RNA for further study.
Localization and expression of RBP-P
RNA binding activity of RBP-P
We next sought to confirm the interaction between RBP-P and glutelin RNA in a cell extract using RNA immunoprecipitation (RNA-IP). IP was performed using extracts made from seed sections subjected to formaldehyde crosslinking to preserve in vivo RNA–protein interactions. Extracts were incubated with Protein A resin bound with antibody to RBP-P or, as a control, GFP antibody. Care was taken to minimize RNA degradation and non-specific binding by adding RNase inhibitors and heparin, respectively. RNAs bound to immunoprecipitated proteins were extracted with TRIzol and used for cDNA synthesis. The presence of glutelin cDNA was then assessed by PCR using glutelin specific primers. Figure 5b shows that glutelin RNA is detected in the starting extract material as well as the RBP-P IP sample, but not in the GFP IP control. To confirm RNA binding activity was specific for glutelin, PCR was performed with actin gene specific primers. Actin RNA, however, could not be significantly detected in either the RBP-P or GFP RNA-IP. These results are in agreement with our earlier findings that indicate RBP-P indeed binds glutelin RNA and this interaction is specific.
To further investigate the apparent lack of interaction with the coding region alone, additional crosslinking experiments were conducted with DIG-labeled RNA of each individual exon, as well as exons 2 and 3 together with their intervening intron (E2IE3). Each RNA tested exhibited binding to His-RBP-P, although the intensity of the DIG-derived signal varied. Crosslinking reactions containing RNAs of exons 1 and 2 had a weak DIG signal, while reactions with exons 3 and 4 showed greater reactivity (Fig. 6b). The finding that exon 3 interacts with RBP-P is in agreement with results from the initial Northwestern blot experiments which used exon 3 RNA as a control probe to identify glutelin zipcode RBPs (Fig. 1b, lower panel). The highest level of RNA binding, however, was seen in the reaction containing exons 2 and 3 together (E2IE3) and mirrored that seen with the 3′ UTR.
Although results from the UV crosslinking experiments indicate RBP-P has a high level of binding activity with glutelin 3′ UTR and E2IE3 RNAs, competition experiments were conducted to verify that the intensity of the DIG-labeled RNA signals was an accurate representation of binding strength and not biased because of the use of UTP-labeled RNA. For this, DIG-labeled 3′ UTR RNA was incubated with equimolar amounts of unlabeled competitor RNAs and subjected to UV crosslinking with RBP-P as above. As seen in Fig. 6c, the presence of unlabeled CDS, exon 1 or exon 2 RNA did not noticeably compete with 3′ UTR for RBP-P binding, which is in agreement with the undetectable or weak signal seen in the initial crosslinking experiments using these RNAs (Fig. 6a, b). The same is true for full length glutelin RNA, exons 3 and 4, and zipcode 1, which all demonstrated reduced RBP-P binding activity as compared to 3′ UTR (Fig. 6a, b) and moderately affected 3′ UTR RNA binding (Fig. 6c). Although all three zipcode regions were found to interact with RBP-P (Fig. 6a), it appears there is greater affinity for zipcodes 2 and 3, which were both able to significantly compete with 3′ UTR RNA (Fig. 6c). Of note, however, is the relatively low signal seen in the initial RBP-P crosslinking reaction with zipcode 3 as compared to 3′ UTR (Fig. 6a), which highlights the importance of doing such competition experiments when assessing binding strength. E2IE3 RNA, which contains exons 2 and 3 and the adjoining intron, was also able to effectively compete with 3′ UTR for RBP-P binding (Fig. 6c). A comparable amount of DIG signal was seen in the crosslinking reactions containing either E2IE3 or 3′ UTR RNA (Fig. 6b), and together with the competition results, indicates RBP-P has a high affinity for both RNAs. Of interest is the finding that neither exon 2 nor exon 3 alone had as strong an interaction with RBP-P as E2IE3, suggesting that the intron between these two exons is important for binding.
The results from the in vitro RNA binding assays indicate RBP-P indeed binds to glutelin RNA, with the strongest activity seen with the 3′ UTR, which contains zipcode 2, and the intron between exons 2 and 3. Because introns and 3′ UTRs of plant mRNAs are generally U- or AU-rich (Graber et al. 1999; Lorković et al. 2000), we next explored whether RBP-P exhibits a preference for U- and A-rich sequences. UV crosslinking competition assays were performed with His-RBP-P and DIG-labeled 3′ UTR glutelin RNA in the presence of 5-, 10-, and 100-fold mole excess of unlabeled homoribopolymers. As seen in Fig. 6d, RBP-P binding to 3′ UTR RNA is only obviously competed for by increasing amounts of poly(U), indicating a preference for this ribopolymer but not poly(A). It should be mentioned that examination of less saturated images revealed similar results. There does seem to be an overall reduced signal in the poly(A) competition experiments as compared to the other ribopolymers; however, this may actually be due to the affinity of poly(A) for U-rich regions in the DIG-labeled glutelin 3′ UTR RNA.
Unlike our successful identification of RBPs that specifically bound to the prolamine zipcode by affinity chromatography, similar efforts failed to identify specific RBPs using the glutelin zipcodes as bait. We then utilized in vitro transcribed glutelin zipcode RNA (zipcode regions 1 and 2, Fig. 1a) as a probe in Northwestern blot experiments. An approximately 55 kDa protein which appeared to exhibit greater binding affinity for the zipcode RNA compared to the control (Fig. 1b) was subsequently identified as RBP-P, a putative oligouridylate binding protein first characterized as a prolamine zipcode RBP (Crofts et al., 2010). Follow up immunodepletion experiments using antibodies specific to RBP-P demonstrated this protein is indeed responsible for the glutelin RNA binding activity seen in Northwestern blots (Fig. 5a). To confirm this interaction under in vivo-like conditions, RNA-IP was done using antibodies to RBP-P after chemical crosslinking of seed sections. As expected, PCR amplification of cDNA synthesized from bound RNAs in RBP-P immunoprecipitates confirmed the presence of glutelin RNA, but not that of a negative control (Fig. 5b).
Although classified as a potential RBP that binds to the prolamine zipcode (Crofts et al. 2010), similar RNA-IP studies could not detect significant amounts of prolamine RNA in RBP-P immunoprecipitates (results not shown). Although it is possible that the association of RBP-P with prolamine zipcode RNA was a spurious interaction, we believe this to be unlikely as the original affinity chromatography studies were conducted in the presence of 5 mg/ml heparin to create very stringent binding conditions so as to avoid false positives (Crofts et al. 2010). One possible explanation for these apparent paradoxical observations is that RBP-P antibody cannot recognize RBP-P when bound to prolamine RNA. This possibility has been suggested from studies on the co-localization of RBPs bound to the prolamine zipcode RNA. For example, RBP-Q is detected in immunoprecipitates generated by antibodies to either RBP-A or RBP-K, indicating that these three proteins are bound to prolamine zipcode RNA (Yang et al. 2014)). In reciprocal IP experiments with RBP-Q antibodies, however, neither RBP-A or RBP-K are detected, suggesting that RBP-Q, when associated with prolamine zipcode RNA, is sterically inaccessible to its antibody due to interactions with other proteins. Genetic and biochemical evidence suggest the prolamine and glutelin RNA pathways are interrelated and thus likely share some common elements such as trans-acting factors (Crofts et al. 2004; Doroshenk et al. 2012). Therefore, we cannot exclude the possibility that RBP-P is also involved in prolamine RNA localization.
Examination of RBP-P at the amino acid level reveals it is most similar to three previously characterized RBPs, V. faba AKIP1 and Arabidopsis UBA2a and UBA2b, sharing approximately 50 % sequence identity (Fig. 2b). All four proteins contain two RRMs and as a group appear to be unique to higher plants (Lorković and Barta 2002). Some overall similarity is seen with metazoan hnRNPs (Li et al. 2002; Bove et al. 2008), which are proposed to be involved in a multitude of functions including splicing, stability, export, localization, and translation (Dreyfuss et al. 2002; Krecic and Swanson 1999). The role of hnRNP-like proteins in plants, however, remains largely unresolved.
Studies show AKIP1 is localized to the nucleus in guard cells and reorganizes into distinct foci upon treatment with the hormone abscisic acid, which is believed to regulate certain aspects of post-transcriptional RNA metabolism (Li et al. 2002; Ng et al. 2004). Such nuclear speckles have been observed to contain splicing factors necessary for pre-mRNA processing and are often localized near sites of active transcription (Reddy et al. 2012). Within the nucleus, AKIP1 was found to interact with a stress-related mRNA, leading to the conclusion that this RBP may be involved in a stress response pathway (Li et al. 2002). Arabidopsis UBA2 proteins are also believed to play a role in stress response. Similar to AKIP1, both UBA2a and UBA2b are nuclear localized and upon treatment with abscisic acid aggregate into nuclear speckles (Lambermon et al. 2002; Riera et al. 2006; Bove et al. 2008). Furthermore, UBA2 transcript expression increases upon mechanical wounding and constitutive overexpression of either protein leads to an increase in the expression of senescence and stress associated genes as well as phenotypic characterizations of cell death (Bove et al. 2008; Kim et al. 2008).
At the molecular level, the functions of AKIP1 and UBA2 proteins are unknown. Arabidopsis UBA2a was identified from a yeast two-hybrid screen as an interacting partner of Nicotiana plumbaginifolia UBP1, a predominately nuclear RBP containing three RRMs that preferentially binds oligouridylates, which are prevalent in plant introns and 3′ UTRs (Lambermon et al. 2000, 2002). Studies suggest UBP1 may play multiple roles within the cell, including splicing and mRNA stabilization via binding to 3′ UTRs (Lambermon et al. 2000, 2002). Both UBA2 proteins also show a preference for oligouridylates (Lambermon et al. 2002; Bove et al. 2008). Compared to UBP1, overexpression of UBA2 proteins did not have any effect on splicing efficiency, but did increase levels of reporter mRNAs, indicating a possible role in RNA stability along with UBP1 (Lambermon et al. 2002).
The current work demonstrates rice RBP-P shares characteristics with its plant homologues, including nuclear localization (Fig. 3) and binding specificity for oligouridylates (Fig. 6d). Functionally, though, results from this study suggest a role in seed storage protein gene expression. RBP-P is also found in the cytoplasm (Fig. 3) and its temporal expression patterns coincide with that of glutelin RNA and protein (Fig. 4). Immunoprecipitation studies further provide evidence that RBP-P associates with glutelin RNAs under in vivo-like conditions (Fig. 5b) and in vitro RNA binding assays demonstrate it binds to all three glutelin zipcode regions depicted in Fig. 1a (Fig. 6a). The latter result is in contrast to experiments with glutelin RNA containing coding sequence only or the vector control, for which little to no binding is evident, indicating there is specificity (Fig. 6a). The binding of RBP-P to glutelin RNA, and in particular to the zipcode regions, indicates this protein may play a role in one or more aspects of storage protein biosynthesis, such as RNA splicing, nuclear export, transport, localization, stability, and/or translational regulation.
As is the case with prolamine, many studies have found that when multiple zipcodes are present in a single RNA species, redundant signals are required for proper localization to the target destination and that in some instances, different zipcodes are required for particular steps (Jambhekar and Derisi 2007). With glutelin, not all of the localization elements are required, as zipcode regions 1 or 3 (Fig. 1a) are each sufficient for proper RNA transport to the cis-ER, while region 2 results in partial localization (Washida et al. 2009a). Combined with the lack of binding activity of RBP-P for glutelin CDS RNA (Fig. 6), it suggests that RNA structure is more important than nucleotide sequence for recruiting trans-acting factors, a finding that is evident in other organisms as well (Marchand et al. 2012; Pratt and Mowry 2013).
Given that RBP-P was detected in both the nucleus and cytosol (Fig. 3), the possibility remains that it possesses multiple activities that are influenced by subcellular localization, a characteristic of metazoan RBPs (Giorgi and Moore 2007). Within the nucleus, RBP-P may participate in transcription, pre-mRNA processing, RNA stability, or export. Although in vitro RNA binding assays demonstrate a strong preference for glutelin 3′ UTR RNA, RBP-P also exhibits high binding activity to RNA containing exons 2 and 3 as well as their connecting intron (E2IE3, Fig. 6b). This is in contrast to binding assays with RNA of exons 2 and 3 alone which resulted in a significant reduction in binding, suggesting the intron is important for this interaction. Plant introns are enriched in U and AU content (Lorković et al. 2000) and RBP-P shows a clear preference for poly(U) homoribopolymers (Fig. 6d). This may indicate that RBP-P is involved in pre-mRNA splicing, which has been demonstrated for N. plumbaginifolia UBP1, though not the more closely related UBA2 proteins (Lambermon et al. 2000, 2002). Preferential binding to glutelin 3′ UTR and poly(U) homoribopolymers, which are also prevalent in plant 3′ UTRs (Graber et al. 1999), may also indicate RBP-P is involved in mRNA stabilization (Fig. 6). This has been observed in the case of Arabidopsis UBA2 and UBP1 proteins, which when overexpressed, increase steady state levels of reporter mRNAs (Lambermon et al. 2000, 2002). UBP1 in particular interacts with 3′ UTRs in vitro and is thought to protect RNAs from degradation (Lambermon et al. 2000). The importance of plant UTRs in mRNA stability has been demonstrated and critical elements within these regions have been identified (Park et al. 2012, and references therein).
Interestingly, neither AKIP1 nor the UBA2 proteins could be reliably detected outside of the nucleus (Li et al. 2002; Ng et al. 2004 Lambermon et al. 2002; Riera et al. 2006; Bove et al. 2008), while UBP1, which was found to interact with UBA2a in the nucleus (Lambermon et al. 2000, 2002), has more recently been observed to re-localize to cytoplasmic granules when exposed to heat stress (Weber et al. 2008). This suggests that although showing a high degree of sequence identity, these proteins may be functionally diverse. It is possible that this difference is the result of variations in their auxiliary domains, which as previously mentioned are thought to influence RBP RNA binding activity, protein–protein interactions, and localization (Biamonti and Riva 1994). Although all have a C-terminus rich in glycine, RBP-P contains an N-terminus that is alanine- and glutamate-rich, while AKIP1 and UBA2 proteins each have a glutamate-rich N-terminus (Fig. 2; Li et al. 2002; Bove et al. 2008; Lambermon et al. 2002). As the N-terminus of RBP-P appears to be important for binding activity to glutelin RNA (Fig. 7), it is conceivable this domain regulates other aspects of RBP-P function or localization. Another possibility for the difference is post-translational modifications, which are proposed to regulate RBP subcellular localization, RNA binding activity, and function (Glisovic et al. 2008). RBP-P has a predicted molecular weight of 49.3 kDa, but routinely migrates to a position above the 55 kDa molecular weight marker on denaturing SDS-PAGE gels (Fig. 3a). In some instances, a second, slightly larger immunoreactive band is apparent (Fig. 5a). Similarly, Arabidopsis UBA2a has a predicted molecular weight of 51 kDa, but was visible only as a 66 kDa protein band (Lambermon et al. 2002). It remains to be determined the cause of these larger than expected proteins, but it could be due to post-translational modifications. For example, phosphorylation of AKIP1 by protein kinase AAPK is required for binding to its target mRNA (Li et al. 2002). It would be of interest to determine whether this is indeed the case for RBP-P. Although our results suggest RBP-P plays a role in both nuclear and cytosolic gene expression, further work is necessary to determine its exact function and in particular, whether it is involved in storage protein RNA localization.
We thank Tin Yan Lau and Sethu Sivasubramanian for their technical assistance. Mass spectrometry analysis of samples was done by the University of Idaho Environmental Biotechnology Institute. This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Grant No. 21380008 to T.K.) and by the National Science Foundation (Grant Nos. DBI–0605016, IOS–1021699, and Intergovernmental Personnel Act funds to T.W.O.). A.J.C. is grateful for financial support from the Japanese Society for the Promotion of Science in the form of a Grant-in-Aid for Young Scientists and to Akita Prefectural University for providing access to equipment. Preparation of the manuscript took placed while T.W.O. was on IR/D as an Intergovernmental Personnel Act (IPA) rotator at the National Science Foundation.